Memory processes in retrospective symptom (over ...

FACULTY OF PSYCHOLOGY AND EDUCATIONAL SCIENCES

HEALTH PSYCHOLOGY

Memory processes in retrospective symptom (over)reporting

Marta Walentynowicz Dissertation offered to obtain the degree of Doctor of Psychology (PhD) Supervisor:

Prof. Dr. Omer Van den Bergh

Co-supervisors:

Prof. Dr. Ilse Van Diest Prof. Dr. Filip Raes

2016

Marta Walentynowicz: Geheugenprocessen bij retrospectieve (over)rapportering van symptomen Proefschrift aangeboden tot het verkrijgen van de graad van Doctor in de Psychologie, 2016 Promotor: Prof. Dr. Omer Van den Bergh Copromotoren: Prof. Dr. Ilse Van Diest, Prof. Dr. Filip Raes Zelf-gerapporteerde somatische symptomen spelen een voorname rol in de gezondheidszorg. Ze beïnvloeden het gedrag en de beslissingen van zowel patiënten als van medisch specialisten. Deze beslissingen zijn vaak gebaseerd op de retrospectieve beoordeling van eerdere symptoomervaringen. Echter, zowel concurrente als retrospectieve symptoom-ratings zijn kwetsbaar voor verschillende vertekenende factoren die leiden tot onjuiste (vaak overschatte) retrospectieve symptoomrapporteringen. Het doel van het huidige doctoraatsproject was om somatische herinneringen en de rol van geheugenprocessen in retrospectieve symptoomrapportering te onderzoeken. Volgens het “dual-process”perspectief, is symptoomrapportering het resultaat van het samenspel tussen sensorisch-perceptuele en affectief-motivationele componenten van een somatische ervaring. Bijgevolg kan de differentiële verwerking van die componenten een van de factoren zijn die geheugenprocessen beïnvloeden. Op basis van deze aannames veronderstelden we dat een vertekende klachtenrapportering ontstaat als reactie op een relatief onevenwicht van de twee componenten, vooral wanneer sterke negatieve affectieve informatie de sensorische component overschaduwt, mogelijk resulterend in een minder gedetailleerde herinnering van de perceptuele details. Omdat een relatieve dominantie van affectieve over sensorische verwerking van somatische informatie kan worden verwacht bij de individuen met een overreactief evaluatiesysteem, werden de voorgestelde hypotheses getest bij de personen die frequente, vaak medisch onverklaarde, lichamelijke klachten rapporteren (habitual symptom reporting; HSR) en hoog scoren voor negatieve affectiviteit. De eerste studie (Hoofdstuk 2) hanteert een psychometrische benadering om de latente structuur van symptoomrapportering te onderzoeken. In lijn met eerdere bevindingen werd symptoomrapportering best verklaard door een bifactor model bestaande uit een algemene factor en een aantal symptoomspecifieke factoren, die zouden kunnen worden geïnterpreteerd als een afspiegeling van respectievelijk affectieve en sensorische componenten van symptoomervaringen. Om de factoren die retrospectieve symptoomrapportering beïnvloeden te onderzoeken, gebruikten de volgende vier studies een experimentele benadering met gestandaardiseerde procedures om aversieve lichamelijke symptomen (pijn, kortademigheid) in gecontroleerde laboratoriumomstandigheden te induceren. Zowel zelf-gerapporteerde als van psychofysiologische reacties op somatische prikkels werden gelijktijdig gemeten. Retrospectieve ratings werden verzameld twee (Studies 2-4) en vier (Studie 5) weken na de somatische ervaringen. In de eerste twee experimentele studies (Hoofdstuk 3 en 4), werden gelijktijdig en retrospectieve ratings van de symptomen vergeleken in- en tussen-proefpersonen in een onderzoek met studenten en een studie met patiënten met medisch onverklaarde kortademigheid (MUD) en gezonde controles. Deze toonden aan dat (a) geheugenvertekeningen onmiddellijk beginnen te werken na de somatische gebeurtenis, (b) de intensiteit van zowel gelijktijdige als retrospectieve ratings hoger zijn in HSR/MUD, en (c) dit laatste effect gemedieerd wordt door de affectieve toestand die geassocieerd was met een somatische ervaring (dat wil zeggen, negatief affect en angst). In de volgende experimentele studies (Hoofdstuk 5 en 6), werden symptoominducties gecombineerd met een verwerkingsmanipulatie (processing focus; PF), respectievelijk tijdens encodering en tijdens herinnering, om te onderzoeken of de PF op ofwel sensorisch-perceptuele of affectiefmotivationele aspecten van somatische ervaring de retrospectieve symptoomrapportering kunnen beïnvloeden. Het manipuleren van de PF bij encodering, met name met focus op de affectieve componenten resulteerde in zowel een verhoogde affectieve respons tijdens de symptoominducties en als in verhoogde retrospectieve rapportering van kortademigheid (maar niet van pijn). Anderzijds had het manipuleren van retrieval geen invloed op de manier waarop symptomen of affectieve reacties werden opgeroepen. Ten slotte werd het onderwerp van somatische herinneringen ook benaderd vanuit een autobiografisch geheugenperspectief (Hoofdstuk 7) waaruit bleek een verminderde geheugenspecificiteit van gezondheidsgerelateerde autobiografische herinneringen vastgesteld werd bij patiënten met MUD in vergelijking met gezonde controles. Deze studies benadrukken duidelijk het belang van de manier waarop somatische ervaringen verwerkt worden voor retrospectieve symptoom (over)rapportering. In het bijzonder werd een vertekening vastgesteld bij een verhoogde aandacht voor de affectieve aspecten van een aversieve somatische ervaring. Deze bevindingen dragen bij aan het begrijpen van vertekeningen in klachtenrapportering in het algemeen, maar ook van de grotere kwetsbaarheid voor retrospectieve symptoom overrapportering bij individuen met

een overreactief evaluatiesysteem (hoge mate van negatieve affectiviteit). In het laatste hoofdstuk wordt een gedetailleerd overzicht van onze bevindingen en van de beperkingen van de gerapporteerde studies besproken en worden aanbevelingen voor toekomstig onderzoek gepresenteerd.

Marta Walentynowicz: Memory processes in retrospective symptom (over)reporting Thesis submitted to obtain doctoral degree in Psychology, 2016 Supervisor: Prof. Dr. Omer Van den Bergh Co-supervisors: Prof. Dr. Ilse Van Diest, Prof. Dr. Filip Raes Self-reported somatic symptoms play a primary role in the health care system, guiding the behavior and decisions of both patients and medical specialists. These decisions are often based on retrospective evaluations of past symptom episodes. However, both momentary and memory-based symptom ratings are vulnerable to various biasing factors leading to inaccurate (often overestimated) symptom recall. The aim of the current doctoral project was to investigate somatic memories and the role of memory processes in retrospective symptom reporting. According to the “dual-process” perspective, symptom reporting results from the interplay between sensory-perceptual and affective-motivational components of a somatic experience. Consequently, differential processing of those components could be one of the factors affecting memory processes. Based on those assumptions, we hypothesized that biased symptom reporting emerges in response to a relative imbalance of the two components, especially when strong negative affective information overshadows a sensory component, potentially resulting in less detailed memory of sensory details. Because a relative dominance of affective over sensory processing of bodily signals could be expected among individuals with an overreactive evaluative system, the advanced hypotheses were tested among the individuals who report frequent, often medically unexplained, somatic experiences (habitual symptom reporting; HSR) and score high on negative affectivity (NA). The first study (Chapter 2) adopted a psychometric approach to test the latent structure of symptom reporting. In line with earlier findings, symptom reporting was best explained by a bifactor model comprising one general and several symptom specific factors, which could be interpreted as reflecting the affective and sensory components of symptom experiences, respectively. To examine the factors affecting retrospective symptom reporting, the next four studies adopted an experimental approach with standardized procedures to induce aversive bodily symptoms (pain, dyspnea) in controlled laboratory conditions. Assessment of both self-reported and psychophysiological responses to somatic stimuli took place concurrently during symptom inductions. Retrospective ratings were collected up to two (Studies 2-4) and four (Study 5) weeks after the somatic experiences. In the first two experimental studies (Chapters 3 and 4), concurrent and retrospective responses to the symptoms were compared within- and between-subjects in a study with students and a study with patients with medically unexplained dyspnea (MUD) and healthy controls. This demonstrated that (a) memory biases start to operate immediately after the somatic event, (b) intensity ratings of both concurrent and retrospective ratings are more elevated in HSR/MUD, and (c) the latter effect is mediated by the affective state associated with a somatic experience (i.e., state NA and anxiety). In the following experimental studies (Chapters 5 and 6), symptom inductions were combined with the processing focus (PF) manipulation at encoding and at retrieval to investigate whether directing PF to either sensory-perceptual or affective-motivational aspects of somatic experience can influence retrospective symptom reporting. The manipulation of PF at encoding led to differences in affective responses as well as to the memory bias, such that the affective PF resulted in both increased affective responses to symptom inductions and increased retrospective ratings of dyspnea (but not of pain). On the other hand, the manipulation at retrieval did not influence the way symptoms or affective responses were recalled. Finally, the topic of somatic memories was also approached from an autobiographical memory perspective (Chapter 7) showing reduced memory specificity of health-related autobiographical memories in patients with MUD compared to healthy controls. Taken together, these studies clearly emphasize the importance of affective processing in retrospective symptom reporting. Specifically, biased symptom recall was related to an increased focus on the affective aspects of a distressing somatic experience. Consequently, this could explain not only biased symptom reporting in general, but also the greater vulnerability to retrospective symptom overreporting among individuals with an overreactive evaluative response system (higher level of NA). A detailed discussion of these findings, together with limitations of the reported studies and recommendations for future research are presented in the last chapter.

Acknowledgments

This thesis marks the end of a very important step on my academic / life journey. And it was quite an adventure, full of ups and downs, twists and turns. An accurate evaluation of this period should take all those moments into account. However, this time I will allow myself for some (positive) memory bias, focus on the beautiful moments, and thank all the people who made this adventure so exciting.

First of all, I would like to thank my supervisor, Omer, for giving me the opportunity to start this adventure in Leuven and for supporting and guiding me along the way. From the very beginning you have fostered my independence and self-confidence – I could explore, learn from my own mistakes, but also consult you whenever I got lost. I am really grateful that you supported my ideas (going to Gdańsk), understood my weaknesses (a/the), gave me an extra push in the back when I doubted myself, and taught me so much during those years. I value your expertise, passion and high standards for research and I am really proud that I could learn and grow under your wings. Also many thanks to Angela for the kindness and the amazing international dinner parties.

I was also very lucky to have two great co-supervisors, Ilse and Filip, who always found time for me in their busy schedules. Ilse, thank you for all the help with psychophysiology, your valuable feedback, your interest in my future, and the thought-provoking chats about research, opportunities and work-life balance. Filip, thank you for introducing me to the fascinating world of autobiographical memory, the guidance with the AMT development, your valuable comments, and the possibility to apply for the PDM under your supervision. I really hope that we will be able to make that project happen at some point.

I would also like to express my gratitude to the supervisory committee – Michael Witthöft, Sibylle Petersen, Lukas Van Oudenhove, Eva Ceulemans and Walter Schaeken. Thank you for your valuable feedback and interesting comments that helped to shape the project. I am also grateful to Maria Kleinstäuber and Pierre Philippot, who accepted to become the members of the jury.

I would also like to thank the people who contributed to my work throughout those years. My master students and interns – Anne-Marie Verlinden, Hannie Tay, Sander De Deygere, Sebastiaan Pessers – it was a pleasure to supervise you. Thank you for your interest, effort and help with data collection.

During my PhD, I have spent many months in the pneumology unit of UZ Gasthuisberg. Linda, my angel, thank you for your kindness, smile, and support with the recruitment of patients. Also many thanks to Geert Celis for his help with the equipment and the room arrangements and to all the specialists for having me around and accepting my lab-on-wheels.

I would also like to thank Mariola Bidzan and Małgorzata Lipowska for visiting us in Leuven and inviting us for their conference in Gdańsk.

I also had the pleasure to work with Michael Witthöft, Fabian Jasper, Lukas Van Oudenhove and Jan Tack on the functional dyspepsia project – thank you all for the inspiring collaboration. I am especially grateful to Michael and Fabian who hosted me in Mainz and taught me so much about CFA.

I would also like to thank all the OGPs, old and new, for the fun retreats, the delicious IDPs, the labmeetings, and the congresses. Johan, Andreas, Katleen, Thomas J, Elke, Meike, Ann, Sibylle, Steven, Maaike, Mathias, Hassan, Rena, Ali K, Elena, Ruth, Tom, Jonas, Nathalie, Kim, Stéphanie, Kai, Emma, Thomas R, Imke, Ali G, Nils, Michaela, Josef, Stefan, Nadia, Christine, Erik, it was great to interact with you every day. Ruth, Steven, Joanna, Nathalie, Jonas, Nadia, Christine, thank you for being great office mates. Jonas, thanks for the stat support and the co-authorship of our office jungle. Katleen, thanks for the support with the rebreathing paradigm and the translations. Andreas & Stefan, thank you for giving me the opportunity to practice my teaching skills. Thomas & Meike, thanks for the help with psychophysiology. Tom, thanks for the help with the AMTs. Jeroen and Mathijs, you are my superheroes! Thank you for your time and support, the Polish jokes, Mac vs. Windows discussions, your optimism and laughs. Liesbet, Martine, An, and Lin, thank you for your assistance and care. Dropping by to get a stamp or to discuss the multiple-choice exams frequently turned into very interesting chats, which I will miss! Finally, thanks to the CLEPpers and the other members of CfE team for being great neighbors, it was nice to interact with you in the coffee room and during the retreats.

But of course those five years were not only about PhD and PSI! Elena, Maaike, Mathias, Hassan, Ali – thank you for being wonderful colleagues and friends. I will never forget our BBQs (especially those Iranian dishes by Elham, Ali, and Hassan), trips, drinks, lunches, dinners, sushi boats, and receptions.

Elena, Conor, Brey, Fabio, Erwan, Brigitte – you made my time in Leuven so great that it simply became my second home. I really appreciate the memories that we created throughout those years – from drinks accompanying our board games (at least the ones that survived), open-air movie nights in the streets of Brussels and BBQs to drinks celebrating very important events including graduations, weddings and babies!

Maaike, thank you for being one of the brave MUSketeers and such a good friend. Luckily our wine & cheese nights made the MUS search more pleasant. Cheers to the mutual “good” influence and feedback!

Elena, you have been present in my PhD life from the very beginning, when you became my OGP godmother. Thank you for all the support that I received from you, for those evenings in PATC with Kriek and Gageleer, for long discussions about everything and nothing (and MUS of course). I could always rely on you, whether you were in Leuven or London, whether it was about PhD or life. Thank you for everything!

Also thanks to Yasemin for all our coffee breaks, chats, and walks, Marieke for the punching and kicking during kickfun, irreplaceable style advice, and all those candles, and Koen for introducing me to this interesting country.

Other friends, even though far away in geographical terms, were always there for me via chats, Skype, visits in Leuven and reunions back home or in Manchester, Copenhagen, Bilbao/Neuruppin, etc. I owe special thanks to Asia, Giba, Aga & Piotrek, Marta & Okoń, Pumba, Toudi, Zuzia, Klaudia, Aga & Pyra, Karolina, Emma, Peter, and Annamaria (just to mention a few). Many thanks and hopefully see you soon!

Mamo, Tato, dziękuję Wam za wszystko. Za to, że wspieraliście moje najbardziej szalone pomysły (no może poza tym wypadem w Pireneje) i że zawsze we mnie wierzyliście. Dzięki Wam wiem, że nie ma rzeczy niemożliwych, tylko wymagają one trochę więcej czasu. Jesteście dla mnie wzorem do naśladowania. Typie, Babciu, Dziadku – dziękuję Wam za wsparcie, za rozmowy przez Skypa, za czas razem w Gdańsku. Niestety nie mogę być blisko Was na co dzień, ale jesteście w moim sercu i zawsze z wytęsknieniem czekam na chwile, kiedy się zobaczymy na żywo. Lepszej rodziny nie mogłam sobie wymarzyć! Kocham Was!

Lene, Guy en Yannick, hartelijk bedankt voor jullie interesse in mijn werk, voor de ondersteuning, en voor de mooie momenten samen. Bedankt dat jullie mij verwelkomen in jullie familie en dat jullie Filip met mij willen delen.

Filip, schatje, I cannot thank you enough for everything that you have done for me. The day, that our pathways crossed so unexpectedly, was the best and the most life-changing moment of my life. Thank you for being there for me 24/7, for helping me to face and overcome my worries and fears, for teaching me the power of now and the importance of self-confidence. But also for creating and breaking habits, for thinking out of the box, for your opinions and advice, and of course for dancing tango with me. You have showed me that as a team we can achieve everything we desire and made me realize that all of the world’s opportunities are out there waiting for us. My PhD is ready now, so let’s go and take them! I love you baby!

Thank you! Dank u wel! Dziękuję!

Table of Contents Chapter 1 General introduction ................................................................................................................ 1 1.1.

Memory ............................................................................................................ 2

1.2.

Somatic memories ............................................................................................ 4

1.2.1. Symptom schemata .................................................................................... 5 1.2.2. Assessment of somatic memories .............................................................. 6 1.2.3. Sources of somatic memory biases ............................................................ 8 1.2.4. Proposed explanatory mechanisms .......................................................... 11 1.3.

Memory processes in symptom (over)reporting: Dual-process perspective .. 24

1.3.1. Medically unexplained symptoms ........................................................... 25 1.3.2. Dual-process perspective on symptom overreporting in MUS ................ 27 1.4.

Research aims ................................................................................................. 29

1.4.1. General aim .............................................................................................. 29 1.4.2. Overview of studies ................................................................................. 30 Chapter 2 Sensory and affective components of symptom perception: A psychometric approach . 35 2.1.

Introduction .................................................................................................... 36

2.2.

Methods .......................................................................................................... 39

2.2.1. Participants ............................................................................................... 39 2.2.2. Measures .................................................................................................. 40 2.2.3. Data analysis ............................................................................................ 40 2.3.

Results ............................................................................................................ 41

2.3.1. The latent structure of somatic symptoms in the CSD ............................ 41 2.3.2. The latent structure of somatic symptoms in the PHQ-15 ....................... 43 2.3.3. Association between the bifactor models (CSD and PHQ-15) and NA .. 44 2.4.

Discussion ...................................................................................................... 45

2.5.

Acknowledgments .......................................................................................... 47

Chapter 3 Was it so bad? The role of retrospective memory in symptom reporting ...................... 49 3.1.

Introduction .................................................................................................... 50

3.2.

Methods .......................................................................................................... 52

3.2.1. Participants ............................................................................................... 52 3.2.2. Measures .................................................................................................. 53 3.2.3. Apparatuses and physiological recordings............................................... 54 3.2.4. Procedure ................................................................................................. 55 3.2.5. Data analysis ............................................................................................ 56 3.3.

Results ............................................................................................................ 57

3.3.1. Sample characteristics .............................................................................. 57 3.3.2. StudyPain ................................................................................................. 57 3.3.3. StudyDyspnea .......................................................................................... 59 3.3.4. Affective responses .................................................................................. 60 3.4.

Discussion ...................................................................................................... 62

3.5.


Chapter 4 Retrospective memory for symptoms in patients with somatic symptom disorder ......... 67 4.1.

Introduction .................................................................................................... 68

4.2.

Methods .......................................................................................................... 70

4.2.1. Participants ............................................................................................... 70 4.2.2. Measures .................................................................................................. 71 4.2.3. Apparatuses and physiological recordings............................................... 73 4.2.4. Procedure ................................................................................................. 73 4.2.5. Data analysis ............................................................................................ 74 4.3.

Results ............................................................................................................ 75

4.3.1. Dyspnea perception and breathing behavior ............................................ 76 4.3.2. Retrospective dyspnea rating ................................................................... 78 4.3.3. The mediating role of state anxiety .......................................................... 79 4.3.4. The peak-end effect.................................................................................. 80 4.4.

Discussion ...................................................................................................... 81

4.5.


Chapter 5 Ways of encoding somatic information and their effects on retrospective symptom reporting .................................................................................................................................. 85 5.1.

Introduction .................................................................................................... 86

5.2.

Methods .......................................................................................................... 88

5.2.1. Participants ............................................................................................... 88 5.2.2. Measures .................................................................................................. 89 5.2.3. Apparatuses and physiological recordings............................................... 90 5.2.4. Manipulation of processing focus ............................................................ 91 5.2.5. Procedure ................................................................................................. 91 5.2.6. Data analysis ............................................................................................ 92 5.3.

Results ............................................................................................................ 93

5.3.1. Sample characteristics .............................................................................. 93 5.3.2. Physiological responses ........................................................................... 93 5.3.3. State symptoms and affective responses .................................................. 94 5.3.4. Retrospective average symptom ratings .................................................. 98 5.4.

Discussion ...................................................................................................... 99

5.5.

Acknowledgments ........................................................................................ 101

Chapter 6 Recalling somatic memories: The effect of processing focus during retrospective symptom reporting ............................................................................................................... 103 6.1.

Introduction .................................................................................................. 104

6.2.

Methods ........................................................................................................ 106

6.2.1. Participants ............................................................................................. 106 6.2.2. Measures ................................................................................................ 107 6.2.3. Manipulation of processing focus .......................................................... 108 6.2.4. Procedure ............................................................................................... 109 6.2.5. Data analysis .......................................................................................... 111 6.3.

Results .......................................................................................................... 112

6.3.1. Responses at T0 ..................................................................................... 112 6.3.2. Manipulation checks .............................................................................. 115 6.3.3. Retrospective ratings: T1 vs. T0 ............................................................ 117 6.3.4. Retrospective ratings: T2 vs. T1 ............................................................ 120 6.4.

Discussion .................................................................................................... 121

6.4.1. Responses to symptom inductions ......................................................... 122 6.4.2. Retrospective ratings of symptoms and affective responses.................. 123 6.5.


Chapter 7 The specificity of health-related autobiographical memories in patients with somatic symptom disorder ................................................................................................................. 127 7.1.

Introduction .................................................................................................. 128

7.2.

Method ......................................................................................................... 130

7.2.1. Participants ............................................................................................. 130 7.2.2. Measures ................................................................................................ 131 7.2.3. Procedure ............................................................................................... 133 7.2.4. Data analysis .......................................................................................... 133 7.3.

Results .......................................................................................................... 134

7.3.1. Multiple mediator model........................................................................ 134 7.4.

Discussion .................................................................................................... 136

7.5.


Chapter 8 General discussion ................................................................................................................ 139 8.1.

Summary of the findings .............................................................................. 139

8.2.

Bias in retrospective symptom reporting ..................................................... 141

8.2.1. Recall bias appears immediately ............................................................ 141 8.2.2. Recall bias can increase over time ......................................................... 142 8.2.3. Sensory and affective processes in retrospective symptom reporting ... 146 8.3.

Clinical implications .................................................................................... 149

8.3.1. Reducing symptom overreporting.......................................................... 150 8.3.2. Improving symptom assessment ............................................................ 151 8.4.

Strengths and limitations .............................................................................. 153

8.5.

Unresolved issues and further research perspectives ................................... 155

8.5.1. The relevance/weight of specific moments in symptom evaluation ...... 155 8.5.2. Processing focus and memory consolidation ......................................... 156 8.5.3. Memory reconsolidation ........................................................................ 158 8.5.4. The role of autobiographical memory specificity .................................. 159 8.5.5. Memory enhancement ............................................................................ 159

References ............................................................................................................................. 161 Appendix A ............................................................................................................................ 179 Appendix B ............................................................................................................................ 183 Appendix C ............................................................................................................................ 188 Appendix D ............................................................................................................................ 190 Appendix E ............................................................................................................................. 192 Appendix F ............................................................................................................................. 195 Appendix G ............................................................................................................................ 197 Appendix H ............................................................................................................................ 199 Appendix I .............................................................................................................................. 201

List of Tables

Table 2-1.

Goodness of fit for the four different models tested with the CSD (N = 1054).

41

Table 2-2.

Goodness of fit for the four different models tested with the PHQ15 (N = 1053).

43

Table 3-1.

Means and standard deviations for state symptoms, state NA, state anxiety and threat value, and significant effects of repeatedmeasures ANOVA for all dependent variables.

62

Table 4-1.

Group comparisons characteristics.

trait

76

Table 4-2.

Means and standard deviations for state symptoms, state NA and state anxiety in control (n = 22) and patient (n = 22) groups.

76

Table 4-3.

Number (and percentages) of participants per group (controls and patients) who preferred the short or the long trial on each of the forced choice questions assessing the peak-end effect.

80

Table 5-1.

Means and standard deviations for state NA, symptoms, SAM (valence, arousal, and control) and threat ratings for each processing focus and significant effects of repeated-measures ANOVAs.

96

Table 6-1.

Means and standard deviations for the symptom induction trials and significant effects of repeated-measures ANCOVA for the state dependent measures and symptom ratings at T0 (N = 90)

114

Table 6-2.

Pearson product-moment coefficients (r) between the focus ratings during pain induction and affect and symptom ratings at T0: state NA, valence, arousal, threat, averaged concurrent, and immediate pain ratings (N = 90).

116

Table 6-3.

Pearson product-moment coefficients (r) between the focus ratings during dyspnea induction and affect and symptom ratings at T0: state NA, valence, arousal, threat, averaged concurrent, and immediate dyspnea ratings (N = 90).

116

Table 6-4.

Means and standard deviations of the affect and symptom ratings during the pain and the dyspnea induction trials for 3 processing focus manipulation groups (sensory, affective, undirected) at 3 measurement moments (T0, T1, T2).

118

Table 7-1.

Demographic and personality trait characteristics of patients with SSD compared with control group.

135

of

demographic

and

personality

Table 7-2.

Means and standard deviations for indices of autobiographical memory specificity by group.

136

Table A-1.

The Checklist for Symptoms in Daily Life (CSD).

181

Table A-2.

Results of the exploratory factor analysis with varimax rotation (N = 1288). Items of the Checklist for Symptoms in Daily Life are presented with factor loadings for the extracted factors. Cronbach’s alphas, eigenvalues, and the percentage of explained variance for each extracted factor are included.

182

Table B-1.

Goodness of fit for the four different models tested with the CSD for males (n = 184) and females (n = 870).

184

Table B-2.

Goodness of fit for the four different models tested with the PHQ15 for males (n = 184) and females (n = 869).

184

Table F-1.

Means and standard deviations for the processing focus (PF) manipulation groups and F ratios of one-way ANOVAs for the trait characteristics and self-report variables for baseline (BL), pain induction trial (PT) and dyspnea induction trial (DT) at T0 (N = 90).

195

Table I-1.

Pearson product-moment coefficients (r) between the indices of reduced autobiographical memory specificity (specific, categoric, and extended memories) and the main psychological variables (depression and rumination) for the whole sample (N = 54), and the SSD (n = 30) and control group (n = 24) separately.

201

List of Figures

Figure 2-1.

Overview of the models tested with the CSD data. Residual terms of manifest indicator variables not shown.

38

Figure 2-2.

Overview of the models tested with the PHQ-15 data. Residual terms of manifest indicator variables not shown.

39

Figure 2-3.

Bifactor model of somatic symptoms in the CSD with standardized factor loadings (left) and with negative affectivity (NA; right). Single headed arrows represent factor loadings; factor loading coefficients printed in bold are significant at p < .05. Double-headed arrows represent latent correlation coefficients; all correlation coefficients printed in bold are significant at p < .001; residual terms of manifest variables not shown.

42

Figure 2-4.

Bifactor model of somatic symptoms in the PHQ-15 with standardized factor loadings (left) and with negative affectivity (NA; right). Single headed arrows represent factor loadings; factor loading coefficients printed in bold are significant at p < .05. Double-headed arrows represent latent correlation coefficients; all correlation coefficients printed in bold are significant at p < .001; residual terms of manifest variables not shown.

44

Figure 3-1.

Mean values and standard errors of concurrent pain ratings (0-100) for high and low habitual symptom reporters (HSR) during the short (left) and the long trial (right). Whiskers denote standard errors.

58

Figure 3-2.

Mean averaged concurrent and retrospective pain ratings (0-100, left) and dyspnea ratings (0-100, right) for high and low habitual symptom reporters (HSR). Whiskers denote standard errors.

58

Figure 3-3.

Mean averaged concurrent and immediate pain ratings (left) and dyspnea ratings (right) for short and long trial. Whiskers denote standard errors.

59

Figure 3-4.

Mean values and standard errors of concurrent dyspnea (0-100), mean fractional end-tidal concentration of CO2 (FetCO2) and minute ventilation for high and low habitual symptom reporters (HSR) in baseline, rebreathing and recovery phase for the short (left) and the long trial (right).

61

Figure 4-1.

Mean values and standard errors of concurrent dyspnea (0-100), minute ventilation, and fractional end-tidal concentration of CO2 (FetCO2) for controls and patients with MUD in baseline, rebreathing and recovery phase for the short (left) and the long trial (right). Whiskers denote standard errors.

77

Figure 4-2.

Mean averaged concurrent and retrospective dyspnea ratings (0-100) for controls and patients with MUD. Whiskers denote standard errors.

78

Figure 4-3.

Multiple mediator models for short dyspnea trial (left) and long dyspnea trial (right). The panels show direct and indirect effects of a group (patients/controls) on the retrospective dyspnea ratings, mediated by state anxiety and concurrent dyspnea ratings. The model coefficients are reported in unstandardized form.

79

Figure 5-1.

Mean values and standard errors of fractional end-tidal concentration of CO2 (FetCO2, %), minute ventilation (l/min), and respiratory rate (breaths/min) for high and low HSR during dyspnea induction. Data from two sessions are aggregated. To obtain a detailed picture of change over time, data are displayed per 30-s intervals; baseline phase: 0-60 s, induction phase: 60-210 s; recovery phase: 210-360 s.

95

Figure 5-2.

Mean values and standard errors of state negative affect (NA) for sensory and affective processing focus (PF) during baseline, pain induction and dyspnea induction trial as a function of PF manipulation order.

97

Figure 5-3.

Mean retrospective average pain ratings (0-100, left) and dyspnea ratings (0-100, right) for sensory and affective processing focus manipulation. Whiskers denote standard errors.

98

Figure 6-1.

Schematic overview of the study protocol. The order of symptom induction trials was counterbalanced across participants (not shown). Note: PF = processing focus.

110

Figure 6-2.

The interaction of Induction Trial with habitual symptom reporting (HSR) scores for state NA (left), valence (middle), and threat ratings (right) during induction trials at T0. The trial effects are plotted at the different levels of HSR scores (average, +1 SD, -1SD).

113

Figure 6-3.

Mean values and standard errors of ratings of focus on sensations (focus-sens), emotions (focus-emo), and on other aspects of experience (focus-other) in sensory, affective, and undirected processing focus (PF) manipulation during the retrieval (T1) of the pain (left) and dyspnea induction (right).

117

Figure 6-4.

Mean values and standard errors of average pain (left panels) and dyspnea ratings (right panels) in sensory, affective, and undirected processing focus (PF) manipulation for T0 and T1 (top panels), and T1 and T2 (bottom panels).

119

Figure 6-5.

The interaction of Time (T1/T2) with the habitual symptom reporting (HSR) scores for valence ratings of the dyspnea induction trial.

120

Figure B-1.

Bifactor model of somatic symptoms in the CSD with standardized factor loadings (left) and with negative affectivity (NA; right) in male sample. Single headed arrows represent factor loadings; factor loading coefficients printed in bold are significant at p < .05. Double-headed arrows represent latent correlation coefficients; all correlation coefficients printed in bold are significant at p < .001; residual terms of manifest variables not shown.

185

Figure B-2.

Bifactor model of somatic symptoms in the CSD with standardized factor loadings (left) and with negative affectivity (NA; right) in female sample. Single headed arrows represent factor loadings; factor loading coefficients printed in bold are significant at p < .05. Double-headed arrows represent latent correlation coefficients; all correlation coefficients printed in bold are significant at p < .05; residual terms of manifest variables not shown.

186

Figure B-3.

Bifactor model of somatic symptoms in the PHQ-15 with standardized factor loadings (left) and with negative affectivity (NA; right) in male (top) and female samples (bottom). Single headed arrows represent factor loadings; factor loading coefficients printed in bold are significant at p < .05. Double-headed arrows represent latent correlation coefficients; all correlation coefficients printed in bold are significant at p < .001; residual terms of manifest variables not shown.

187

Figure C-1.

Simple (panels a and c) and multiple (panel b) mediator models for long pain trial (panel a), short dyspnea trial (panel b), and long dyspnea trial (panel c). The panels show direct and indirect effects of a group (high/low HSR) on the retrospective symptom ratings, mediated by state NA (all panels) and concurrent symptom ratings (panel b). The model coefficients are reported in unstandardized form.

189

Figure E-1.

Mean values and standard errors of heart rate (beats per minute) for high and low HSR during the pain (left panel) and the dyspnea (right panel) induction trials. Data from two sessions are aggregated. To obtain a detailed picture of change over time, data are displayed per 30-s intervals. Pain induction: baseline phase: 060 s, induction phase: 60-125 s; recovery phase: 125-185 s; dyspnea induction: baseline phase: 0-60 s, induction phase: 60-210 s; recovery phase: 210-360 s.

192

Figure E-2

Mean retrospective pain ratings (0-100, left) and dyspnea ratings (0-100, right) for high and low habitual symptom reporters (HSR) in sensory and affective processing focus conditions. Top panels represent average ratings, middle panels – peak ratings, and bottom panels – end ratings. Whiskers denote standard errors.

193

Figure E-3.

Mean retrospective end pain ratings (0-100) for high and low habitual symptom reporters (HSR) as a function of processing focus manipulation order (sensory-affective/affective-sensory). Whiskers denote standard errors.

194

Figure F-1.

The interaction of Time with the T0 (upper panels) and T1 ratings (lower panels) of pain/dyspnea (T0/T1 pain/dyspnea) scores for average symptom ratings for pain (left panels) and dyspnea (right panels) ratings.

196

General introduction

1 Chapter 1 General introduction

Somatic symptoms play a primary role in the health care system. They are the main reason for patients to decide about utilizing primary health care, while for medical specialists they serve as a major source of information for further testing, clinical diagnosis and informed treatment choice. The decisions of both patients and doctors are based not only on current characteristics and intensity of symptoms (How do you feel right now?), but also, and even to a greater extent, on the retrospective evaluation of bodily symptoms (How did you feel today/last week? How did you feel when you used this medication the previous time?). The way previously experienced somatic events are remembered can have serious and enduring implications for future actions taken by both patients and medical specialists. The recalled unpleasantness of the procedure may lead to the avoidance of certain medical procedures necessary not only for treatment, but also for prevention (Erskine, Morley, & Pearce, 1990; Kent, 1985; Redelmeier, Katz, & Kahneman, 2003). Moreover, as retrospective symptom ratings inform both patient and doctor/researcher on the progress of therapeutic intervention, inaccurate symptom reports may either reduce the adherence to treatment when the symptoms are underreported, or decrease the satisfaction with provided medical care when the improvement remains unnoticed due to symptom overreporting. On the other hand, doctors may decide to unnecessarily increase dosages, assess the symptoms as untreatable, or regard the patient as troublesome (Page & Wessely, 2003). Finally, biased symptom recall may also influence the results of clinical studies assessing effectiveness of therapies and treatments, leading to inaccurate conclusions. Contrary to common belief that recalled symptom ratings correctly reflect prior symptom levels, memory for symptoms is relatively inaccurate (Broderick et al., 2008; Giske, Sandvik, & Røe, 2010; Linton & Melin, 1982) and the retrospective overreporting of experienced symptom intensity is often demonstrated. The reason for this has its source in the complexity of memory processes. Somatic memories are not the unbiased imprints of the state of the body at a given moment of time, but both encoding and retrieval processes can be influenced by various emotional, cognitive, and contextual factors, leading to a modified and often inaccurate retrospective symptom reporting. 1

Chapter 1

As inaccurate symptom memory can results in serious consequences, the primary focus of this thesis is the role of perceptual-cognitive and affective processes in retrospective symptom reporting. Previous studies have repeatedly shown that such processes (e.g., attention, anticipation, illness beliefs, emotions, and personality) modulate the way somatic information is perceived. Those psychological mechanisms can result in a biased perception of bodily symptoms, as seen for example in people with a tendency to report frequent occurrence of physical symptoms in the daily life or medically unexplained symptoms (see Rief & Broadbent, 2007; Rief & Martin, 2014; Van den Bergh, Bogaerts, & Van Diest, 2015, for extensive reviews). However, the question of whether and how memory processes influence retrospective symptom (over)reporting remains largely unanswered. This chapter will first focus on memory processes and biases, especially in the context of somatic memories. After a brief introduction to the topic of memory in general, I will focus on somatic memories in particular. First, their effect on symptom perception through symptom schemata will be explained, followed by a description of memory assessment methods together with a summary of research findings regarding the variability of retrospective symptom ratings in different populations. Moreover, a number of theoretical models explaining retrospective memory biases will be presented. In the last section, I will put forward a working model for this thesis, in which the role of memory processes in symptom reporting will be outlined. Special attention will be directed towards the processes leading to symptom overreporting, especially among individuals characterized by the habitual reporting of bodily complaints, that is, non-clinical habitual symptom reporters and patients with medically unexplained symptoms. The chapter will be closed with the research questions and the outline of the thesis. 1.1. Memory In general, memory is a process of encoding, retention, and retrieval of information. Memories can be classified as explicit and implicit (e.g., Squire, 2004). The latter do not require conscious awareness during formation or retrieval and include the following subtypes: priming, procedural (e.g., cognitive and motor skills), associative (e.g., conditioning), and non-associative memories (e.g., sensitization). While sensitization and conditioning have been widely examined in the domain of symptom experience and chronification, in the current project the focus is predominantly put on self-reported somatic experiences, thus on verbalized memories belonging to the explicit memory class. In contrast to implicit memory,

2


explicit memory can be actively and consciously searched and recollected as well as described to other people. It comprises two independent but related systems: episodic and semantic memory systems (Tulving, 1972; see Tulving, 2002; Wheeler, Stuss, & Tulving, 1997, for reviews). The first system refers to conscious recollection of specific personal events within their defined spatio-temporal context (experience-near knowledge) and contains the summary records of sensory-perceptual-conceptual-affective processing that distinguished a specific experience (Tulving, 1983). By contrast, the latter enables the acquisition and storage of culturally-shared acontextual general knowledge about situations, objects, and relations. Whereas both episodic and semantic systems are involved in memory retrieval, for example in a form of retrospective ratings, their degree of contribution may depend on a number of factors, such as the retention interval and the availability of details (Robinson & Clore, 2002a). Recently, personal semantics were proposed as an additional, intermediate form of explicit memory (e.g., Renoult et al., 2016; Renoult, Davidson, Palombo, Moscovitch, & Levine, 2012). Personal semantics refer to the knowledge of one’s past and share the qualities of both episodic (highly personal) and semantic systems (detached from the acquisition context). They include memories of repeated events and autobiographical knowledge, which are also the components of autobiographical memory. The autobiographical memory system, representing a long-term accumulation of personal knowledge, is also situated within explicit memory. In this memory system, episodic memories are embedded within more general selfrelated semantic knowledge structures, making autobiographical memory a more encompassing system than episodic memory, which includes only personally experienced past events (e.g., Conway & Pleydell-Pearce, 2000; Conway, 2005, 2009). The basic model of autobiographical memory, the self-memory system model (SMS) of Conway and PleydellPearce (2000), defines autobiographical memories as transitory dynamic mental constructions generated from an underlying knowledge base. This hierarchically structured autobiographical knowledge base consists of three levels of representations, in order of descending generality: lifetime periods, general events, and event-specific knowledge (episodic memories; Conway, 2005). The highest level refers to the extended periods of time (e.g., when I went to the highschool), the intermediate (general events) summarizes repeated or single events in form of conceptual categories (e.g., taking exams), while the lowest level represents more concrete sensory-perceptual details of a specific event. Because the levels are nested, accessing specific memory often implies the activation of higher levels of the hierarchy. The construction of a specific memory happens via two different processes, namely direct and 3

Chapter 1

generative retrieval. Direct retrieval leads to spontaneous and automatic recall of a specific memory, when the event-specific knowledge is directly activated. On the other hand, in effortful and hierarchical generative retrieval, the mnemonic cues initiate and specify topdown search processes, which spread activation across the knowledge base, from the general event representation to the event-specific knowledge. The formation and retrieval of memories involves three main stages: encoding, storage (or consolidation), and retrieval (also known as recall or recollection). Encoding refers to various processes (e.g., attention) by which the incoming information is transformed into a memory representation. Once a mental representation consisting of some aspects of the experience is formed, its memory trace needs to be stabilized. In this storage stage, the encoded material can be either consolidated/stabilized (e.g., via rehearsal) or be subject to decay/forgetting. The final memory stage takes place when the information from the past is retrieved in response to a cue eliciting the memory for this experience (e.g., retrospective selfreport measure). As previously mentioned, memories are subject to modulating influences of various cognitive, emotional, and contextual factors, which can operate at all stages of memory processing. Memory distortions due to, for example, quality and focus of attention, previous beliefs and schemata, emotional state and personality characteristics, retention interval, and retrieval cues, as well as the mechanisms underlying them will be discussed in the context of somatic memories at a later stage (see 1.2.4.). In the following section, the focus will be narrowed down to somatic memories: their effect on symptom perception via the symptom schemata, memory biases related to them, and mechanisms that can account for those memory distortions and link them with symptom reporting. 1.2. Somatic memories Somatic memories in the context of this thesis refer to memories of actually experienced somatic symptoms1. The way individuals remember past symptom episodes can affect decisions about future treatment choices, with recalled unpleasantness of the noxious experience usually leading to the avoidance of certain procedures (e.g., Erskine et al., 1990; Kent, 1985; Redelmeier et al., 2003). Memories of aversive somatic experiences are also closely associated with subsequent symptom reporting. In fact, remembered symptom

1

Other terms used to denote those memories in this thesis and in the research domain are symptom memories, pain/dypnea/other symptom memories, and memories for symptoms.

4


intensity has been shown to be a better predictor of future symptom ratings of the same symptom episode than the actual initial symptom ratings (Gedney & Logan, 2006; Noel, Chambers, McGrath, Klein, & Stewart, 2012a). 1.2.1. Symptom schemata One of the pathways by which somatic memories can modulate symptom perception is through symptom schemata. Symptom schemata (or symptom representations) are based on repeated experiences with symptom episodes and capture the similarities between various past symptom occurrences. They have a multidimensional character (e.g., Leventhal, Leventhal, & Contrada, 1998) and their formation and salience may by affected by an interaction of various cognitive and affective processes (see Petersen, van den Berg, Janssens, & Van den Bergh, 2011 for an integrative model on illness and symptom representation). A model of symptom perception which put a particular emphasis on memory structures and symptom representations was proposed by Brown (2004). According to this model, existing symptomrelated knowledge stored in memory in the form of symptom representations/schemata and the level of their activation plays a crucial role in symptom perception and misperception. More precisely, the activation of symptom representations in memory influences the attentional selection of a primary attentional system (PAS) in the early stages of somatic processing by generating a number of perceptual hypotheses from which the most active is selected. The hypothesis denotes a possible interpretation of a somatic stimulus and its activation depends on several factors, including the nature of somatic stimulus, present goals, learning experience with the hypothesis, and activation of memory traces associated with it (R. J. Brown, 2006). The selected hypothesis guides the organization of incoming information into primary representations, which subsequently influence the perceptual and control processes. This approach also accounts for misperception of symptoms, which is thought to originate

from

excessive

activation

of

inappropriate

or

rogue

symptom

representations/schemata in the memory, causing the PAS to choose unsuitable bodily signals during attentional selection. Various factors can affect the formation of maladaptive symptom representations (R. J. Brown, 2004). Memory traces, which become the basis for the rogue representations, are created not only by personal experiences with organic pathology, physical reactions to emotional states, or traumatic events, but are also stored in the absence of any direct illness experience. Indirect exposure, such as observation of the others, verbal

5

Chapter 1

suggestions, and sociocultural models of illness can create memory representations that are parallel to the ones experienced directly by the self. While relevant symptom representations are automatically and effortlessly activated via the PAS in response to well-known stimuli, novel sensations for which the system does not yet possess appropriate schemata require the control of a higher-order, more conscious system – the secondary attentional system (SAS). In such cases, the SAS influences the perceptual process indirectly via the PAS by biasing the activation levels of the symptom representations. Overactivation of symptom representations in memory, which is caused by conscious and recurrent allocation of attention to symptoms by the SAS, may ultimately promote the experience of symptoms in the absence of the actual peripheral physical stimuli, as in placebo and nocebo responses, leading to the development of medically unexplained symptoms (MUS). In summary, repeated experiences of somatic events stored in memory in the form of symptom schemata can modulate the way incoming bodily signals are perceived and interpreted, highlighting the crucial role of somatic memories in symptom perception. Because the reconstructive nature of memory makes it susceptible to various biases, it is important to be aware of the factors that can affect memory for symptoms. Before proceeding with the detailed description of somatic memory biases, the various methodologies and standards applied to assess memory performance will be introduced. 1.2.2. Assessment of somatic memories The growing attention to the studies investigating symptom memory resulted in a wide variety of methodologies and comparison standards used to assess memory performance. The extent to which remembered past somatic episodes correspond with the initial somatic experiences (sometimes referred to as memory accuracy) is mostly assessed by comparing retrospective symptom ratings with momentary assessments collected over a certain period of time or during a somatic experience itself. Because concurrent self-reports are perceived as less biased and more “objective” than memory-based ratings, the aggregation of such ratings is often considered as the golden standard for the comparisons with retrospective ratings (but see Conner & Barrett, 2012, for a different approach). It should be noted that in the context of research on somatic memories, retrospective symptom overreporting denotes the discrepancy between momentary ratings given during the somatic experience and retrospective, memory-based ratings of the same experience, such that the retrospective ratings are higher compared to the momentary ratings. In the experimental 6


studies, symptom overreporting most frequently concerns the ratings of symptom intensity. A similar approach is also applied in the studies of this doctoral project: The bias in retrospective symptom reporting is understood as the discrepancy between retrospective and momentary ratings of symptom intensity. Momentary symptom ratings can be collected either in real-life or naturalistic settings, often in the form of diary protocols, or in response to a controlled induction of somatic symptoms during experimental studies. The studies applying the first type are noticeably more abundant. Diaries are brief self-administered questionnaires used to collect multiple data over extended time periods. Diary protocols used in research on symptom reporting differ substantially in many aspects, including (a) diary format (paper-pencil; e.g., Eich, Reeves, Jaeger, & Graff-Radford, 1985; Giske et al., 2010; Lefebvre & Keefe, 2002 vs. electronic diaries; e.g., Bennett, Amtmann, Diehr, & Patrick, 2012; Sohl & Friedberg, 2008; Stone, Schwartz, Broderick, & Shiffman, 2005), (b) study duration and the schedule of recall moments (end-of-day; e.g., Bennett, Patrick, Lymp, Edwards, & Goss, 2010; Broderick et al., 2008; Houtveen & Oei, 2007; Jensen, Mardekian, Lakshminarayanan, & Boye, 2008 vs. endof-week/study; e.g., Broderick et al., 2008; Howren, Suls, & Martin, 2009; Larsen, 1992), and (c) sampling frequency (once a day; e.g., Bennett et al., 2010; Giske et al., 2010 vs. multiple momentary assessments; e.g., Friedberg & Sohl, 2008; Jensen et al., 2008; Stone et al., 2005). While the older diary studies relied predominantly on paper-pencil diaries and were often faced with the issue of low compliance undermining their validity (Stone, Shiffman, Schwartz, Broderick, & Hufford, 2002), the technological advancements over the past decades enabled a more effective and precise data collection. Newer procedures including ecological momentary assessment (Shiffman, Stone, & Hufford, 2008) and experience sampling methods (ESM; Larson & Csikszentmihalyi, 1983) using electronic format and high sampling frequency were developed to collect (close to) real-time self-reports. Such sampling strategies give the possibility to assess the variability in symptom intensity and the changes in symptom ratings over time. While improvements in diary research have enabled the collection of real-time selfreports of the frequency and intensity of symptoms and their changes over time in the environment natural for the participants, measurement and control of the actual physiological state tend to be poor in such a setting (which may substantially improve in the future with the introduction of the new wearable biosensors). Laboratory-based research using standardized protocols to induce symptoms overcomes this limitation. Several standardized experimental manipulations exist to induce different types of somatic experience (e.g., pain, fatigue, 7

Chapter 1

dyspnea) with control over its intensity, quality, and duration, together with a simultaneous measurement of physiological responses to sensory stimulus and of concurrent symptom ratings. Unfortunately, only a limited number of studies so far have used experimental symptom inductions to investigate the correspondence between the concurrent and retrospective rating of symptom intensity (e.g., Bogaerts et al., 2012; Kahneman, Fredrickson, Schreiber, & Redelmeier, 1993; Redelmeier & Kahneman, 1996; Redelmeier et al., 2003), even though this format permits better control over concurrent symptoms to assess more clearly which factors influence somatic memories. 1.2.3. Sources of somatic memory biases Previous research focused on the discrepancy between concurrently experienced symptoms and overall retrospective evaluation of somatic experiences (also referred to as the memory-experience gap; Miron-Shatz, Stone, & Kahneman, 2009) showing inaccuracies in somatic memories. Overreporting of past symptoms in diverse populations is reported in both children (Van Den Brink, Bandell-Hoekstra, & Huijer Abu-Saad, 2001) and adult patients (Broderick et al., 2008; Eich et al., 1985; Giske et al., 2010; Stone, Broderick, Shiffman, & Schwartz, 2004), as well as healthy individuals (Cohen et al., 2001; Gedney & Logan, 2006). Even though the majority of studies have suggested retrospective overestimation, some findings have also demonstrated a high agreement between recalled and momentary ratings (Bolton, 1999; Jamison, Raymond, Slawsby, McHugo, & Baird, 2006) as well as an underestimation of recalled symptoms (Bąbel, 2016; Brodie & Niven, 2000; Eli, Baht, Kozlovsky, & Simon, 2000). Various factors have been proposed to lead to biased recall of somatic memories including characteristics of the somatic sensations during encoding and retrieval, the affective state accompanying them, the retention interval, cognitive heuristics, as well as characteristics of the individual. First, retrospective symptom ratings can be affected by characteristics of somatic experiences, such as the variability of real-time symptom levels (Lefebvre & Keefe, 2002; Sohl & Friedberg, 2008; Stone et al., 2005), their momentary intensity (Feine, Lavigne, Thuan Dao, Morin, & Lund, 1998; Giske et al., 2010; Hunter, Philips, & Rachman, 1979; Jamison, Sbrocco, & Parris, 1989; Sohl & Friedberg, 2008), and their acuteness/chronicity (Giske et al., 2010; Hunter et al., 1979; Linton & Melin, 1982; Roche & Gijsbers, 1986). However, the findings concerning the impact of those properties on subsequent retrospective symptom ratings are inconclusive. For example, symptom flare-ups resulting from the variability in momentary intensity were associated with both symptom overreporting (e.g., 8


Stone et al., 2005) and more accurate representation of experienced symptoms (e.g., Jamison et al., 1989; Linton & Melin, 1982). In contrast to the mixed findings regarding symptom characteristics during encoding, the effect of actually experienced symptoms at the time a past symptom episode is being retrieved, has received robust support (Eich et al., 1985; Lefebvre & Keefe, 2002; Meek, Lareau, & Anderson, 2001; W. B. Smith & Safer, 1993; Tasmuth, Estlanderb, & Kalso, 1996). More specifically, high symptom intensity at recall was found to be associated with overestimation of past symptom experiences. Second, the affective state during a symptom episode is an important predictor of the recalled symptom ratings. The studies exploring this relationship have shown that a crucial role is played by negative emotional experiences including state anxiety (Eli et al., 2000; Gedney, Logan, & Baron, 2003; Noel, Chambers, McGrath, Klein, & Stewart, 2012b), distress (Everts et al., 1999; Jamison et al., 1989), and negative affect (Gedney & Logan, 2004, 2006). Unpleasant emotional reactions during symptom experience were consistently related to an increase in symptom ratings at retrieval. The impact of affective reactions on memory processes will be discussed more thoroughly later in this chapter. Another factor that might bias retrospective ratings during retrieval is the retention interval: the time between the somatic event and the recall. Specifically, as time frames are wider, a decrease in the correspondence between memory-based ratings and the initially reported symptoms and a gradual increase in overestimation of recalled ratings can be observed (Broderick et al., 2008; Houtveen & Oei, 2007). While ratings given after short retention periods are presumed to be more accurate, for example due to the greater accessibility of recent experiences in episodic memory (Robinson & Clore, 2002a), retrospective overreporting of experienced symptoms can be also observed quite shortly after the end of the aversive somatic experience (Redelmeier & Kahneman, 1996; Redelmeier et al., 2003), suggesting that some recall biases (e.g., episodic memory biases) may operate promptly. One of such biases is due to cognitive heuristics related to salience and recency, resulting in an effect known as the peak-end (PE) effect (Kahneman et al., 1993). The PE “rule” assumes that the retrospective global evaluation of an experience is predominantly determined by two distinctive moments, namely the moment with the highest intensity (peak) and the final moment of the experience, while the duration of the total event is relatively neglected. The role of this heuristic in symptom memory was studied not only in the laboratory (Bogaerts et al., 2012; Kahneman et al., 1993), but also in naturalistic settings, such as during medical examinations (Redelmeier & Kahneman, 1996; Redelmeier et al., 9

Chapter 1

2003) and childbirth (Chajut, Caspi, Chen, Hod, & Ariely, 2014). The vast majority of studies support the notion that the weighted average of peak and final moments is a better predictor of the retrospective symptom rating, compared with simple averaging across all ratings given throughout the experience. For example, in the seminal study of Kahneman and colleagues (1993), participants provided continuous discomfort ratings while being exposed to painful sensations induced with ice-cold water during two trials. In the short trial, the hand was immersed in cold water of 14°C for 60 s. In the long trial, the initial 60-s immersion was extended by additional 30 s, when the temperature gradually increased to still unpleasant but less painful 15°C. When inquired about the discomfort caused by both trials, the long trial was regarded as less painful than the short one, indicating that even though the total time of discomfort caused by the stimulus was extended, the final moments of lower intensity affected the retrospective evaluation by decreasing the overall discomfort ratings. Similar effects were demonstrated also with the other types of bodily sensations, such as dyspnea (Bogaerts et al., 2012). In an experimental study on dyspnea perception and memory processing in patients with medically unexplained dyspnea (MUD) and healthy participants (Bogaerts et al., 2012), the authors adapted the Kahneman’s (1993) design to create a more disorder-relevant experience. Instead of pain induction with cold water immersion, the experimental manipulation included two within-subject dyspnea induction trials using a rebreathing paradigm, consisting of 60-s room air phase and 150-s rebreathing phase. The latter leads to a gradual increase in PCO2, minute ventilation, and a feeling of dyspnea. The short trial ended at the end of the rebreathing phase, thus at the peak level of dyspnea. In the long trial, a mild end was included by extending the induction with 150-s recovery period, during which participants were breathing room air through the breathing system. The relative evaluation after experiencing both trials was assessed using a forced-choice format. While the majority of healthy participants expressed preference for the long trial and rated the short trial retrospectively as more distressing, confirming the PE effect for dyspnea, such pattern was not observed among the patients with MUD. In other words, patients with MUD did not show a peak-end effect. Finally, the differences in retrospective symptom reporting can be related to individual differences in personality characteristics, which can influence cognitive processes during all stages of memory formation and predispose to the overestimation of unpleasant bodily sensations. Personality traits which have a biasing effect on memory of somatic episodes include negative affectivity (NA; Aronson, Feldman Barrett, & Quigley, 2006; Larsen, 1992; Levine & Safer, 2002; Watson & Pennebaker, 1989), pain catastrophizing (Lefebvre & Keefe, 10


2002; Noel, Rabbitts, Tai, & Palermo, 2015; Sohl & Friedberg, 2008), anxiety (Kent, 1985; Suls & Howren, 2012), and depression (Suls & Howren, 2012). Those biasing effects on symptom reporting may be related to various underlying mechanisms including increased attention and hypervigilance to bodily signals, problems with disengaging attention from sensations, negative evaluation of experiences, misattributions, as well as a tendency to ruminate. The role of psychological processes in symptom perception was first emphasized in the seminal work of Pennebaker (1982), who postulated that increased self-focused attention and negative reporting biases observed with higher levels of NA can predispose to misperception and increased reporting of somatic sensations. This approach was further elaborated by Suls and Howren (2012) who demonstrated that anxious and depressive components of NA, due to the distinctive cognitive-affective biases associated with them, have unique impact on selective encoding and retrieval of somatic memories. More specifically, anxiety was associated with inflated frequency reports of momentary symptoms due to heightened attentional bias to threatening stimuli. On the other hand, recall bias, selffocus, rumination, and impaired disengagement from negative information associated with depression was related to overreporting of the frequency of previously experienced (recalled) symptoms. In summary, memory for symptoms seems to be relatively inaccurate and subject to a number of biases during both encoding and retrieval. Even though the findings are sometimes equivocal in terms of outcomes (i.e., over- vs. underestimation), it is necessary to account for different sources of bias while investigating retrospective symptom reporting. Reported inconsistencies may very likely originate from various research methods used (e.g., diaries vs. experiments). More reliable results can be expected from studies employing experimental designs, in which a somatic event, its intensity, frequency, and duration can be standardized and measured. 1.2.4. Proposed explanatory mechanisms In this section, a number of general memory models explaining the mechanisms underlying memory biases will be presented. Whereas the models are of more basic, thus not symptom-specific, character, they will be described within the context of symptom memories and related to the somatic memory biases described above. First, the role of emotions in memory encoding and retrieval will be discussed, followed by an elaboration on the differential function of episodic and semantic memories in retrospective self-report from the perspective of the accessibility model (Robinson & Clore, 2002a). Subsequently, the impact 11

Chapter 1

of memory retrieval processes on the recollection of specific autobiographical memories (Conway & Pleydell-Pearce, 2000) and the factors which may interrupt those processes (Williams et al., 2007) will be brought into focus. Finally, the role of sensory-perceptual and affective-motivational processes in symptom reporting will be addressed. This dual-process perspective will be also proposed as a working model in this thesis. 1.2.4.1. Emotions and memory Emotions exert a great influence on cognitive processing (see e.g., Dolan, 2002, for a review), including memory. Two dimensions of emotional information, valence (positive vs. negative) and arousal (calm vs. arousing), were found to have differential effects on memory processes and systems. With regard to the first dimension, positive and negative emotions affect the way information is remembered, especially during memory encoding (see Kensinger, 2009, for a review; Storbeck & Clore, 2011). Generally, positive emotions have been linked to greater memory distortions than the negative ones, which is probably caused by the type of processing focus engaged while attending to the information. Positive emotions tend to enhance a more schematic, relational, heuristic-based processing, during which the individuals are more prone to the recall of false memories (Storbeck & Clore, 2005, 2011; Storbeck, 2013) and focus more on global rather than local features of stimuli (Gasper & Clore, 2002). Negative mood encourages another type of information processing, which is more item-specific and analytical (e.g., Storbeck & Clore, 2005), and results in opposite effects, that is, fewer false memories and greater focus to certain details, such as visual specificity (Kensinger, Garoff- Eaton, & Schacter, 2007). The mood-related shift to a particular processing style might be modulated by certain factors, including self-relevance of the processed information and emotion-related individual characteristics. A few studies suggested that individuals with NA-related personality traits show a reversed pattern of responses. For example, while false memories are usually observed in response to positive stimuli (Storbeck & Clore, 2005, 2011; Storbeck, 2013), an increased occurrence of false memories for negative material, compared to positive and neutral, was found in individuals with trait anxiety (Toffalini, Mirandola, Coli, & Cornoldi, 2015) and clinical depression (Joormann, Teachman, & Gotlib, 2009). Also in the context of somatic symptoms, negative mood should lead to a more analytical processing style than positive mood. However, the opposite pattern could be present among those with high levels of trait NA. Partial support for this hypothesis was demonstrated in a study of Constantinou et al. (2016) who showed that while induced negative affect may lead to more item-specific 12


processing, more schematic processing in response to the negative mood induction was actually observed among individuals habitually reporting symptom occurrences in daily life (habitual symptom reporters), who also tend to be high on trait NA (Bogaerts et al., 2015). Arousal elicited by the emotional experiences was also found to affect the probability of remembering them. Enhancement of memory for emotionally arousing materials has been observed in a vast amount of studies using various paradigms and stimuli (Cahill & McGaugh, 1995; Christianson & Loftus, 1991; Dolcos, LaBar, & Cabeza, 2004; Hamann, Ely, Grafton, & Kilts, 1999; LaBar & Phelps, 1998). Moreover, the arousal component of emotion can influence different stages of memory processing, such as encoding, consolidation, and retrieval. Arousing materials have higher chances of being successfully encoded, because of their effect on perception and attention during stimuli acquisition (Anderson & Phelps, 2001; Christianson & Loftus, 1991). The mediation theory of emotional memory enhancement (Talmi, 2013; Talmi, Schimmack, Paterson, & Moscovitch, 2007) further suggests that appraisal of an event as arousing may also elicit the allocation of cognitive resources other than attention. More specifically, early long-term memory of emotional stimuli will be also affected by their distinctiveness (emotional stimuli stand out when presented together with nonemotional ones) and tighter organization (a thematic link shared between the emotional stimuli is more automatically grasped). Once encoded, arousing material seems more likely to be consolidated into a stable and lasting memory: Compared to the memories of non-emotional, neutral information which decrease over time, emotional memories become enhanced following a delay (Cahill, Babinsky, Markowitsch, & McGaugh, 1995; Cahill & McGaugh, 1995, 1998; Hamann et al., 1999; Sharot & Phelps, 2004). From a neural perspective, the role of amygdala activation in the modulation of memory consolidation has received robust support from numerous studies (e.g., Dolcos, LaBar, & Cabeza, 2005; LaBar & Cabeza, 2006; McGaugh, 2004, 2006). The focus on the consolidation process and the role of the amygdala is incorporated in the modulation model of memory enhancement (McGaugh, 2004). It posits that secretion of stress hormones (e.g., adrenaline and cortisol) due to the elicited arousal leads to the activation of the amygdala, which is responsible for augmenting the memory traces during consolidation. Finally, while the arousal-related effects have been well defined for encoding and consolidation processes, the evidence concerning the retrieval phase is less conclusive (see Buchanan, 2007, for a review). It is important to note that enhancement of memory caused by arousal does not affect all aspects of the memories equally. More specifically, it was often found that arousal 13

Chapter 1

promotes better memory for the gist, or the central aspects of arousing negative information, but not for peripheral details (see Kensinger, 2009, for a review). According to one of the theories concerned with the memory-narrowing effects of arousal, this trade-off could be attributable to the emotional salience of processed information (Mather & Sutherland, 2011). For example, arousal leads to enhanced processing of salient stimuli, which in turn augments memory for such stimuli, regardless of whether they represent the gist or the specific details (Kensinger, Garoff-Eaton, & Schacter, 2006; Kensinger et al., 2007). Finally, it should be mentioned that emotion enhancement does not necessarily indicate improvements in memory accuracy. Whereas highly arousing events were shown to increase the subjective sense of recollection, it was not always coupled with objective accuracy of those memories (e.g., Rimmele, Davachi, Petrov, Dougal, & Phelps, 2011; Sharot, Delgado, & Phelps, 2004). This means that emotions can lead to more vivid memories, which are subjectively perceived as clearly remembered, but not necessarily are correct, as is, for example, the case with flashbulb memories (e.g., Talarico & Rubin, 2003; see also Phelps & Sharot, 2008, for a review). 1.2.4.2. Accessibility of memory The recollection of somatic memories involves both episodic and semantic explicit memory systems (Brodie & Niven, 2000; Terry, Brodie, & Niven, 2007; Terry, Niven, Brodie, Jones, & Prowse, 2008). However, their degree of contribution to retrospective symptom ratings depends on a number of factors including the retention interval and the availability of details (Geng, Chen, Lam, & Zheng, 2013; Robinson & Clore, 2002a). To delineate the distinctive role of those two systems in the discrepancy between the momentary and retrospectively recalled ratings, Robinson and Clore (2002a) proposed the accessibility model. While it was initially developed to understand biases in ratings of emotions (Robinson & Clore, 2002b), this model was recently adapted to the context of self-reported somatic symptoms (Conner & Feldman Barrett, 2012). The model suggests that the memoryexperience gap is attributable to changes in the accessibility of different types of knowledge, which tap different retrospective memory systems, namely, episodic and semantic memory. Individuals are expected to use those sources of information which are most relevant to the current evaluation and still accessible. In this respect, momentary ratings are expected to reflect experiential knowledge generated in the course of experience, which is episodic, eventspecific, and situated in a particular time and context. When the experiential knowledge becomes inaccessible, for example, during retrospective self-reports, the necessary 14


information will be retrieved from either episodic or semantic memory, introducing different sources of bias. As a result, in self-reports that are based on information from episodic memory, such as the short term recall reports (e.g., Day Reconstruction Method; Kahneman, Krueger, Schkade, Schwarz, & Stone, 2004), episodic memory biases can be expected (e.g., the PE effect; Chajut et al., 2014; Houtveen & Oei, 2007). In contrast, with longer time frames, when episodic details become less available, self-reports might be guided to a greater extent by semantic memory, thus, by information that is more conceptual and decontextualized (Robinson & Clore, 2002b). These reconstructed retrospective ratings might be influenced by subjective beliefs, sematic knowledge, and other individual differences (Bieg, Goetz, & Lipnevich, 2014; Chajut et al., 2014; Robinson & Barrett, 2010). However, even though the accessibility of episodic and semantic memory is believed to depend on the passage of time, as the episodic details fade away faster, it remains unclear when the shift between episodic and semantic memory retrieval takes place. Robinson and Clore (2002b) suggested that time frames shorter than the last few weeks would engage an episodic retrieval strategy, while longer ones would be related to a semantic one. This proposition was further specified in a study investigating the recall of cold pressor pain (Terry et al., 2007), in which the authors found that recollections of pain after 2 weeks involved predominantly episodic memory (the “remembered” pain) and to smaller degree semantic memory (the “known” pain). Moreover, according to Geng and colleagues (2013), the shift (or watershed) between the short and long retention interval happens between 3 to 7 weeks after the experience. In general, it is difficult to pinpoint the exact time frame in which the shift happens. Nevertheless, it can be assumed that in the time period close to the event, evaluations are based on episodic information, while judgments over a long retention interval are constructed using both episodic and semantic information (Brodie & Niven, 2000; Geng et al., 2013; Terry et al., 2008). Furthermore, the accessibility of episodic and semantic information is not only timedependent, but might be also related to the availability of details or features. If the available information does not contain concrete details about the past experience, for example, due to poor encoding or reduced memory specificity, situation-specific or belief-based semantic knowledge could fill-in this gap and have greater impact on the reconstruction process. It that case, personal semantic memories incorporating specific illness- and symptom-related beliefs (or symptom schemata) due to previous symptom experiences or certain personality characteristics (e.g., NA and anxiety) could bias the retrospective ratings leading to symptom overreporting (e.g., Houtveen & Oei, 2007; Larsen, 1992), especially over longer retention 15

Chapter 1

intervals (Robinson & Clore, 2002a). In addition, a recent study showed that recalling more remote memories, a tendency observed among dysphoric individuals, has been related to the deficits in accessibility to specific episodic memories and the retrieval of more overgeneral memories (Falco, Peynircioglu, & Hohman, 2015), which will be discussed further below. 1.2.4.3. Memory specificity The recollection of specific memories can be also influenced by memory retrieval processes and people vary in the ability to retrieve detailed features of remembered experience. Especially overgeneral memory (OGM; Williams et al., 2007), defined as the difficulty to retrieve specific personal memories of a past event, seems to have detrimental effects individuals’ well-being. OGM, also indicated as reduced autobiographical memory specificity (rAMS), was previously observed in a range of psychopathological disorders, including depression, posttraumatic stress disorder (PTSD), and eating disorders (see Williams et al., 2007). Recently, it was also found among patients with chronic pain (Liu et al., 2014). Nonetheless, rAMS is not only a correlate of depression and other emotional disorders, but it is also proposed as an important vulnerability factor for the unfavorable course of psychopathology, predicting the onset and recurrence of the disorder, illness duration, severity of symptoms, and treatment outcomes (Brittlebank, Scott, Williams, & Ferrier, 1993; Sumner et al., 2011; see Sumner, Griffith, & Mineka, 2010, for a review and meta-analysis).

The

assessment

of

memory

specificity

typically

involves

the

Autobiographical Memory Test (AMT; Williams & Broadbent, 1986), in which cue-words are used to prompt the recall of specific memories related to one’s personally experienced past events. rAMS is viewed as the difficulty in retrieving requested specific memories (e.g., “last month, when I attended a funeral of my uncle” to the cue “sad”) or a tendency to recall non-specific memories that cover broader categories of events (e.g., “every time my mother is angry at me”). Various factors which appeared to contribute to or underlie rAMS have been incorporated into one comprehensive model: the CaR-FA-X model (Williams et al., 2007; see also Sumner, 2012). The model is based on assumptions of the basic model of autobiographical memory, the self-memory system model (SMS) of Conway and PleydellPearce (2000) described in Section 1.1. According to the CaR-FA-X model (Williams et al., 2007; see also Sumner, 2012, for a review), overgeneral memories are formed when generative retrieval processes are interrupted and prematurely terminated at higher levels of

16


memory representations as a result of one or more interacting processes: Capture and Rumination (CaR), Functional Avoidance (FA), or impaired eXecutive control (X). The capture phenomenon (CaR) refers to circumstances in which generative memory retrieval is aborted and “captured” at more general levels of the memory hierarchy. As a consequence, the progression to more concrete, specific content is delayed and the retrieved memories are characterized by conceptual and overgeneral content. It is proposed that such capture arises when highly self-relevant retrieval cues activate the network of self-related general information or schemata, promoting the further activation of other highly connected conceptual self-schemata (Dalgleish et al., 2003). In line with this view, a link between cue self-relevance and rAMS has been demonstrated in studies among different populations including patients in remission from major depressive disorder (MDD; Crane, Barnhofer, & Williams, 2007; Spinhoven, Bockting, Kremers, Schene, & Williams, 2007, Study 1), borderline personality disorder (Spinhoven et al., 2007, Study 2), and non-clinical dysphoric participants (Matsumoto & Mochizuki, 2016). Moreover, the susceptibility to capture is increased among individuals with a tendency to engage in ruminative thinking, understood as repetitive and passive self-focus on symptoms of distress and on their possible causes and implications (Nolen-Hoeksema, 1991). Particularly, thinking in an abstract, evaluative manner seems to be most maladaptive (Watkins, 2008) as posited by the processing mode theory of rumination (e.g., Watkins, Moberly, & Moulds, 2008; see also Watkins & Nolen-Hoeksema, 2014). This theory distinguishes between two modes of ruminative self-focus: abstract, characterized by decontextualized, analytical, and evaluative thinking about causes, meanings, consequences of symptoms and concrete, related to non-evaluative and experiential attention to specific sensory characteristics of an event. A number of experimental studies have investigated the relationship between the two modes of ruminative processing and OGM in both clinical (Crane, Barnhofer, Visser, Nightingale, & Williams, 2007; Watkins, Teasdale, & Williams, 2000; Watkins & Teasdale, 2001, 2004) and non-clinical populations (Raes, Watkins, Williams, & Hermans, 2008). It was consistently demonstrated that instructing participants to think about the “why?” aspects, thus the meanings, consequences, and causes of one’s feelings (i.e., abstract/analytical mode) was more detrimental for memory specificity than the focus

on

concrete,

sensory-perceptual,

“how?”

aspects

of

one’s

feelings

(i.e.,

concrete/experiential mode). Moreover, the latter is not only linked with the reduction of OGM, but also with other positive outcomes such as better social problem solving (Watkins & Moulds, 2005), less negative global-judgments (Rimes & Watkins, 2005), less emotional 17

Chapter 1

reactivity and negative mood related to failure experiences (Watkins et al., 2008; Watkins, 2004), and decreased negative generalization (Van Lier, Vanbrabant, Vervliet, & Raes, 2016; Van Lier, Vervliet, Vanbrabant, Lenaert, & Raes, 2014). Finally, also correlational studies gave support to the relationship between rAMS and rumination (Debeer, Hermans, & Raes, 2009; Sumner et al., 2014; but see Smets, Griffith, Wessel, Walschaerts, & Raes, 2013). The association with rAMS seems to be especially the case of the analytical and abstract type of repetitive or ruminative thinking (or “brooding” Treynor, Gonzalez, & Nolen-Hoeksema, 2003). The second factor, functional avoidance, posits that rAMS may be caused by (cognitive) avoidance of (retrieving) aversive and intense specific memories in order to diminish the impact of negative affect associated with those memories (Williams et al., 2007). This theorizing was stimulated by the evidence showing that heightened levels of OGM are associated with a history of traumatic experiences (e.g., Hermans et al., 2004; Kuyken & Brewin, 1995; McNally, Perlman, Ristuccia, & Clancy, 2006). Because the recollection of specific traumatic memories can lead to re-experience of negative affect, avoiding detailed retrieval was proposed to develop as a reaction to (early) adversity and trauma (Williams et al., 2007). The mixed findings in this domain suggest that while the occurrence of early traumatic events is a plausible vulnerability factor, it is neither sufficient, nor necessary for developing rAMS (see Sumner, 2012, for a review of the role of early adversities). Nonetheless, the affect-regulation hypothesis of functional avoidance, which posits that rAMS can be considered a broader cognitive avoidance strategy, has received considerable support from a growing number of studies. In line with this idea, rAMS shows correspondence with avoidant coping styles as assessed by a variety of avoidance measures (Hermans, Defranc, Raes, Williams, & Eelen, 2005), and can be observed especially under threatening conditions (Debeer, Raes, Williams, & Hermans, 2011). Furthermore, avoiding the retrieval of specific memories appears to be a functional and protective strategy to regulate one’s affective state, at least on the short-term. More specifically, a less specific retrieval style was associated with a lower emotional distress to mild aversive experiences, such as failure on an exam (Hermans et al., 2008) or solving a difficult puzzle (Raes, Hermans, de Decker, Eelen, & Williams, 2003; Raes, Hermans, Williams, & Eelen, 2006, Study 1). However, extended and less flexible use of this retrieval style may develop into a categoric retrieval strategy and become maladaptive (Raes et al., 2006, Study 2; Williams et al., 2007).

18


The third factor, executive control, refers to deficits in executive resources that prevent the successful retrieval of specific memories (Dalgleish et al., 2007). Such resources are necessary to retain retrieval goals, inhibit inappropriate representations, and maintain the focus of attention during retrieval. Numerous studies focused on studying the association between difficulties with executive capacity and rAMS in various populations were summarized in a comprehensive review by Sumner (2012), who showed that reduced memory specificity is related to deficits in inhibition, working memory capacity, verbal fluency, and the ability to update and maintain information in working memory. Those deficits may affect the retrieval of specific memories by disrupting the ability to organize a successful generative retrieval search or to dismiss irrelevant information. It is important to note that the importance of the outlined mechanisms in the occurrence of rAMS may depend on the characteristics of clinical populations or contexts. For example, while functional avoidance mechanisms (affect regulation) can have more influence on memory specificity in trauma-exposed samples (Dalgleish, Rolfe, Golden, Dunn, & Barnard, 2008), impaired executive control may have a more critical function in rAMS among patients with MDD (Sumner et al., 2014). Furthermore, particular mechanisms may lead to rAMS independently or by interacting with each other (Barnhofer, Crane, Spinhoven, & Williams, 2007; Sumner et al., 2014). However, more experimental studies examining how multiple mechanisms simultaneously affect memory specificity are needed to fully understand the interactions within the model. Even though the phenomenon of OGM has been predominantly investigated in the context of depression and other emotional disorders, it might be also relevant in understanding memory biases observed in somatic patient groups, such as medically unexplained symptoms (MUS). The proposed vulnerability factors for MUS include, for example, increased occurrence of early lifetime adversities (“FA” mechanism) among patients qualifying for MUS (Bohn, Bernardy, Wolfe, & Häuser, 2013; Creed et al., 2012; Kuwert, Braehler, Freyberger, & Glaesmer, 2012; Nelson, Baldwin, & Taylor, 2012). Furthermore, deficits in executive control (“X” mechanism) have also been reported in this patient population, including deficits in response inhibition, cognitive flexibility and working memory (Aizawa et al., 2012; Correa, Miró, Martínez, Sánchez, & Lupiáñez, 2011; Solberg Nes, Roach, & Segerstrom, 2009). rAMS in this patient population would indicate that these patients rely on a more categoric retrieval style, thus recall summaries of somatic experiences instead of specific symptom occurrences. This lack of access to sensory-perceptual details of

19

Chapter 1

past somatic experiences could have an important effect on retrospective symptom reporting, leading to less detailed and more biased recollections of somatic memories. 1.2.4.4. Dual-process perspective to symptom reports According to both behavioral (parallel processing model; Leventhal & Everhart, 1979) and traditional neurobiological models (Craig, 2003; Price, 2000), ratings of somatic symptoms result from the interplay of two components of the somatic sensation: sensoryperceptual and affective-motivational. The sensory-perceptual component refers to the distinct sensory characteristics of bodily experiences including quality, intensity, location, and duration of the stimulus. On the other hand, the affective-motivational component conveys the feelings induced by experiences together with a motivational drive to behaviorally react or adjust (to) its effect. Consequently, somatic symptoms can be perceived as “homeostatic emotions” which play a crucial role in maintaining the integrity of the body (Craig, 2003), such that the state and the needs of the organism will determine the affective-motivational evaluation of the somatic input. For example, a heat stimulus may be perceived as pleasant when one feels cold, and unpleasant when feels hot. The dual-process perceptive incorporates the assumptions of parallel processing model (Leventhal, Brown, Shacham, & Engquist, 1979; Leventhal & Everhart, 1979), which posit that the sensory and affective aspects of somatic information are processed in parallel, contributing to the global symptom evaluation. Consequently, retrospective symptom ratings might be influenced by the way somatic experience is attended to and processed. Specifically, focusing on concrete, objective and nonemotional aspects of somatic information (sensoryperceptual processing focus; sensory PF) could lead to more detailed encoding of sensoryperceptual features of bodily changes. On the other hand, focusing on the affective aspects of a somatic experience (affective-motivational processing focus; affective PF), which is usually unpleasant and aversive, may direct attention towards symptom-related distress which could bias symptom ratings. Moreover, given the limited capacity of attention, it can be assumed that selecting one of the processing foci could be associated with a decreased attention to the other. As a consequence, increased affective processing could happen at the cost of detailed sensory-discriminative processing leading to less detailed encoding and memory of the perceptual details of the experience. Furthermore, as the sensory and affective aspects are not completely independent, they might influence each other (e.g., a parallel-serial model of pain affect; Price, 2000). Finally, the impact of the two components on symptom ratings may change with the passage of time due to the characteristics of the components. More 20


specifically, while the relative influence of the affective component on symptom memory tends to increase over time (Gedney et al., 2003; Gedney & Logan, 2004; Kent, 1985), the sensory details are forgotten faster (Reyna & Brainerd, 1995). This could suggest, that biases observed with the extended retention times might be related to the affective component. A range of studies have investigated the outcomes of processing somatic information in a sensory-perceptual and/or an affective-motivational manner. Attending to the perceptual, sensory details was often shown to be related to lower symptom ratings (Ahles, Blanchard, & Leventhal, 1983; Crane & Martin, 2003; Haythornthwaite, Lawrence, & Fauerbach, 2001; Johnston, Atlas, & Wager, 2012; Keogh, Hatton, & Ellery, 2000; Keogh & Herdenfeldt, 2002; Leventhal et al., 1979; Suls & Fletcher, 1985). For example, Crane and Martin (2003) conducted a diary study among individuals suffering from irritable bowel syndrome (IBS). On separate days, the patients were requested to evaluate the severity of gastrointestinal symptoms after a processing style manipulation, which included a neutral (sensory) focus, an illness context focus, and three control conditions. The authors reported that focusing on the symptoms in a neutral way led to lower symptom ratings compared to the control conditions, while attending to the illness context resulted in higher symptom ratings, probably due to the activation of illness-schemata. Keogh and colleagues adapted an experimental approach in their studies by inducing a painful experience via the cold pressor task (Keogh et al., 2000; Keogh & Herdenfeldt, 2002). The findings indicated that lower sensory pain ratings were related to attending towards the pain rather than avoiding it (Keogh et al., 2000). Moreover, attending towards the physical sensations (sensory PF) evoked by the painful task was more advantageous in this respect than focusing on the emotional aspects (affective PF) (Keogh & Herdenfeldt, 2002). However, this beneficial effect of sensory PF was found only for men. In contrast, women did not profit from the sensory PF and responded to the affective PF with higher affective pain ratings (compared to sensory PF). Except for gender, the efficacy of a sensory PF also seems to be modulated by experience-related stress (Baron, Logan, & Hoppe, 1993; Cioffi, 1991; Logan, Baron, & Kohout, 1995) and anxiety-related personality traits, for example health anxiety (Hadjistavropoulos, Hadjistavropoulos, & Quine, 2000), fear of pain (J. Roelofs, Peters, van der Zijden, & Vlaeyen, 2004), and catastrophizing (Michael & Burns, 2004). In a laboratory setting, Michael and Burns (2004) induced two consecutive painful experiences in chronic pain patients by means of the cold pressor task. The trials were separated by a betweensubject manipulation of the processing focus. The findings suggest that the effect of sensory and affective PF on changes in pain threshold and tolerance was related to the level of pain 21

Chapter 1

catastrophizing. Affective PF was associated with decreased pain threshold and tolerance among high catastrophizers, but had no visible impact among the low catastrophizers. Conversely, sensory PF led to an increase in pain threshold and tolerance among the low group, but had no effects on high catastrophizers. Similar findings were demonstrated in a study of Bogaerts and colleagues (2008), who examined interoceptive accuracy and retrospective symptom reporting among nonclinical high and low reporters of medically unexplained symptoms. The participants experienced two experimentally induced dyspnea episodes, which were framed in either neutral way (respiratory sensations) or in a symptom context (respiratory symptoms). Whereas in the neutral frame no group-related differences could be observed with respect to interoceptive accuracy, the symptom context led to increased retrospective ratings and reduced accuracy only in the high symptom reporters, who also tend to score high on trait NA. It may be of relevance to look at the influence of different processing styles on memory in other domains such as reality monitoring (discrimination between perceived and imagined events) and source monitoring (identification of source of information). Numerous studies by Johnson and collaborators investigated the effects of two types of rehearsal, factual and affective, on memory characteristics (Hashtroudi, Johnson, Vnek, & Ferguson, 1994; Mather & Johnson, 2003; Suengas & Johnson, 1988). In their understanding, a factual (perceptual) review comprises a focus on perceptual and contextual features of the experience, while an affective (apperceptive) review involves focusing on feelings, thoughts and ideas, together with reactions and difficulties experienced during the event. Thinking about the affective characteristics of an event was found to have various consequences on memory performance. First, affective rehearsal led to memories with fewer sensory and contextual details as compared with perceptual rehearsal, suggesting that concentration on thoughts and feelings related to the event results in decreased access to sensory and contextual information (Suengas & Johnson, 1988). Second, affective review was associated with lower recall accuracy for experienced events (Hashtroudi et al., 1994). Finally, while perceptual review led to better recall of details, apperceptive review resulted in a higher proportion of inferences about the previously read story (Mather & Johnson, 2003). Additionally, participants reviewing the story in an affective manner relied to a greater extent on schematic knowledge during recognition judgments. Based on those findings, it can be concluded that affective rehearsal impacts memory for experienced events by decreasing the accessibility of sensory and contextual information, while simultaneously supporting recall of elaborations and inferences. This reduced access to specific memory details may further lead to increased 22


schema-reliance during memory retrieval and possible schema- and belief-based misattributions (Robinson & Clore, 2002a). In the abovementioned studies, a type of review was experimentally manipulated during memory consolidation and retrieval. However, affective processing may be also induced by various situational (highly emotional events, social isolation, depressed mood) or dispositional factors (self-consciousness, trait anxiety and rumination). People with such personality traits have a tendency to focus on their emotional reactions to the events and for that reason may be more susceptible to schematic memories and misinterpretations resulting from them. In general, the findings from various research domains presented above suggest that symptom ratings might be influenced by the trade-off between a focus on affective components and a focus on perceptual components of an event. A sensory PF is often perceived as more beneficial, as it usually leads to lower symptom ratings. On the other hand, focusing on the emotional reactions to the experience may result in negative outcomes, such as higher symptom ratings or lower tolerance to aversive stimuli, especially among individuals with anxiety-related personality traits, who may be more responsive to negative affective stimuli. While the majority of findings demonstrated the effects of processing focus manipulation at the encoding phase, this manipulation at a later stage of memory processing (i.e., retrieval) could also have an impact on the way symptom memories are remembered. Each time a (symptom) memory is retrieved, it is malleable (e.g., Schwabe, Nader, & Pruessner, 2014). If persons are instructed to focus on the affective aspects of the retrieved memory, it is likely that those aspects are more strongly reconsolidated, or added to the memory. As a consequence, guiding the way people think about the past events may change the way they remember them. The overview of theoretical accounts and research findings concerning somatic memories provided in this section demonstrates that symptom ratings, both momentary and retrospective, are not an unbiased report of the physiological responses of the physical body. Retrospective ratings are affected not only by the characteristics of somatic stimulus, but also by a number of factors including affective states, individual differences, and how the individuals attend to or process the incoming information at different stages of memory formation. In the following section, the knowledge derived from the described literature will be incorporated to propose a working model that will be explored in this thesis to gain more insight into the role of memory processes in somatic symptoms (over)reporting.

23

Chapter 1

1.3. Memory processes in symptom (over)reporting: Dual-process perspective The assessment of somatic symptoms in both clinical and research settings frequently relies on self-report which is prone to various biasing factors. Inaccuracies in symptom reporting are often shown in studies concerned with symptom memory, taking the form of retrospective symptom overreporting (memory-experience gap). Various psychological mechanisms have been proposed to affect symptom reporting, including attention, symptom schemata, illness beliefs, expectancy, affective cues, comparison processes, personality traits, and also memory (see Rief & Broadbent, 2007; Van den Bergh et al., 2015, for detailed reviews). Numerous models have considered the role played by such cognitive-affective processes in (biased) reporting and (mis)perception of symptom in general (e.g., Pennebaker, 1982), and of specific symptoms, such as pain (Keefe, Rumble, Scipio, Giordano, & Perri, 2004; Lethem, Slade, Troup, & Bentley, 1983; Vlaeyen & Linton, 2000, 2012) and respiratory symptoms (De Peuter et al., 2004; Janssens, Verleden, De Peuter, Van Diest, & Van den Bergh, 2009; von Leupoldt & Dahme, 2007b). Until now, a considerable amount of attention was placed on attentional processes, illness beliefs, and personality traits in the abovementioned models. However, as symptom ratings often tend to be retrospective (How did you feel?), they also depend to a great extent on memory. For this reason, a more explicit focus on memory processes in symptom reporting could possibly shed light on the mechanisms underlying retrospective symptom (over)reporting. While a number of potential explanatory mechanisms for somatic memory bias have been outlined in the previous sections, our investigations to be reported here will most predominantly focus on and test a dual-process perspective on symptom reporting. In this view, we propose that the way sensory and affective components of somatic experiences are processed, stored, and retrieved could account for the biases in retrospective symptom reporting. More specifically, biased symptom reporting could be a consequence of (a) less detailed processing of somatic information, leading to reduced specificity of sensory component, and (b) an overreactive evaluative system which increases the attention to and the impact of symptom-related distress, negative emotions and affective cues, resulting in stronger influence of negative affective component. In order to understand how memory processes influence symptom overreporting, the proposed model will be first explored among individuals with a tendency to overreport2

2

A feature characteristic for high HSR is increased symptom reporting/overreporting of symptoms occuring in daily life. The questionnaires measuring HSR typically include an extended list

24


symptoms in the daily life, known as habitual symptom reporting (HSR). HSR is characterized by self-reports of frequent somatic symptom experiences that cannot be attributed to a known physiologic dysfunction. Moreover, it is associated with a tendency to experience increased levels of negative affect (NA) and anxiety (Bogaerts et al., 2015). With regard to symptom ratings, compared to low HSR individuals, those with high levels of trait HSR tend to retrospectively recall more symptoms than it would be expected from momentary ratings. This retrospective overreporting was also found to increase over longer retention intervals (Houtveen & Oei, 2007). The insight gained from investigating the processes related to retrospective symptom overreporting in this group is expected to complement existing models concerned both with symptoms with clear physiological origin (e.g., Janssens et al., 2009; von Leupoldt & Dahme, 2007b) and with symptoms without sufficient underlying medical explanation, so called medically unexplained symptoms (e.g., R. J. Brown, 2004; Rief & Broadbent, 2007). 1.3.1. Medically unexplained symptoms Medically unexplained symptoms (MUS) are very common in the society, with the symptoms of 20-50% of patients in both primary (see Haller, Cramer, Lauche, & Dobos, 2015, for a review) and secondary care (Carson et al., 2000; Nimnuan, Hotopf, & Wessely, 2001; Perkin, 1989; Reid, Wessely, Crayford, & Hotopf, 2001) remaining without a clear medical explanation. Due to an extensive and complex diagnosis and treatment, MUS are a serious burden not only for the individual, but also for the society, resulting in increased use of health service resources, more primary care and specialists visits, in addition to higher inand outpatient costs (Barsky, Orav, & Bates, 2005), which are comparable to anxiety and depressive disorders (Konnopka et al., 2012). Additionally, a prolonged diagnostic procedure and unknown illness status reduces patient’s quality of life, satisfaction with patient-doctor contact, and increases chances for absenteeism. The development and presence of MUS have been related to a number of demographic, environmental, and psychological factors, including female gender (Barsky, Peekna, & Borus, 2001; Kroenke & Spitzer, 1998; Nimnuan, Hotopf, et al., 2001; Steinbrecher, Koerber, Frieser, & Hiller, 2011), early lifetime experiences with chronic illness

of somatic symptoms, with the final score based on the aggregation of both frequency ratings per symptom and the number of indicated different symptoms. For this reason, habitual reporting of symptoms in daily life is often viewed as symptom overreporting in terms of frequency ratings. However, overreporting among high HSR is can be seen for both types of measures: frequency and intensity (e.g., Bogaerts et al., 2008, 2015; Constantinou et al., 2013).

25

Chapter 1

and MUS (Hotopf, 2002; Hotopf, Mayou, Wadsworth, & Wessely, 1999), and traumatic experiences, as well as physical, psychological, and sexual abuse (Creed et al., 2012; see Nelson et al., 2012; K. Roelofs & Spinhoven, 2007, for extensive reviews). Moreover, a robust association has been observed between certain personality traits such as NA and the presence of elevated symptom frequency reports and MUS (Bogaerts et al., 2015; Van Diest et al., 2005). Also, the symptoms of anxiety and depression often coincide with MUS in the general population (van Eck van der Sluijs et al., 2015) and in the primary care (de Waal, Arnold, Spinhoven, Eekhof, & van Hemert, 2005; Kroenke, 2003a; Löwe et al., 2008), while psychiatric comorbidities with anxiety and depressive disorders are frequently reported for patients with functional syndromes (de Waal, Arnold, Eekhof, & van Hemert, 2004; Henningsen, Zimmermann, & Sattel, 2003; Wessely, Nimnuan, & Sharpe, 1999). The psychometric research exploring the latent structure of symptom reporting, in general as well as of disorder-specific symptoms, suggests that MUS can be viewed as a dimensional rather than a categorical phenomenon (e.g., Jasper, Hiller, Rist, Bailer, & Witthöft, 2012; Van Oudenhove et al., in press). This means that symptom reporting can be represented as a continuum that stretches from mild bodily symptoms in daily life, such as in the non-clinical cases of HSR, to a serious somatic symptom disorders, including irritable bowel syndrome, chronic fatigue syndrome (CFS), or fibromyalgia (Fink, Rosendal, & Olesen, 2005; Katon et al., 1991; R. C. Smith & Dwamena, 2007). In the previous Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev.; DSM–IV–TR; American Psychiatric Association, 2000), the extreme end of the MUS continuum was a defining criterion of somatoform disorders. However, the recent revision (5th ed.; DSM–5; American Psychiatric Association, 2013) introduced a number of substantial modifications concerning this disorder, including the change of term to somatic symptom disorder (SSD). Moreover, the diagnostic criteria were updated and currently require not only the presence of at least one persistent, distressing, and impairing somatic symptom, but also the fulfillment of psychological criteria, that is, excessive feelings, thoughts, and behaviors linked to those symptoms. As a result, in the current classification the distinction between the medically explained and unexplained somatic symptoms was removed, and greater emphasis was placed on the cognitive and emotional evaluation of bodily complaints, which is in line with models of MUS that accentuate the central role of perceptual-cognitive processes in the development and maintenance of somatic syndromes (e.g., R. J. Brown, 2004; Rief & Barsky, 2005).

26


1.3.2. Dual-process perspective on symptom overreporting in MUS3 Current psychological models on somatic symptoms, both explained and unexplained, include among other cognitive factors, the role of memory processes in symptom reporting. For example in the model of Brown (2004), MUS are expected to arise from a chronic activation of memory schemata/symptom representations. Nevertheless, a very limited number of studies have focused on possible biases in retrospective symptom reporting among individuals with MUS, both in mild (non-clinical high HSR) and more severe form (patients with MUS/SSD). With the exception of the previously mentioned diary study of Houtveen and Oei (2007), which demonstrated retrospective overreporting of daily life symptoms among the high HSR individuals, most of the studies concerning memory processes in MUS were concentrated on either memory deficits (e.g., Dick, Verrier, Harker, & Rashiq, 2008; Glass, Park, Minear, & Crofford, 2005; Niemi, Portin, Aalto, Hakala, & Karlsson, 2002; Sephton et al., 2003) or on attentional bias towards body-/illness-related information (Lim & Kim, 2005; Martin, Buech, Schwenk, & Rief, 2007; Wingenfeld, Terfehr, Meyer, Löwe, & Spitzer, 2013; Witthöft, Gerlach, & Bailer, 2006), but not on memory of actually experienced somatic symptoms. While research explicitly focusing on retrospective symptom overreporting in MUS is lacking, a number of findings concerning symptom perception and memory in individuals with HSR/MUS suggest that a dual-process perspective may elucidate, at least to some extent, a putative mechanism underlying symptom overreporting. This approach posits that somatic symptom ratings arise from the interplay of sensory-perceptual and affective-motivational components of a somatic experience. Consequently, overreporting of somatic symptoms could originate from two factors: (a) reduced specificity of a sensory component due to a less detailed information processing, and (b) stronger influence of affective component because of an affective system being more reactive to emotional cues, symptom-related distress, and negative feelings. With regard to the first aspect, growing evidence from multiple experimental studies has documented several specific characteristics of symptom perception and memory among 3

Even though the omission of the criterion regarding medical explanation for the somatic complaints may be beneficial for clinical practice, it was advised to maintain this division for research purposes by specifying the MUS as symptoms which “are not better explained by a general medical condition” (Rief & Martin, 2014, p. 343). However, as MUS still remains the widely used term in the symptom research, the terms “medially unexplained symptoms (MUS)” and “somatic symptom disorder (SSD)” will be used interchangeably throughout this thesis, although they refer to different concepts/editions of DSM. In this thesis, both terms are used to refer to self-reported bodily complaints which cannot be completely explained by an organic dysfunction.

27

Chapter 1

individuals with HSR/MUS. These include (a) lower correspondence between self-reported symptom ratings and relevant physiological responses (Bogaerts et al., 2008; Bogaerts, Van Eylen, et al., 2010), (b) elevated vulnerability to nocebo effects (Van den Bergh, Stegen, & Van De Woestijne, 1997), (c) a lack of the peak-end heuristic for somatic episodes, typically found in healthy participants (Bogaerts et al., 2012), and (d) reduced differentiation among different somatic sensations together with increased bias due to a priori knowledge when classifying them (Petersen, Van Staeyen, Vögele, von Leupoldt, & Van den Bergh, 2015). Taken together, those findings indicate a less detailed sensory-perceptual processing of bodily information, which may impact how somatic episodes are experienced and remembered. Second, the predictions concerning the characteristics of the affective component can be based on a well-documented relationship between HSR and NA (Bogaerts et al., 2015). As already mentioned, there is a strong relationship between elevated reports of somatic symptoms and NA (Kirmayer, Robbins, & Paris, 1994; Larsen, 1992; Van Diest et al., 2005; Watson & Pennebaker, 1989). This association is often explained by the symptom-perception hypothesis (Watson & Pennebaker, 1989), which postulates that high NA individuals are more hypervigilant to signs of threat, what leads to cautious scanning of both external and internal environment for possible cues. Increased self-focused attention and negative reporting biases predispose to easier perception of somatic sensations and to more eager interpretation of them as signs of illness. Those misinterpretations, in addition to an overreactive evaluative system (Hariri, Bookheimer, & Mazziotta, 2000; Yiend, 2010), may increase the relative impact of affective processes compared to sensory-perceptual ones during somatic experiences, resulting in increased symptom reporting. Even though this theory assumes that high NA persons are overreactive to minor physical dysfunctions, there is substantial experimental evidence that physiological changes are not necessary to evoke symptom experience (Bogaerts, Janssens, De Peuter, Van Diest, & Van den Bergh, 2010; Constantinou, Bogaerts, Van Diest, & Van den Bergh, 2013; Van den Bergh, Stegen, & Van De Woestijne, 1998). In fact, it has been shown that symptom overreporting in high HSR/NA persons is automatically triggered by negative affective cues such as foul-smelling odor or negative pictures in the absence of corresponding changes in peripheral physiology (Bogaerts, Janssens, et al., 2010; Constantinou et al., 2013; Van den Bergh et al., 1998). Also, this effect was not mediated by the experienced negative affective state of the subject, but moderated independently by valence and arousal of the cues (Constantinou et al., 2013). Moreover, negative context during which bodily sensations are experienced results in both higher symptom ratings and lower interoceptive accuracy among 28


high NA and high HSR (Bogaerts et al., 2005; Van den Bergh et al., 2004), suggesting a substantial influence of affective context on the processing of somatic information. Altogether, these findings suggest a critical role of affective cues in symptom acquisition and experience in general, and that, specifically in high HSR and high NA persons, the association between negative emotional states and symptom reporting is more easily formed and subsequently activated. This is in line with schema theory (R. J. Brown, 2004), which assumes that well-learned schemata are automatically activated by contextual and environmental cues and can be misinterpreted as a current symptom experience. In this case, affective cues activate symptom schemata or somatic memories more easily and high HSR may have difficulties to inhibit this activation, resulting in more biased symptom reporting. Moreover, increased reliance on the self-schemata from the semantic knowledge can be also stimulated by the lack of access to the episodic memories, caused by either the lack of detailed information concerning the somatic experience or the passage of time (Robinson & Clore, 2002a). In summary, the evidence presented in this section suggests that somatic memories in high HSR/MUS are dominated by the affective component, while the sensory-perceptual information processing in this group is supposed to be less detailed. This reduced specificity of sensory details together with increased recollection of the experienced unpleasantness may be proposed as a mechanism underlying the symptom overreporting not only in general, but also among the individuals who experience and report frequent occurrences of (often unexplained) bodily symptoms. 1.4. Research aims 1.4.1. General aim The main goal of the present doctoral project is to investigate somatic memories and the role of memory processes in retrospective symptom (over)reporting. Based on the dualprocess perspective on symptom reporting (e.g., Craig, 2003; Leventhal & Everhart, 1979; Price, 2000), we put forward the hypothesis that differential processing of sensory and affective components of a somatic experience is one of the mechanisms underlying biases in symptom reporting. Specifically, we propose that when somatic experiences with a strong negative emotional component are combined with reduced specificity of perceptual details, the affective information can dominate over the less detailed sensory information leading to more intense symptom reports. Such conditions are likely to be present especially among the

29

Chapter 1

individuals with an overreactive evaluative system, among whom negative affectivemotivational processing of somatic information could overshadow the sensory-perceptual component when reporting the experienced symptoms. As such characteristics apply to individuals with high frequency of self-reported symptoms (high HSR) and patients with MUS, individuals fulfilling the criteria for high HSR/MUS will be selected to explore the hypotheses stated in this project. Symptom reporting, due to its dimensional structure (Jasper et al., 2012), can be represented as a continuum ranging from mild bodily complaints to severe disorders. Consequently, individuals who display a tendency to experience mild symptoms in daily life (high HSR) can be treated as a non-clinical proxy to clinical samples of patients who suffer from MUS. As a result, the studies focused on this non-clinical group could also improve the understanding of the mechanisms underlying the development and maintenance of MUS. Due to a general lack of experimental studies on memory processes in (retrospective) symptom reporting in MUS, the focus was first put on establishing a more solid data base on the role of symptom memory in symptom (over)reporting in high HSR and patients with MUS. In most of our studies, two types of bodily symptoms, dyspnea and pain, were induced within-subject in controlled laboratory conditions in order to explore (1) the extent to which retrospective evaluations of somatic experiences resemble the momentary ratings, and (2) the changes in symptom reporting over time. Assessment of both self-reported and psychophysiological responses to somatic stimuli took place concurrently during symptom inductions. The aggregated momentary ratings of symptom intensity given during the experience served as a standard for comparison with retrospective ratings concerning the same event, which were collected up to two (Studies 2-4) and four (Study 5) weeks after the somatic experiences. Moreover, the possible differences in retrospective symptom reporting related to HSR were examined by between-subject comparisons, in which the responses of low HSR were considered as a standard for comparisons. A second – related - research question explored the relative role of sensory-perceptual and affective-motivational processing focus in symptom (over)reporting. In some studies we therefore manipulated the processing focus at different phases of memory formation. 1.4.2. Overview of studies Study 1: Sensory and affective components of symptom perception: A psychometric approach In a first study, we adopted a psychometric approach to investigate the latent structure of symptom reporting in a healthy sample. Psychological and neurobiological models propose 30


that symptom perception results from the interplay between sensory-perceptual and affectivemotivational components of symptoms. Recently, psychometric analyses have pointed out that the latent structure of symptom reporting is best explained by a bifactor model consisting of one general and several symptom-specific factors (Witthöft, Fischer, Jasper, Rist, & Nater, 2016; Witthöft, Hiller, Loch, & Jasper, 2013). Moreover, it has been suggested that the general factor represents the affective component, while the symptom-specific factors are related to the sensory features. The primary aim of this study was to replicate those findings with another self-report symptom questionnaire and to explore the construct validity of proposed latent factors by measuring their association with trait NA. The secondary goal was to validate the self-report symptom questionnaire, the Checklist for Symptoms in Daily Life, which was used throughout the project to select the participants according to their habitual symptom reporting levels.

Study 2: Was it so bad? The role of retrospective memory in symptom reporting The primary goal of this experimental study was to investigate the role of memory processes in retrospective symptom reporting and to explore whether the differences in retrospective symptom ratings are related to HSR. To examine how different somatic experiences are remembered, female participants, high or low in HSR, took part in two laboratory studies, during which they experienced either dyspnea (via the rebreathing task) or pain (via the cold pressor task). Each study consisted of two trials in order to investigate the peak-end effect: One terminated at peak distress, while the other included an additional recovery phase to end at a less distressing level. Symptom ratings were collected on four moments: concurrently (along with respiratory measures), immediately after each trial, at the end of the experiment, and after two weeks. Based on the previous findings (Houtveen & Oei, 2007), we have expected a gradual increase in recalled symptom ratings over time, with this being more pronounced in the high HSR group. Following the PE rule, it was hypothesized that short trials will be retrospectively rated as more intense than the long trials. Moreover, this effect was expected to be present in low, but not high, HSR (Bogaerts et al., 2012; Houtveen & Oei, 2007).

Study 3: Retrospective memory for symptoms in patients with somatic symptom disorder This experimental study was built upon our previous study (Study 2) and aimed to investigate the retrospective memory for dyspnea in the clinical sample. Additionally, it served as a replication of the earlier finding showing that patients with medically unexplained 31

Chapter 1

dyspnea (MUD) did not rely on the PE heuristic while evaluating the dyspneic experience (Bogaerts et al., 2012). To investigate how patients remember the dyspnea episodes, we have induced two dyspnea trials by means of the rebreathing task in female MUD patients and healthy controls. The dyspnea ratings were collected continuously during symptom induction (concurrent with respiratory measures), after the experiment, and after two weeks in order to measure the correspondence between retrospective and momentary ratings. The peak-end effect was tested with forced-choice questions measuring relative preference for the trials. It was hypothesized that the discrepancy between the recalled and momentary ratings will be greater in the patient group. Moreover, the PE effect, indicated by a preference for the long above the short trial, was expected in the control, but not in the patient group (cf. Bogaerts et al., 2012).

Study 4: Ways of encoding somatic information and their effects on retrospective symptom reporting The aim of this study was to investigate whether focusing on sensory-perceptual or affective-motivational aspects of somatic experience could influence retrospective symptom ratings. For this purpose, we developed a processing focus (PF) manipulation, which included a detailed explanation of the differences between the two types of processing focus together with processing-specific instructions. In the sensory PF condition, participants were requested to focus on the bodily sensations in a neutral manner, while in the affective PF condition the emphasis was placed on the emotional responses to the stimuli. In order to examine the effects of such manipulation of symptom ratings, we have induced dyspnea (rebreathing task) and pain (cold pain via a thermode) during two experimental sessions in healthy female participants selected for the study based on their HSR levels. The manipulation of processing focus (sensory and affective) was within-subject and took place at the encoding phase, thus just before the symptom induction. Dyspnea and pain ratings were collected during the sessions and after 2 weeks. We predicted the affective PF to result in more emotion-laden/less sensory processing of bodily information, leading to higher symptom ratings than the sensory PF, with retrospective symptom ratings increase over time in the affective PF condition. Also, the high HSR group was expected to benefit relatively more from this PF manipulation than the low HSR group, as a sensory PF tends to be more beneficial for anxious or oversensitive individuals (e.g., J. Roelofs et al., 2004).

32


Study 5: Recalling somatic memories: The effect of processing focus during retrospective symptom reporting The memory formation is a complex process, which can be influences at different stages: encoding, consolidation, and retrieval. While attention was often given to encodingrelated biases, less is known about factors affecting retrospective symptom ratings at later stages of memory processing. In this study, we investigated whether guiding the processing focus to different aspects of somatic experience during memory retrieval has an impact on the way symptoms are remembered and, consequently, on the retrospective symptom reporting. To this end, female participants varying in HSR participated in three sessions. During the first session, they rated experienced symptoms while being exposed to a dyspneic (rebreathing task) and painful stimulation (cold pain via a thermode). After two weeks, they were requested to think back about experienced stimulations while focusing on either sensoryperceptual details or emotions that they experienced during the first session. The retrospective symptom ratings were collected after the manipulation and again after two weeks.

Study 6: The specificity of health-related autobiographical memory in patients with somatic symptom disorder The final study approached the problem of somatic memories from a different perspective and concentrated on one aspect of autobiographical memories, namely their specificity. The difficulty to retrieve specific memories is related to a range of emotional disorders (Williams et al., 2007) and was recently found also among patients with chronic pain (Liu et al., 2014). We were interested in examining whether the deficits in retrieving specific health-related autobiographical memories could be also present in patients suffering from MUS. To investigate this issue, we have first developed a modified version of Autobiographical Memory Test (AMT; Williams & Broadbent, 1986), in which cue words associated with health were used to elicit the memory retrieval. This modified AMT was completed by patients with MUD and healthy controls, together with other questionnaires measuring depressive symptoms and rumination. We hypothesized that the patients would retrieve fewer specific and more categoric memories in response to health-related cue words compared to controls.

33

Chapter 1

34

Sensory and affective components of symptom perception: A psychometric approach

2Chapter 2 Sensory and affective components of symptom perception: A psychometric approach

Abstract Psychological accounts of symptom perception distinguish sensory-perceptual and affectivemotivational components of symptoms, while psychometric analyses suggest a bifactor model describing the latent structure of symptom reporting. To corroborate the view that the general and symptom-specific factors of a bifactor model represent affective and sensory components, respectively, confirmatory factor analyses were performed on the Patient Health Questionnaire-15 and the Checklist for Symptoms in Daily Life completed by students (N = 1054). Additionally, the association of latent factors with negative affectivity (NA) was explored. For both questionnaires, the best fit was found for a complex bifactor model with one general and several symptom-specific factors. NA yielded strong associations with the general factor, but weaker with the somatic symptom-specific factors in both questionnaires. The observed latent structure supports a distinction between sensory-perceptual and affectivemotivational components, while the association between the NA and the general factor confirms the affective tone of the latter.

Based on: Walentynowicz, M., Witthöft, M., Raes, F., Van Diest, I., & Van den Bergh, O. (in preparation). Sensory and affective components of symptom perception: A psychometric approach.

35

Chapter 2

2.1. Introduction Self-reported somatic symptoms suggest the presence of an unusual, often pathophysiological condition of the organism. However, the process of symptom perception relies on a complex and constructive integration of multiple sources of information whereby interoceptive (bottom-up) information from within the body interacts with cognitive-affective (top-down) processes, such as attention, expectations, memory and beliefs (R. J. Brown, 2004; Rief & Broadbent, 2007; Van den Bergh et al., 2015). Traditional neurobiological approaches (Craig, 2003) have suggested that different structures in the brain process sensoryperceptual and affective-motivational components of a somatic experience. The first component conveys the distinct details about the intensity and spatiotemporal characteristics of internal sensations, while the latter includes the feelings towards the sensations promoting a behavioral drive to adjust (to) the stimulus. Bodily symptoms can therefore be seen as “homeostatic emotions” necessary to maintain the integrity of the body (Craig, 2003), in which the affective evaluation that is associated with sensory information depends on the state and the needs of the organism. For example, relieving thirst may be associated with different levels of pleasantness depending on the degree of water deprivation. The importance of the sensory-perceptual and the affective-motivational components of a somatic experience may also be reflected in psychometric research, which explored the latent structure of somatic symptom reports. The latent structure can be specified by various measurement models, which include, in order of ascending complexity: (a) a general factor model, in which all items load on one factor; (b) a correlated factor model comprising correlated symptom-specific factors (e.g., a pain factor, a cardiorespiratory factor); (c) a hierarchical model, in which the associations between the lower-order symptom-specific factors are accounted for by a higher-order general factor; and (d) a bifactor model, which consists of a general symptom factor and several symptom-specific factors. The bifactor model was recently shown to yield the best model fit for the latent structure of medically unexplained symptoms based on the Patient Health Questionnaire-15 (PHQ-15) in various samples, including college students (Witthöft et al., 2016), the general population, and primary care patients (Witthöft, Hiller, et al., 2013). Also the latent structures underlying symptoms of the irritable bowel syndrome (Jasper, Egloff, Roalfe, & Witthöft, 2015) and functional dyspepsia (Van Oudenhove et al., in press) yielded a bifactor model with a general somatization factor and three symptom-specific factors, outperforming other models.

36


In the bifactor model, each symptom shares both general and specific components of systematic variance. In other words, each symptom is explained by two latent factors: a general factor and orthogonal symptom-specific factors. The above studies not only showed that the bifactor model outperforms other, less complex models, but may also suggest a possible meaning of those latent (general and symptom-specific) factors. For example, the general factor may refer to the affective component of symptom experience, while the symptom-specific factors are related to the sensory-perceptual features. This is suggested by a strong relationship between the general factor and psychological constructs, such as health anxiety, somatosensory amplification, somatoform dissociation, and fatigue (Jasper et al., 2015; Witthöft et al., 2016), while symptom-specific factors do not show such a relationship. One of the important individual differences associated with elevated symptom reporting is negative affectivity (NA), a tendency to experience negative mood or affect and to perceive situations as threatening (Watson & Pennebaker, 1989). A robust association of trait NA and symptom reporting is typically observed and oscillates around r = .50 in both clinical (Kroenke, 2003b; Wessely et al., 1999) and non-clinical populations (Van Diest et al., 2005). However, not all symptoms are equally associated with NA: stronger associations are observed with vague, systemic complaints than with specific, localized symptoms, and for symptoms that are more severe and distressing (Van Diest et al., 2005). Because NA is a general psychological trait that is strongly related to symptom reporting, investigating its association with the latent factors in the bifactor model could provide more evidence for the construct validity of this model and help with the interpretation of the statistical models. Taking into account previous findings connecting the general factor to psychological constructs capturing specific concerns about symptoms, a strong relationship would also be expected with NA as a broad trait overarching several more specific expressions. However, the association of the latent structure underlying symptom reports with NA was so far not formally tested. The primary goals of this study were to replicate previous findings concerning the latent structure of symptom reporting, and to explore the construct validity of the proposed latent variables by examining their association with trait NA. To this end, the latent structure of symptom reporting was first investigated not only with the previously used PHQ-15 (Kroenke, Spitzer, & Williams, 2002), but additionally with the Checklist for Symptoms in Daily Life (CSD). The CSD is a self-report measure of habitual symptom reporting, which comprises more respiratory, neurological, and psychological symptoms than the PHQ-15. Four different models depicted in Figures 2-1 (CSD) and 2-2 (PHQ-15) were compared: 37

Chapter 2

a general factor model, a correlated factor model, a hierarchical model, and a bifactor model. The best fit was expected for the bifactor model consisting of both the general symptom reporting factor, which is related to every symptom, and the orthogonal symptom-specific factors (e.g., cardiorespiratory, psychological) representing the unique variance components of the extracted subscales of the questionnaires. Second, the association between the general and symptom-specific factors of both questionnaires and the trait NA was explored to investigate the construct validity of the proposed latent structure. Following the assumption that the general factor represents the affective component of symptom reporting, it was proposed that NA will show a stronger association not only with the psychological factor of the CSD, which taps unpleasant and anxious feelings, but also with the general factor of both the CSD and the PHQ-15, compared to the other symptom-specific factors.

Figure 2-1. Overview of the models tested with the CSD data. Residual terms of manifest indicator variables not shown.

38


Figure 2-2. Overview of the models tested with the PHQ-15 data. Residual terms of manifest indicator variables not shown. 2.2. Methods 2.2.1. Participants Data were collected during 3 consecutive group testing sessions (2013-2015) by means of a web-based survey among all first-year psychology students (N = 1054; 82.5% women; Mage = 18.69, SDage = 2.61) from the University of Leuven, Belgium, in return for course credit. All group testing sessions were approved by the Multidisciplinary Ethics Committee of the Faculty of Psychology and Educational Sciences of the University of Leuven.

39

Chapter 2

2.2.2. Measures 2.2.2.1. Checklist for Symptom in Daily Life (CSD) The CSD is a self-report measure of habitual symptom reporting based on the Psychosomatic Symptom Checklist (Wientjes & Grossman, 1994). The frequency of 39 symptoms in the past year is assessed on a 5-point scale (never, seldom, sometimes, often, very often). This scale was previously used to screen for habitual symptom reporting (Bogaerts, Janssens, et al., 2010; Bogaerts et al., 2015; Constantinou et al., 2013). In the present sample Cronbach’s alpha was .92. Based on the results of the principal component analysis (PCA) from the validation study (see Appendix A), 5 items with unclear loadings (hands tremble, stomach feels blown up, toe or leg cramps, cold hands or feet, and burning feeling in the eyes) were removed from the analysis. The CSD and validation data are included in Appendix A. 2.2.2.2. Patient Health Questionnaire-15 (PHQ-15) The PHQ-15 (Kroenke et al., 2002) is a widely used self-report questionnaire measuring the distress related to 15 somatic symptoms over the previous 4 weeks. The 3-point scale ranges from not bothered at all to bothered a lot. The Dutch version was used (van Ravesteijn et al., 2009). Cronbach’s alpha in the present sample was .72. 2.2.2.3. Positive and Negative Affect Schedule (PANAS) The Dutch version (Engelen, De Peuter, Victoir, Van Diest, & Van den Bergh, 2006) of the PANAS (Watson, Clark, & Tellegen, 1988) was used. It assesses to what extent positive and negative adjectives apply to participants’ feelings in general on a 5-point Likert scale ranging from 1 (not at all) to 5 (very much). For the present study, only scores on NA were investigated. Cronbach’s alpha in the present sample was .88. 2.2.3. Data analysis Confirmatory factor analysis (CFA) was performed on the CSD and the PHQ-15 data with Mplus 7 software (Muthén & Muthén, 2012). The models were estimated with the robust mean- and variance-adjusted weighted least squares (WLSMV) procedure, which is based on the matrix of tetrachoric correlations. As such correlations can be biased by low cell frequencies (M. B. Brown & Benedetti, 1977), rarely used response categories were collapsed to reach the frequencies of minimum 5% in each cell. Due to the sensitivity of chi-square test to the complexity of the model and the sample size, the model fit was evaluated with other descriptive fit measures, such as the root mean square error approximation (RMSEA) and the 40


comparative fit index (CFI). Following the recommendations of Hu and Bentler (1999), RMSEA values close to .06 and CFI values close to .95 are treated as the indices of a good model fit. However, as CFI cutoff of .95 is sometimes perceived as too restrictive, the models with CFI values larger than .90 may be accepted, especially in research concerning psychological traits (Beauducel & Wittmann, 2005). Four different models, that is, a general factor, a correlated group factor, a hierarchical, and a bifactor model were tested and compared with the chi-square difference tests. All coefficients are reported in the standardized form. 2.3. Results 2.3.1. The latent structure of somatic symptoms in the CSD First, a general factor model was tested. This model consisted of 34 symptoms loading on one latent factor (G-SOM). The model yielded poor fit indices (Table 2-1). Second, a correlated factor model was examined. The model was specified with five correlated latent factors previously determined by the PCA (see Appendix A), that is, a cardiorespiratory factor, a psychological factor, a neurological factor, a cold/flu factor, and a cerebral factor. The correlations between all factors were significant and moderate to strong, with rs ranging from .30 to .76, all ps < .001. The model fit indices were approaching the cut-off values. A similar fit was obtained by a hierarchical model, specified by five symptom-specific lowerorder factors and one general higher-order factor. Finally, a bifactor model (see Figure 2-3, left panel) was tested. Model specifications allowed every item to load on a general factor as well as on one of the five orthogonal symptom-specific factors. This model showed an adequate fit on RMSEA, while the values on CFI were approaching adequacy cut-off scores.

Table 2-1. Goodness of fit for the four different models tested with the CSD (N = 1054). Model I

Model II

Model III

Model IV

General factor

Correlated

Hierarchical

Bifactor model

model

factor model

model

χ 2 (df)a

7120.08 (527)

2719.92 (517)

2875.11 (522)

2352.71 (493)

CFI

.719

.906

.900

.921

RMSEA

.109

.064

.065

.060

.061-.066

.063-.068

.057-.062

90% CI RMSEA .107-.111 a

All χ 2 values were highly significant, p < .001.

41

Chapter 2

Model comparisons performed by means of the chi-square difference tests revealed that the correlated factor model resulted in a better fit than both the general factor model, χ2(10) = 1508.33, p < .001, and the hierarchical model χ2(5) = 96.76, p < .001. Based on the RMSEA and the CFI values, the model fit of the bifactor model (RMSEA = .060, CFI = .921) was better than the one of the correlated factor model (RMSEA = .064, CFI = .906).

Figure 2-3. Bifactor model of somatic symptoms in the CSD with standardized factor loadings (left) and with negative affectivity (NA; right). Single headed arrows represent factor loadings; factor loading coefficients printed in bold are significant at p < .05. Double-headed arrows represent latent correlation coefficients; all correlation coefficients printed in bold are significant at p < .001; residual terms of manifest variables not shown. Unfortunately, due to the fact that the correlated factor model is not nested within the bifactor model, the model comparison with the chi-square difference tests was not possible. However, the supplementary analysis with Maximum Likelihood (ML) estimation and model

42


comparison based on the Bayesian information criterion (BIC) showed that the bifactor model (BIC = 82649) outperformed the correlated factor model (BIC = 82834).4 2.3.2. The latent structure of somatic symptoms in the PHQ-15 The latent structure of somatic symptoms in the PHQ-15 was tested with four CFA models. A general factor model included 13 symptoms 5 and resulted in the poor fit (Table 2-2). A correlated factor model was specified with four correlated latent factors (pain-, gastrointestinal-, cardiopulmonary-, and fatigue-related symptoms), which showed moderate to strong correlations with rs ranging from .45 to .67, all ps < .001. The model fit indices were good and comparable to the fit obtained by a hierarchical model with four lower-order and one higher-order factor. Finally, a bifactor model (see Figure 2-4, left panel) including four orthogonal symptoms-specific factors and one general factor showed not only an excellent fit (CFI = .991; RMSEA = .020; 90% CI [.007, .029]), but also fitted the data significantly better than the hierarchical model, χ2(5) = 51.57, p < .001, as revealed by the chi-square difference tests. Table 2-2. Goodness of fit for the four different models tested with the PHQ-15 (N = 1053)a. Model I

Model II

Model III

Model IV

General factor

Correlated

Hierarchical

Bifactor model

model

factor model

model

χ 2 (df)b

504.50 (65)

147.87 (59)

144.17 (61)

78.70 (56)

CFI

.825

.965

.967

.991

RMSEA

.080

.038

.036

.020

.030-.045

.028-.044

.007-.029

90% CI RMSEA .074-.087 a b

One participant did not complete the PHQ-15. All χ 2 values were significant, p < .05.

4

Even though the non-nested models cannot be compared with chi-square difference tests, the BIC can be used to compare such models. The BIC statistic can be obtained in the models using ML estimation, but it is not available in the WLSMV estimation, which was used in the current study. For this reason, the supplementary analysis of both models with ML estimation was performed and the BIC values were subsequently compared. The smaller BIC values indicate a better-fitting model. The BIC difference of >10 represents strong evidence for meaningful differences between the models, while >100 means a decisive evidence (Kass & Raftery, 1995). 5 As in the previous CFA analyses (Witthöft et al., 2015; Witthöft, Hiller, et al., 2013), two items were excluded from the analysis because of a very low base rate (fainting spells) and a genderspecific content (menstrual problems).

43

Chapter 2

2.3.3. Association between the bifactor models (CSD and PHQ-15) and NA The strength of the association between NA and different factors in the bifactor model was tested separately for the CSD and the PHQ-15. In the CSD bifactor model, NA was strongly associated with both the CSD general symptom factor, r = .59, p < .001, and one symptom-specific psychological factor, r = .56, p < .001 (see Figure 2-3, right panel). A weak but significant correlation with a CSD symptom-specific neurological factor (r = -.12, p < .001) was also observed. No significant associations between NA and other symptomspecific factors were found (all rs < .05, ps > .05). The model fit was acceptable: χ²(521) = 2410.10, p < .001; RMSEA = .059 (90% CI: .056-.061); CFI: .920. In the PHQ-15 bifactor model, NA was only associated with the PHQ-15 general symptom factor, r = .53, p < .001 (see Figure 2-4, right panel). No significant correlation was observed between NA and any of the symptom-specific factors (all rs < .14, ps > .05). The model fit was good, χ²(64) = 85.74, p = .036; RMSEA = .018 (90% CI: .005-.027); CFI: .992.

Figure 2-4. Bifactor model of somatic symptoms in the PHQ-15 with standardized factor loadings (left) and with negative affectivity (NA; right). Single headed arrows represent factor loadings; factor loading coefficients printed in bold are significant at p < .05. Double-headed arrows represent latent correlation coefficients; all correlation coefficients printed in bold are significant at p < .001; residual terms of manifest variables not shown.

44


Because the associations between the NA and the different types of symptoms could differ between genders (Van Diest et al., 2005), we explored possible effects of gender on the relationship between NA and the different factors in the bifactor models. For the data from both investigated questionnaires, the model fit indices were acceptable to excellent with CFIs between .925 and .991. The pattern of associations for males and females resembled the one observed in the full sample, that is, strong correlations between the NA and (a) the general symptom factors, (b) the CSD psychological factor with rs for both ranging from .52 to .59, all ps < .001. The associations between NA and other symptom specific factors were weak (all rs < .20) and, except for the CSD neurological factor, nonsignificant. Detailed information on fit indices and associations is provided in Appendix B. 2.4. Discussion This psychometric study aimed at empirically testing the view that symptom perception involves a sensory-perceptual and affective-motivational component. The specific aims were twofold: First, to replicate the previously reported finding regarding the latent structure of somatic symptoms in a different population and with two symptom questionnaires, differing to some extent in scope and content. Second, to further investigate the construct validity of the observed factors by examining their associations with trait NA. Compared to the alternative models, the bifactor model with one general symptom factor and several symptoms-specific factors revealed the best fit for symptom reporting measured by the two self-report instruments. Whereas previous studies investigating this issue analyzed symptom distress as assessed by the PHQ-15 (Kroenke et al., 2002), we added a questionnaire (CSD) consisting of a largely different sample of symptoms including a broad range of cardiorespiratory, psychological, neurological, common cold, and cerebral symptoms. This questionnaire therefore covers more respiratory- and anxiety-related items than the PHQ-15, and also it uses a different time frame (the last year vs. the last 4 weeks). Moreover, it collects frequency ratings of symptoms in the daily life, while the PHQ-15 assesses the distress caused by the symptoms. In the current study, almost all symptoms included in both bifactor models showed significant loadings on both the general factor and one of the symptom-specific factors, indicating that nearly each symptom was simultaneously determined by two sources of systematic variance. With the bifactor model outperforming alternative structural models in the analyses based on both the PHQ and the CSD, these results are consistent with the earlier findings and add to the growing psychometric evidence in favor of the bifactor model

45

Chapter 2

approach adopted in structural modeling of symptom reporting (Jasper et al., 2015; Thomas & Locke, 2010; Witthöft et al., 2016; Witthöft, Hiller, et al., 2013). In addition, exploratory analyses performed separately for male and female participants did not indicate any major gender-related differences in the latent structure of symptom reporting and the associations between the NA and symptom factors. The findings of this study also provide additional support to the interpretation of the latent factors of this model. Strong associations have been previously observed between on the one hand the general factor (but not symptom-specific factors) and psychological traits such as health anxiety, somatosensory amplification, depressive symptoms (Witthöft et al., 2016), and somatoform dissociation on the other hand (Jasper et al., 2015). Moreover, the general factor was also related to emotion regulation strategies, including a negative association with cognitive reappraisal and a positive correlation with symptom focused rumination (Witthöft, Loch, & Jasper, 2013). Those findings suggest that the general symptom factor is closely related to the affective processes and represents the affectivemotivational component of symptom perception. The observed association of trait NA with the general factor and the psychological factor, but not other symptom-specific factors, corroborates this conclusion. As NA is a general vulnerability factor to develop symptom reporting unrelated to physiological dysfunction (Bogaerts et al., 2015), understanding the mechanisms underlying the close association between this personality trait and increased symptom reporting is crucial. Two perspectives on this link were advanced. The first considers the role of, for example, (para)sympathetic activity (Jarrett et al., 2003; but see also Houtveen & van Doornen, 2007). The second perspective suggests that NA triggers schematic processing of symptoms leading to a biased symptom processing (Bogaerts, Janssens, et al., 2010; Constantinou et al., 2013; Van den Bergh et al., 1998). A significant and high correlation between the trait NA and the general symptom factor (but not with the symptomspecific factors) seems to favor the latter perspective, in which NA could act as an amplifier of existing (negative) symptom representations in memory (schemata). Excessive activation of the symptom schemata is perceived by some theories as a core process in the distorted perception of somatic symptoms and the development of habitual symptom reporting (R. J. Brown, 2004; Van den Bergh et al., 1998). The finding that somatic symptom reporting is related not only to the sensoryperceptual, physical aspects of somatic experiences, but also to the affective-motivational processes involved in the evaluation of bodily sensations appears to be compatible with recent changes included in the DSM-5 (American Psychiatric Association, 2013) regarding the 46


criteria for somatic symptom disorder. The new classification accentuates the distress caused by persistent somatic symptoms, and includes excessive feelings, thoughts, and behaviors in response to those symptoms. It is important to realize that this recent conceptualization does not require, as the DSM-IV somatoform disorder, a lack of organic explanation for the presented symptoms, but focuses on the preponderant affective-motivational responses to those symptoms. In this respect, the bifactor model psychometrically endorses this approach by showing that symptom perception is influenced by both a general symptom factor representing the affective-motivational aspects of symptom perception, but also by several symptom-specific factors. The latter may reflect sensory-perceptual component and are likely disorder-specific. For example, the IBS diagnosis will show stronger association with the gastrointestinal than with respiratory symptoms (Witthöft, Hiller, et al., 2013). A limitation of the current work is the sample, which in the current study consisted of young, healthy, and predominantly female students. Moreover, even though a bifactor model was previously supported in studies involving primary care patients (Witthöft, Hiller, et al., 2013) and patients with epilepsy (Thomas & Locke, 2010), further studies are needed to replicate and extend the construct validity of the latent factors in the bifactor model found in the present homogenous samples of university students (Jasper et al., 2015; Witthöft et al., 2016) to clinical samples. In summary, the current study replicates and extends findings regarding the latent structure of somatic symptoms, and psychometrically supports the division between sensoryperceptual and affective-motivational components of symptom perception. Moreover, the association between the NA and the general factor provides further evidence that the latter factor mainly represents the affective component or the distress factor of symptom perception. 2.5. Acknowledgments The study was funded by Grant OT/10/027 from the University Research Council of the University of Leuven.

47

Chapter 2

48

Was it so bad? The role of retrospective memory in symptom reporting

3Chapter 3 Was it so bad? The role of retrospective memory in symptom reporting

Abstract Retrospective symptom reports are an important source of information in both laboratory and clinical settings. The present study investigated memory for experimentally induced pain and dyspnea in high and low habitual symptom reporters (HSR). Healthy women (N = 48; 24 high/24 low HSR) participated in two laboratory studies. One study included two pain episodes (cold pressor task), the other study included two dyspnea episodes (rebreathing task). Pain and dyspnea ratings were collected (1) continuously during symptom inductions, (2) after each trial, (3) immediately after the experiment, and (4) at 2-week follow up. Symptom ratings, negative affect (NA) and anxiety measures were also completed following each trial. While the retrospective pain ratings were higher in the high compared with the low HSR group (p = .01), both groups rated recalled dyspnea higher relative to concurrent dyspnea (p < .001). A further increase in bias over time was only found for dyspnea in high HSR (p = .02). Moreover, dyspnea induction was associated with higher state NA (p = .03) and anxiety (p = .007) than pain induction. Our findings show that even though memory for pain and dyspnea is overall distorted, the extent of bias in symptom recall clearly differs between symptoms and groups. The observed increase of dyspnea reporting over time may have important implications for diagnostic assessment based on symptom reporting.

Based on: Walentynowicz, M., Bogaerts, K., Van Diest, I., Raes, F., & Van den Bergh, O. (2015). Was it so bad? The role of retrospective memory in symptom reporting. Health Psychology, 34, 1166 -1174.

49

Chapter 3

3.1. Introduction Retrospective descriptions of somatic experiences are important sources of information for health care professionals and can influence clinical diagnosis and treatment choice. Interestingly, a considerable number of patients in both primary (Barsky et al., 2005; Khan, Khan, Harezlak, Tu, & Kroenke, 2003) and secondary health care (Carson et al., 2000; Nimnuan, Rabe-Hesketh, Wessely, & Hotopf, 2001) tend to report symptoms in the absence of underlying physical dysfunction (often called medically unexplained symptoms; MUS). In the recently proposed DSM-5 (American Psychiatric Association, 2013) most of these patients would meet criteria for somatic symptom disorder (SSD), which emphasizes the presence of persistent distressing somatic symptoms, as well as excessive thoughts, feelings and behaviors linked to those symptoms. Various studies have explored the perceptualcognitive processes underlying symptom overreporting in this group (Rief & Broadbent, 2007; Van den Bergh et al., 2015), but few studies have focused on the role of memory processes herein. Although research on memory for symptoms often results in contradictory findings, one consistent conclusion is that memory for symptoms is relatively inaccurate and mostly results in retrospective overreporting of experienced symptoms (Broderick et al., 2008; Giske et al., 2010; Linton & Melin, 1982). Several sources of bias have been identified: (1) variability of real-time symptom levels (Lefebvre & Keefe, 2002; Sohl & Friedberg, 2008; Stone et al., 2005), (2) symptom intensity (Feine et al., 1998; Giske et al., 2010; Hunter et al., 1979; Sohl & Friedberg, 2008), (3) emotional state during symptom experience (Everts et al., 1999; Gedney & Logan, 2004), (4) symptom intensity during recall (Eich et al., 1985; Lefebvre & Keefe, 2002; Meek et al., 2001; W. B. Smith & Safer, 1993), (5) time since actual symptom episode (Broderick et al., 2008; Houtveen & Oei, 2007), and (6) cognitive heuristics, such as the peak-end (PE) effect (Kahneman et al., 1993). The PE effect assumes that the retrospective evaluation of an experience is predominantly determined by two distinctive moments, the one with the highest intensity (peak) and the final (end) part of the episode, with relative duration neglect, meaning that the actual duration of the experience has a limited influence on the global retrospective evaluation. The influence of this heuristic on symptom memory was confirmed not only in the laboratory (Bogaerts et al., 2012; Kahneman et al., 1993), but also in naturalistic settings, such as during medical examinations (Redelmeier & Kahneman, 1996; Redelmeier et al., 2003) and childbirth (Chajut et al., 2014). Finally, psychological factors may inflate both concurrent (catastrophizing, Lefebvre &

50


Keefe, 2002; Sohl & Friedberg, 2008; anxiety, Suls & Howren, 2012) and retrospective (depression, Suls & Howren, 2012) symptom ratings. Also negative affectivity (NA) has been found to be strongly related to memory distortions for symptoms (Levine & Safer, 2002; Safer, Levine, & Drapalski, 2002). Similarly, trait NA is associated with a tendency to attend more to somatic information (Stegen, Van Diest, Van de Woestijne, & Van den Bergh, 2001) and to interpret it as threatening (Stegen, Van Diest, Van de Woestijne, & Van den Bergh, 2000), as well as to overreport symptoms during recall (Larsen, 1992). Even though memory processes play an important role in biased symptom reporting, only a limited number of studies have explored memory for bodily symptoms among patients with MUS or non-consulting high habitual symptom reporters (HSR). Bogaerts et al. (2012) investigated the PE effect in the memory for dyspnea in patients with medically unexplained dyspnea (MUD) and healthy participants. Dyspneic experience was experimentally induced with two rebreathing trials: One ended at the peak of dyspnea, while in the other a recovery phase was added to assure a milder end. The expected PE effect was observed among healthy participants, but not in the MUD group. Because the patients displayed a slower recovery in self-reported dyspnea, which could not be accounted for by differences in respiratory physiology, it is suggested that perceptual-cognitive processing of aversive sensations among patients with MUD differs from healthy people. Investigating a non-consulting high HSR group, Houtveen and Oei (2007) conducted a diary study and found that, compared to averaged concurrent symptom reports, both high and low HSR reported experiencing more symptoms during recall. However, only high HSR showed a gradual increase in estimation of experienced symptoms with longer time frames. Moreover, biased recall in high HSR was not related to the hypothesized sources of bias, that is, the PE effect and symptom variability. Taken together, delayed recovery in symptom reports but not in physiological dysfunction, as well as bias in retrospective symptom reporting, suggest distorted and less detailed perceptual-cognitive processing of symptom experiences in persons with MUS. In view of this limited set of findings, the present study aimed to advance our understanding of the role of perceptual-cognitive biases affecting retrospective symptom reports. The primary goals of this study were to investigate whether the retrospective symptom reports are subject to recall biases leading to increased symptom reporting and whether such biases are larger for high HSR. To this end, retrospective memory for two experimentally induced and well-controlled aversive bodily sensations, pain and dyspnea, was examined. We selected participants high and low on HSR and administered the two aversive sensations within subject to examine the generality of the findings across symptom types. In 51

Chapter 3

one study (StudyPain), a painful experience was induced by means of the cold pressor task (CPT), while in the other (StudyDyspnea), dyspnea was induced via a rebreathing paradigm (Read, 1967). Each study consisted of two trials in order to investigate the peak-end effect: One terminated at peak distress, while the other included an additional recovery phase to end at a less distressing level. Participants rated their concurrent symptom levels while being exposed to aversive stimuli during the trials, which were followed by three retrospective ratings of induced symptoms and affective responses. Based upon the arguments described above, the following hypotheses were tested in each study separately: (1) Retrospective symptom ratings were expected to be higher than averaged concurrent ratings, with this effect being more pronounced in high HSR; (2) Recalled symptom reports were expected to increase over time in high HSR, but not in low HSR; (3) According to the PE rule, short trials were expected to be retrospectively rated as more intense than the long trials. However, this effect was hypothesized to be present in low, but not high, HSR (Bogaerts et al., 2012; Houtveen & Oei, 2007). Possible differences between the two symptom types were investigated in an exploratory manner, thus no specific hypotheses were formulated regarding these differences. 3.2. Methods 3.2.1. Participants Forty-eight healthy students (all women), aged 18 – 27 years, participated in both experiments in return for two course credits or 15 euros. They were selected after screening for habitual symptom reporting via the Checklist for Symptoms in Daily Life (CSD; Walentynowicz, Witthöft, Raes, Van Diest, & Van den Bergh, 2016). Predefined cut-off scores were used to select high (≥ 100; n = 24) and low (≤ 75; n = 24) habitual symptom reporters. Cut-off scores were based on upper and lower quartiles of the scores on this questionnaire found in large samples from the same population (Bogaerts et al., 2008). Prior to the experiment, participants completed the CSD a second time; Only participants who still met the cut-offs were included. Exclusion criteria were any self-reported chronic illness (e.g., pulmonary, cardiovascular, gastrointestinal, neuromuscular diseases), acute illnesses, fever or headache, major psychiatric condition, diabetes, recent arm fracture or wrist sprain prior to participating, earlier frostbite, and pregnancy. The experimental protocol was approved by the Multidisciplinary Ethics Committee of the Faculty of Psychology and Educational Sciences of the University of Leuven.

52


3.2.2. Measures 3.2.2.1. Habitual symptom reporting Habitual symptom reporting was assessed using the Checklist for Symptoms in Daily Life (CSD; Walentynowicz, Witthöft, et al., 2016). Participants rated how often they experienced 39 listed symptoms in the past year on a 5-point Likert scale (never, seldom, sometimes, often, very often). The total score (range: 39 – 195) was used to select high/low HSR; reliability (Cronbach’s α) exceeded .95 in our sample. 3.2.2.2. Negative affectivity Trait and state Negative Affectivity (NA) were assessed with the Dutch version of the Positive and Negative Affect Schedule. The PANAS consists of 20 positive and negative adjectives for which participants had to indicate (on a 5-point Likert scale ranging from not at all to very much) to which extent they felt that way in general (trait) or now (state). Good reliability and validity have been reported (Engelen et al., 2006; Watson et al., 1988). 3.2.2.3. State symptom checklist At baseline and after every symptom induction trial in both studies, a state symptom checklist was administered. Participants had to rate to which extent they experienced each of 12 symptoms now (baseline) or during the past trial on a 5-point rating scale ranging from 0 (not at all) to 4 (very much). This symptom list included: chest tightness, pounding of the heart, stomach or abdominal cramps, headache, fatigue, not able to breathe deeply, rapid heartbeat, nausea, dizziness, muscular pain, dyspnea, pain. State symptom checklists showed acceptable and good internal consistency, with Cronbach’s alphas ranging from .70 to .86. 3.2.2.4. State anxiety and threat A numerical rating scale (NRS) was used to evaluate the level of anxiety (1 = not anxious at all, 9 = very anxious) at the baseline and after every symptom induction trial in both studies. Additionally, after every symptom induction trial, a NRS concerning the threat value of each trial (1 = not threatening at all, 9 = very threatening) was administered. 3.2.2.5. Concurrent symptom ratings During each symptom induction trial, concurrent symptom ratings were collected on a 0-100 computerized scale. The scale was presented as a vertical bar in the middle of the screen. Different levels of the experienced pain/dyspnea, based on a modified Borg scale (Borg, 1982), were verbally described on its right side: none (0), very slight (10), slight (20),

53

Chapter 3

moderate (30), fairly severe (40), severe (50), very severe (60), very severe (70), very severe (80), very, very severe (90), intolerable (100). In StudyPain, perceived pain was rated continuously with a scroll wheel (sampling every second), while in StudyDyspnea perceived dyspnea was rated every 10 s (after auditory cue) with a mouse click. 3.2.2.6. Retrospective symptom ratings In both studies, the retrospective ratings of symptoms experienced during each trial were collected at three moments: immediately after each trial (immediate rating), at the end of the experimental session (delayed rating) and in a two weeks follow-up (follow-up rating). Participants indicated the average symptom level (StudyPain: pain/ StudyDyspnea: dyspnea) experienced during the trial (How much pain/ dyspnea have you experienced on average during this trial?) on a visual analog scale (10 cm) ranging from 0 (no pain/dyspnea) to 100 (maximum pain/dyspnea). The follow-up ratings for both studies were collected on a single occasion two weeks after the last study. 3.2.3. Apparatuses and physiological recordings 3.2.3.1. StudyPain: Pain induction – CPT During the two trials of the CPT, participants immersed their hand in a Plexiglas box (Julabo®) filled with 18 L of water. The water temperature was controlled by an electric immersion cooler, type FT200, and a bath circulator, type ED-19A. This ensured that the temperature could be either maintained at a constant level (12°C) or increased by 2°C in 60 s. The changes in water circulation during temperature manipulation were unnoticeable for the participant. In contrast with Kahneman et al. (1993), who used 14°C to induce pain and 15°C to induce discomfort reduction, temperatures in the current experiment were set at 12°C and 14°C (± 0.3°C), after a pilot study showing a detectable change in discomfort with these temperatures, which was not observed with 14°C and 15°C. The Plexiglas box was placed upon a trolley adjustable in height to provide comfortable access. Before each CPT, participants were asked to hold both hands in the second box (type FT200 Julabo®), in which water was kept at room temperature (20.5°C ± 0.3°C). A 2-min baseline was used to ensure that the skin temperature of the participants was similar before each trial. 3.2.3.2. StudyDyspnea: Dyspnea induction – the rebreathing paradigm Two trials of the standardized rebreathing paradigm (Read, 1967, see also Bogaerts et al., 2012) were used to induce the sensations of dyspnea. During the trials, participants wore a nose clip and breathed through a mouthpiece, connected to the rebreathing bag via a wide 54


vinyl tube and a Y-valve ending on a pneumotachograph (Fleisch no. 2, Lausanne, Switzerland) measuring airflow. The valve allowed to switch between room air and the rebreathing bag, which was initially filled with 5-L gas mixture of 5% CO2 and 95% O2. Breathing in this closed hyperoxic system led to a progressive rise of PCO2, of minute ventilation and of dyspnea, defined as uncomfortable feeling of not having enough air, an urge to breathe, or a feeling of having more difficulty in breathing. Fractional end-tidal concentration of CO2 (FetCO2) was determined using an infrared CO2 monitor (POET RC, Criticare Systems Inc., Waukesha, WI). The exhaled air was sampled close to the mouthpiece. The data from the pneumotachograph and the CO2 monitor were sampled at 20Hz and stored on a computer. All data were stored and analyzed offline to determine the following parameters: minute ventilation (MV) in L/min and FetCO2 in %. The manipulation of the valve was undetectable by the participants to make sure that they depended exclusively on the experienced bodily changes to give their concurrent ratings. 3.2.4. Procedure Selected participants were invited to participate in two experiments examining the relationship between bodily sensations and well-being. One day before the laboratory sessions, participants completed the trait questionnaires (CSD and PANAS) online. Upon arrival in the laboratory, participants signed the informed consent, were informed about the procedure and completed the questionnaires (state symptoms checklist, PANAS, and anxiety). Participants completed both studies on two consecutive days and the order of the studies and the trials within study was counterbalanced across participants. In StudyPain, two CPT trials were administered, one on each hand, with a 7-min intertrial interval. Each trial began with a baseline period, during which participants immersed both hands in room-temperature water. For the short trial, baseline was followed by a cold phase (60 s in 12°C water) after which participants could withdraw their hand. For the long trial, baseline was followed by the same cold phase (60 s in 12°C) with an additional recovery phase (60 s) during which the temperature increased to 14°C (unknown to the participants). The order of trials (short/long, to the dominant/nondominant hand) was counterbalanced across participants. In StudyDyspnea, participants went through two rebreathing trials, the order of which was counterbalanced across participants. A short trial consisted of a baseline (60 s of room-air breathing) and a rebreathing phase (150 s). After 150-s rebreathing, the trial was stopped and participants could breathe freely outside of the rebreathing system. In a long trial, the baseline

55

Chapter 3

(60 s) and rebreathing phases (150 s) were followed by an additional recovery phase (150 s), initiated by unobtrusively switching the valve to room air. Full recovery between the trials was ensured by a 15-min intertrial interval. Respiration was measured throughout each trial. Average pain/dyspnea experienced during the trials was rated after each trial (immediate rating), followed by the ratings of state affect, symptoms and anxiety. After two trials of each study, the average pain/dyspnea ratings were repeated (delayed rating). Two weeks after the second session participants completed an online questionnaire regarding retrospective ratings for both trials in each of the studies (follow-up rating). 3.2.5. Data analysis 3.2.5.1. Manipulation check: Concurrent symptom ratings Concurrent symptom ratings were analyzed to verify the effect of symptom induction and to examine whether symptom reports differed between high and low HSR. Concurrent ratings were divided into equal time segments of 10 s (StudyPain) or 30 s (StudyDyspnea) in order to acquire a detailed picture of the somatic experience. Separate repeated-measures analyses of variance (ANOVAs) were conducted on self-reported symptom ratings (pain/dyspnea) during each trial as dependent variables, with Group and Order of Trials as between-subject factors and Time Segment as a within-subject variable. Moreover, the effect of dyspnea induction on respiratory behavior was investigated by separate repeated-measures ANOVAs on MV and FetCO2 (per 30 s) during each trial as dependent variables, with Group and Order of Trials as between-subject factors and Time Segment as a within-subject variable. 3.2.5.2. Testing hypotheses: Retrospective symptom ratings To assess symptom memory in both studies, separate repeated-measures ANOVAs were performed on symptom ratings (pain/dyspnea) as dependent variables, with Group (high/low HSR) and Order of Trials as between-subjects factors, Trial (short trial/long trial) and Moment of Symptom Assessment (averaged concurrent/immediate/delayed/follow-up) as within-subject factors. The averaged concurrent symptom scores were averaged across the actual symptom reports given during pain and dyspnea inductions per trial. Planned contrasts were used to examine specific time and group effects (C1: averaged concurrent vs. all retrospective ratings to test Hypothesis 1; C2: retrospective ratings during experimental session (immediate, delayed) vs. follow-up ratings to test Hypothesis 2), as well as trial

56


effects (C3: averaged concurrent vs. immediate ratings to test Hypothesis 3). Greenhouse– Geisser corrections were applied when the sphericity assumption was violated. 3.2.5.3. Affective responses In order to investigate the differences in affective responses to different bodily stimuli, data regarding affective states during both studies were analyzed together. Repeated-measures ANOVAs were carried out on state symptom, NA and anxiety ratings as dependent variables, with Induction (StudyPain/StudyDyspnea) and Moment of Measurement (baseline/short trial/long trial) as within-subject factors and Group (high/low HSR) and Order of Studies as between-subject factors. Another repeated-measures ANOVA was conducted on threat value as dependent variable, with Induction (StudyPain/StudyDyspnea) and Moment of Measurement (short/long trial) as within-subject factors and Group (high/low HSR) and Order of Studies as between-subject factors. All analyses were conducted with SPSS 22.0. 3.3. Results 3.3.1. Sample characteristics Low HSR reported less habitual symptoms than high HSR (low: M = 57.13, SE = 2.06; high: M = 114.75, SE = 2.61; t(46) = -17.33, p < .001). High HSR also reported higher trait NA levels than low HSR, t(38.04) = -7.34, p < .001. 3.3.2. StudyPain 3.3.2.1. Manipulation check: Concurrent pain ratings No significant group-related differences were observed for either pain ratings or their change over time during the short trial, while during the long trial high HSR tended to report more pain, F(1, 44) = 2.98, p = .09, !!! = .06 (see Figure 3-1). 3.3.2.2. Testing hypotheses: Retrospective pain ratings Hypothesis 1 (retrospective symptom reporting): Main effects showed that high HSR reported overall more pain than low HSR, F(1, 44) = 4.80, p = .03, !!! = .10 (Figure 3-2, left panel) and that pain was higher when rated retrospectively than concurrently, C1: F(1, 44) = 58.99, p < .001, !!! = .57. Moreover, the latter effect was stronger in high HSR than in low HSR participants, C1 for Group × Moment: F(1, 44) = 6.63, p = .01, !!! = .13; Group × Moment: F(2.43, 106.96) = 3.53, p = .03, !!! = .07. Hypothesis 2 (increase in retrospective symptom reporting over time): Retrospective ratings did not further increase 57

Chapter 3

over time, C2: F(1, 44) = 1.35, p = .25, !!! = .03. Hypothesis 3 (PE effect): Even though the concurrently rated pain was lower in the short trial than in the long trial, pain rating in the immediate rating was higher for the short, compared with the long, trial, C3: F(1, 44) = 17.38, p < .001, !!! = .28 (Figure 3-3, left panel); Trial × Moment: F(1.91, 83.99) = 4.69, p = .01, !!! = .10. This confirmed the peak-end effect, but no group differences appeared for this interaction, C3: F(1, 44) = 1.22, p = .28, !!! = .03; Trial × Group × Moment: F(1.91, 83.99) = .75, p = .47, !!! = .02.

Figure 3-1. Mean values and standard errors of concurrent pain ratings (0-100) for high and low habitual symptom reporters (HSR) during the short (left) and the long trial (right). Whiskers denote standard errors.

Figure 3-2. Mean averaged concurrent and retrospective pain ratings (0-100, left) and dyspnea ratings (0-100, right) for high and low habitual symptom reporters (HSR). Whiskers denote standard errors.

58


Figure 3-3. Mean averaged concurrent and immediate pain ratings (left) and dyspnea ratings (right) for short and long trial. Whiskers denote standard errors.

3.3.3. StudyDyspnea 3.3.3.1. Manipulation check: Concurrent dyspnea ratings Group-related differences were observed for concurrent symptom ratings in both trials, with high HSR reporting more dyspnea during the short, F(1, 41) = 9.53, p < .01, !!! = .19, and the long trial, F(1, 41) = 3.97, p = .05, !!! = .09 (see Figure 3-4). Moreover, this difference became stronger over time during the short trial (Group × Time Segment: F(1.50, 61.40) = 5.81, p < .01, !!! = .12). No group differences were observed for either FetCO2 or MV during the rebreathing trials. 3.3.3.2. Testing hypotheses: Retrospective dyspnea ratings. Hypothesis 1 (retrospective symptom reporting): Retrospective dyspnea ratings were higher compared to averaged concurrent ones, C1: F(1, 43) = 126.63, p < .001, !!! = .75 (Figure 3-2, right panel); Moment of Symptom Assessment: F(2.29, 98.55) = 64.45, p < .001, !!! = .60. However, the expected group differences were not found for dyspnea ratings, F(1, 43) = 1.70, p = .20, !!! = .04. Hypothesis 2 (increase in retrospective symptom reporting over time): A Group × Moment of Symptom Assessment interaction emerged, F(2.29, 98.55) = 3.02, p = .05, !!! = .07, revealing that high HSR gave higher follow-up ratings compared to immediate and delayed ratings than low HSR (C2: F(1, 43) = 6.39, p = .02, !!! = .13). Hypothesis 3 (peak-end effect): Even though the concurrently rated dyspnea did not 59

Chapter 3

differ between the trials, dyspnea in the immediate rating was higher for the short, compared with the long, trial, C3: F(1, 43) = 28.09, p < .001, !!! = .40 (Figure 3-3, right panel); Trial × Moment: F(2.09, 89.90) = 10.37, p < .001, !!! = .19. This confirmed the peak-end effect, but no group differences appeared for this interaction, C3: F(1, 43) = .81, p = .37, !!! = .02; Trial × Group × Moment: F(2.09, 89.90) = .40, p = .68, !!! = .01.6 3.3.4. Affective responses High HSR had higher state NA than low HSR, F(1, 44) = 9.71, p < .01, !!! = .18. The dyspnea induction resulted in higher state NA after the trials than the pain induction (Induction × Moment: F(2, 88) = 3.80, p = .03, !!! = .08). State symptom and anxiety ratings followed a similar pattern, as can be seen in Table 3-1. No group differences regarding the threat value of the trials were observed. However, the short rebreathing trial was overall more threatening than the long trial, while the threat value of pain induction trials did not differ (Induction × Moment: F(1, 44) = 5.18, p = .03, !!! = .11). Moreover, mediation analyses (see Appendix C) showed that for three out of four trials (except for short pain trial) state NA during the trial was a significant mediator of the association between the HSR and retrospective symptom ratings.

6

The effect of the order of studies was additionally investigated by including Order of Studies as a between-subject factor to the repeated-measures ANOVAs reported above. The abovementioned effects did not change. A small tendency towards sensitization was observed for the concurrent and retrospective pain ratings, when the pain induction was preceded by the dyspnea induction (concurrent pain ratings: Time Segment × Order of Studies in short trial, F(1.62, 64.77) = 3.59, p = .04, !!! = .08, and in long trial, F(2.45, 97.87) = 4.60, p = .01, !!! = .10; retrospective pain ratings: Order of Studies, F(1, 40) = 4.00, p = .05, !!! = .09). A tendency towards habituation was found for concurrent ratings of dyspnea in short trial (Time Segment × Order of Studies, F(1, 37) = 3.94, p = .06, !!! = .10). However, no interactions between the Order of Studies and the Group variables were found.

60


Figure 3-4. Mean values and standard errors of concurrent dyspnea (0-100), mean fractional end-tidal concentration of CO2 (FetCO2) and minute ventilation for high and low habitual symptom reporters (HSR) in baseline, rebreathing and recovery phase for the short (left) and the long trial (right).

61

Chapter 3 Table 3-1. Means and standard deviations for state symptoms, state NA, state anxiety and threat value, and significant effects of repeated-measures ANOVA for all dependent variables. Significant effects StudyPain StudyDyspnea Measure Group (F;df) Short Long Short Long Baseline Baseline trial trial trial trial IND*** (78.39; 1) low HSR M 1.38 2.67 2.75 1.67 8.50 8.08 MOM*** SD 1.31 2.08 2.17 1.43 4.94 4.93 (27.87; 1.59) State INDxMOM*** symptoms high HSR (41.74; 1.53) M 7.25 6.83 6.54 7.42 15.13 13.21 HSR*** (35.91; 1) SD 5.24 5.24 3.99 4.68 6.02 6.46 low HSR IND*** (20.49; 1) M 12.38 13.17 13.17 13.33 15.00 14.79 MOM*** (15.70; 2) SD 2.53 3.13 3.47 3.07 4.60 3.62 INDxMOM* (3.80; 2) State NA HSR** (9.71; 1) high HSR M 15.21 16.96 16.71 16.13 20.21 19.50 SD 5.35 6.04 5.69 5.17 6.78 6.54 low HSR IND*** (18.87; 1) M 1.67 2.58 2.58 2.04 3.79 3.71 MOM*** (50.72; 2) SD 0.82 1.67 1.79 1.23 1.96 1.99 MOMxOrder* State (3.78; 2) anxiety high HSR INDxMOM** M 2.58 3.71 4.04 2.88 5.25 4.67 (5.67; 1.77) SD 1.50 2.20 2.22 1.94 2.13 2.16 HSR* (7.30; 1) low HSR IND* (6.50; 1) M 3.54 3.17 4.96 4.25 MOM** (7.40; 1) 2.25 2.04 1.88 2.09 INDxMOM* (5.18; 1) Threat SD Order* (7.06; 1) value high HSR M 4.54 4.54 5.42 4.50 SD 2.59 2.50 2.17 2.47 Note. HSR = habitual symptom reporting; IND = induction; MOM = moment of measurement. *p < .05, **p < .01, ***p < .001.

3.4. Discussion The present study aimed to investigate the course of retrospective memory for two distinct aversive bodily sensations, pain and dyspnea, across a 2-week period and whether individual differences in habitual symptom reporting would moderate this course over time. These sensations were experimentally induced in the laboratory to individuals scoring high or low on HSR. Concurrent symptom ratings collected during the inductions served as a reference point for comparisons with retrospective ratings collected at three fixed time points after the symptom induction. Consistent with previous research, it was found that retrospective memory for symptoms is inaccurate and that the course of bias over time differs between the groups. In addition, differences between the two aversive bodily sensations were also observed. 62


The manipulation checks indicated a successful pain and dyspnea induction. Nonetheless, it was also found that the groups differed in their concurrent symptom perception, with high HSR reporting more intense symptoms than low HSR, especially during dyspnea induction. However, physiological responses to this induction, as measured by MV and FetCO2, were not different. These observations are in line with the findings of Bogaerts et al. (2010) who showed increased concurrent dyspnea ratings in MUD patients compared to healthy controls, despite the lack of differences at the physiological level. Considering retrospective symptom ratings, two important findings emerged: (1) symptoms are retrospectively biased, and (2) the pattern of bias in symptom recall differs between low and high HSR. With regards to the first finding, retrospective ratings for both pain and dyspnea were overall higher than averaged concurrent evaluations. This accords with previous research on the inaccuracy of symptom memory in general, showing that the overall retrospective evaluation of the experience is usually rated as more intense than the averaged concurrent reports (e.g., Broderick et al., 2008; Giske et al., 2010; Houtveen & Oei, 2007). It should be noted, however, that previous studies investigating this discrepancy were mostly based on diary protocols and included longer time frames (starting from one day recollections). To our knowledge, our study is the first to show that the global evaluation is biased already immediately after the symptom episode. This discrepancy is often explained by peak/salience effects (Kahneman et al., 1993; Miron-Shatz et al., 2009; Stone et al., 2005) which suggests that the peak intensity of the experience, due to its aversive and threatsignaling connotation, is more heavily weighted during retrospective assessment of averaged symptoms while other symptom-free moments tend to be disregarded. Findings related to trial differences support this idea: During immediate ratings, the short trials (ending at aversive peak) were consistently rated as more intense than the long trials, including gradual recovery. Interestingly, this difference was found not only when the averaged concurrent ratings of both trials did not differ (dyspnea induction), but also when the long trial caused more concurrent pain than the short one. However, in contrast to earlier findings showing a lack of the PE heuristic in high symptom reporters (patients: Bogaerts et al., 2012; non-clinical HSR: Houtveen & Oei, 2007), no group differences emerged for the PE related findings. While this may be due to methodological differences (diary vs. experimental study, PE effect measured by forced-choice preferences for the short or long trial vs. by actual symptom ratings), it could also suggest that this heuristic is used to the same extent by both groups. Further work is needed to elucidate this issue.

63

Chapter 3

Secondly, the pattern of bias in retrospective symptom reporting differed between the groups. In pain induction, group differences emerged for immediate evaluations in that high HSR reported having experienced more pain than low HSR, while in dyspnea induction both high and low HSR reported similar levels of dyspnea. This pattern of results could be caused by the differences in symptom intensity and affective reactions to symptom manipulations. In the current study, pain induction was associated with less state symptoms, and lower state NA and anxiety levels than dyspnea induction. We assume that the lower intensity and threat value of pain induction allowed for the increased influence of the personality characteristics, such as HSR or trait NA. This interpretation is in line with previous findings showing that trait NA and HSR influence somatic complaints especially when symptoms are ambiguous or low in intensity (Bogaerts et al., 2008; De Peuter et al., 2007; Larsen, 1992; Stegen et al., 1998, 2001; Van Diest et al., 2005). The differences were also observed for the pattern of retrospective symptom reporting over time: while there was no increase in recalled pain over the course of two weeks, dyspnea reports increased over time in high HSR, but not in low HSR group. This corroborates the findings from the diary study by Houtveen and Oei (2007), who observed a rise in retrospective symptoms reporting over time in high HSR participants, but it is currently unclear why increasing bias in symptom recall over time in the high HSR group was specific for dyspnea. One reason could be related to the ratio of sensory and affective processing of bodily information used during retrospective recall. Sensory and affective aspects of somatic information are processed in parallel (Leventhal & Everhart, 1979), but a focus on one component may decrease attention to the other. In contrast to sensory processing of bodily signals, which leads to more detailed perception and reduced symptom ratings (Cioffi, 1991; Crane & Martin, 2003), affective processing of usually unpleasant and aversive symptom experience may lead to negative affect biasing both symptom perception and retrospective memory (Bogaerts et al., 2008; Michael & Burns, 2004). It is plausible that high HSR, who also have elevated NA, focus relatively more on the affective aspects of the somatic experience, reducing the influence of the actual sensory input. This way, the level of NA and anxiety experienced during symptom induction could affectively color the memory of experience, resulting in reporting bias. This explanation is to some extent supported by the findings from the mediation analyses, which showed that the relationship between the HSR and retrospective symptom ratings was mediated by the state NA. Moreover, because the pain induction was not as distressing as the dyspnea induction, the negative biasing of the affective component might have been attenuated in this condition. Further research is needed to explore 64


the processing styles adopted by high HSR and their possible influence on symptom memory bias in equally distressing somatic conditions. The findings from this study make several contributions to the current literature. First, as one of the first studies it has investigated memory for symptoms in high HSR by means of experimentally controlled symptom inductions. The inclusion of both concurrent symptom ratings and physiological responses during controlled symptom inductions together with retrospective evaluations of the same experience extend our knowledge of symptom perception and memory, without being undermined by possible physiological differences. Second, retrospective symptom ratings were found to be substantially biased already immediately after the end of experience, with relatively little change thereafter. This is important for studies assessing the accuracy of symptom memory with methods such as the Ecological Momentary Assessment (EMA, Shiffman, Stone, & Hufford, 2008) or the Experience Sampling Method (ESM, Larson & Csikszentmihalyi, 1983). In the majority of studies using the EMA or the ESM, such immediate ratings would be considered concurrent and relatively unbiased, while our data show that, by then, biases have had most of their effect already. Finally, by investigating two types of symptoms, pain and dyspnea, in two different groups, our findings point to the role of particular characteristics of the aversive bodily sensations that, in interaction with individual differences, determine retrospective memory for symptoms. The present study has some limitations. First, the current study used a healthy female HSR sample, which may limit the generalizability of the findings to a clinical MUS population. However, given that our findings concerning dyspnea perception were largely in line with the previous study using the rebreathing paradigm in patients with MUD (Bogaerts et al., 2012), it could be expected that other cognitive processes (i.e., memory) are comparable in those two groups. Second, different concurrent assessment procedures (continuous vs. every 10 s) were applied in both studies. Nonetheless, because the participants were inquired to rate an average symptom experience, the influence of the trial duration and frequency of assessment should not have a great impact on the ratings. In conclusion, the present study documents retrospective memory inaccuracy for symptoms, it replicates the peak-end bias in two different bodily sensations, and it extends our understanding of symptom memory in HSR. The observed increase of retrospective dyspnea reporting over time in high HSR corroborates the role of perceptual-cognitive and memory processes underlying HSR and MUS. Future research is needed to more narrowly

65

Chapter 3

specify the precise mechanisms underlying the observed symptom memory distortions in high HSR. 3.5. Acknowledgments The authors want to thank Anne-Marie Verlinden for her assistance in data collection. The study was funded by Grant OT/10/027 from the University Research Council of the University of Leuven.

66

Retrospective memory for symptoms in patients with somatic symptom disorder

4Chapter 4 Retrospective memory for symptoms in patients with somatic symptom disorder

Abstract Clinical assessment and diagnostic processes heavily rely on retrospective symptom reports. We investigated retrospective memory for symptoms and the peak-end effect for dyspnea in somatic symptom disorder patients with medically unexplained dyspnea (MUD) and healthy participants. Female MUD patients (n = 22) and matched healthy controls (n = 22) participated in two dyspnea induction trials (short and long). Dyspnea ratings were collected: (1) continuously during symptom induction (concurrent with respiratory measures), (2) immediately after the experiment, and (3) after 2 weeks. Symptoms, negative affect, and anxiety were assessed at baseline and after every trial. The mediating role of state anxiety in symptom overreporting was assessed. The peak-end effect was tested with forced-choice questions measuring relative preference for the trials. Compared to controls, dyspnea induction resulted in more symptoms (p < .001) and higher anxiety (p < .05) ratings in the patient group. Also higher concurrent dyspnea ratings (p = .001) and minute ventilation (p < .05) were observed. In both groups, retrospective overreporting compared to concurrent measurement (p < .001) was found, which did not change over 2 weeks. Elevated retrospective symptom reports were mediated by both state anxiety and concurrent dyspnea ratings during the induction. Patients did not show a peak-end effect, whereas controls did. In conclusion, elevated symptom reporting in the patient group appeared already during concurrent ratings and was not further biased in retrospective reports compared to controls. Affective responses play an important role in symptom overreporting and memory biases.

Based on: Walentynowicz, M., Bogaerts, K., Stans, L., Van Diest, I., Raes, F., & Van den Bergh, O. (in preparation). Memory for symptoms in patients with somatic symptom disorder. 67

Chapter 4

4.1. Introduction The recent revision of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) introduced important modifications regarding the concept of somatoform disorders (American Psychiatric Association, 2013). First, the latter term was replaced by somatic symptom disorder (SSD) and, second, the diagnostic criteria were modified, so that they not only require the presence of persistent distressing somatic symptoms, but also include psychological criteria, such as excessive thoughts, feelings and behaviors linked to those symptoms. Several specific perceptual-cognitive and affective characteristics related to symptom perception have been thoroughly investigated and documented for patients qualifying for SSD (Rief & Broadbent, 2007; Rief & Martin, 2014; Van den Bergh et al., 2015). However, little attention was given so far to the role of symptom memory, despite the fact that clinical assessments and questionnaire studies rely on it. In the general patient population, symptom memory is frequently characterized by low accuracy and overestimation (Broderick et al., 2008; Linton & Melin, 1982), which are related to several factors including symptom intensity (Feine et al., 1998; Giske et al., 2010; Sohl & Friedberg, 2008) and variability of symptoms (Lefebvre & Keefe, 2002; Sohl & Friedberg, 2008), time since the episode (Broderick et al., 2008; Houtveen & Oei, 2007), symptom intensity during recall (Eich et al., 1985; Lefebvre & Keefe, 2002; Meek et al., 2001; W. B. Smith & Safer, 1993), and cognitive heuristics. For example, the peak-end (PE) effect holds that the retrospective evaluation of a somatic episode is determined by the most distressing (peak) and the final (end) moments of the episode, and less so by its duration (Bogaerts et al., 2012; Chajut et al., 2014; Kahneman et al., 1993; Redelmeier & Kahneman, 1996; Redelmeier et al., 2003; Walentynowicz, Bogaerts, Van Diest, Raes, & Van den Bergh, 2015). Moreover, psychological factors such as anxiety (Suls & Howren, 2012), negative affectivity (NA; Larsen, 1992; Levine & Safer, 2002; Walentynowicz et al., 2015), catastrophizing (Lefebvre & Keefe, 2002; Sohl & Friedberg, 2008), and depression (Suls & Howren, 2012) play a role. Little is known on how bodily sensations are encoded and stored in memory in SSD patients. Only a few studies applying either diary or experimental designs were carried out among non-clinical high habitual symptom reporters (HSR). In a diary study on symptoms in daily life, Houtveen and Oei (2007) compared concurrent and recalled symptom ratings in non-clinical low and high HSR groups. They observed an overestimation of recalled compared to concurrent symptoms in both groups, but only in high HSR retrospective

68


symptom ratings gradually increased over time. Focusing on a similar sample of high and low HSR, Walentynowicz et al. (2015) experimentally induced pain and dyspnea and followed up symptom ratings over time. They found that pain was retrospectively more overreported by high compared to low HSR, while recalled dyspnea, initially overreported by both groups, increased over time only in high HSR. Interestingly, this biased recall in high HSR was mediated by the affective reactions to the dyspneic experience. Finally, a study using experimental dyspnea inductions showed that clinical SSD patients with medically unexplained dyspnea (MUD) did not exhibit the PE effect, whereas healthy controls did (Bogaerts et al., 2012). Overall, these studies indicate that retrospective memory for symptoms is more distorted in persons with non-clinical high HSR and in SSD patients than in healthy controls. While the mechanisms underlying biased symptom perception and memory are still unclear, the available evidence suggests that affective factors play a role in perceptual and retrospective distortions of symptom ratings. Both neurobiological (Craig, 2003) and behavioral (Leventhal et al., 1979) research suggests that processing somatic sensations involves a sensory-perceptual and an affectivemotivational dimension. The sensory-perceptual dimension covers aspects such as intensity, quality and spatio-temporal characteristics of the experience, while the affective-motivational component refers to its distressing, aversive aspects representing the drive to behaviorally act to reduce (the impact of) the somatic experience. Because SSD patients are in general more anxious and score higher on trait NA (Han et al., 2004; Wan, Stans, Bogaerts, Decramer, & Van den Bergh, 2012), they may possess an over-reactive affective-motivational system (Hariri et al., 2000; Yiend, 2010) resulting in a relative dominance over the sensoryperceptual system when responding to and encoding an aversive somatic episode (Wan et al., 2012). As a consequence, the affective-motivational dimension would affect symptom ratings relatively more (Leventhal & Everhart, 1979), but its impact would also become more dominant over time as memory traces of sensory-perceptual details are short-lasting and more prone to forgetting than global gist traces (Reyna & Brainerd, 1995). The abovementioned findings suggest that the absence of the PE effect in SSD may be its critical marker. However, only one study as of now has shown that SSD patients did not exhibit the PE effect while healthy persons did (Bogaerts et al., 2012). Therefore, we wanted to replicate this intriguing finding and subsequently investigate underlying mechanisms, namely whether affective responses to symptom inductions critically mediates retrospective memory distortion.

69

Chapter 4

In this study, we focused on retrospective memory for dyspnea among patients with MUD, also known as behavioral dyspnea (Bogaerts et al., 2012; Han et al., 2008). MUD is characterized by a number of complaints in different bodily systems, such as urge to breathe, chest tightness, anxiety, and fatigue that do not originate from an underlying cardiovascular or respiratory disorder. The symptoms, experienced as distressing and disruptive, are associated with excessive worrying, anxiety, and frequent medical consultations (Han et al., 2004, 2008; Wan et al., 2012). Dyspnea was experimentally induced by means of two rebreathing trials (Read, 1967), one ending at the most intense level of dyspnea (short trial), while the other additionally including a partial recovery period (long trial). Relative preference for the trials, measured by forced-choice questions, tested the PE effect (Bogaerts et al., 2012; Kahneman et al., 1993). Moreover, participants rated experienced dyspnea concurrently during the induction trials, immediately after the experiment and after two weeks. Based on abovementioned arguments, it was predicted that: (1) The patients would rate concurrent dyspnea as more intense than healthy controls; (2) Retrospective dyspnea ratings would be higher than the concurrent ratings, with a relatively stronger recall bias among the patients; (3) This recall bias in the patient group would be mediated by the affective responses to the dyspnea trials; (4) Retrospective ratings would gradually increase over time only among the patients; (5) The PE effect, indicated by a preference for the long above the short trial (Bogaerts et al., 2012; Kahneman et al., 1993), would be observed in the control, but not in the patient group (replicating Bogaerts et al., 2012). 4.2. Methods 4.2.1. Participants Patients with MUD (n = 22, all women) and healthy controls (n = 22, all women) participated in the study in return for 15 euros (see Table 4-1 for comparison of basic demographic and personality trait characteristics). The patients were recruited from the outpatient pneumology clinic of the Leuven University Hospital (Gasthuisberg), to which they were referred by their general practitioner or a medical specialist due to multiple somatic complaints such as dyspnea, breathing distress, fatigue, and numbness. The patients were classified as having MUS after (1) a systematic medical work-up procedure to exclude

70


physiological causes for those complaints7, including respiratory (e.g., asthma and COPD) or cardiovascular dysfunction; and (2) a systematic interview, namely the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition Axis I Disorders, administered by a qualified psychologist confirming psychological criteria of SSD. Four patients reported use of medication. Control participants were recruited from the general population and were matched for gender, age, BMI, and educational level. The groups did not differ with regard to marital status and occupation. Exclusion criteria included a self-reported history of pulmonary, cardiovascular, gastrointestinal, or neuromuscular disease, or other medical conditions that likely affect respiratory capacity, such as acute illnesses, fever, or flu. Participants were also excluded if they currently suffered from major psychiatric condition other than SSD (self-reported via a general item), were pregnant or lactating. Participants were required to refrain from coffee, tea, or alcohol after midnight before participating and not to smoke for at least 2 hours before the experiment. The study was approved by the Medical Ethics Committee of the University Hospital of the University of Leuven and took place between August 2012 and April 2014. 4.2.2. Measures 4.2.2.1. Habitual symptom reporting The Checklist for Symptoms in Daily Life (CSD; Walentynowicz, Witthöft, et al., 2016) was used to assess HSR. The scale consists of 39 symptoms for which participants indicated how often they experienced them in the past year on a 5-point Likert scale (never, seldom, sometimes, often, very often). The total score (range: 39 – 195) was used to select low HSR for the control group. The test was reliable in the present sample (Cronbach’s α = .96). 4.2.2.2. Negative affectivity Trait and state NA were measured with the Dutch version of the Positive and Negative Affect Schedule (PANAS). Participants rated, on a 5-point Likert scale ranging from not at all to very much, to what extent they felt 10 positive and 10 negative emotions in general (trait) or now (state). Good reliability and validity have been reported (Engelen et al., 2006;

7

Work-up procedure typically includes the following tests: histamine provocation test, spirometry, chest x-ray, electrocardiogram, echocardiogram, D-dimer, computed tomography of the thorax, cycle-ergometry, tilt table test, electroencephalogram, evoked potentials, electromyogram, brain scan (nuclear magnetic resonance), gastroscopy, gastrointestinal tract x-ray, manometry, esophageal pH-impedance monitoring, abdominal echogram, feces and urine test, and blood test.

71

Chapter 4

Watson et al., 1988). In the current sample, both trait and state versions were reliable (Cronbach’s alphas ranging from .83 to .92). 4.2.2.3. Depression Depression was measured with the Dutch version of Beck Depression Inventory-II (BDI-II; Van der Does, 2002), a 21-item questionnaire assessing current cognitive, affective and physical symptoms of depression on a 0-3 scale. The Cronbach’s alpha in the present sample was .93. 4.2.2.4. Symptom and affect assessment Complaints at baseline and after each dyspnea induction trial were assessed by a state symptom checklist with a 5-point rating scale ranging from not at all to very much. Participants were asked to indicate to which extent they experienced each of 46 symptoms during the trial. Internal consistency of the state symptom checklists was good, with Cronbach’s alphas ranging from .93 to .96. Additionally, a single item numerical rating scale (NRS) evaluating the level of anxiety (1 = not anxious at all, 9 = very anxious) was administered at baseline and after each trial. During the follow-up session, participants evaluated the overall unpleasantness of the laboratory session on NRS (-4 = very unpleasant, 4 = very pleasant). 4.2.2.5. Concurrent dyspnea ratings During each rebreathing trial, concurrent dyspnea was rated with a mouse click every 10 s (after auditory cue) on a 0-100 computerized scale displayed as a vertical bar in the middle of the screen. Different levels of the dyspnea, described on its right side, were based on a modified Borg scale (Borg, 1982): none (0), very slight (10), slight (20), moderate (30), fairly severe (40), severe (50), very severe (60), very severe (70), very severe (80), very, very severe (90), intolerable (100). 4.2.2.6. Retrospective dyspnea ratings The retrospective evaluations of average dyspnea experienced during each trial were collected at two recall moments: at the end of the experimental session (delayed rating) and after two weeks (follow-up rating). Participants indicated the average dyspnea level experienced during the trial (How much dyspnea have you experienced on average during this trial?) on a visual analog scale (10 cm) ranging from 0 (no dyspnea) to 100 (maximum dyspnea).

72


4.2.2.7. Forced choice questions Following Bogaerts et al. (2012), the PE effect was assessed after the two trials with forced choice questions inquiring which trial the participants preferred (see Table 4-3 for question description). For each of the questions, participants could choose between the first and the second trial, except for the PE-Duration, which included the option “no difference”. According to the PE effect, the experience in which the peak intensity is followed by a period of lower intensity (long trial) should be preferred to the experience ending on the most intense moment (short trial). 4.2.3. Apparatuses and physiological recordings Two trials of the rebreathing paradigm (Bogaerts et al., 2012; Read, 1967; Walentynowicz et al., 2015) were used as a dyspnea induction. During the trials, participants wore a nose clip and breathed through a mouthpiece, connected to the rebreathing bag with a wide vinyl tube and a Y-valve ending on a pneumotachograph (Fleisch no. 2, Lausanne, Switzerland) measuring airflow. The valve enabled to change between room air and the rebreathing bag, which was filled with 5-L gas mixture of 5% CO2 and 95% O2 beforehand. Breathing in this closed hyperoxic system resulted in a progressive increase of PCO2, of minute ventilation and of dyspnea, defined as uncomfortable feeling of not having enough air, an urge to breathe, or a feeling of having more difficulty to breathe. The exhaled air was sampled close to the mouthpiece in order to determine the fractional end-tidal concentration of CO2 (FetCO2), measured by an infrared CO2 monitor (POET RC, Criticare Systems Inc., Waukesha, WI). The data from the pneumotachograph and the CO2 monitor were sampled at 20Hz, stored on a computer, and analyzed offline to define the following parameters: minute ventilation (MV) in L/min and FetCO2 in %. As the switching of the valve was not noticeable by the participants, their concurrent dyspnea ratings depended exclusively on the experienced physical changes. 4.2.4. Procedure Before arriving to the laboratory, participants were asked to complete a set of trait questionnaires at home (CSD, PANAS, BDI-II). Upon arrival, participants were informed that they would inhale three different air mixtures. After signing informed consent, participants completed the state PANAS, the symptom checklist, and rated their anxiety level. Afterwards, participants were familiarized with the procedure (concurrent rating system) and equipment (mouthpiece, nose clip) during a practice trial with room air, followed by the two rebreathing trials. The order was counterbalanced across participants. The short trial comprised a baseline 73

Chapter 4

phase (60 s of room-air breathing) and a rebreathing phase (150 s), after which the participants took out the mouthpiece and the trial was terminated. The long trial consisted of the same baseline and rebreathing phases as the short trial, which were followed by an additional recovery phase (150 s), during which participants were breathing room air through the system while continuing to provide concurrent dyspnea ratings. The discrete manipulation of the valve was unnoticeable to participants. A 15-min intertrial interval between the two rebreathing trials was included to allow full recovery. Respiration was measured continuously throughout both trials. After completion of both trials, participants were informed that additional information was required prior to the final trial. They were presented with forced choice preference questions, after which the experimenter told them that the third trial was not necessary after all. As a last step, the average dyspnea experienced during the trials was measured (delayed ratings). At home, two weeks after the laboratory session, participants completed the followup questionnaires including the retrospective average dyspnea ratings for both trials (followup ratings) and the question about the overall unpleasantness of the laboratory session. 4.2.5. Data analysis Independent sample t tests were conducted to investigate the group differences in demographic variables, trait measures and retrospective unpleasantness ratings. Group (patients/controls) and Order of Trials were included as between-subject factors in all repeated-measures analyses of variance (ANOVAs) reported next. Group differences in state symptoms, NA and anxiety ratings were examined with repeated-measures ANOVAs with Trial (baseline/short trial/long trial) as a within-subject factor. To test Hypothesis 1, concurrent self-reported and respiratory (FetCO2 and MV) responses during each trial were averaged per 30 s for each participant and examined in separate repeated-measures ANOVAs with Time Segment as a within-subject variable. Dyspnea ratings on different moments were compared in repeated-measures ANOVA with Dyspnea

Trial

(short

trial/long

trial)

and

Moment

of

Assessment

(averaged

concurrent/delayed/follow-up) as within-subject factors. For this analysis, concurrent dyspnea reports were averaged per trial. Planned contrasts were applied to assess specific time and group effects (C1: concurrent vs. all retrospective ratings to test Hypothesis 2; C2: retrospective ratings during experimental session (delayed) vs. follow-up ratings to test Hypothesis 4). Greenhouse–Geisser corrections were applied when the sphericity assumption was violated. 74


Multiple mediator models were applied to short and long dyspnea trials to examine whether state affect experienced during the trials mediates the association between the group (patients/controls) and the retrospective (follow-up) dyspnea ratings (Hypothesis 3). We have focused on state anxiety rather than on state NA, as the latter did not change after the induction trials in either group as compared to baseline. As it is possible that concurrently perceived dyspnea also affects the retrospective ratings, both state anxiety and concurrent dyspnea ratings were simultaneously included as mediators and tested in a single parallel multiple mediator model using the bootstrapping procedure (Preacher & Hayes, 2008). The advantages of this method, which was designed for small sample sizes, include the estimation not only of total effects, but also of specific indirect effects (paths a1b1 and a2b2) of each mediator. The 95% confidence intervals of the effects were derived with 5000 bootstrap resamples. The model coefficients, direct and indirect effects are reported in the unstandardized form (Hayes, 2013). In order to assess the PE effect within each group and the differences between the groups with regard to forced choice questions, Pearson chi-square tests were applied (Hypothesis 5). Data were analyzed with IBM SPSS Statistics (Version 22.0) and the PROCESS Procedure for SPSS (Hayes, 2013). 4.3. Results Table 4-1 shows the demographic and psychological characteristics of the participants. No differences between the groups were found for age and BMI. However, patients scored higher on habitual symptoms, trait NA, and BDI-II. Patients also reported more symptoms at baseline and after the trials compared to controls, F(1, 40) = 28.96, p < .001, !!! = .42 (Table 4-2). Even though state NA (as measured by the NA-scale) was overall higher for patients compared to controls, F(1, 40) = 34.97, p < .001, !!! = .47, it did not increase after the trials in either group. On the other hand, the patient group not only reported higher anxiety in general (as measured by the single item NRS), but their anxiety also increased to a greater extent after the rebreathing trials compared to the control group, Group × Trial, F(1.48, 59.08) = 3.79, p = .040, !!! = .09. The retrospective unpleasantness ratings did not differ between the groups, t(42) = -1.38, p = .17 and showed that both patients (M = .86, SE = .41) and controls (M = .09, SE = .38) perceived the laboratory session as neutral.

75

Chapter 4 Table 4-1. Group comparisons of demographic and personality trait characteristics. Variable

Controls

Patients

t test

(n = 22)

(n = 22)

Age (yrs)

37.59 (9.94)

36.86 (9.58)

t(42) = .247

BMI (kg/m2)

22.07 (2.38)

23.53 (4.05)

t(33.95) = -1.46

CSD

62.68 (7.82)

115.50 (17.51)

NA

16.14 (4.63)

29.45 (7.40)

t(42) = -7.57***

BDI

4.50 (4.48)

19.32 (9.41)

t(30.06) = -6.67***

t(29.05) = -12.92***

Note. *p < .05, **p < .01, ***p < .001.

Table 4-2. Means and standard deviations for state symptoms, state NA and state anxiety in control (n = 22) and patient (n = 22) groups. Trial Statistics Significant F Variable Group Baseline Short trial Long trial effects 51.05 62.36 61.77 Trial*** 15.28 Controls a b b State (4.40) (12.36) (12.02) Group*** 28.96 symptoms 72.64 89.36 86.09 Patients (16.11)c (26.96)d (25.36)d 12.45 12.55 12.32 Group*** 34.97 Controls (2.69)a (2.13)a (1.86)a State NA 18.05 18.77 18.59 Patients b b (5.00) (6.14) (5.85)b State 1.55 2.50 2.27 Trial*** 23.85 Controls a b ab anxiety (.96) (1.50) (1.24) Group*** 17.54 Trial×Group* 3.79 Patients 2.59 4.73 4.41 b d d (1.62) (2.41) (2.02) Note. *p < .05, **p < .01, ***p < .001.

!!! .28 .42

.47

.37 .31 .09

4.3.1. Dyspnea perception and breathing behavior Patients reported significantly more dyspnea than controls during both the short, F(1, 40) = 13.60, p = .001, !!! = .25, and the long trial, F(1, 40) = 15.05, p < .001, !!! = .27 (see Figure 4-1, upper panels). This difference between the groups became stronger over time in both trials, Group × Time Segment interaction in the short, F(1.70, 68.17) = 13.73, p < .001, !!! = .26, and in the long trial, F(2.68, 107.28) = 4.85, p = .005, !!! = .11.

76


Figure 4-1. Mean values and standard errors of concurrent dyspnea (0-100), minute ventilation, and fractional end-tidal concentration of CO2 (FetCO2) for controls and patients with MUD in baseline, rebreathing and recovery phase for the short (left) and the long trial (right). Whiskers denote standard errors. As shown in the middle panels of Figure 4-1, the patients also had a higher mean MV during both trials than the controls in both the short and long trial, F(1, 39) = 4.32, p = .044, !!! = .10, and F(1, 40) = 5.64, p = .022, !!! = .12, respectively. Similar to the pattern of dyspnea ratings, this difference became stronger over time both in the short and long trial: 77

Chapter 4

Group × Time Segment interaction, F(1.31, 51.07) = 4.97, p = .022, !!! = .11, and F(2.05, 82.04) = 3.12, p = .048, !!! = .07, respectively. Furthermore, when the long trial was preceded by the short one, participants reached higher peak levels of MV during the long trial (Order × Time Segment interaction for the long trial, F(2.05, 82.04) = 4.11, p = .019, !!! = .09). With regard to the FetCO2 levels, no group-related differences were found (see Figure 4-1, bottom panels).8 It is interesting to note that the differences between groups observed in concurrent dyspnea ratings and MV were significantly related to state anxiety experienced during dyspnea induction when it was introduced as a covariate in the abovementioned analyses. 4.3.2. Retrospective dyspnea rating Patients reported overall more dyspnea than controls, F(1, 40) = 12.82, p = .001, !!! = .24 (Figure 4-2). Interestingly, retrospective dyspnea ratings were higher than the averaged concurrent ones, indicating a general dyspnea overreporting (C1: F(1, 40) = 82.04, p < .001, !!! = .67, Moment of Assessment: F(2, 80) = 55.05, p < .001, !!! = .58). However, the expected interaction with Group was not found, Group × Moment: F(2, 80) = 1.50, p = .23, !!! = .04. Patients and controls overreported the experienced dyspnea to the same extent, C1 for Group × Moment: F(1, 40) = 1.55, p = .22, !!! = .04. Also, the retrospective ratings did not further increase over time in either group, C2 for Group × Moment: F(1, 40) = 1.40, p = .24, !!! = .03.

Figure 4-2. Mean averaged concurrent and retrospective dyspnea ratings (0-100) for controls and patients with MUD. Whiskers denote standard errors. 8

Significant Order × Time Segment interactions for both short, F(2.17, 82.54) = 7.21, p = .001, !!! = .16, and long trial, F(3.14, 125.69) = 3.12, p = .027, !!! = .07, indicated a lower FetCO2 level during the baseline of the short trial when the long trial was given first, and higher peak FetCO2 and higher FetCO2 during recovery phase of the long trial when it was administered second.

78


4.3.3. The mediating role of state anxiety In both short and long dyspnea trials (Figure 4-3), group predicted both follow-up dyspnea ratings, state anxiety (path a1) and concurrent dyspnea (path a2), while state anxiety (path b1) and concurrent dyspnea (path b2) predicted follow-up dyspnea ratings after controlling for group. Consequently, state anxiety and concurrent dyspnea ratings could be treated as candidates for mediators in both trials. In the short trial, specific indirect effects for both state anxiety, a1b1 = 10.41 (95% CI [1.25, 22.40]) and concurrent symptom ratings, a2b2 = 11.80 (95% CI [4.18, 23.10]) on follow-up ratings were significant. The effects did not differ in size, with the point estimate for the contrast between the two indirect effects, -1.38 (95% CI [-18.02, 13.06]), not significantly different from zero. A similar pattern was found for the long trial, with significant specific indirect effects on follow-up ratings for both state anxiety, a1b1 = 13.81 (95% CI [4.03, 29.42]) and concurrent symptom ratings, a2b2 = 8.35 (95% CI [.57, 20.74]). The point estimate for the contrast between the two indirect effects, 5.47 (95% CI [-12.78, 25.38]), indicated that the effects did not differ in size.9

Figure 4-3. Multiple mediator models for short dyspnea trial (left) and long dyspnea trial (right). The panels show direct and indirect effects of a group (patients/controls) on the retrospective dyspnea ratings, mediated by state anxiety and concurrent dyspnea ratings. The model coefficients are reported in unstandardized form.

9

As the observed group differences in trait psychological characteristics could also affect the association between group and retrospective dyspnea ratings, additional analyses including trait NA and depression as mediators were performed in all reported multiple mediator models. Indirect effects through either trait NA or depression were found nonsignificant. In addition, they did not affect the previosly reported results.

79

Chapter 4

4.3.4. The peak-end effect The frequencies on each of the forced choice questions are shown in Table 4-3. Significant group differences were found only for the question concerning the greatest discomfort, with the PE effect only emerging in the control group (the short trial chosen as the one causing greatest discomfort). Within-group analyses also showed that the control group chose the short trial as causing greatest distress at the peak and greatest dyspnea at the peak (marginal significance). The PE effect was overall absent in the patient group.

Table 4-3. Number (and percentages) of participants per group (controls and patients) who preferred the short or the long trial on each of the forced choice questions assessing the peak-end effect. Question df, χ2 Short Long No df, χ2 trial trial difference PE-Preference1: Which trial would you prefer to repeat tomorrow? Controls 1, .73 9 (40.9) 13 (59.1) 1, .10 Patients 1, 1.64 8 (36.4) 14 (63.6) PE-Preference2: Which trial would you pick for today’s third trial? Controls 1, .73 9 (40.9) 13 (59.1) 1, .09 Patients 1, .18 10 (45.5) 12 (54.5) PE-Discomfort: Which trial caused greatest discomfort? Controls 1, 8.91** 18 (81.8) 4 (18.2) 1, Patients 1, .00 11 (50.0) 11 (50.0) 4.96* PE-Duration: Which trial lasted longer? Controls 2, 12.64** 2 (9.1) 15 (68.2) 5 (22.7) 2, Patients 2, 27.91*** 1 (4.5) 19 (86.4) 2 (9.1) 2.09 PE-Max discomfort: Which trial caused greatest distress at peak? Controls 1, 4.55* 16 (72.7) 6 (27.3) 1, Patients 1, .18 12 (54.5) 10 (45.5) 1.57 PE-Max Dyspnea: Which trial caused the greatest amount of dyspnea at peak? Controls 1, 2.91a 15 (68.2) 7 (31.8) 1, 2.32 Patients 1, .18 10 (45.5) 12 (54.5) 2 Note. Significant χ values in the first column indicate the within group differences in the choice for the short versus long trial. Significant χ2 values in the last column indicate the between group differences in the choice for the short versus long trial. a p < .10, *p < .05,**p < .01.

80


4.4. Discussion The present study investigated the role of perceptual-cognitive biases in retrospective symptom reporting in SSD patients complaining of medically unexplained dyspnea. To this end, concurrent and retrospective ratings of experimentally induced dyspneic experiences in patients with MUD and healthy controls were compared. Additionally, physiological and affective reactions to the stimuli were measured. In accordance with previous findings, this study showed that patients with MUD reported higher concurrent and retrospective dyspnea ratings than healthy controls. However, compared to concurrent ratings, retrospective overreporting was not larger in patients than in controls and the symptom ratings did not further increase over time in either group. The lack of PE effect was confirmed among patients with MUD. The manipulation checks showed that dyspnea was successfully induced in both groups. However, the patients with MUD reported higher levels of concurrently experienced dyspnea than the controls, with the differences increasing over the course of the episode and leading to a more intense peak in the patient group. This group difference in self-report coincided with higher minute ventilation in the patient group, replicating previous findings (Bogaerts, Van Eylen, et al., 2010; Wan et al., 2012). Patients felt more anxious not only at baseline, but also after the induction trials, which is also consistent with earlier findings (Han et al., 2004; Wan et al., 2012). Interestingly, when state anxiety experienced during the trials was taken into account, the group-related differences in both self-reported and physiological measures were no longer significant. It seems that patients with MUD, being more prone to experience negative affective states, exhibited a more anxious response to the dyspnea induction, which in turn led to an increase in MV and more elevated dyspnea. However, the elevated MV did not impact the level of FetCO2, which is a critical respiratory parameter for dyspnea. This pattern of data suggests that the differences between patients and controls in concurrent dyspnea ratings were largely driven by symptom induction-related anxiety. Given the fact that the patients with MUD can be characterized by affective hypersensitivity to dyspnea (Wan et al., 2012), it can be assumed that affective processes played a crucial role in dyspnea perception in this group (see De Peuter et al., 2004, for review). Considering memory for dyspnea, a general overestimation of experienced dyspnea was present in both groups, confirming earlier findings showing that symptoms are retrospectively rated as more intense compared to the averaged concurrent ratings (Broderick et al., 2008; Giske et al., 2010; Walentynowicz et al., 2015). In line with the concurrent

81

Chapter 4

ratings of dyspnea, the retrospective ratings were overall higher in the patients than in the controls. Based on the dual-process perspective, it was predicted that patients would respond with stronger affective-motivational responses to dyspnea (Wan et al., 2012), resulting in a relatively greater impact of the unpleasantness aspect on their dyspnea ratings compared to the sensory-perceptual aspect. We expected this would lead to better encoding of aversive information, subsequently resulting in elevated retrospective dyspnea ratings. However, contrary to our expectations and to previous findings (Walentynowicz et al., 2015), the degree of overresponding did not differ between the groups. One possible explanation of this difference is that the stronger affective-motivational responses in patients resulted in a more intense, faster and deeper, breathing pattern as indicated by the elevated MV. This may also have produced more salient sensory-perceptual information to encode during the somatic experience, hence attenuating the relative impact of the affective-motivational response (Leventhal et al., 1979). The degree in which affective responses influence physiological reactions and, subsequently, sensory input, may be larger in this patient group than in nonclinical high HSR. Indeed, in a previous study (Walentynowicz et al., 2015) with the same dyspnea induction paradigm, high HSR individuals showed an increase in dyspnea overreporting over time, but they did not differ in MV during the dyspnea induction. This interpretation may also shed light on another unanticipated finding, namely that retrospective dyspnea ratings did not increase over the course of two weeks in the patient group (as we found in non-clinical high HSR reporters Walentynowicz et al., 2015) and remained at a similar level in both groups. This prediction was based on the assumption that memory for sensory-perceptual details decays at a faster rate than global gist traces representing experienced distress (Reyna & Brainerd, 1995). However, as relatively more sensory-perceptual input (elevated MV) was produced by the patients than by the controls during the encoding of experience, the hypothesized relative dominance of affective compared to sensory information in symptom memory of patients may have been reduced. Our mediation analyses are largely consistent with this interpretation: Both the concurrent level of experienced dyspnea and state anxiety were equally strong and significant mediators of the association between the experimental group and retrospective dyspnea ratings. Interestingly, the two predictors of retrospective symptom ratings were also highly intercorrelated, making it difficult to disentangle their relative effects (short dyspnea trial: r(42) = .71, p < .001; long dyspnea trial: r(42) = .75, p < .001). Another relevant finding in this respect is that the participants of the current study evaluated participating in the experiment as rather neutral, whereas participation in a previous 82


study using identical dyspnea inductions was retrospectively evaluated as unpleasant (Walentynowicz et al., 2015). It is therefore likely that retrospective memory in this study was not dominated by aversive feelings during retrieval and, thus, did not affect the retrospective ratings in a negative way. All in all, the results of this study show that dyspnea memory after two weeks accurately reflected the initial retrospective ratings (but not the concurrent ones). Nonetheless, we established the absence of the PE effect in patients while it clearly appeared in the healthy controls, replicating an earlier study (Bogaerts et al., 2012) and confirming the differences between SSD patients and healthy persons in the way they encode and/or retrieve memory for symptoms. While the mechanisms underlying this intriguing difference remain unclear, the replicability of this finding suggests a robust phenomenon, although our results are less consistent across the entire set of preference questions compared to what was found in a previous study (Bogaerts et al., 2012). One reason for these weaker effects may be related to the interpretation advanced above, namely that the elevated anxiety of patients during the dyspnea induction also produced more intense physiological responses, resulting in a better encoding of detailed sensory-perceptual input. For the PE effect to occur, the sensory input seems required at least to some extent: In the somatic experience, a clear differentiation is needed between a peak and a subsequent phase representing a gradual decline towards a less intense end. One can imagine that a PE effect may fail to occur when the experience is dominated by a non-differentiated distress response to dyspnea, and may gradually appear when more sensory-perceptual details become available, weakening to some extent the difference between patients and controls. A key strength of the present study was an inclusion of concurrent and retrospective symptom ratings over a substantial follow-up period, together with assessing physiological responses to a standardized and validated rebreathing paradigm used to induce symptoms. Exploring the issue of memory for symptoms in both clinical and control group is an important addition to the previous studies, which very often focused only on the patient groups (Broderick et al., 2008; Jamison et al., 2006). However, the current study also has some limitations. Even though experimental control over the dyspneic experience is necessary to investigate its precise course, such manipulation may not resemble a prototypical dyspneic episode occurring in naturalistic settings. Moreover, retrospective ratings were collected only at the end of the experiment and not immediately after the trial, while it was previously shown that biased symptom reporting already occurred at this early stage (Walentynowicz et al., 2015). However, as the difference between immediate and delayed ratings was previously

83

Chapter 4

reported to be minor, it can be assumed that immediate ratings in the current study would resemble the delayed ones. The current findings add to a growing body of literature on cognitive and affective processes characterizing patients with SSD. They highlight the important role of affective state in both concurrent and retrospective overreporting of dyspnea and confirm the absence of a PE effect in retrospective symptom evaluations in SSD patients. This might suggest that the way these patients encode and recall a somatic event may be a robust and critical marker of their condition. However, further investigations are needed to fully understand the underlying mechanisms. 4.5. Acknowledgments The study was funded by Grant OT/10/027 from the University Research Council of the University of Leuven.

84

Ways of encoding somatic information and their effects on retrospective symptom reporting

5Chapter 5 Ways of encoding somatic information and their effects on retrospective symptom reporting

Abstract Retrospective symptom reports tend to overestimate actual symptom intensity. We investigated how focusing on sensory-perceptual or on affective-motivational aspects of a somatic experience influenced retrospective symptom reports in high and low habitual symptom reporters (HSR). Dyspnea (rebreathing) and pain (cold pain) were induced during two experimental sessions in healthy women varying in habitual symptom reporting (N = 45; 21 high HSR, 24 low HSR). Within-subject manipulation of sensory and affective processing focus (PF) took place at the encoding phase. Dyspnea and pain ratings were collected during the sessions and after 2 weeks. Breathing behavior was recorded during dyspnea trials, while affective state and symptom measures were collected after each trial. Compared to pain, dyspnea induction was perceived as more unpleasant, arousing, and threatening (ps < .001). Affective PF resulted in higher arousal (p < .01) and threat ratings (p = .01) than sensory PF. Affective PF also led to an increase in retrospective dyspnea ratings over the course of two weeks (p = .039), which was not observed for pain, nor for dyspnea after sensory PF. The processing focus during symptom encoding may explain previously observed bias in retrospective symptom reporting. The results are relevant to understand the mechanisms underlying overreporting of symptoms and medically unexplained symptoms.

Based on: Walentynowicz, M., Raes, F., Van Diest, I., & Van den Bergh, O. (in preparation). Ways of encoding information and their effects on retrospective symptom reporting. 85

Chapter 5

5.1. Introduction Retrospective symptom memory is of critical importance in both medical care and in health-related research. It is a major source of information for medical doctors during consultations, and symptom questionnaires are abundantly used in research to assess the health status of respondents. Inaccurate retrospective symptom memory can adversely affect medical diagnosis, compliance, or outcome evaluation. Previous research has documented various memory biases leading to inaccurate and often overestimated symptom ratings among patients (Broderick et al., 2008; Giske et al., 2010; Walentynowicz, Bogaerts, et al., 2016) as well as among healthy individuals (Gedney & Logan, 2006; Walentynowicz et al., 2015). Those biases start immediately after the actual somatic experience (Walentynowicz et al., 2015; Walentynowicz, Bogaerts, et al., 2016) and are related to the duration of the recall period (Broderick et al., 2008; Houtveen & Oei, 2007; Walentynowicz et al., 2015), to characteristics of the somatic sensations (Eich et al., 1985; Giske et al., 2010; Lefebvre & Keefe, 2002; W. B. Smith & Safer, 1993; Sohl & Friedberg, 2008; Walentynowicz et al., 2015), and of the individual. The latter includes negative affectivity (NA), anxiety, catastrophizing, and depression (Larsen, 1992; Lefebvre & Keefe, 2002; Suls & Howren, 2012; Walentynowicz et al., 2015). Symptom reports are affected by the interplay of sensory-perceptual and affectivemotivational components of the somatic sensation (Craig, 2003; Leventhal & Everhart, 1979; Price, 2000). While the sensory component refers to distinct sensory characteristics of somatic experiences such as quality, intensity, location, and duration, the affective one refers to feelings induced by experiences together with a motivational drive to behaviorally react or adjust (to) its effect. Because the relative contribution of both components may impact symptom reports (Leventhal et al., 1979), focusing on the affective aspects of a somatic experience may direct attention towards symptom-related distress and cause more elevated, possibly biased, symptom ratings. In line with this argument, it was previously shown that affective responses to somatic experiences partially mediate symptom overreporting (Walentynowicz et al., 2015; Walentynowicz, Bogaerts, et al., 2016). On the other hand, attending to the sensory details has been shown to result in lower symptom ratings (Ahles et al., 1983; Crane & Martin, 2003; Keogh et al., 2000; Leventhal et al., 1979). However, the efficacy of a sensory PF seems to be modulated by anxiety-related individual characteristics, for example health anxiety (Hadjistavropoulos, Hadjistavropoulos, & Quine, 2000), fear of

86


pain (J. Roelofs et al., 2004), and habitual symptom reporting (Bogaerts et al., 2008), with advantages of a sensory PF appearing with increased anxiety levels. Taking the abovementioned into account, the manipulation of PF towards sensory aspects should be most beneficial for individuals scoring high on trait NA, who are considered to possess an overreactive affective system (Hariri et al., 2000; Yiend, 2010) which may increase the relative impact of affective processes compared to sensory-perceptual ones during somatic experiences. Indeed, high trait NA persons tend to be high habitual symptom reporters in daily life (HSR; Bogaerts et al., 2015; Han et al., 2004; Houtveen & Oei, 2007; Walentynowicz et al., 2015; Wan et al., 2012), they tend to overreport induced bodily symptoms (Walentynowicz et al., 2015), and they are more likely to develop somatic symptom disorder (SSD; American Psychiatric Association, 2013). Previous research investigating symptom perception and memory in high NA/high HSR groups showed biased processing of bodily information, manifesting as lower correspondence between self-reported symptom ratings and relevant physiological responses (Bogaerts et al., 2008; Bogaerts, Van Eylen, et al., 2010), as a lack of the peak-end heuristic for somatic episodes (Bogaerts et al., 2012; Walentynowicz, Bogaerts, et al., 2016), and as reduced differentiation between various bodily sensations (Petersen, von Leupoldt, & Van den Bergh, 2015). These studies point to less detailed sensory processing of bodily information. Other studies documented stronger affective responses towards somatic experiences in high HSR and in patients with SSD (Walentynowicz et al., 2015; Walentynowicz, Bogaerts, et al., 2016; Wan et al., 2012), and these stronger responses have been shown to mediate symptom overreporting (Walentynowicz et al., 2015; Walentynowicz, Bogaerts, et al., 2016). In general, less detailed sensory processing of somatic information along with a stronger influence of negative affective responses may bias retrospective symptom reporting in this group towards overreporting. As a result, shifting focus away from the affective to the sensory aspects of one’s responses to bodily sensations might reduce the impact of negative affective processes, and in return lower symptom ratings. The purpose of the current study was to examine the mechanisms underlying retrospective symptom overreporting, and more specifically, the role of processing focus during symptom episodes. To this end, two types of PF were induced during the experience of two symptoms, namely pain via a cold stimulus and dyspnea via a rebreathing manipulation in participants high and low on HSR (Bogaerts, Van Eylen, et al., 2010; Read, 1967; Walentynowicz et al., 2015). Participants were instructed to focus either on the sensory

87

Chapter 5

features in a neutral way (sensory PF) or on the feelings and emotions induced by the somatic experiences (affective PF). Symptoms and affective responses were measured during and immediately after the induction of somatic sensations, and after two weeks. Based on the arguments outlined above, we expected the affective PF to result in more emotion-laden/less sensory processing of bodily information, leading to higher symptom ratings than the sensory PF. Moreover, as the relative influence of the affective component on the symptom memory tends to increase over time (Gedney et al., 2003; Gedney & Logan, 2004; Kent, 1985), while sensory details are forgotten faster (Reyna & Brainerd, 1995), the retrospective symptom ratings were predicted to increase over time in the affective PF condition. Finally, as a sensory PF may be hypothesized to be more beneficial for anxious or oversensitive individuals (Baron et al., 1993; Hadjistavropoulos et al., 2000; J. Roelofs et al., 2004), the high HSR group was expected to benefit relatively more from this PFmanipulation than the low HSR group. 5.2. Methods 5.2.1. Participants The sample consisted of forty-five healthy women (Mage = 20.07, SDage = 2.10, range = 17 – 25) selected for a two-part study based on their scores on the Checklist for Symptoms in Daily Life (CSD; Walentynowicz, Witthöft, et al., 2016) measuring habitual symptom reporting. Two extreme groups of high (≥ 100; n = 21) and low (≤ 75; n = 24) HSR were formed based on the previously used cut-off scores (Bogaerts et al., 2008; Constantinou et al., 2013; Walentynowicz et al., 2015). The CSD was re-administered after the first session to confirm the allocation of the participants, and only the ones meeting the cut-offs on both measurements were included in the final sample. Participants were excluded if they self-reported a chronic medical condition or psychiatric disorder, current acute illness, recent pain or conditions affecting the dominant arm, or pregnancy. The experimental protocol was approved by the Multidisciplinary Ethics Committee of the Faculty of Psychology and Educational Sciences of the University of Leuven. Participants received either course credits or 20 euros for their participation.

88


5.2.2. Measures 5.2.2.1. Habitual symptom reporting The Checklist for Symptoms in Daily Life (CSD; Walentynowicz, Witthöft, et al., 2016) was used to assess the frequency of 39 symptoms in the past year on a 5-point Likert scale (never, seldom, sometimes, often, very often). The test was reliable in the present sample (Cronbach’s α = .97). 5.2.2.2. Negative affectivity Negative affectivity, both trait and state, was measured with the Positive and Negative Affect Schedule (PANAS; Engelen et al., 2006; Watson et al., 1988). Participants indicated to which extent 20 positive and negative adjectives apply to their emotions now (state) or in general (trait) on a 5-point Likert scale ranging from not at all to very much. In the current sample, reliability of both trait and state versions was good (Cronbach’s alphas ranging from .76 to .87). 5.2.2.3. State symptom checklist Participants assessed the presence of 14 complaints during the symptom induction trial on a 5-point rating scale ranging from 1 (not at all) to 5 (very much). This symptom list included: chest tightness, pounding of the heart, headache, fatigue, not able to breathe deeply, rapid heartbeat, dizziness, muscular pain, dyspnea, pain, tingling sensation, stinging sensation, cold, burning sensation. State symptom checklists showed acceptable and good internal consistency, with Cronbach’s alphas ranging from .76 to .90. 5.2.2.4. Affect and threat ratings Experienced affect was measured after each induction trial via a computerized ninepoint version of the Self-Assessment Manikin (SAM; Bradley & Lang, 1994) with the dimensions of valence (1 = unpleasant, 9 = pleasant), arousal (1 = very calm, 9 = very arousing), and control (1 = low control, 9 = high control). The SAM is a pictorial method in which levels of the affective dimensions are represented by human-like pictographs. Together with SAM scales, a single item numerical rating scale (NRS) assessing the threat value of each trial (1 = not threatening at all, 9 = very threatening) was administered. 5.2.2.5. Retrospective pain/dyspnea ratings Symptoms (pain/dyspnea) experienced during each trial were retrospectively evaluated at three moments: immediately after each trial (immediate rating), at the end of the

89

Chapter 5

laboratory session (delayed rating) and after 2 weeks (follow-up rating). Three visual analog scales (10 cm) ranging from 0 (no pain/dyspnea) to 100 (unbearable pain/dyspnea) were used to measure (a) average symptom level, (b) symptom level at its most intense moment, and (c) symptom level at the end of the trial. 5.2.3. Apparatuses and physiological recordings 5.2.3.1. Cold pain induction Painful stimuli were delivered to the ventral surface of the dominant forearm via a 30 × 30 mm thermode (Medoc, TSA, Ramat Yishau, Israel). This device assures a controllable and precise delivery of thermal stimuli. The baseline temperature was set at 32°C. During pain induction trial, the stimulus temperature decreased to a plateau temperature of 2°C. After a 5-s plateau, the temperature of the thermode returned to the baseline temperature. The rates of decline and rise were equal and set at a constant of 0.5°C/s. 5.2.3.2. Dyspnea induction Dyspnea was induced by means of a rebreathing paradigm (Bogaerts et al., 2012; Read, 1967; Walentynowicz et al., 2015). The rebreathing system consisted of an airtight plastic bag containing 5 L of a decompressed air mixture of 5% CO2 in 95% O2 fixed to a three-way valve, which allowed to switch between room air and the bag. During the trial, participants wore a nose clip and (re)breathed through the mouthpiece, which was attached to the three-way valve through an antibacterial filter, a pneumotachograph (Fleisch no. 2, Lausanne, Switzerland) measuring airflow and a wide vinyl tube. The exhaled air was sampled via a small tube close to the mouthpiece and sent to an infrared CO2 monitor (CAPNOGARD ETCO2 Monitor; Novametrix Medical Systems Inc; Wallingford, CT) to determine fractional end-tidal concentration of CO2 (FetCO2). All data were sampled at 20 Hz, stored on a computer and analyzed offline to extract the respiratory parameters: minute ventilation (MV) in L/min, FetCO2 in %, and respiratory rate (RR) in breaths/min. Due to equipment failure, the FetCO2 data of 7 participants (5 low HSR/2 high HSR) were not recorded. The data were divided into 3 induction phases (baseline: 0-60 s, induction: 60210 s, recovery: 210-360 s) and the mean values of FetCO2, MV, and RR were calculated for each of those phases.

90


5.2.3.3. Physiological measures Heart rate (HR) was measured during both symptom induction trials. Three 8 mm Ag/AgCl electrodes attached to participants’ chest, two on left and right shoulders and one below the left lower rib, recorded cardiac activity. The signal was sent to a Coulbourn V7504 bioamplifier (Allentown, PA) and sampled at 1000 Hz. The signal was manually processed offline and converted to beats per minute (BPM) with the PSPHA software (De Clercq, Verschuere, De Vlieger, & Crombez, 2006). Symptom induction trials were divided into 3 induction phases: baseline, induction, and recovery (pain induction: 0-60 s, 60-125 s, 125-185 s; dyspnea induction: 0-60 s, 60-210 s, 210-360 s), and mean HR was calculated for each of those phases. 5.2.4. Manipulation of processing focus The sensory-perceptual and affective-motivational PF instructions were given to participants prior to each symptom induction. The instructions started with the detailed explanation of the differences between two types of processing focus followed by processingand trial-specific instructions. In the sensory PF condition, participants were asked to: “Focus on all bodily sensations and on how they change during this somatic experience, no matter how weak or strong they are. Observe how they change from moment to moment in the course of time. Try to notice them in a neutral manner (like an objective, neutral observer) without assessing whether they are good or bad.” In the affective PF condition, the following instructions were provided: “Focus on all emotions and feelings that you experience and on how they change during this somatic experience, no matter how weak or strong they are. Try to notice how this stimulus makes you feel, how you react to these experiences and how they change.” The full instructions are included in Appendix D. 5.2.5. Procedure Selected participants were invited to a two-part study investigating mechanisms of subjective well-being. The study consisted of two laboratory sessions, each followed by retrospective ratings. The only difference between the sessions was the within-subject manipulation of processing focus during the symptom induction trials (all individuals did one session with the sensory PF and one with the affective PF manipulation). The time between the two parts of the study was never shorter than 4 weeks (M = 7.23, SD = 4.21, range = 4 – 24.14 weeks). The order of PF manipulation and the order of symptom induction trials were fully counterbalanced across all subjects.

91

Chapter 5

The first session started with participants signing an informed consent and completing the state PANAS. The experimenter explained the procedure and attached the electrodes for HR measurement. Two symptom induction trials were administered in counterbalanced order. In the pain induction trial, cold pain was delivered to the dominant hand via a thermode attached to participants’ forearm with a Velcro strap. To assure that the baseline skin temperature was standardized across participants and trials, the trial started with a 60-s baseline at the temperature of 32°C, followed by the cold phase (65 s), during which the temperature decreased to a 2°C plateau at the decline rate of 0.5°C/s. After a 5-s plateau, the temperature returned to baseline at the rise rate of 0.5°C/s (60 s). Dyspnea was induced with the rebreathing paradigm, consisting of the baseline phase (60 s of room-air breathing), the rebreathing phase (150 s), during which participants where breathing into the rebreathing bag, and the recovery phase (150 s of room-air breathing). Participants were informed that the trials would start with a baseline phase during which only physiological recording took place, and that the symptom induction would only start once the instructions (focus on sensations/emotions) appeared on the screen. Before each symptom induction, participants were instructed to focus on either physical sensations or emotional responses during the trial. The PF manipulation instructions were presented on the computer screen and verbally repeated by the experimenter. All subjects were asked to paraphrase the instructions and confirm that they understood them. After each trial, participants completed retrospective symptom ratings (immediate), state PANAS, SAM, and the threat question. At the end of the laboratory session, retrospective symptom ratings (delayed) for both trials were collected. Participants completed the trait questionnaires (CDS, PANAS) online after the session. Two weeks after the laboratory session, follow-up questionnaires were completed online and participants were invited for the second session. 5.2.6. Data analysis 5.2.6.1. Physiological responses To examine how the physiological responses differed between the groups and PF manipulations, separate repeated-measures analyses of variance (RM ANOVAs) were conducted on FetCO2, MV, RR, and HR during the dyspnea induction trial and on HR during both pain and dyspnea induction trials as dependent variables. Induction Phase (baseline/induction/recovery) and PF Manipulation (sensory/affective) were included as

92


within-subject variables, while Group (high/low HSR) and Order of PF Manipulation (sensory-affective/affective-sensory) served as between-subject factors. Group and PF manipulation differences in physiological reactivity were examined with planned contrasts, C1: baseline versus induction phase, C2: baseline versus recovery phase, C3: induction versus recovery phase. 5.2.6.2. State affect and symptom measures Group and PF manipulation effects on state variables, that is, NA, symptom, SAM and threat ratings, were investigated with RM ANOVAs with PF Manipulation and Induction Trial (pain/dyspnea; baseline/pain/dyspnea in state NA analysis) as within-subject factors and Group and Order of PF Manipulation as between-subject factors. 5.2.6.3. Retrospective symptom ratings Pain and dyspnea ratings (average, peak and end) were analyzed in a 2 × 3 × 2 × 2 mixed-factorial ANOVA, with PF Manipulation and Time (immediate/delayed/follow-up) as within-subject variables and Group and Order of PF Manipulation as between-subject variables. Planned contrast examined specific Time, Group and PF Manipulation effects (C4: ratings during the experimental session (immediate, delayed) vs. follow-up ratings). Greenhouse–Geisser corrections were applied when the sphericity assumption was violated. All analyses were conducted with SPSS 23.0. 5.3. Results 5.3.1. Sample characteristics High HSR reported more habitual symptoms (high: M = 112.81, SE = 1.97; low: M = 64.63, SE = 1.71; t(43) = 18.54, p < .001) and higher trait NA levels (high: M = 27.10, SE = 1.64; low: M = 16.46, SE = 1.04; t(43) = 5.61, p < .001) than low HSR. 5.3.2. Physiological responses The respiratory parameters are shown in Figure 5-1. The levels of all respiratory parameters were higher during the induction (rebreathing) phase than during the baseline phase: FetCO2, C1 for Induction Phase, F(1, 34) = 1831.27, p < .001, !!! = .98, Induction Phase, F(1.70, 57.74) = 892.06, p < .001, !!! = .96; MV, C1 for Induction Phase, F(1, 41) = 132.73, p < .001, !!! = .76, Induction Phase, F(2, 82) = 86.72, p < .001, !!! = .68, RR, C1 for Induction Phase, F(1, 41) = 26.61, p < .001, !!! = .39, Induction Phase, F(1.30,

93

Chapter 5

53.44) = 31.12, p < .001, !!! = .43. In the recovery phase, FetCO2 decreased to the level lower than during the baseline phase, C2 for Induction Phase, F(1, 34) = 40.86, p < .001, !!! = .55, while MV remained at a similar level as during the induction phase, C3 for Induction Phase, F(1, 41) = .01, p = .94, !!! = .00, and RR was higher compared to the induction phase, C3 for Induction Phase, F(1, 41) = 20.95, p < .001, !!! = .34.10 The pattern of change for HR in the dyspnea induction was similar to the one observed in the RR, F(1.43, 58.48) = 23.05, p < .001, !!! = .36, while in the pain induction the HR was marginally higher during the induction and returned to baseline level in the recovery phase, C1 for Induction Phase, F(1, 41) = 6.04, p = .018, !!! = .13, Induction Phase, F(2, 82) = 2.73, p = .072, !!! = .06 (see Figure E-1 in Appendix E). Group-related differences were found for FetCO2 and MV in all phases of the dyspnea induction, with lower FetCO2, F(1, 34) = 7.87, p = .008, !!! = .19 and higher MV, F(1, 41) = 4.91, p = .032, !!! = .11 observed among high HSR compared to low HSR. Moreover, FetCO2 in high HSR decreased from the induction to the recovery phase to a larger extent than in low HSR, C3 for Induction Phase × Group, F(1, 34) = 5.10, p = .031, !!! = .13, interaction Induction Phase × Group, F(1.70, 57.74) = 3.32, p = .051, !!! = .09. Moreover, high HSR had higher HR compared to low HSR during the pain induction, F(1, 41) = 6.16, p = .017, !!! = .13, and marginally higher during the dyspnea induction, F(1, 41) = 3.51, p = .067, !!! = .08. Induction of PF affected only RR, with the affective PF resulting in higher RR, F(1, 41) = 4.22, p = .046, !!! = .09, compared with the sensory PF (Maff = 15.20, SEaff = .61; Msens = 14.55, SEsens = .61). 5.3.3. State symptoms and affective responses As presented in Table 5-1, high HSR reported higher state NA and state symptoms (checklist) than low HSR; state NA: F(1, 41) = 8.16, p = .007, !!! = .17, state symptoms: F(1, 41) = 9.02, p = .005, !!! = .18. They also tended to perceive the symptom induction trials as more threatening, F(1, 41) = 3.89, p = .055, !!! = .09. However, no group differences were found with respect to valence, arousal and control ratings of the induction trials.

10

Based on the analyses performed on the data averaged per phase, it may seem that no recovery took place for MV and RR. However, a visual inspection of the respiratory parameters displayed on Figure 5-1 shows that these averages represent a gradual decrease in MV and RR during the recovery phase (which emerges significantly when including repeated measures during recovery).

94


Figure 5-1. Mean values and standard errors of fractional end-tidal concentration of CO2 (FetCO2, %), minute ventilation (l/min), and respiratory rate (breaths/min) for high and low HSR during dyspnea induction. Data from two sessions are aggregated. To obtain a detailed picture of change over time, data are displayed per 30-s intervals; baseline phase: 0-60 s, induction phase: 60-210 s; recovery phase: 210-360 s.

95

.22 .16 .58 .23 .09 .17 .32 .14 .18 .10 .35

11.70 (1,41) 7.75 (1,41) 57.58 (2,82) 12.49 (2,82) 4.14 (2,82) 8.16 (1,41) 19.38 (1,41) 6.40 (1,41) 9.02 (1,41) 4.63 (1,41) 21.79 (1,41)

7.22 (1,41) .15 59.85 (1,41) .59 3.89 (1,41) .09

9.21 (1,41) .18 37.20 (1,41) .48 29.51 (1,41) .42

!!!

F(df)

SAM (valence, arousal, and control) and threat ratings for each processing

Affective PF Significant effects BL Pain Dyspnea 12.38 18.50 (6.06) 22.13 (6.87) PF** (1.93) PF×Order** 16.67 19.19 (5.47) 25.24 (8.00) IND*** (5.89) PF×IND*** PF×IND×Order* Group** State Low HSR 26.79 (5.63) 28.79 (8.80) 26.21 (7.71) 30.58 (8.54) IND*** symptoms PF×IND×Order* High HSR 32.10 (6.88) 35.67 (8.69) 29.95 (7.61) 35.38 (7.72) Group** Group×Order* Valence (1-9) Low HSR 4.63 (1.61) 3.50 (1.22) 4.63 (1.84) 3.17 (1.17) IND*** High HSR 4.81 (1.63) 3.76 (1.84) 4.48 (1.75) 3.29 (1.15) Arousal (1-9) Low HSR 3.58 (1.86) 5.54 (1.79) 4.58 (1.86) 6.25 (1.36) PF** High HSR 4.38 (2.11) 6.05 (1.60) 5.19 (1.72) 6.14 (1.96) IND*** Control (1-9) Low HSR 4.71 (2.29) 3.33 (1.95) 4.08 (1.84) 3.04 (1.90) IND*** High HSR 4.67 (2.39) 2.86 (2.03) 4.29 (2.22) 2.81 (1.50) Threat (1-9) Low HSR 2.75 (1.62) 4.88 (2.42) 3.54 (1.98) 5.42 (2.08) PF* High HSR 3.76 (1.95) 5.62 (1.47) 3.95 (1.99) 6.43 (1.50) IND*** Groupa Note. BL = baseline; HSR = habitual symptom reporting; IND = induction trial; PF = processing focus. a p < .10, *p < .01, **p < .05, ***p < .001.

Table 5-1. Means and standard deviations for state NA, symptoms, focus and significant effects of repeated-measures ANOVAs. Measure Group Sensory PF BL Pain Dyspnea State NA Low HSR 13.54 14.46 (4.32) 17.38 (5.58) (4.44) High HSR 17.48 17.76 (5.77) 22.05 (5.85) (4.92)


Compared to pain induction, dyspnea induction trials were evaluated as more unpleasant, F(1, 41) = 21.79, p < .001, !!! = .35, arousing, F(1, 41) = 37.19, p < .001, !!! = .48, and threatening, F(1, 41) = 59.85, p < .001, !!! = .59. During those trials, participants experienced also higher state NA, F(2, 82) = 57.58, p < .001, !!! = .58, more symptoms, F(1, 41) = 19.38, p < .001, !!! = .32, and lower sense of control, F(1, 41) = 29.51, p < .001, !!! = .42. With regard to the processing focus manipulation, affective PF was more arousing and threatening for the participants than the sensory PF; arousal, F(1, 41) = 9.21, p = .004, !!! = .18; threat, F(1, 41) = 7.22, p = .01, !!! = .15. It also resulted in higher state NA during the symptom induction trials, F(1, 41) = 11.70, p = .001, !!! = .22. However, state NA ratings were influenced not only by the type of PF manipulation, but also by the order in which they were presented (Figure 5-2). Reduction of state NA from the first to the second session was observed when the affective PF manipulation was first, but this did not occur when the sensory PF manipulation came first (PF Manipulation × Trial × Order: F(2, 82) = 4.14, p = .019, !!! = .09).

Figure 5-2. Mean values and standard errors of state negative affect (NA) for sensory and affective processing focus (PF) during baseline, pain induction and dyspnea induction trial as a function of PF manipulation order.

97

Chapter 5

5.3.4. Retrospective average symptom ratings Groups did not differ in the average retrospective symptom reports, neither for pain, F(1, 41) = 1.37, p = .25, !!! = .03, nor for dyspnea ratings, F(1, 41) = 1.61, p = .21, !!! = .04 (see Figure E-2 in Appendix E). Also no group differences were found for symptom intensity at the peak and the end (Figure E-2 in Appendix E). In general, no main effects of PF manipulations on pain and dyspnea ratings were found. However, the average dyspnea ratings increased over the course of two weeks in the affective PF as compared to the sensory PF condition, C4 for PF Manipulation × Time, F(1, 41) = 4.56, p = .039, !!! = .10; interaction PF Manipulation × Time, F(2, 69.99) = 3.15, p = .057, !!! = .07 (Figure 5-3, right panel). Such interaction was not observed for the retrospective average pain ratings (Figure 5-3, left panel), which in contrast showed a small tendency to decrease over time, F(1.50, 61.35) = 2.53, p = .10, !!! = .06.11 Detailed results concerning the retrospective peak and end ratings are provided in Appendix E.

Figure 5-3. Mean retrospective average pain ratings (0-100, left) and dyspnea ratings (0-100, right) for sensory and affective processing focus manipulation. Whiskers denote standard errors.

11

The order of symptom indiction trials (pain-dyspnea/dyspnea-pain) on the retrospective symptom ratings was additionally investigated by including the Order of Induction Trials as a between-subject factor to the RM ANOVAs outlined above. However, as the order of trials was counterbalanced across the sessions and the subjects, its effects had to be explored separately for the sensory and the affective sessions. The abovementioned results did not change, and no interactions between the Order of Induction Trials and the Group variables were found.

98


5.4. Discussion The current study examined whether focusing on sensory compared to affective aspects of a somatic experience could be one of the mechanisms explaining previously reported distortions in retrospective symptom reporting. To this end, participants high and low on habitual symptom reporting were instructed to focus on either the sensory-perceptual aspects (sensory PF) or on feelings and emotions (affective PF) evoked by experimentally induced pain and dyspnea. The type of PF was manipulated within subject, and its effect on symptom ratings and affective responses was measured immediately, after a short delay and after 2 weeks. Our study showed that an affective PF resulted in stronger affective responses during the somatic episode followed by an increase in retrospective symptom reporting over time. However, this effect was only observed for the dyspnea episode, which was perceived as more unpleasant, arousing, and threatening than the pain induction. This shows that the impact of PF manipulation depends on the aversiveness of the stimulus. The manipulation checks confirmed our predictions regarding symptom induction and experimental group differences. First, not only was the dyspnea induction perceived as more unpleasant, arousing, and threatening, it also resulted in a lower sense of control and higher state NA than the pain induction. This supported previous findings showing that experimentally induced dyspnea is overall more anxiogenic than pain induced by a cold stimulus (Walentynowicz et al., 2015). Second, high HSR reported higher state NA and symptoms ratings than low HSR in response to the symptom inductions, confirming that high HSR react more strongly to experimentally induced symptoms and experience them as more aversive (Bogaerts et al., 2008; Walentynowicz et al., 2015). As in a previous study (Walentynowicz et al., 2015), the groups did not differ in intensity of their physiological reactions to the symptom induction. However, the high HSR group differed from the low HSR with regard to the overall levels of FetCO2, minute ventilation, and heart rate, which was possibly related to the general distress caused by the experimental context. In general, these findings are consistent with other research showing that NA predisposes individuals to interpret somatic stimuli as threatening (Stegen et al., 2000). Interestingly, whereas in studies that do not manipulate the processing style it is typically found that high HSR report more symptoms and higher state NA than low HSR groups (Bogaerts, Janssens, et al., 2010; Constantinou et al., 2013; Walentynowicz et al., 2015), such overall difference in retrospective symptom ratings and affective responses (as measured by SAM) was not found in the present study. As the focus of participants’ attention

99

Chapter 5

was manipulated by the PF instructions, the degree to which they could use their “default processing focus” may have been reduced, resulting in no overall differences between the groups in symptom ratings. Even though the impact of a different processing focus on retrospective ratings did not appear immediately on the day of the laboratory session, it did show up after two weeks, and only for the dyspnea induction. While ratings in the sensory PF remained stable over time, the affective PF resulted in the expected increase in symptom ratings after two weeks. However, this effect was only present in retrospective dyspnea ratings, whereas the pain ratings did not change. Several mechanisms may play a role in this selective effect of PF. First, the relative effect of an affective component of a somatic experience on symptom memory tends to increase over time. This suggests that the affective responses to symptoms and anxiety-related psychological characteristics of individuals influence long-term symptom memory more than the sensory-perceptual details of somatic experience (Gedney et al., 2003; Gedney & Logan, 2004; Kent, 1985). Second, memory for sensory details tends to decay at a faster rate (Reyna & Brainerd, 1998), again giving more room for the influence of the affective features. In the present study, focusing on the affective component of experience enhanced the self-reported symptoms and physiological reactions to the stimuli, as indicated by higher arousal and threat ratings, as well as higher respiratory rate in dyspnea trials. Thus, it seems that attending to emotions evoked by unpleasant and arousing dyspnea induction led to the better perception and encoding of negative emotional memory traces at the expense of sensory details (Leventhal et al., 1979). This, together with a greater impact of the affective component in long-term symptom memory, has likely caused elevated retrospective dyspnea ratings. Finally, an interesting pattern of results emerged with regard to the order effects. It was shown that state NA did not differ between the two sessions when the affective PF was introduced on the second session. In contrast, negative affect was reduced when the sensory PF was induced on a subsequent session. As the emotions experienced during the affective PF sessions were more aversive than during the sensory PF, this finding may indicate that habituation to unpleasant bodily sensations is impaired when the focus is directed to their aversive or threatening features. This may have important consequences for exposure therapy: Our results suggest that the effectiveness of the therapy may be hindered when the affectivemotivational processing focus dominates the sensory one. The present study makes several noteworthy contributions to the growing symptom memory research. First, we investigated the effect of processing focus not only on how symptoms are perceived, but also on how they are remembered by inclusion of a longer time 100


frame for retrieval. Second, previous studies examining the factors influencing retrospective symptom reporting have assessed memory accuracy (Houtveen & Oei, 2007; Linton & Melin, 1982; Walentynowicz et al., 2015), but did not manipulate the way symptoms were encoded in the first place. Our PF manipulation impacted not only how symptoms were experienced, but also encoded in memory. This study was limited by the absence of a measure assessing the amount of actual focus on sensory or affective aspects of experience, which was not possible due to a within-subject study design. Nevertheless, the reported effects of affective PF on emotional responses suggest that participants followed the instructions. Another limiting factor is the lack of a default condition without manipulation. This means that we cannot establish whether the symptom ratings increased or decreased after PF manipulation compared to ratings under a natural, non-manipulated condition. To conclude, this work extends our knowledge of cognitive and affective processes influencing symptom reporting by showing that attention to different components of an aversive somatic experience affects not only emotional distress but also increasingly biases retrospective symptom memory with longer delays. This may explain why individuals with overreactive affective response system, such as habitual symptom reporters or patients with SSD, overreport experienced symptoms when retrospectively assessed. 5.5. Acknowledgments The authors want to thank Hannie Tay for her assistance in data collection. The study was funded by Grant OT/10/027 from the University Research Council of the University of Leuven.

101

Chapter 5

102

Recalling somatic memories: The effect of processing focus during retrospective symptom reporting

6Chapter 6 Recalling somatic memories: The effect of processing focus during retrospective symptom reporting

Abstract Retrospective symptom ratings are subject to various biasing factors operating at different stages of memory processing. While attention was often given to encoding-related biases, less is known about factors affecting self-reported symptom at later memory stages. As a result, the present study aimed to explore whether guiding the attention to different aspects of somatic experience during memory retrieval has an impact on retrospective symptom reporting. Ninety female students varying in habitual symptom reporting (HSR) participated in two laboratory sessions (T0 and T1) followed by an online follow-up assessment (T2). At T0, cold pain and dyspnea were experimentally induced. Two weeks later, at T1, processing focus (sensory/affective/undirected PF) was manipulated between-subject before memory retrieval. Pain/dyspnea ratings and affective responses to induced bodily stimuli were collected on both laboratory sessions (T0 and T1) and after 2 weeks (T2). Dyspnea induction resulted in more symptoms and more intense negative affective responses than pain induction (ps < .001). Retrospective ratings of dyspnea were higher at T1 than at T0 (p = .003). The observed changes in retrospective ratings were affected by neither PF manipulation nor HSR. No changes in retrospective ratings were found for pain. Except for an increase in threat ratings at T1 recall, the indices of affective responses were not overreported at T1 or T2. The findings show that a local temporary manipulation of PF after the memory traces were stabilized seems to have a limited effect on the recall biases.

Based on: Walentynowicz, M., Raes, F., Van Diest, I., & Van den Bergh, O. (in preparation). Recalling somatic memories: The effect of processing focus during retrospective symptom reporting. 103

Chapter 6

6.1. Introduction Retrospective self-reported symptoms are important in both non-clinical and clinical settings and may have great consequences for future decisions on help-seeking, diagnosis, and treatment. Retrospective symptom ratings are based on memory processes that may importantly influence and modify how a past somatic event is reported. Due to the (re)constructive nature of these memory processes, retrospective symptom reports can be influenced and biased by a number of factors present at the various stages of memory processing (Broderick et al., 2008; Gedney & Logan, 2004; Giske et al., 2010). Much attention has been given to factors exerting effects at early stages of memory formation, that is, encoding and early phases of consolidation. These included intensity and variability of somatic stimulus (Feine et al., 1998; Giske et al., 2010; Lefebvre & Keefe, 2002), cognitive heuristics, such as the peak-end effect (e.g., Bogaerts et al., 2012; Chajut et al., 2014; Kahneman et al., 1993) and accompanying negative affective states (e.g., Eli et al., 2000; Gedney et al., 2003; Gedney & Logan, 2004; Noel et al., 2012b; Walentynowicz et al., 2015). A related factor playing a crucial role at this stage is how individuals process the somatic information during the initial encounter with the stimulus. Findings from a number of experimental studies show that focusing on sensory aspects of the experience in a nonemotional way results in lower symptom ratings than focusing on emotional responses induced by the somatic experience (Ahles et al., 1983; Haythornthwaite et al., 2001; Keogh et al., 2000; Leventhal et al., 1979). Conversely, focusing on the affective aspects of a somatic experience was associated with elevated symptom ratings, especially in participants frequently reporting symptoms in daily life (Bogaerts et al., 2008) and characterized by anxiety-related personality traits, such as pain catastrophizing (Michael & Burns, 2004). In a previous study (Walentynowicz, Raes, Van Diest, & Van den Bergh, 2016), we manipulated the processing focus (PF) at the encoding stage of an experimentally induced somatic experience to either sensory-perceptual or affective-motivational aspects and assessed retrospective symptom reports. The affective PF led to stronger negative affective responses. Moreover, symptom ratings increased over a 2-week period, but only for the most distressing somatic experience (dyspnea) in the affective PF condition. The differential effects of PF during encoding on retrospective symptom reports can be explained within the dual-process perspective on symptom reports (e.g., Craig, 2003; Leventhal & Everhart, 1979; Price, 2000). According to this view, the global evaluation of bodily experiences relies on the parallel processing of two aspects: sensory-perceptual and

104


affective-motivational. The focus on affective aspects could (a) increase the salience of the stimulus-related distress, and (b) interfere with detailed encoding of the perceptual features. Consequently, a pronounced negative affective component may influence symptom reporting in the direction of overreporting. From this perspective, biased symptom reporting could be expected particularly among individuals characterized by both strong and negative affective responses to bodily stimuli and reduced specificity of sensory processing of somatic information. This combination of characteristics is distinctive for individuals with a tendency to frequently report somatic complaints in daily life, known as habitual symptom reporting (HSR). First, HSR shows a robust relationship with trait negative affectivity (Bogaerts et al., 2015) and intense negative responses to somatic experiences (Walentynowicz et al., 2015; Wan et al., 2012). Second, a less detailed sensory processing is suggested by poorer correspondence between perceived symptoms and relevant physiological parameters (Bogaerts et al., 2008; Bogaerts, Van Eylen, et al., 2010) or decreased differentiation between various bodily sensations (Petersen, von Leupoldt, et al., 2015). As a result, less specific sensory processing combined with a more profound impact of negative affective responses may result in biased retrospective symptom reporting in HSR. While substantial evidence exists on encoding-related biases in retrospective symptom reports, relatively less is known about the factors which could affect the symptom ratings after they are encoded and consolidated. For example, some studies demonstrated an effect of concurrent symptom intensity, with higher levels causing elevated retrospective symptom reports (e.g., Eich et al., 1985; Meek et al., 2001), while Bąbel (2015) showed that recalled anxiety is associated with recalled pain ratings. Also the retention interval seems to influence the recalled symptom ratings, with gradual increase in symptom overreporting observed with longer time frames (Broderick et al., 2008; Houtveen & Oei, 2007). The present study aimed to investigate whether retrospective reports of experienced symptoms can be also affected at a later stage of memory processing, after the memory was consolidated. More specifically, we examined whether guiding the processing focus to different aspects of a past somatic experience at the retrieval stage could influence the way symptoms are recalled resulting in changes in retrospective symptom reporting. To examine the effects of PF during retrieval, we used the same methodology as in Walentynowicz, Raes, et al. (2016; Chapter 5). Two types of symptoms, pain and dyspnea, were experimentally induced in participants with varying levels of habitual symptom reporting (T0). Two weeks later (T1), before the retrieval of memories related to the experiences at T0, participants’ memory recall was directed towards one of two processing foci in a between-subject design. 105

Chapter 6

Participants were cued to concentrate either on the sensory-perceptual features of the recollections (sensory PF) or on the emotions elicited by the symptom induction at T0 (affective PF). An undirected PF, with no explicit recall instructions, was included as a control condition. The ratings of the symptom induction trials, that is, pain/dyspnea ratings and affective responses, were collected on both laboratory sessions and after 2 weeks (T2). The dual-process perspective outlined above assumes that bias in self-reported symptoms may result from the combination of strong negative affective component and reduced perceptual details of a somatic experience. Accordingly, an increase in recalled ratings of symptoms and affect from T0 to T1/T2 (overreporting) was expected to appear in the conditions increasing the salience of the affective component, especially when this component represents negative and aversive feelings caused by stimulus-related distress. Consequently, overreporting was expected (a) in the dyspnea induction compared to pain induction, because of its elevated aversiveness, (b) in the affective PF condition, compared to the other conditions, due to an increased salience of the affective component, and (c) more in persons with elevated levels of HSR, whose responses to induced stimuli tend to be strong and negative. The hypotheses will be tested separately for pain and dyspnea ratings, with critical tests including main effects of PF and HSR as well as their interaction. 6.2. Methods The methods used in this study are largely the same as in Chapter 5. The present description summarizes these methods and highlights the specific aspects. 6.2.1. Participants Ninety healthy female participants (Mage = 20.24; SDage = 2.59; range 18-28 years) were selected for a two-part study, based on their habitual symptom reporting levels as measured by the Checklist for Symptoms in Daily Life (CSD; Walentynowicz, Witthöft, et al., 2016). To assure the full range of HSR in this study, four groups were created for selection purposes. The groups were based on the quartiles of the total CSD score (range: 39195; cut-offs: ≤ 75, n = 27; 76 - 89, n = 18; 90 - 99, n = 18; ≥ 100, n = 27), obtained from the initial collective testing data (N = 371). Even though the planned group size was not possible due to unbalanced response rate, the scores of the final sample were normally distributed (range = 51-144; M = 86.80; SD = 19.44). Exclusion criteria were a self-reported chronic medical condition or a psychiatric disorder, a current acute illness, recent pain or a condition affecting the non-dominant arm, 106


and pregnancy. The experimental protocol was approved by the Multidisciplinary Ethics Committee of the Faculty of Psychology and Educational Sciences of the University of Leuven. Participant remuneration included course credits or 14 euros. 6.2.2. Measures 6.2.2.1. Habitual symptom reporting The frequency of symptoms in the past year was measured with the Checklist for Symptoms in Daily Life (CSD; Walentynowicz, Witthöft, et al., 2016) on a 5-point Likert scale (never, seldom, sometimes, often, very often). Cronbach’s alpha in the present sample was .94. 6.2.2.2. Negative affectivity The Positive and Negative Affect Schedule (PANAS; Engelen et al., 2006; Watson et al., 1988) was used to assess both trait and state negative affectivity (NA). Participants rated on a 5-point Likert scale ranging from not at all to very much the extent they feel 20 positive and negative emotions in general (trait) or now (state). Cronbach’s alphas for both state and trait versions ranged from .69 to .89. 6.2.2.3. State symptom checklist The occurrence of 14 symptoms at baseline and during the symptom induction trials was assessed on a 5-point scale ranging from 1 (not at all) to 5 (very much). Cronbach’s alphas ranged from .59 to .87. 6.2.2.4. Affect and threat ratings At baseline and after each induction trial, participants used a computerized nine-point version of Self-Assessment Manikin (SAM; Bradley & Lang, 1994) to rate their affective state. Three pictorial scales were used: valence (1 = unpleasant, 9 = pleasant), arousal (1 = very calm, 9 = very arousing), and control (1 = low control, 9 = high control). The threat value of each trial was assessed after each induction with a single item numerical rating scale (NRS; 1 = not threatening at all, 9 = very threatening). 6.2.2.5. Processing focus ratings Three NRSs were used to measure the extent to which the participants focused their attention on different aspects during the symptom induction trials. The scales included “focus on bodily sensations” (Focus-sens), “focus on emotions” (Focus-emo), and “focus on other

107

Chapter 6

aspects”, such as the surrounding, the equipment, or other thoughts (Focus-other). The 9-point scales ranged from 1 (not focused at all) to 9 (very focused). 6.2.2.6. Concurrent pain/dyspnea ratings Concurrent pain/dyspnea was rated continuously with a scroll wheel during each symptom induction on a 0–100 computerized scale (sampling every second). The scale was displayed as a vertical bar in the middle of a screen, with verbal descriptions of symptom levels based on a modified Borg scale (Borg, 1982) on its right side: none (0), very slight (10), slight (20), moderate (30), fairly severe (40), severe (50), very severe (60), very severe (70), very severe (80), very, very severe (90), intolerable (100). 6.2.2.7. Retrospective pain/dyspnea ratings Retrospective ratings of pain/dyspnea experienced during each trial were collected at three moments: immediately after each trial (T0-immediate rating), at the end of the laboratory session (T0-delayed rating), after PF manipulation (T1 rating), and at a follow-up (T2 rating). The average symptom level was assessed with visual analog scales (VAS; 10 cm) ranging from 0 (no pain/dyspnea) to 100 (unbearable pain/dyspnea). 6.2.3. Manipulation of processing focus The processing focus was induced between-subjects before the memory retrieval at T1. All instructions were displayed on a computer screen. For the sensory-perceptual and the affective-motivational PF groups, instructions included a comprehensive explanation of the two types of PF (see Appendix D). Afterwards, participants’ attention was directed to either sensory or affective aspects of retrieved memories by processing- and symptom inductionspecific instructions. In the undirected PF group, no explicit attempt was made to direct participants’ focus during symptom memory recall. Exact instructions are presented below: 6.2.3.1. Sensory-perceptual PF Hopefully it is clear that you can focus on the sensations in your body in various ways. During this session we will ask you to think back about our first lab session. Recall the trial with cold stimulus/breathing trial and try to pay attention (like an objective, neutral observer) to the different sensations that occurred when the cold stimulus was delivered to the inside of your arm with the thermode/ you were breathing through the mouthpiece. We want to ask you now to describe all sensations in detail – how they changed from moment to moment in the course of time during this somatic experience, no matter how weak or strong they were.

108


6.2.3.2. Affective-motivational PF Hopefully it is clear that you can focus on the sensations in your body in various ways. During this session we will ask you to think back about our first lab session. Recall the trial with cold stimulus/breathing trial and try to pay attention to the different emotions that occurred when the cold stimulus was delivered to the inside of your arm with the thermode/ you were breathing through the mouthpiece. We want to ask you now to describe all emotions and feelings in detail – how they changed from moment to moment in the course of time during this somatic experience, no matter how weak or strong they were. 6.2.3.3. Undirected PF During this session we will ask you to think back about our first lab session. Recall the trial with cold stimulus/breathing trial and try to pay attention to the different experiences that you had when the cold stimulus was delivered to the inside of your arm with the thermode/ you were breathing through the mouthpiece. We want to ask you now to describe in detail all memories of what you experienced during that trial. 6.2.4. Procedure The study consisted of three sessions: a symptom induction lab session (T0), a memory retrieval laboratory session (T1), and a memory retrieval online session (T2), each session separated in time by a period of 2 weeks (see Figure 6-1 for the schematic overview of the procedure). During the first laboratory session (T0), the participants were informed about the procedure, signed an informed consent, and completed the state questionnaires (T0 baseline NA, symptoms, SAM). Two symptom induction trials followed in an order that was counterbalanced across participants. The cold pain induction trial consisted of a 30-s baseline (32°C), a 65-s cold phase (60-s decrease of temperature to 2°C with 5-s plateau at 2°C), and a 60-s recovery phase, when the temperature returned to baseline of 32°C. The dyspnea induction trial started with a baseline phase (60 s of room-air breathing), followed by a rebreathing phase (150 s of rebreathing into a bag), and a recovery phase (150 s of room-air breathing). During all abovementioned phases, the participants were requested to rate their concurrently experienced pain/dyspnea. Both induction trials started with a 30-s baseline for physiological recordings (data not included) and participants were informed that the symptom induction trial would only start once the vertical rating scale for concurrent symptom ratings appears on the screen. Immediately after each trial, participants completed retrospective pain/dyspnea ratings (T0-immediate ratings) and state measures (T0 NA, symptoms, SAM, threat and focus ratings). At the end of the laboratory session, retrospective pain/dyspnea 109

Chapter 6

Figure 6-1. Schematic overview of the study protocol. The order of symptom induction trials was counterbalanced across participants (not shown). Note: PF = processing focus. ratings (T0-delayed ratings) were collected. Trait questionnaires (CDS, PANAS) were filled out online later that day. In the second laboratory session (T1), the random assignment of participants to the PF manipulation groups (sensory PF, n = 31; affective PF, n = 31; and undirected PF, n = 28) took place. Participants were requested to recall both symptom induction trials from the first session (T0). The following procedure was used to retrieve memories of both pain and dyspnea induction trials: First, the PF manipulation instructions were presented on a computer screen and verbally repeated by the experimenter. All subjects were asked to paraphrase the instructions and confirm that they understood them. They were given 10 minutes to describe their memories of the symptom induction trial from T0 in a written form. Afterwards,

110


participants were requested to rate how they experienced symptom inductions at T0 using the same rating scale formats as during T0. Those recalled ratings included retrospective pain/dyspnea (T1 ratings) and recalled state affect and symptom ratings (T1 NA, symptoms, SAM, threat ratings). Finally, as a manipulation check, participants rated the focus of their attention during the retrieval session (T1 focus ratings). Two weeks after the second laboratory session, the second memory retrieval session took place online (T2). Apart from the processing focus manipulation, which was not re-induced at T2, the procedure and the questionnaires were identical to T1. After completing the T2 follow-up questionnaires, participants were debriefed. For the description of apparatuses used for symptom induction, see Chapter 5. 6.2.5. Data analysis The analyses started with the investigation of symptom and affective responses to symptom inductions at T0, followed by the comparison of retrospective responses at T1 with the ratings at T0, and exploration of changes in retrospective responses over time between T2 and T1. In all repeated-measures analyses of covariance (RM ANCOVAs) the CSD score was included as a continuous predictor (after centering to the mean). 6.2.5.1. Responses at T0 All state variables (NA, symptoms, affect, threat ratings, and focus ratings) were analyzed with RM ANCOVAs with Induction Trial (baseline/pain/dyspnea12) as a withinsubject factor, and Order of Induction Trials as a between-subject factor. Concurrent symptom ratings during induction trials were averaged per trial. The differences between the initial retrospective ratings of average pain/dyspnea and averaged concurrent ratings were examined with RM ANCOVAs with Time (averaged concurrent/T0 immediate/T0 delayed) as a within-subject factor, and Order of Induction Trials as a between-subject factor. Additionally, Pearson product-moment correlation coefficients (r) were calculated separately for pain and dyspnea trials to examine the relationships between the processing focus ratings at T0 and the affect and symptom ratings at T0 (state NA, valence, arousal, threat, averaged concurrent, and immediate symptom ratings).

12

As the processing focus ratings and threat ratings were collected only after the induction trials, and not at the baseline, in the analyses concerning those variables the within-subject factor Induction Trial has 2 levels: pain/dyspnea.

111

Chapter 6

6.2.5.2. Retrospective ratings (T1 vs. T0, T2 vs. T1) To control for possible differences between the PF manipulation groups (sensory/affective/undirected) at T0, one-way ANOVAs were applied to compare the age, BMI, CSD, NA, and all state variables collected at T0 between the groups. RM ANOVAs on T1 focus ratings with Induction Trials as a within-subject factor and PF Manipulation as a between-subject factor served as a PF manipulation check. The effect of PF manipulation on the recall of affective responses and symptom ratings (checklist, average pain and dyspnea) was tested with RM ANCOVAs performed separately for the pain and the dyspnea induction trials, with Time (T0 immediate13/T1) as a within-subject factor and PF Manipulation as a between-subject factor. In the abovementioned RM ANCOVAs, the T0 immediate ratings (centered to the mean) were included as a second continuous predictor to account for the possible differences at T0. To examine whether the retrospective ratings further changed over time, the RM ANCOVAs for affective responses and symptom data were repeated with Time (T1 and T2 responses) as within-subject factor, and the T1 ratings as an additional continuous predictor. The Greenhouse–Geisser corrections were applied when the sphericity assumption was violated. All analyses were conducted with SPSS 23.0. 6.3. Results 6.3.1. Responses at T0 6.3.1.1. T0 State affect and symptoms RM ANCOVAs performed on state ratings at T0 as dependent variables demonstrated the differences between dyspnea and pain inductions in the expected directions (see Table 6-1 for means and standard deviations). That is, during the dyspnea induction trial, participants experienced higher state NA, F(1.87, 162.34) = 36.35, p < .001, !!! = .30, more symptoms (checklist), F(1.76, 152.90) = 146.96, p < .001, !!! = .63, and lower sense of control, F(2, 164) = 35.79, p < .001, !!! = .30 than during the pain induction. Moreover, compared to the pain induction, the dyspnea induction was evaluated as more unpleasant,

13

The analyses on pain/dyspnea ratings at T0 demonstrated that immediate and delayed ratings did not differ from each other. For this reason, only T0 immediate ratings, collected promptly after the end of the induction, were chosen as a comparison standard for subsequent analyses regarding the time effects (T1 vs. T0 comparison).

112


F(2, 174) = 140.60, p < .001, !!! = .62, arousing, F(2, 174) = 30.73, p < .001, !!! = .26, and threatening, F(1, 87) = 40.13, p < .001, !!! = .32. As expected, negative affective responses (state NA, unpleasantness, threat) as well as symptom ratings (via checklist) were more pronounced as HSR scores increased. More specifically, the HSR variable (as a continuous predictor) had a significant main effect on state symptoms (checklist), F(1, 87) = 22.38, p < .001, !!! = .21, state NA, F(1, 87) = 13.71, p < .001, !!! = .14, and valence, F(1, 87) = 4.72, p = .033, !!! = .05. However, the effects of HSR on affective responses depended on the induction trial: HSR was significantly related to higher state NA and unpleasantness during the baseline and the dyspnea induction, but not during the pain induction (Figure 6-2, left and middle panels). In detail, a significant interaction with Induction Trial specified the effect of HSR on state NA, F(1.87, 162.34) = 3.12, p = .050, !!! = .04, per trial: baseline, F(1, 87) = 15.33, p < .001, !!! = .15, dyspnea, F(1, 87) = 11.38, p = .001, !!! = .12; valence, F(2, 174) = 3.22, p = .044, !!! = .04, per trial: baseline, F(1, 87) = 13.23, p < .001, !!! = .13, dyspnea, F(1, 87) = 4.16, p = .045, !!! = .05. Similarly, HSR influenced threat ratings during the dyspnea induction, but not during the pain induction (Figure 6-2, right panel), F(1, 87) = 5.36, p = .023, !!! = .06, dyspnea: F(1, 87) = 7.28, p = .008, !!! = .08.

Figure 6-2. The interaction of Induction Trial with habitual symptom reporting (HSR) scores for state NA (left), valence (middle), and threat ratings (right) during induction trials at T0. The trial effects are plotted at the different levels of HSR scores (average, +1 SD, -1SD). 6.3.1.2. T0 Pain/dyspnea ratings RM ANCOVAs indicated that average pain was rated higher immediately after the induction (M = 33.72, SD = 18.54) and at the end of the experiment (M = 34.47, SD = 19.63) than concurrently (M = 23.64, SD = 13.60), F(2, 172) = 51.83, p < .001, !!! = .38. The immediate and delayed pain ratings did not significantly differ from each other, F(1, 113

Chapter 6

86) = .49, p = .49, !!! = .01. Similarly, the average dyspnea ratings given immediately after the trial (M = 39.10, SD = 21.00) and at the end of the experiment (M = 40.67, SD = 22.36) were higher compared to averaged concurrent ones (M = 24.71, SD = 15.57), F(1.64, 141.25) = 82.30, p < .001, !!! = .49. No significant difference was found between the immediate and the delayed dyspnea ratings, F(1, 86) = 2.44, p = .12, !!! = .03. No main effects or interactions with the HSR scores were found for pain and dyspnea ratings. Table 6-1. Means and standard deviations for the symptom induction trials and significant effects of repeated-measures ANCOVA for the state dependent measures and symptom ratings at T0 (N = 90). Measure Baseline Pain trial Dyspnea trial Significant effects a b c State NA (5-50) 15.48 (3.54) 16.84 (4.35) 19.94 (6.03) IndTrial*** HSR*** IndTrial × HSR* a b c State symptom (14-70) 18.18 (2.94) 25.40 (5.62) 29.67 (7.61) IndTrial*** HSR*** Valence (1-9) 6.40 (1.20)a 4.37 (1.72)b 3.41 (1.61)c IndTrial*** HSR* IndTrial × HSR* Arousal (1-9) 4.02 (1.62)a 5.16 (1.71)b 5.64 (2.06)b IndTrial*** a b c Control (1-9) 4.88 (1.68) 3.78 (1.85) 3.21 (1.98) IndTrial*** a b Threat (1-9) 3.70 (1.92) 5.14 (2.14) IndTrial*** IndTrial × HSR* T0 Focus Focus-sens (1-9) 6.96 (1.70)a 6.87 (1.52)a Focus-emo (1-9) 4.60 (1.94)a 5.21 (2.19)b IndTrial** a b Focus-other (1-9) 3.10 (2.19) 3.76 (2.33) IndTrial* IndTrial × HSR* Note. NA = negative affectivity, IndTrial = induction trial, HSR = habitual symptom reporting, Focussens = focus on sensations, Focus-emo = focus on emotions, Focus-other = focus on other aspects; Due to the software failure, n for the control scale was 85. Means with different superscripts are significantly different at p < .05. *p < .05, **p < .01, ***p < .001.

6.3.1.3. T0 Focus With regard to the processing focus ratings during the induction trials at T0 (see Table 6-1), RM ANCOVAs revealed that participants’ attention to sensations did not differ between the symptom induction trials. However, participants concentrated to a greater extent on emotions and on the other aspects of experience during dyspnea than during pain (Focusemotions: F(1, 86) = 8.81, p = .004, !!! = .09; Focus-other: F(1, 86) = 4.98, p = .030, !!! = .05). No main effects or interactions with HSR scores were found for the ratings of focus.

114


Moreover, the ratings of focus at T0 were shown to significantly correlate with a number of affective and symptom measures during both pain (Table 6-2) and dyspnea inductions (Table 6-3). Whereas focus on sensations during pain induction was not related to any affect or symptom ratings, during dyspnea induction it was positively associated with arousal, threat and immediate dyspnea ratings. On the other hand, focus on emotions was related to a number of variables, such as NA, threat, and symptom ratings in both pain and dyspnea inductions. Finally, the ratings of focus on other aspects showed an association with valence, arousal, and threat ratings only during pain induction in a direction opposite to the one of focus on emotions, thus higher focus on other aspects was related to lower unpleasantness, arousal, and threat of the painful stimulus. In addition, the T0 focus ratings in general did not show any association with HSR (rs < .11, ps > .28), with an exception of the focus on other aspects during pain induction, which correlated significantly with HSR, r(90) = .217, p = .04). 6.3.2. Manipulation checks A series of one-way ANOVAs confirmed that age, BMI, CSD, NA and most of the dependent measures at T0 did not significantly differ between 3 PF manipulation groups. The only significant group difference was found for Focus on sensations at T0, F(2, 87) = 3.93, p = .023. Post-hoc comparisons using Bonferroni correction indicated that the mean score was significantly lower in the sensory PF than in the undirected PF condition, while the affective PF did not significantly differ from the other conditions. The means and F tests are displayed in Table F-1 in Appendix F. T1 Focus ratings during retrieval were examined with three RM ANOVAs as a manipulation check for PF manipulation. Surprisingly, the ratings of focus on sensations, emotions, and other aspects of experience did not significantly differ between the PF manipulation groups (see Figure 6-3). This lack of differences was observed in the retrieval of both pain induction and dyspnea induction trials. However, the affective PF resulted in significantly lower focus on sensations during the dyspnea induction retrieval compared to the sensory and undirected PF, while in the pain induction retrieval, the focus on sensations did not differ the PF manipulation groups, as specified by the Induction Trial × PF Manipulation interaction for the T1 focus on sensations, F(2, 87) = 4.23, p = .018, !!! = .09. Finally, as for T0, the T1 focus on emotions was significantly higher in the retrieval of the dyspnea compared to the pain induction, regardless of the PF manipulation, F(2, 87) = 9.79, p = .002, !!! = .10. 115

Chapter 6 Table 6-2. Pearson product-moment coefficients (r) between the focus ratings during pain induction and affect and symptom ratings at T0: state NA, valence, arousal, threat, averaged concurrent, and immediate pain ratings (N = 90). Variables

1

2

3

4

5

6

7

8

9

1. Focus-sens

-

2. Focus-emo

.11

-

3. Focus-other

-.54***

-.33**

-

4. State NA

.11

.41***

-.19

-

5. Valence

-.11

-.31**

.35**

-.65***

-

6. Arousal

.13

.24*

-.24*

.55***

-.43***

-

7. Threat

.10

.32**

-.27*

.68***

-.64***

.54***

-

8. T0 average

-.08

.25*

-.19

.61***

-.60***

.31**

.66***

-

9. T0 immediate

-.10

.32**

-.12

.49***

-.55***

.24*

.50***

.79***

-

M

6.96

4.6

3.1

16.84

4.37

5.16

3.7

23.64

33.72

SD

1.70

1.94

2.19

4.35

1.72

1.71

1.92

13.60

18.54

Note. Focus-sens = focus on sensations, Focus-emo = focus on emotions, Focus-other = focus on other aspects, NA= Negative Affectivity, T0 average = average pain ratings, T0 immediate = immediate pain ratings. *p < .05, **p < .01, ***p < .001.

Table 6-3. Pearson product-moment coefficients (r) between the focus ratings during dyspnea induction and affect and symptom ratings at T0: state NA, valence, arousal, threat, averaged concurrent, and immediate dyspnea ratings (N = 90). Variables

1

2

3

4

5

6

7

8

9

1. Focus-sens

-

2. Focus-emo

.23*

-

3. Focus-other

-.40***

.15

-

4. State NA

.15

.34**

-.07

-

5. Valence

-.05

-.17

.12

-.65***

-

6. Arousal

.25*

.06

-.21

.42***

-.25

-

7. Threat

.22*

.33**

-.05

.73***

-.66***

.42***

-

8. T0 average

.16

.25*

-.09

.55***

-.53***

.25*

.57***

-

9. T0 immediate

.22*

.29**

.08

.55***

-.55***

.58**

.57***

.75***

-

M

6.87

5.21

3.76

19.94

3.41

5.64

5.14

24.71

39.10

SD

1.52

2.19

2.33

6.03

1.61

2.06

2.14

15.57

21.00

Note. Focus-sens = focus on sensations, Focus-emo = focus on emotions, Focus-other = focus on other aspects, NA= negative affectivity, T0 average = average dyspnea ratings, T0 immediate = immediate dyspnea ratings. *p < .05, **p < .01, ***p < .001.

116


Figure 6-3. Mean values and standard errors of ratings of focus on sensations (focus-sens), emotions (focus-emo), and on other aspects of experience (focus-other) in sensory, affective, and undirected processing focus (PF) manipulation during the retrieval (T1) of the pain (left) and dyspnea induction (right).

6.3.3. Retrospective ratings: T1 vs. T0 6.3.3.1. T1: PF effects on recalled affect and symptoms In the recall of pain, significantly higher ratings at T1, compared to T0, were observed for state symptoms (checklist), F(1, 83) = 7.34, p = .008, !!! = .08, and threat ratings, F(1, 83) = 19.93, p < .001, !!! = .19, while the ratings of perceived control increased over time, F(1, 78) = 3.98, p = .050, !!! = .05 (see Table 6-4). A tendency was found for state NA ratings to be higher at T1 compared to T0, F(1, 83) = 3.54, p = .064, !!! = .04. In line with the expectations, the affective PF resulted in a greater increase in the recalled state NA ratings than the sensory PF, as indicated by the Time × PF Manipulation interaction, F(2, 83) = 3.04, p = .053, !!! = .07. However, the retrieval of other affect and symptom ratings did not differ significantly between the PF manipulation groups. The recalled ratings of valence and arousal did not significantly differ from the T0 ratings, and no significant effects or interactions for HSR were found for the examined variables.

117

Chapter 6 Table 6-4. Means and standard deviations of the affect and symptom ratings during the pain and the dyspnea induction trials for 3 processing focus manipulation groups (sensory, affective, undirected) at 3 measurement moments (T0, T1, T2). Measure

Pain induction

Dyspnea induction

Sensory F

Affective

Undirected

Sensory

Affective

Undirected

T0

17.16 (3.27)

16.45 (4.90)

16.93 (4.86)

19.35 (4.54)

19.87 (6.65)

20.68 (6.84)

T1

16.74 (3.92)

18.16 (6.16)

17.68 (5.28)

19.52 (5.10)

20.45 (6.87)

21.25 (7.75)

T2

17.00 (4.32)

16.87 (5.10)

17.29 (6.55)

19.77 (6.05)

19.77 (6.00)

20.43 (7.57)

T0

26.00 (5.45)

24.42 (5.18)

25.82 (6.30)

28.39 (7.12)

29.87 (7.77)

30.86 (7.99)

T1

26.94 (5.69)

26.71 (5.89)

26.25 (5.80)

28.61 (6.56)

31.19 (8.41)

31.07 (8.75)

T2

26.29 (7.02)

25.87 (6.30)

27.64 (7.10)

28.13 (8.10)

29.68 (8.46)

30.29 (9.54)

T0

4.13 (1.57)

4.39 (1.73)

4.61 (1.89)

3.39 (1.54)

3.58 (1.86)

3.25 (1.40)

T1

4.03 (1.58)

4.58 (1.82)

4.11 (1.45)

3.29 (1.37)

3.55 (2.03)

3.04 (1.45)

T2

3.55 (1.21)

4.13 (1.86)

4.32 (1.79)

2.71 (1.01)

3.03 (1.62)

3.21 (1.93)

T0

5.13 (1.46)

5.16 (2.02)

5.18 (1.66)

5.55 (2.26)

5.68 (2.07)

5.71 (1.86)

T1

5.16 (1.37)

5.35 (1.66)

5.04 (1.90)

5.94 (1.81)

6.10 (2.07)

5.79 (2.13)

T2

6.94 (1.57)

6.52 (1.81)

6.50 (1.95)

5.81 (2.06)

5.97 (2.15)

5.82 (2.16)

T0

3.50 (1.75)

4.31 (2.02)

3.39 (1.75)

3.17 (1.80)

3.53 (2.30)

2.82 (1.68)

T1

3.68 (1.54)

4.41 (2.43)

4.00 (2.16)

3.60 (2.03)

3.50 (2.18)

3.21 (1.99)

T2

3.77 (1.79)

4.35 (2.12)

4.18 (1.98)

3.03 (1.72)

3.58 (2.41)

3.25 (1.90)

T0

3.97 (1.94)

3.74 (2.14)

3.36 (1.62)

5.19 (2.06)

5.23 (2.38)

5.00 (2.04)

T1

4.45 (1.82)

4.16 (2.13)

4.14 (1.56)

5.81 (1.70)

5.45 (2.39)

5.89 (1.87)

T2

4.45 (2.13)

4.06 (2.31)

4.11 (1.89)

5.77 (1.65)

5.13 (2.47)

5.61 (2.03)

State NA

State symptoms

Valence

Arousal

Control

Threat

With regard to the recalled dyspnea episode, affect and symptom ratings did not significantly differ from the T0 ratings, except for arousal (marginally) and threat ratings. Both increased from T0 to T1: arousal, F(1, 83) = 3.71, p = .058, !!! = .04; threat, F(1, 83) = 15.62, p < .001, !!! = .16. Contrary to the expectations, the PF manipulation did not influence the recall of affect and symptom ratings of the dyspnea induction. Also, no significant effects were found for HSR.

118


Additionally, in the recall of both inductions, the increase from T0 to T1 was dependent on the values of the ratings at T0, such that the increase was more pronounced with lower levels of T0 ratings. 6.3.3.2. T1: PF effects on recalled dyspnea and pain ratings RM ANCOVAs performed on the average pain ratings revealed that retrospective average pain ratings at T1 did not significantly differ from the ratings at T0 (Figure 6-4, top left panel). Also, no significant main effects or interactions with the PF manipulation or the HSR were found. The effects concerning the dyspnea ratings were in line with our expectations (Figure 6-4, top right panel): Retrospective average dyspnea ratings at T1 were higher than at T0, F(1, 83) = 9.57, p = .003, !!! = .10. However, neither PF nor HSR had a significant effect on retrospective symptom ratings. Additionally, T0 symptom ratings were found to interact with Time for both pain, F(1, 83) = 5.96, p = .017, !!! = .07, and dyspnea ratings, F(1, 83) = 10.37, p = .002, !!! = .11, such that the increase from T0 to T1 was observed for lower, but not higher, levels of T0 ratings (see Figure F-1 in Appendix F).

Figure 6-4. Mean values and standard errors of average pain (left panels) and dyspnea ratings (right panels) in sensory, affective, and undirected processing focus (PF) manipulation for T0 and T1 (top panels), and T1 and T2 (bottom panels).

119

Chapter 6

6.3.4. Retrospective ratings: T2 vs. T1 6.3.4.1. T2: PF effects on recalled affect and symptoms The comparison of the retrospective affect and symptom ratings of the pain episode at T2 and T1 (Table 6-4) showed that arousal recalled at T2 was significantly higher than at the T1 recall, F(1, 83) = 58.92, p < .001, !!! = .42. The PF manipulation at T1 did not influence the recall of affect and symptom ratings of the pain induction at T2. Moreover, no significant main effects or interactions with HSR were found for the examined variables. The recalled affect and symptom ratings of the dyspnea episode did not significantly differ from the T1 ratings, with the exception of valence ratings. The unpleasantness recalled at T2 was greater than at T1, F(1, 83) = 6.34, p = .014, !!! = .07, and this difference was more pronounced with higher levels of HSR, Time x HSR: F(1, 83) = 4.84, p = .031, !!! = .06 (Figure 6-5). Contrary to expectations, the PF manipulation at T1 did not influence the recall of affect and symptom ratings of the dyspnea induction at T2. Also, no other significant effects were found for the HSR. In addition, in the recall of both episodes, the change in T2 compared to T1 ratings was dependent on the values of the ratings at T1, probably illustrating regression to the mean with more extreme ratings at T1 being closer to the average on T2.

Figure 6-5. The interaction of Time (T1/T2) with the habitual symptom reporting (HSR) scores for valence ratings of the dyspnea induction trial.

120


6.3.4.2. T2: PF effects on recalled dyspnea and pain ratings The retrospective average pain and dyspnea ratings at T2 did not significantly differ from the ratings at T1 (Figure 6-4, bottom panels). Also, no significant main effects or interactions with the PF manipulation or the HSR were found for either pain or dyspnea recall at T2. Moreover, T1 symptom ratings were found to interact with Time for both pain, F(1, 83) = 13.74, p < .001, !!! = .14, and dyspnea, F(1, 83) = 6.75, p = .011, !!! = .08. Again this may suggest a regression to the mean (lower ratings at T1 were associated with greater increases at T2, and higher ratings at T1 were related to greater decreases at T2; see Figure F1 in Appendix F). 6.4. Discussion The primary goal of the present study was to examine the susceptibility of retrospective symptom ratings to biasing factors at the moment of memory retrieval. More specifically, it was explored whether directing attention to sensory-perceptual versus affective-motivational aspects of somatic memories during recall of a symptom episode could influence the way experienced symptoms and affective responses are remembered and reported. To investigate this issue, participants with varying levels of habitual symptom reporting were requested to focus on either sensory or affective aspects of memories while recalling two aversive bodily stimuli (pain and dyspnea), which were experimentally induced 2 weeks before the retrieval session. An undirected PF, with no explicit recall instructions, was included as a control condition. An additional follow-up rating session was scheduled after another 2 weeks. Symptoms and affective responses to stimuli were assessed on every session. Based on the dual-process perspective, an increase in recalled symptom and affect ratings was expected in conditions raising the salience of negative affective component, thus (a) following the affective PF manipulation, (b) in the dyspnea induction, and (c) with a higher level of HSR. Our findings demonstrated an increase in retrospective symptom reporting of dyspnea ratings over the period of 2 weeks from T0 to T1. Contrary to our expectation, the observed changes in retrospective ratings at T1 were affected neither by the manipulation of processing focus at retrieval, nor by HSR, even though both focus on emotions and HSR were related to higher initial ratings of negative affective reactions at T0. Moreover, whereas participants retrospectively rated the experienced symptoms at T1 as more

121

Chapter 6

threatening than initially reported at T0, other indices of affective responses were not overreported during T1 and T2 recall. 6.4.1. Responses to symptom inductions Consistent with previous findings, our study demonstrated that the intensity of both symptoms and affective reactions elicited by somatic stimuli is affected by the symptom induction procedure as well as by individual differences. First, the induction of dyspnea elicited stronger and more negative responses than pain induced by cold temperature, as shown by affective reactions (i.e., state NA, unpleasantness, arousal, threat) and reported symptoms (see also Walentynowicz et al., 2015; Walentynowicz, Raes, et al., 2016). Second, the effects of aversive somatic stimulation were modulated by individual differences in HSR, which were included as a continuous predictor. In line with the earlier findings (Bogaerts et al., 2008; Walentynowicz et al., 2015; Walentynowicz, Raes, et al., 2016), symptom ratings as well as affective responses (state NA, unpleasantness, and threat) were higher with increasing levels of HSR. Moreover, this association showed up at baseline and in response to dyspnea, but not in response to pain induction. The relationship observed at baseline, thus before the experimental stimulus induction, could be explained by the robust association of HSR with trait NA (Bogaerts et al., 2015), which predisposes the individuals with higher levels of NA to report more somatic symptoms in general. On the other hand, the association between HSR and affective responses in the dyspnea, but not in the pain trial, could be related to the qualitative differences between the two types of stimuli. Indeed, because dyspnea was overall considered more distressing, it may have affected HSR/NA individuals relatively stronger (Stegen et al., 2000). Additionally, the associations between trait NA and bodily complaints are not equal for all symptoms: stronger associations are observed with vague, systemic complaints (e.g., dyspnea) than with specific, localized symptoms (e.g., pain in the wrist), as well as for symptoms that are more severe and distressing (Van Diest et al., 2005). The way participants spontaneously focused on the different aspects of the somatic experience also differed between the symptom inductions. During dyspnea, participants focused on emotions and on other aspects (external environment) more than during the pain induction. Obviously this follows from the fact that dyspnea was experienced as more aversive. Correlational analyses further showed a positive relationship between spontaneous focus on emotions and symptom reports as well was negative affective responses for pain and dyspnea induction. Only for dyspnea, a spontaneous focus on sensations was also related to higher arousal and symptoms. It seems that when the experience is threatening or arousing, 122


the focus is not only on emotional but also on sensory aspects, suggesting a stronger internal focus in general. Conversely, in less threatening conditions (pain induction), only the focus on emotions is related to the affective responses. Finally, in both trials the self-reported focus on bodily sensations was greater than on the other components (emotions or external environment), which probably results from the experimental procedure: Measuring concurrently the experienced symptoms (pain and dyspnea) during the experience obviously induces to some extent increased attention towards the sensory details. 6.4.2. Retrospective ratings of symptoms and affective responses In general, the pattern of retrospective symptom reporting observed in this study closely resembles the findings of our previous studies investigating memory for dyspnea and pain (Walentynowicz et al., 2015; Walentynowicz, Raes, et al., 2016; Walentynowicz, Bogaerts, et al., 2016). First, biased retrospective reporting of symptoms appeared immediately after termination of the aversive stimulus. Even though the participants were explicitly requested to provide a rating of the average pain/dyspnea experienced during the trial, immediate and delayed retrospective ratings were higher than the averaged concurrent ratings, confirming previous observations (Walentynowicz et al., 2015; Walentynowicz, Bogaerts, et al., 2016). Obviously, providing a memory-based rating requires a retrospective integration of all elements of a past event. In line with assumptions underlying the peak-end effect (e.g., Kahneman et al., 1993; Redelmeier & Kahneman, 1996), this integration process does not depend on a simple mathematical averaging of moment-by-moment ratings, but instead implies different weighting of various elements of the experience. In that respect, it is not unexpected that the act of providing retrospective ratings, even immediately following the experience, results in different ratings compared to simple averaging. Demonstrating that the differences in symptom ratings are affected by the type of rating (memory-based vs. concurrent) rather than by time between the event and the rating could have implications for the studies interested in a measurement of experience-near states (e.g., current psychological or somatic states) such as Ecological Momentary Assessment (EMA; e.g., Smyth & Stone, 2003). To assure a precise assessment of immediate experiences, the prompt questions should cue the responses with “right now/ at this moment”. Otherwise, even with very short time frames like “in the last 5 minutes”, the retrospective recall biases cannot be avoided due to aggregation processes. Second, in line with our expectations, the changes in retrospective ratings given over a longer time period depended on the type of symptom induction. While an increase in

123

Chapter 6

retrospective reporting was observed for the more distressing symptom (dyspnea), retrospective pain ratings remained at the same level after 2 weeks at T1, replicating our previous observations (Walentynowicz et al., 2015; Walentynowicz, Raes, et al., 2016). Further changes in retrospective ratings after additional 2 weeks at T2 were not observed for either pain or dyspnea, suggesting that once a somatic experience is consolidated in memory, little or no change follows anymore over a longer period. Turning to the predictions concerning the effects of processing focus manipulation and individual differences (HSR), the increase in retrospective symptom reporting was not related to either of those factors. First, no differences between retrospective symptom reports in the three PF conditions (sensory, affective, and undirected) were observed. Contrary to our expectations, inducing affective PF before memory retrieval did not affect retrospective ratings of experienced symptoms. Moreover, except for a single effect on the state NA ratings in pain induction, affective PF did not influence the strength of the affective responses or distress related to the recalled symptom inductions. This suggests that PF during the retrieval phase does not have the same impact on the symptom reporting as the PF manipulation during encoding (Walentynowicz, Raes, et al., 2016). As a consequence, the manipulation of PF may be considered as more effective when it directs the attention during the actual encounter rather than during the reconstructive process. Those memory enhancing effects could be related to the increased arousal elicited by focusing on the emotions during the encoding stage, which was repeatedly shown to have a strong effect on the encoding and consolidation processes memories (e.g., Cahill & McGaugh, 1998; Sharot & Phelps, 2004; Talmi, 2013). The lack of PF effects on the affective responses during the retrieval phase could be one of the explanations of the absence of PF differences in symptom recall in this study. Second, even though HSR was related to more negative affective and symptom responses to induced sensations, possibly leading to a more salient affective component of symptom experience, it did not result in symptom overreporting at recall. This is in contrast to the previous study, in which the increase in dyspnea reporting was only present among participants scoring high on HSR (Walentynowicz et al., 2015). While this inconsistency could be related to the selection of the participants (dichotomous vs. continuous scales for HSR), the positive relationships between HSR and symptom ratings (via checklist) and affective responses to the symptom inductions at T0 demonstrated in this study speak against this interpretation. Also, the observation that neither concurrent nor retrospective symptom ratings were related to the levels of HSR is consistent with earlier findings showing that manipulating the processing focus during encoding wiped out individual differences in 124


symptom reports related to the HSR levels (Walentynowicz, Raes, et al., 2016). It is therefore possible that also in the current study the use of PF manipulation, this time at the retrieval phase, reduced the effects of HSR on the recall of symptom ratings. Overall, the PF manipulation did not seem to affect the way symptom ratings and affective responses accompanying the somatic experience were recalled. Moreover, investigating the effect of self-reported focus at T1 as a manipulation check showed that, even though higher focus on sensations was reported in the sensory PF condition compared to affective or undirected PF, and higher focus on emotions was observed during the affective PF than in the other conditions, those differences were not statistically significant. This suggests that the manipulation used during the retrieval phase was not strong enough or successful in guiding information processing in an intended direction, even though all participants were requested to rephrase the instructions and confirmed that they understood them. In most of the conditions participants indicated a high focus on sensory aspects of the experience during the retrieval process. One of the possible reasons for this dominance of sensory focus may be that it is the focus adapted spontaneously during the initial symptom experience, which further may have been unintentionally reinforced due to concurrent symptom ratings. Nevertheless, it is still surprising that the focus on emotions was not stronger, especially taking into account the assumption that over longer retention intervals the affective component seems to dominate over the sensory one in the event memory (Gedney et al., 2003; Gedney & Logan, 2004; Kent, 1985). The failure of PF manipulation at the retrieval phase to influence the recall of symptom and affective ratings could indicate that the biases leading to symptom overreporting operate earlier in memory formation, for example, during encoding, as found in our previous study, or during consolidation. It is plausible that once a memory trace is formed and consolidated, a local temporary manipulation is not strong enough to modify it. Episodic and autobiographical memories can be modified through a process known as memory reconsolidation (see Schwabe, Nader, Pruesser, 2014, for a review). When a memory trace becomes reactivated, it may re-enter the unstable state, during which it is vulnerable to change. However, a simple reactivation is not enough to destabilize and update the memories. For example, memory reconsolidation will not take place when no or little new information in present during memory retrieval (Sevenster, Beckers, & Kindt, 2012; Wichert, Wolf, & Schwabe, 2013) or when reactivation takes place without a post-recall stressor (Schwabe & Wolf, 2010). Because the temporary PF manipulation used in this study did not lead to either increased distress or learning new information, which could update the memory for past 125

Chapter 6

somatic experience, it is likely that the reactivated memory trace was not subject to reconsolidation processes. However, to draw a valid conclusion about the role of PF during retrieval, it is necessary to overcome the limitations of this study. First, a manipulation of PF could be improved by including more prompt questions which could facilitate or further promote the intended processing focus during the retrieval of somatic memories. Moreover, the selfreports of the adapted focus could be collected by means of a more extended questionnaire rather than a numerical rating scale. In conclusion, attending to different components of a somatic event during memory retrieval, thus after the memory traces were stabilized, seems to have a limited effect on the recall biases. Moreover, retrospective symptom ratings were found to increase over time only for aversive and distressing symptoms, giving an additional support the important role of affective factors in symptom reporting. Consequently, the situational and psychological factors increasing the salience of the affective component during symptom experience itself could play an important role in mechanisms underlying symptom (over)reporting. 6.5. Acknowledgments The authors want to thank Sander De Deygere for his assistance in data collection. The study was funded by Grant OT/10/027 from the University Research Council of the University of Leuven.

126

The specificity of health-related autobiographical memories in patients with somatic symptom disorder

7Chapter 7 The specificity of health-related autobiographical memories in patients with somatic symptom disorder

Abstract Reduced autobiographical memory specificity (rAMS) is related to a range of emotional disorders and is considered a vulnerability factor for an unfavorable course of pathology. The present study investigated whether the specificity of health-related autobiographical memories is reduced in patients with somatic symptom disorder (SSD) with medically unexplained dyspnea complaints, compared to healthy controls. Patients with SSD have persistent distressing somatic symptoms that are associated with excessive thoughts, feelings, and behaviors. Female SSD patients (n = 30) and matched healthy controls (n = 24) completed a health-related Autobiographical Memory Test, the Beck Depression Inventory, the Ruminative Response Scale, and rumination scales concerning bodily reactions. Depressive symptoms and rumination were assessed because both variables previously showed associations with rAMS. Patients with SSD recalled fewer specific (F(1, 52) = 13.63, p = .001) and more categoric (F(1, 52) = 7.62, p = .008) autobiographical memories to healthrelated cue words than healthy controls. Patients also reported higher levels of depressive symptoms and rumination (all ts > 3.00, ps < .01). Importantly, the differences in memory specificity were independent of depressive symptoms and trait rumination. To conclude, the present study extends findings on rAMS to a previously unstudied sample of patients with SSD. Importantly, the presence of rAMS could not be explained by increased levels of depressive symptoms and rumination. We submit that rAMS in this group reflects how health-related episodes and associated symptoms are encoded in memory. .

Based on: Walentynowicz, M., Raes, F., Van Diest, I., & Van den Bergh, O. (in press). The specificity of health-related autobiographical memories in patients with somatic symptom disorder. Psychosomatic Medicine.

127

Chapter 7

7.1. Introduction According to the recent DSM-5 (American Psychiatric Association, 2013), somatic symptom disorder (SSD) is characterized by persistent distressing somatic symptoms which are associated with disproportionate and excessive feelings, thoughts, and behaviors, and that result in substantial disruption of functioning. In many cases, patients with SSD report symptoms that cannot be explained by a physiologic dysfunction. Studies focusing on perceptual-cognitive processes in patients with SSD and in persons scoring high for habitual symptom reporting unrelated to disease showed specific characteristics distinguishing them from healthy persons (Rief & Martin, 2014; Van den Bergh et al., 2015). First, sensory-perceptual processing of somatic information tends to be less detailed, as suggested by (a) the absence of peak-end heuristic for somatic episodes (which states that symptom evaluation is predominantly influenced by the most intense (peak) and the final (end) moments and less so by the duration; Bogaerts et al., 2012); (b) diminished correspondence between self-reported complaints and related physiological reactions (Bogaerts, Van Eylen, et al., 2010); and (c) poorer differentiation between various somatic sensations and stronger influence of earlier knowledge during categorization (Petersen, Van Staeyen, et al., 2015). Second, negative emotional responses to bodily stimuli are more intense in these groups (Walentynowicz et al., 2015; Wan et al., 2012) and seem to mediate the observed overreporting of recalled symptoms (Walentynowicz et al., 2015). Focusing on the affective information at the expense of detailed encoding of sensory-perceptual features of somatic episodes can influence the way bodily symptoms are remembered, which may manifest itself in reduced specificity of retrospective memory. The difficulty to retrieve specific personal memories of a past event, termed as reduced autobiographical memory specificity (rAMS) or overgeneral memory (OGM) was previously found in a range of psychopathological disorders, most importantly depression and PTSD (see Williams et al., 2007). Memory specificity is typically assessed using an emotional cue-word procedure, known as the Autobiographical Memory Test (AMT; Williams & Broadbent, 1986). Especially depressed patients find it difficult to retrieve specific memories on the AMT. More often than healthy controls, they recall non-specific or overgeneral memories (e.g., “every time other people hurt me” to the cue “disappointed”) rather than the requested specific memories (e.g., “three weeks ago, when John called to tell me he would not be coming over for my birthday”). rAMS (or OGM) is not only a concomitant of depression, but is also considered as a relatively stable marker of an

128


unfavorable course of psychopathology, impacting severity of symptoms, illness duration, and treatment success (see Sumner et al., 2010, for a review and meta-analysis). According to the CaR-FA-X model (Williams et al., 2007; see also Sumner, 2012; Sumner et al., 2014), three mechanisms contribute to rAMS, alone or in interaction: Capture and Rumination (CaR), Functional Avoidance (FA) and impaired eXecutive control (X). The first factor (CaR) refers to situations in which memory retrieval is disrupted and “captured” at a more general level in the memory hierarchy (see Self-Memory Model; Conway & PleydellPearce, 2000). In such instances, highly self-relevant cue words may activate networks of self-related general information, which hinder the retrieval of specific memories (Dalgleish et al., 2003) as observed in clinical groups (Crane, Barnhofer, & Williams, 2007; Spinhoven et al., 2007). Moreover, rumination, in the form of analytical and abstract repetitive thinking (or “brooding”; Treynor et al., 2003) has been shown to further promote this capture, as demonstrated by correlational (Debeer et al., 2009; Sumner et al., 2014) and experimental studies (e.g., Crane, Barnhofer, Visser, et al., 2007; Raes et al., 2008; Watkins & Teasdale, 2001, 2004). Consequently, one remains “stuck” in a cyclic retrieval of general self-related information, and progression towards more concrete, specific memory content is delayed. Second, functional avoidance occurs when specific recollections of adverse experiences are avoided in order to reduce the impact of negative affect associated with those memories (Williams et al., 2007). In line with this affect-regulation hypothesis, rAMS is associated with avoidant coping styles as assessed by a variety of avoidance measures (Debeer et al., 2011; Hermans et al., 2005). While in the short-term, a less specific retrieval style may be more functional and beneficial, as it is related to a lower emotional distress to a mild aversive experience (Hermans et al., 2008; Raes et al., 2003, 2006), this general retrieval style may become maladaptive when used for a longer time (Williams et al., 2007). The final “X”-factor relates to the impairments in executive resources that prevent successful retrieval of specific memories, including the deficits in working memory capacity, inhibition, and verbal fluency (Dalgleish et al., 2007). Patients with SSD respond to the experience of somatic symptoms with maladaptive thoughts and emotions, which can affect perceptual and mnemonic processes related to bodily sensations in different ways. With regard to symptom perception, the tendency to experience symptoms as intense, noxious, and disturbing may result in an increased attentional focus to somatic changes, leading to increased perception of bodily sensations and symptom reporting, as postulated by the theory of somatosensory amplification (Barsky & Wyshak, 1990). However, in case of memory retrieval, the functional avoidance hypothesis (CaR-FA-X 129

Chapter 7

model; Williams et al., 2007) suggests that the specific recollections of aversive events are avoided as a means of affect regulation. Consequently, because patients perceive somatic experiences as threatening and aversive, they may be more inclined to avoid retrieving highly specific details of those memories, as these would evoke associated anxiety and intense negative emotions. As a result, reduction of symptom-related distress could be achieved at the expense of memory specificity. Such functional avoidance, together with less detailed sensory processing described above, could lead to rAMS in this patient group. However, until now this aspect of memory in SSD patients has not been investigated. The purpose of the present study was to investigate the specificity of autobiographical memory in patients with SSD. It was previously shown that the cues relevant to one’s concerns tend to prompt overgeneral memories (Barnhofer et al., 2007; Spinhoven et al., 2007). Accordingly, cue words associated with health were used to elicit the retrieval of specific health-related autobiographical memories, which were assumed to be highly relevant to this group. Moreover, because comorbid emotional disorder is often associated with SSD (Henningsen et al., 2003), we also assessed depression and rumination in order to control for these variables. We hypothesized that patients will retrieve fewer specific and more categoric autobiographical memories in response to health-related cues compared to controls. The focus was placed on those indices of rAMS, because previous research has shown that among the non-specific memories the categoric subtype is a marker of pathology or vulnerability for pathology (e.g., Barnhofer, Jong-Meyer, Kleinpaß, & Nikesch, 2002; Williams & Dritschel, 1992). In addition, we also expected higher levels of rumination and depression in the patient (vs. control) group. 7.2. Method 7.2.1. Participants The data used in this sample are derived from a larger two-part questionnaire and experimental study (see Chapter 4) investigating memory for dyspnea in patients with SSD who particularly suffered from medically unexplained dyspnea (MUD; n = 30; all women). Healthy controls were matched for age, gender, BMI, and educational level (n = 24; all women). Patients were recruited from the outpatient pulmonology clinic of the Leuven University Hospital (Gasthuisberg) and were classified as having MUD after (1) a systematic medical work-up procedure which excluded physiological causes for the multiple somatic complaints such as dyspnea, breathing distress, fatigue, and numbness; and (2) a systematic

130


interview, namely the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition Axis I Disorders, administered by a qualified psychologist, which excluded psychiatric reasons for experienced dyspnea other than somatization disorder. The assessment of psychological criteria of SSD was based on the new classification criteria of DSM-5 (American Psychiatric Association, 2013). A validated instrument to directly measure the psychological characteristics of SSD became only recently available (Toussaint et al., 2016). Exclusion criteria included a self-reported history of pulmonary, cardiovascular, gastrointestinal, or neuromuscular disease, or other medical conditions that likely affect respiratory capacity, such as acute illnesses, fever, or flu. Participants were also excluded if they currently suffered from mental disorders other than SSD (self-reported via a general item), were pregnant or lactating. Five patients reported used of medication, including protonpump inhibitors (pantoprazole), beta blockers (propranolol), selective serotonin reuptake inhibitors (escitalopram, sertraline), and benzodiazepines (alprazolam). Because the inclusion or exclusion of the medication-taking patients did not influence the results of the study, these patients were retained in the final sample. The study was approved by The Medical Ethics Committee of the University Hospital of the University of Leuven and took place between August 2012 and April 2014. 7.2.2. Measures 7.2.2.1. Health-related Autobiographical Memory Test (h-AMT) Autobiographical memory specificity (AMS) for health-related cue words (h-AMS) was assessed with use of the Autobiographical Memory Test (Williams & Broadbent, 1986) adapted for health-related memories. Five positive (recover, health, cure, vaccination, treatment) and five negative (disease, flu, feverish, bacterium, headache) health-related cue words were presented in alternating order. The two word groups were selected and matched on emotional extremity, imageability, familiarity and relatedness to health, following a twostep procedure: First, a pool of potential health-related cue words was generated (N = 78). Second, a sample of undergraduate psychology students (N = 141) rated each item on valence, imageability, familiarity, and relatedness to health/illness. Six words were repeated (reliability test). A list of 5 positive and 5 negative words that were matched for all the rated characteristics was constructed. Participants were instructed to recall a specific event related to the cue word, which was read aloud by the researcher. A specific memory was defined as a memory about a personally experienced event that happened at a particular time and place and that lasted less than 1 day. Then, examples of both specific and overgeneral memories were 131

Chapter 7

provided. Three practice words (relaxation, doctor, active) were used before the test to familiarize participants with the procedure. The time given to retrieve a specific memory to each cue was 60 s and the responses were audiotaped. If the answer was non-specific, the participants were prompted to retrieve a memory of a specific event. In case of an ambiguous response, a clarification question was asked. The complete Dutch instructions together with the English translation are included in Appendix G. Responses were coded as specific if they referred to a personal memory of an event, which happened more than 7 days before the testing day and lasted less than one day. Otherwise, the memories were coded as non-specific and further subdivided into overgeneral categoric (events occurring more than once, e.g., “Every time when I went to physiotherapist”), overgeneral extended (events lasting more than one day, e.g., “When I was recovering after the surgery”), same events, omissions (no response provided), and no memories (responses not referring to the past; associations or references to the future). In line with previous studies (Raes, Schoofs, Griffith, & Hermans, 2012; Williams et al., 2007), specific and categoric first memories were used as indices of rAMS. For inter-rater reliability, a random sample of audiotaped responses of 20 participants (37%) was evaluated by an independent rater blind to participant’s group. There was high agreement between the two raters (Κ = .845). 7.2.2.2. Depression Depressive symptoms were measured with the Dutch version of Beck Depression Inventory-II (BDI-II; Van der Does, 2002). This 21-item questionnaire uses a 0-3 scale to assess current cognitive, affective and physical symptoms of depression. Cronbach’s alpha in the present sample was .92. 7.2.2.3. Rumination on sadness Ruminative thinking in response to sad and depressed mood was assessed with the brooding subscale of Ruminative Response Scale (RRS; Treynor et al., 2003; see also Griffith & Raes, 2014). Responses to five items describing thoughts about possible causes and consequences of one’s mood are given on 4-point Likert scale (almost never to almost always). The Dutch version was used (Raes et al., 2009; Schoofs, Hermans, & Raes, 2010). Cronbach’s alpha in the present sample was .62.

132


7.2.2.4. Rumination on bodily reactions The modified RRS was used to measure ruminative responses to bodily sensations and symptoms. In the adapted version (h-RRS) participants rated the frequency of thoughts regarding possible causes, meanings, and consequences in response to bodily sensations and symptoms. Similarly to the RRS (Treynor et al., 2003), the h-RRS consists of two factors: body brooding and body reflection. Details concerning items and scale construction are described in Appendix H. Both subscales were reliable in the present study, with Cronbach’s alpha of .93 for brooding and .87 for reflection. 7.2.3. Procedure Participants were invited to a study investigating the influence of respiratory challenges on breathing behavior and subjective well-being. The study consisted of two unrelated parts: a questionnaire study and an experimental study. The latter part is beyond the scope of the present paper and is reported separately (see Chapter 4). Before the laboratory session, participants completed a series of trait questionnaires at home, including the BDI-II. Other administered questionnaires including the Checklist for Symptoms in Daily Life (Wientjes & Grossman, 1994), the Positive and Negative Affect Schedule (Watson et al., 1988), and the Perseverative Thinking Questionnaire (Ehring et al., 2011) are not reported in this chapter. In the laboratory, each participant was informed about the procedure and signed the informed consent. After completing the demographic information sheet, the questionnaire study took place, during which the h-AMT was administered in an oral form, followed by the trait rumination questionnaires. Once completed, the experimental study followed. 7.2.4. Data analysis The group differences in demographic variables and trait measures were compared using independent sample t tests for continuous data and chi-square tests for categorical data. Group differences in memory specificity were examined with 2 (Group: patients/controls) × 2 (Cue valence: positive/ negative) mixed-design analyses of variance. Multiple mediator models were applied to specific and categoric memories to examine whether the group differences in rAMS could be attributed to the differences in psychological characteristics known to be closely related to rAMS (i.e., depression and rumination). Instead of performing multiple testing with a series of simple mediation models, multiple mediator models were used in which the mediators are included simultaneously in a single integrated model (Hayes, 2013). As a result, the estimated effect of a specific mediator is conditional upon the other mediators in the model. To perform the analyses, scores on the BDI-II, RRS133

Chapter 7

brooding, and h-RRS brooding were simultaneously included as mediators and tested in a single parallel multiple mediator model using the bootstrapping procedure (Preacher & Hayes, 2008). This procedure, designed for small sample sizes, allowed to estimate the indirect effects of the group on both specific and categoric memories through each of the mediators, as well as the direct effects of the group. The 95% confidence intervals of the effects were derived with 5000 bootstrap re-samples. Direct and indirect effects are reported in unstandardized form (Hayes, 2013). Data were analyzed with IBM SPSS Statistics 23.0 and PROCESS Macro for SPSS (Hayes, 2013). 7.3. Results Demographic and personality characteristics of both groups are presented in Table 71. The groups did not differ with regard to demographic characteristics, but patients did score significantly higher on the BDI-II, the RRS-brooding, and both subscales of the h-RRS as compared to controls. Most crucially, group differences emerged also with regard to memory specificity (see Table 7-2). Patients with SSD, compared to controls, retrieved fewer specific, F(1, 52) = 13.63, p = .001, !!! = .21, and more categoric memories, F(1, 52) = 7.62, p = .008, !!! = .13. In line with previous findings indicating unidimensionality of AMT (Griffith et al., 2009), the cue valence had no effect on either specific, F(1, 52) = .77, p = .39, !!! = .02, or categoric memories, F(1, 52) = 2.09, p = .16, !!! = .04. 7.3.1. Multiple mediator model The analyses above showed significant group differences in the indices of rAMS, depressive symptoms, and trait rumination. Moreover, the number of specific memories correlated negatively with depressive symptoms, r(54) = -.34, p = .012, as well as with brooding on bodily sensations and symptoms, r(54) = -.34, p = .011 (for a complete list of the unadjusted correlations between the indices of rAMS and the psychological measures, see Table I-1 in Appendix I). To investigate whether the group differences in rAMS could be associated with differences in depression and rumination, multiple mediator models were used. With regard to specific memories, the bootstrap results indicated that the total effect of group on specific memories (total effect, b = -2.28, p = .001, 95% CI [-3.53, -1.04]) remained significant when all mediators were included in the model (direct effect of group, b = -1.76, p = .048, 95% CI [-3.51, -0.02]). Moreover, neither the total indirect effect of group on specific memories through the three mediators, nor the specific indirect effects of each of the proposed 134


mediators were significant (all bs < .70, ps > .31). Similar results were observed for the multiple mediator model using categoric memories as the dependent variable. The total effect of group on categoric memories (total effect, b = .73, p = .008, 95% CI [.20, 1.25]) remained significant when the mediators were included in the model (direct effect of group, b = .76, p = .046, 95% CI [.01, 1.51]), with neither total nor specific indirect effects reaching significance (all bs < .14, ps > .47). These findings indicate that depressive symptomatology and brooding subscales of rumination did not mediate the relationship between the group and rAMS.

Table 7-1. Demographic and personality trait characteristics of patients with SSD compared with control group. Variable Patients Controls Statistics (n = 30) (n = 24) Mean age (SD) 38.27 (9.03) 37.42 (9.84) t(52) = .33, p = .74 Working, n (%) 23 (76.7) 20(83.3) χ2(1, N = 54) = 0.37, p = .55 Marital status, n (%) χ2(3, N = 54) = 0.96, p = .81 Married or co-habiting 19 (63.3) 15 (62.5) Single 6 (20) 6 (25) Divorced 4 (13.3) 3 (12.5) Widowed 1 (3.3) 0 (0) Educational level, n (%) χ2(2, N = 54) = 2.88, p = .24 High school 11 (36.7) 4 (16.7) College 10 (33.3) 12 (50) University 9 (30) 8 (33.3) BDI-II (SD) 18.27 (9.34) 5.13 (4.92) t(45.67) = 6.64, p < .001 RRS-brooding (SD) 11.33 (2.90) 9.17 (2.01) t(52) = 3.10, p = .003 h-RRS brooding (SD) 15.23 (5.11) 9.13 (2.86) t(47.09) = 5.55, p < .001 h-RRS reflection (SD) 15.73 (4.00) 12.17 (3.46) t(52) = 3.45, p = .001 Note. BDI-II = Beck Depression Inventory; RRS = Ruminative Response Scale; h-RRS = Ruminative Response Scale to bodily sensations and symptoms.

135

Chapter 7 Table 7-2. Means and standard deviations for indices of autobiographical memory specificity by group. Statistics Group Patients (n = 30) Controls (n = 24) M

SD

M

SD

Significant effects Group

Specific memories Total 7.13 2.87 9.42 1.06 Positive cues 3.73 1.51 4.67 .76 Negative cues 3.40 1.61 4.75 .53 Overgeneral categoric memories Group Total .93 1.17 .21 .59 Positive cues .33 .55 .13 .45 Negative cues .60 .77 .08 .28 Overgeneral extended memories Group Total 1.00 1.23 .21 .51 Positive cues .43 .63 .13 .34 Negative cues .57 .86 .08 .28 Same eventa .07 .25 .00 .00 a No memory .63 1.22 .13 .34 a Omission .23 .63 .04 .20 a For the categories with low response rates only total scores are provided.

F (p)

!!!

13.63 (.001)

.21

7.62 (.008)

.13

8.70 (.005)

.14

7.4. Discussion Our study investigated memory specificity in SSD patients. The results indicated that our sample of SSD patients showed reduced memory specificity when recalling health-related experiences compared to healthy controls. One of the possible explanations of this finding may be related to the way patients with SSD process information about their somatic sensations and complaints. As bodily sensations consist of both sensory-perceptual and affective-motivational components (Craig, 2003; Leventhal & Everhart, 1979; Price, 2000), SSD patients may encode and store fewer sensory elements while focusing on the affective features of somatic episodes. In consequence, patients can be expected to experience difficulties in retrieving detailed and specific memories in response to health-related cue words indicating reduced specificity of autobiographical memories for health-related episodes. The first-time observation of rAMS in SSD patients in the current study is consistent with findings showing biased processing of somatic information (Bogaerts et al., 2012; Bogaerts, Van Eylen, et al., 2010; Petersen, Van Staeyen, et al., 2015), but adds to it that memory processes may also contribute to biased retrospective symptom reporting in clinical interviews and questionnaire-based symptom assessments. Future studies are necessary to investigate whether the rAMS is also a prognostic factor for the onset, 136


maintenance, recurrence of SSD, similar to its role in predicting the course of affective disorders (Sumner et al., 2010). The well-documented association between depression and rAMS (see Williams et al., 2007, for a review) together with the high prevalence rate of comorbid emotional disorder in SSD patients (Henningsen et al., 2003) may suggest that the observed deficits in memory specificity originated from increased depressive symptoms in the patient group. However, the relationship between group and rAMS remained significant after controlling for levels of depressive symptoms and trait rumination. This is in line with other studies showing that the retrieval of less specific memories is independent of trait depression in chronic pain (Liu et al., 2014), in borderline patients (Spinhoven et al., 2007), and in individuals with a history of child sexual abuse (Hauer, Wessel, Geraerts, Merckelbach, & Dalgleish, 2008). The current findings also inform about possible mechanisms underlying rAMS in SSD patients. The CaR-FA-X model (Williams et al., 2007) describes that when self-related information is processed in a ruminative manner, the cognitive resources necessary to perform the memory search are “captured” at more general levels, disrupting the further retrieval of more specific memories. In particular, this effect is found for the maladaptive form of repetitive thought, namely brooding (Schoofs et al., 2010; Watkins, 2008). The higher levels of rumination observed in our patient group (vs. controls), could therefore suggest “(Capture and) Rumination” as a possible underlying mechanism of rAMS in this group. However, the mediation analyses did not support this interpretation, as neither affective nor health-related brooding mediated this relationship. This suggests that rumination is not a key factor explaining the deficit in memory specificity in this particular group. However, as activation of state ruminative processing may be needed to observe such relationship (Crane, Barnhofer, Visser, et al., 2007; Raes et al., 2012), more research is necessary to further delineate this association, since we only relied on a trait measure of different forms of rumination. According to the Car-FA-X model, factors underlying rAMS can interact or be active to a different extent in different groups. This implies that other mechanisms than rumination, that is, functional avoidance or executive control could also influence memory specificity in patients with SSD. It was indeed shown that patients with SSD tend to perceive bodily stimuli as more aversive and threatening which may promote avoidance (Wan et al., 2012). Also deficits in executive control (X), including response inhibition, cognitive flexibility and working memory have been reported in patients qualifying for SSD (Aizawa et al., 2012; Correa et al., 2011; Solberg Nes et al., 2009).

137

Chapter 7

In the present study, the widely used questionnaires measuring memory specificity (AMT) and rumination (RRS) were tailored to SSD patients. While the adjustment of AMT to health-related context was expected to elicit more OGM (Barnhofer et al., 2007; Spinhoven et al., 2007), two consequences of this content adaptation should be mentioned. First, the healthrelated cue words might be more concrete than the affective ones used in the standard AMT which could have enhanced the retrieval of specific memories (Hauer et al., 2008). Indeed, even though our clinical sample retrieved fewer specific memories in response to the cue words compared to the control group, the deficits in AMS were less severe than usually observed among depressed/dysphoric individuals. Second, it is uncertain whether rAMS in patients with SSD is limited to health-related word cues or whether it also generalizes to emotional stimuli. However, as the assumptions of CaR-FA-X model are based on more general memory mechanisms, we would predict that similar memory specificity effects would appear regardless of the cue-word used. In sum, this study is the first to report a deficit in memory specificity for health-related cues in patients with SSD. Importantly, this relationship could not be attributed to increased levels of depressive symptoms and rumination in the patient group. While rAMS could affect the way health-related episodes and associated symptoms are remembered, future research is necessary to replicate this and to examine which factors underlie this pattern of retrieval. 7.5. Acknowledgments The authors want to thank Dr. Tom Van Daele for scoring a subset of the AMTs. The study was funded by Grant OT/10/027 from the University Research Council of the University of Leuven.

138

General discussion

8Chapter 8 General discussion

Self-reported symptoms are often the only way to gain information about one’s experience of somatic events, such as an asthmatic attack, recovery after a surgery, or medical/dental procedures. However, symptom ratings are prone to various biases potentially leading to adverse consequences. While a body of research findings exists on cognitive, affective, and situational factors influencing symptom reporting, the role of memory processes in this domain has only recently received growing attention. The aim of the current doctoral project was to investigate somatic memories and, in particular, how memory processes can influence retrospective symptom (over)reporting. The exploration of this issue was grounded within a dual-process perspective (e.g., Craig, 2003; Leventhal et al., 1979; Price, 2000), that posits that symptom ratings result from an interplay of sensory-perceptual and affective-motivational components of a somatic experience. The first aspect includes details about the quality, intensity, location, and duration of the somatic experience, while feelings towards sensations and drive to adjust (to) a stimulus pertains to the latter aspect. In this view, distortions in symptom ratings may emerge in situations characterized by a relative imbalance of those two components. Specifically, we have proposed that increased symptom reporting should be particularly observed when strong negative affective information overshadows a sensory component, potentially resulting in less detailed memory of sensory details. Because a relative dominance of affective over sensory processing of bodily signals could be expected among individuals with an overreactive evaluative system, the hypotheses advanced in this project were explored among the individuals who tend to report frequent somatic experiences (high habitual symptom reporters; HSR) and medically unexplained symptoms (MUS). 8.1. Summary of the findings Different research approaches were applied to investigate retrospective symptom reporting. First, by means of psychometric research (Chapter 2), an analysis of the latent structure of symptom reporting demonstrated that symptom reporting is best explained by a bifactor model comprising one general and several symptom-specific factors. This finding 139

Chapter 8

confirmed previous observations in various clinical and non-clinical populations (Jasper et al., 2015; Thomas & Locke, 2010; Witthöft et al., 2016; Witthöft, Hiller, et al., 2013). Moreover, the observed pattern of associations of trait negative affectivity (NA) with the latent factors further supported the previously advanced interpretation linking the general symptom factor with the affective component of a symptom experience, and the symptom-specific factors with the sensory component (Jasper et al., 2015; Witthöft et al., 2016). The results of this study did not only psychometrically support the dual-process approach to symptom selfreport, but also indicated good validity and reliability of the Checklist for Symptoms in Daily Life, the questionnaire which was used throughout this doctoral project to select participants according to their habitual symptom reporting level. Second, we used experimental studies with controlled symptom inductions to standardize the somatic experience across participants. Assessment of both self-reported and psychophysiological responses to somatic stimuli took place concurrently during symptom inductions. Retrospective ratings were collected up to two (Chapters 3-5) and four (Chapter 6) weeks after the somatic experiences. In the first two experimental studies (Chapters 3 and 4), we compared concurrent and retrospective ratings of induced symptoms within- and betweensubjects. Both studies extended our knowledge about retrospective symptom reporting and the possible underlying mechanisms by showing that (a) memory biases start to operate immediately after the event, (b) the intensity of both concurrent and retrospective ratings is elevated in HSR/MUS, and (c) the relationship between the HSR/MUS and elevated retrospective symptom ratings is mediated by the affective state associated with the somatic experience (i.e., state NA and anxiety). In the subsequent experimental studies (Chapters 5 and 6), the symptom inductions were combined with a manipulation of information processing focus (PF) at encoding and at retrieval. The primary goal of those studies was to investigate whether directing PF to either sensory-perceptual or affective-motivational aspects of somatic experience can influence retrospective symptom reporting. The manipulation of PF at encoding led to differences in affective responses as well as in memory bias. In particular, the affective PF resulted not only in increased affective responses to symptom inductions, but also in an increase in retrospective dyspnea reporting over time. On the other hand, the manipulation of the PF at a later stage of memory processing (i.e., retrieval) did not have an impact on the way symptoms or affective ratings were recalled. Finally, we approached the topic of somatic memories also from a different perspective, namely autobiographical memory (Chapter 7). This study demonstrated reduced 140

General discussion

memory specificity of health-related autobiographical memories in patients with MUS compared to healthy controls. 8.2. Bias in retrospective symptom reporting The experimental studies of this dissertation investigated memory biases in retrospective symptom reporting, operationalized as a discrepancy between concurrent and retrospective ratings of the same somatic experience. Such discrepancy was observed for both pain and dyspnea, with an overall evaluation of the symptom episode being rated as more intense than the averaged momentary ratings of experimentally induced symptoms (Chapters 3, 4, and 6). This is in line with previous findings from clinical settings showing that recalled symptom ratings are often more intense compared to averaged momentary ratings (e.g., Broderick et al., 2008; Giske et al., 2010; Redelmeier & Kahneman, 1996; Redelmeier et al., 2003; Stone et al., 2005). The differences that we observed between retrospective and concurrent self-reported symptoms appeared not only for the ratings given immediately following the episode, but also for the ratings given after longer retention intervals (i.e., two or four weeks). Moreover, an increase in retrospective symptom ratings was observed in a number of cases suggesting the factors that might influence symptom reporting and lead to symptom overreporting. The discussion of those findings is provided in this section. 8.2.1. Recall bias appears immediately The robust finding that retrospective symptom ratings differ from the averaged initial self-reported symptoms is related to an unavoidable dependence on the memory processes when evaluating a past experience. Providing any memory-based response to a question about a somatic event in the past (and a rating of a somatic event that only terminated a moment ago, such as in our immediate tests, is also memory-based) requires a retrospective integration of all the elements of a past episode. This mental aggregation process is not a simple task, and individuals use a variety of different strategies to answer questions about recalled symptoms. For example, a study using a cognitive interviewing technique (Broderick, Stone, Calvanese, Schwartz, & Turk, 2006) has shown that when patients are requested to provide a summary rating of their average pain severity of the past week, 40% did not attempt to calculate the average of the pain experiences, 20% did not use the information from across the entire week, while 28% did not consider the times without pain. This qualitative study supports the assumption that the integration process used during symptom recall is not necessarily a 141

Chapter 8

mathematical average of moment-by-moment ratings. Instead, it suggests that various elements of the experience receive different weights in this mental aggregation, with some not being taken into account at all (Broderick et al., 2006; see Fredrickson, 2000 for a review). This reasoning is expressed in a cognitive heuristic related to saliency and recency known as the peak-end (PE) effect (e.g., Kahneman et al., 1993; Redelmeier et al., 2003; Stone, Broderick, Kaell, DelesPaul, & Porter, 2000). The PE “rule” posits that the global evaluation of a past experience is primarily determined by two distinctive moments: the worst or most intense part of the experience (peak) and the intensity at the final moments of experience (end). While those moments are weighted heavily in the final rating, the duration of the experience has relatively little impact on the evaluation (e.g., Koyama, Koyama, Kroncke, & Coghill, 2004). This heuristic accounted for retrospective symptom ratings not only in clinical settings (Chajut et al., 2014; Redelmeier & Kahneman, 1996; Redelmeier et al., 2003; Stone et al., 2000), but also in the laboratory (Bogaerts et al., 2012; Kahneman et al., 1993). In Chapters 3 and 4 we have described the results of our experimental studies, which replicated the PE effect for experimentally induced pain (Kahneman et al., 1993) and dyspnea (Bogaerts et al., 2012) in non-clinical samples. Moreover, we demonstrated that the PE effect was absent in patients with medically unexplained dyspnea, replicating an earlier study from our lab (Bogaerts et al., 2012). The robust finding regarding the lack of PE effect in patients with MUS suggests that the way those patients encode and/or retrieve somatic memories differs from healthy persons. One of the reasons why the PE effect may not occur may be related to the less clear-cut distinction between the different phases of a somatic experience (e.g., a peak and an end). This might happen when individuals are overwhelmed by affectivemotivational elements of the experience, possibly combined with inadequate processing of sensory-perceptual details. 8.2.2. Recall bias can increase over time Earlier findings from the symptom memory literature demonstrated that longer reporting periods are often associated with a gradual increase in retrospective symptom reporting (e.g., Broderick et al., 2008; Houtveen & Oei, 2007). This observation was partially supported by the results of our experimental studies, which included a longitudinal follow-up assessment up to two (Chapters 3-5) and four (Chapter 6) weeks. However, the accordance with earlier findings depended on the type of symptom induction. While retrospective ratings of dyspnea increased over time in a number of conditions (i.e., Chapter 3, among high HSR; 142

General discussion

Chapter 5, in affective PF; Chapter 6; but not in Chapter 4), retrospective pain ratings given after two/four weeks did not significantly differ from immediate ratings (Chapters 3, 5, and 6). This different course of retrospective symptom ratings could be related to the level of distress associated with the specific somatic experience. In our studies, dyspnea induction was consistently perceived as more aversive than pain induction. By contrast, pain induction used in our experiments was a single/unique experience of localized (i.e., a hand or a small area of a forearm) acute noxious stimulus of a rather low intensity. Because memory for this type of pain was previously reported as less biased (e.g., Roche & Gijsbers, 1986), the characteristics of our noxious stimulus could explain the lack of increase in retrospective ratings of pain over time. 8.2.2.1. The role of the characteristics of the somatic experience Three of the experimental studies reported in this dissertation included a withinsubject induction of two aversive somatic experiences: pain via a cold stimulus and dyspnea via a rebreathing task. As mentioned above, throughout those studies (Chapters 3, 5, and 6), biases in retrospective symptom reports were predominantly observed in the ratings of dyspnea, and less so in self-reported pain. In all studies, the dyspnea induction was rated as more aversive and threatening than the pain induction. Moreover, it elicited higher state NA, lower sense of control, as well as more systemic and threatening symptoms (i.e., rapid heartbeat, dizziness, chest tightness). This suggests that the retrospective bias in symptom reporting emerges mainly with somatic experiences characterized by strong distress/ negative affective aspects. In support of this suggestion, a number of studies has previously demonstrated a robust relationship between increased retrospective symptom reporting and negative emotional states associated with symptom experience including distress (Everts et al., 1999; Jamison et al., 1989), state negative affect (Gedney & Logan, 2006), and state anxiety (Eli et al., 2000; Gedney et al., 2003; Noel et al., 2012b). It should be noted, though, that based on our findings it cannot be claimed that dyspneic episodes are more prone to memory distortions than noxious experiences. In order to compare the effects of dyspnea and pain (representing the different types of bodily symptoms), the magnitude of the two sensations (i.e., intensity and unpleasantness) should be comparable (von Leupoldt & Dahme, 2007a). It is possible that an induction of a more

143

Chapter 8

unpleasant noxious pain stimulus could lead to similar recall biases as observed for dyspnea ratings in our studies. 8.2.2.2. The role of individual differences Retrospective reporting of past symptoms can be also influenced by personal characteristics, particularly the ones modulating the valence and intensity of affective responses to bodily experiences. For example, the personality traits including negative affectivity (Larsen, 1992; Levine & Safer, 2002; Watson & Pennebaker, 1989), anxiety (Kent, 1985; Suls & Howren, 2012), and pain catastrophizing (Lefebvre & Keefe, 2002; Noel et al., 2015) have been frequently related to increased retrospective symptom reports. In this project, the focus was placed on individual differences in habitual symptom reporting, which is characterized by frequent reporting of somatic complaints in daily life. In Chapter 3 we have demonstrated that individuals high on HSR do not only report high frequency of complaints, but also show greater bias in retrospective ratings of experimentally induced dyspnea compared to low HSR. This finding is in line with a previous observation from a diary study (Houtveen & Oei, 2007), that showed that retrospective overreporting of naturally occurring bodily complaints was more pronounced in high than in low HSR. HSR shows a robust relationship with trait NA (Bogaerts et al., 2015), which was also confirmed in our studies (Chapters 3, 4, and 5). Throughout the studies, higher levels of HSR/NA were related to more negative affective responses elicited by the symptom inductions, including higher state NA, anxiety, and perception of threat, which is consistent with earlier findings (e.g., Bogaerts et al., 2008; Han et al., 2004; Stegen et al., 2001; Wan et al., 2012). Interestingly, the mediation analyses performed in Chapters 3 and 4 demonstrated that those elevated affective responses elicited by the symptom induction (i.e., state NA and anxiety, respectively) mediated the observed group-related differences in retrospective symptom reporting. This shows that the effect of individual characteristics such as HSR or MUS on retrospective symptom ratings is due to increased intensity of the affective reactions towards somatic stimuli. In addition, HSR/MUS was also related to higher symptom ratings during the symptom induction as measured by a symptom checklist after the trials and concurrent ratings given during the trials (Chapters 3-6). Nevertheless, stronger self-reported responses to symptom inductions in non-clinical high HSR group were not always accompanied by more intense physiological reactions, as measured by respiratory parameters (minute ventilation, FetCO2) during the dyspnea induction (Chapter 3 and 5). This implies that the HSR-related 144

General discussion

differences in self-reported symptoms are not due to augmented psychophysiological responses. On the other hand - and similar to previous studies (Bogaerts, Van Eylen, et al., 2010; Wan et al., 2012) - group differences in self-reported dyspnea ratings coincided with higher MV in the patient group (Chapter 4). However, these group-related differences in selfreported and physiological reactions were no longer significant once state anxiety during the trials was taken into account. This indicates that the differences between patients and controls in concurrent dyspnea ratings were driven by the anxiety elicited by the dyspneic experience, once again highlighting the influential role of affective processes in symptom reporting (see also De Peuter et al., 2004). 8.2.2.3. The role of the type of information processing The final factor which was shown to influence retrospective symptom ratings in our studies, was the information processing focus (PF). More specifically, in Chapter 5 we showed that directing attention to emotions elicited by a bodily stimulus during the experience (i.e., at encoding) resulted in stronger negative affective responses (i.e., arousal and threat) to symptom induction compared to focusing on sensory details. Affective PF did not only affect the intensity and valence of the affective component during the induction trials, but also resulted in an increase in retrospective dyspnea reporting over time. This finding is consistent with other studies showing that attending to and processing of the distressing aspects of bodily stimulus (affective PF) leads to more negative outcomes (Bogaerts et al., 2008; Crane & Martin, 2003; Michael & Burns, 2004). Moreover, the adverse effects of affective processing focus seem to be confined to (a) the memory encoding phase (cf. Chapter 5 vs. Chapter 6), and (b) conditions characterized by strong distress/ negative affective aspects (cf. dyspnea vs. pain, Chapter 5). It is possible that when the experience is not perceived as threatening nor distressing, attending to the emotions elicited by the experience does not exert comparable negative effects on symptom ratings. Additionally, manipulating the processing focus in our studies seemed to reduce the effect of individual differences in HSR on self-reported symptoms. More specifically, an increase in retrospective dyspnea reporting after the affective PF manipulation was observed in both high and low HSR, while it previously was found only in the high HSR group (Chapter 3, symptom induction without PF manipulation). Similarly, after the sensory PF manipulation, the dyspnea ratings of neither high nor low HSR increased over time. This observation suggests that the biased symptom reporting in high HSR is related to a stronger affective processing focus that they typically apply when encountering somatic stimuli. 145

Chapter 8

Our processing focus manipulation seems to have wiped out the between-group differences underlying high and low HSR. However, more research is needed to investigate the characteristics of information processing focus and its association with HSR. 8.2.3. Sensory and affective processes in retrospective symptom reporting The investigation of retrospective symptom reporting in this doctoral project was grounded upon the dual-process perspective (e.g., Leventhal et al., 1979; Price, 2000), which postulates that a symptom rating is a result of an interplay between the sensory-perceptual and affective-motivational components of a somatic experience. Specifically, we proposed that distortions in retrospective self-reported symptoms are related to the relative dominance of the affective aspects over the sensory ones, potentially resulting in less detailed memory of sensory details. The psychometric and behavioral studies of this thesis tested and gave a considerable support to both of our claims. First, the dual-process account received further confirmation from a psychometric perspective (Chapter 2). Second, through the experimental studies we have demonstrated that both concurrent and retrospective symptom ratings are influenced by the negative affective states associated with a somatic experience. Moreover, those affective responses to bodily stimuli can be modulated/enhanced by factors related to individual differences and information processing focus. The findings discussed above advocate for the critical role of somatic information processing in self-reported symptoms and enrich our understanding of the mechanisms underlying symptom reporting. Additionally, they are also relevant to understand why certain individuals seem to be at a greater risk to exhibit biased symptom reporting. In particular, the dual-process approach could help to elucidate the processes leading to increased self-reported frequency and (according to our findings) intensity of somatic complaints (HSR) or somatic complaints even in the absence of the relevant physiological change (MUS). 8.2.3.1. The role of sensory-perceptual processing According to the dual-process perspective, overreporting of somatic symptoms could originate from reduced specificity of a sensory component due to a less detailed information processing and/or stronger influence of affective component. Even though the first assumption was not explicitly tested in this project, findings from various experimental studies suggest that sensory processing of bodily information is less detailed among individuals with HSR/MUS. This would be consistent with studies demonstrating (a) a poor correspondence between the intensity of self-reported symptoms and relevant physiological changes (Bogaerts et al., 2008; Bogaerts, Van Eylen, et al., 2010), (b) an absence of the peak146

General discussion

end effect for somatic episodes, typically found in healthy participants (Bogaerts et al., 2012; also Chapter 4), which may suggest that retrospective symptom reporting is less determined by the distinct elements of the experience, (c) elevated susceptibility to nocebo effects (Van den Bergh et al., 1997), and (d) lower differentiation among various respiratory sensations together with increased bias due to a priori assumptions when classifying them (Petersen, Van Staeyen, et al., 2015). One of the findings reported in this dissertation adds to this growing body of evidence. Specifically, the novel finding reported in Chapter 7 showed that autobiographical memory specificity for health-related memories was reduced among patients with MUS/SSD compared with the healthy controls, suggesting a less detailed processing of health-related material. Taken together with earlier data, it could indicate that the patients encode and store limited sensory-perceptual details of health-related events. From the dual-process perspective, this could give way to increased impact of affective component, quite certainly negative due to high symptom-related distress and anxiety-related personality traits (e.g., de Waal et al., 2004; Henningsen et al., 2003; Kroenke, 2003a) on symptom ratings. However, future studies should explore the possible effects of limited sensory details and reduced memory specificity on somatic symptom reporting. 8.2.3.2. Affective processes and symptom overreporting in HSR/MUS While the role of sensory-perceptual processing of somatic experience on symptom reporting requires further exploration, the findings of this thesis clearly highlight the important role of dominant affective-motivational processing – especially the perceived aversiveness of the somatic input and the factors that could influence it. In particular, greater sensitivity for negative and threatening information as indicated by trait anxiety (see BarHaim, Lamy, Pergamin, Bakermans-Kranenburg, & van Ijzendoorn, 2007, for a review) and negative affectivity (e.g., Lonigan & Vasey, 2009; Stegen et al., 2001; Watson & Pennebaker, 1989) is of relevance here. Individuals with an over-reactive affective-motivational system (Hariri et al., 2000; Yiend, 2010), for example those high on trait NA or anxiety, might respond with stronger affective-motivational responses to bodily sensations, leading to greater influence of negative affective aspects on symptom reporting. This fits with earlier studies demostrating (a) a strong relationship between habitual symptom reporting and trait NA (Bogaerts et al., 2015), (b) comorbid emotional disorders (anxiety/depression) among patients with MUS/SSD (de Waal et al., 2004; Henningsen et al., 2003; Wessely et al., 1999), and (c) cognitive biases to somatic information observed in those groups (e.g., Gropalis, 147

Chapter 8

Bleichhardt, Hiller, & Witthöft, 2013; Lim & Kim, 2005; Martin, Buech, Schwenk, & Rief, 2007; Witthöft, Gerlach, & Bailer, 2006). This dissertation complemented the existing evidence for the role of the stronger affective component in biased symptom experience in those groups, by showing that high HSR/MUS respond to symptom inductions with stronger negative states, which then result in elevated retrospective symptom reports. Furthermore, the stronger negative affective processing together with less detailed memories in individuals characterized with HSR/MUS could have a combined effect on their symptom ratings, making them more sensitive to the biasing influence of affective states. It is generally known that unpleasant affective cues can enhance the perception/reporting of bodily signals, as demonstrated for experimentally induced dyspnea (Bogaerts et al., 2008; von Leupoldt, Mertz, Kegat, Burmester, & Dahme, 2006), pain (de Wied & Verbaten, 2001; Meagher, Arnau, & Rhudy, 2001; Weisenberg, Raz, & Hener, 1998), and responses to esophageal stimulation (Phillips et al., 2003). Moreover, the negative cues are typically shown to have a more pronounced effect on symptom experience among individuals with high HSR/NA. Particularly in those groups, negative affective cues, such as unpleasant odors or negative pictures, can automatically trigger symptom experiences even in the absence of relevant physiological changes (Bogaerts, Janssens, et al., 2010; Constantinou et al., 2013; Van den Bergh et al., 2001, 1997, 1998; Van den Bergh, Winters, Devriese, & Van Diest, 2002). This does not only highlight the important role of emotional cues and states in somatic experience, but it also indicates that the connection between negative affective states and symptom reporting can be more easily established and activated among those high in HSR/NA. This approach matches the schema-based approach to symptom (mis)perception of Brown (2004), according to which the overactivation of maladaptive symptom schemata, triggered by contextual and environmental cues, is proposed as the mechanisms leading to the development of medically unexplained symptoms (R. J. Brown, 2004). In the case of high HSR/NA, the affective cues could amplify the existing (negative) symptom representations in memory leading to the misinterpretation of incoming somatic stimuli and biased symptom reporting. Moreover, the information processing focused on emotions elicited by the bodily experience (affective processing focus) could make the affective cues more salient, subsequently leading to a distorted symptom experience. Finally, an alternative view on the interaction between the affective and the sensory processing should be mentioned. We have assumed that stronger affective processing could compromise processing of sensory information leading to less detailed somatic memories. This argument was based on the dual-process perspective which posits that the sensory148

General discussion

perceptual and affective-motivational aspects are processed in parallel, possibly implying that the focus on one aspect (e.g., emotions) may weaken the focus on other aspects (e.g., sensory material). However, as those processes are not completely independent (e.g., Price, 2000), it is possible that increased affective processing of aversive bodily information could have another effect on sensory processing. In particular, increased negative affect and arousal could lead to deeper encoding of all the elements of this emotional experience (e.g., Anderson & Phelps, 2001; Talmi, 2013; Talmi, Schimmack, Paterson, & Moscovitch, 2007), including the sensory-perceptual details. However, the effects of arousal on processing and remembering of various items may depend on their salience (arousal-biased competition; Mather & Sutherland, 2011). In the case of individuals with HSR/NA, due to their increased levels of trait anxiety and NA, it could be expected that negative and threatening information will be more salient than perceptual details, leading to the stronger encoding of the affective elements. 8.3. Clinical implications The findings of this doctoral project have several implications for the interventions targeting biases in retrospective symptom reporting, as well as for the assessment of symptoms in both clinical and research settings. Self-reported symptoms are important in the context of health care, as they have an impact on the health/illness-related behaviors. The actions taken by doctors do not only depend on the results of medical examinations and laboratory tests. Very often a diagnostic process, a decision to continue medical testing, and a choice of treatment will depend on how patients describe their complaints. Moreover, the way past somatic experiences are remembered can also affect future decisions of patients themselves. For example, if a medical procedure was memorized as very aversive and threatening, the patient may avoid undergoing this procedure in the future (Erskine et al., 1990; Kent, 1985; Redelmeier et al., 2003). This could have serious consequences not only for treatment, but also for prevention. For this reason, both the help-seeking individuals and the medical specialists should be aware that symptom ratings might be biased under certain conditions and that there are a number of factors that can influence this bias. Along with other studies in the field of symptom perception, this thesis highlights the role of affective states in bias observed in symptom reporting (e.g., De Peuter et al., 2004; Gijsbers van Wijk & Kolk, 1997; Janssens, Verleden, De Peuter, Van Diest, & Van den Bergh, 2009; Watson & Pennebaker, 1989; see

149

Chapter 8

Van den Bergh, Bogaerts, & Van Diest, 2015, for a review). In our studies, perceiving experimentally induced symptoms as unpleasant was related to retrospective overreporting of those symptoms. Similar effects of affective states and traits were also found for symptoms of asthma, with negative affect and catastrophic thinking being associated with overperception of asthma symptoms (De Peuter et al., 2007; De Peuter, Lemaigre, Van Diest, & Van den Bergh, 2008; Janssens et al., 2009). Moreover, affect (both valence and arousal) was found to influence evaluations of daily reported health satisfaction even before concrete symptoms were present (Whitehead & Bergeman, 2013). Finally, the importance of cognitive and emotional aspects of bodily complaints in symptom reporting was also acknowledged in the recent revision DSM-5 (American Psychiatric Association, 2013) concerning somatic symptom disorder. An important modification was related to the inclusion of psychological criteria, that is, excessive feelings, thoughts, and behaviors linked to the experienced symptoms, which are perceived as persistent, distressing, and impairing. The understanding of the factors that bias symptom reporting might help with improving the accuracy of self-reported symptoms, necessary for symptom assessment in both medical and research settings. Moreover, the knowledge about the mechanisms underlying symptom reporting can lead to the development of adequate interventions which could limit the overreporting of past symptoms. 8.3.1. Reducing symptom overreporting The abovementioned observations clearly emphasize the crucial role that is played by the affective states in both concurrent and retrospective symptom reporting. To counteract this negative influence and to reduce the bias in symptom reporting, more attention should be brought to the techniques and interventions which specifically tackle the way patients attend to somatic information and focus on reducing the symptom-related distress. Based on our studies (Chapter 5), it could be suggested that guiding the attention towards the sensory features of the experience and processing them in a neutral way could reduce the bias in retrospective symptom reports. By contrast, allowing the patients to focus primarily on the negative emotions elicited by the somatic event could increase the influence of the negative affective component on self-reported symptoms, leading to more negative evaluation of the experience. In line with those suggestions, increasing the attention to experienced bodily sensations was shown to have beneficial effects on reported symptoms (e.g., Kerstner et al., 2015; Prins, Decuypere, & Van Damme, 2014; Schaefer, Egloff, Gerlach, & Witthöft, 2014). 150

General discussion

Schaefer and colleagues (2014) used the heartbeat perception training to enhance interoceptive accuracy in patients with somatoform disorders and demonstrated that even a single experimental training procedure was able to reduce the severity of concurrently reported symptoms. The study of Kerstner et al. (2015) has examined the effects of a 2-week attribution modification training on health anxiety symptoms as well as on evaluation and memory processes in patients with pathological health anxiety. They showed that training patients to carefully scan their bodies to detect bodily sensations in a neutral way (by replacing the catastrophic symptom attributions with the neutral ones) resulted in lower levels of reported symptoms as well as smaller memory response bias to symptom words. Finally, the effects of attending to bodily sensation in a moment-to-moment, non-judgmental manner (mindfulness intervention) were also investigated among healthy participants (Prins et al., 2014). The effects of mindfulness manipulation (vs. distraction) during a heat pain induction on affective pain ratings were moderated by dispositional pain catastrophizing. In particular, the mindfulness induction led to lower affective pain ratings than the distraction induction among high pain catastrophizers, while the opposite pattern was observed for low pain catastrophizers. In general, focusing on the sensory aspects of a somatic experience in a nonevaluative, neutral way, even by means of short-term interventions, can be perceived as a promising technique to reduce symptom reporting in clinical populations. Nevertheless, as the abovementioned studies explored the effects of such interventions on concurrent symptom experience only, more research is needed to test their impact on symptom memories. In addition, for the manipulation to be effective, it should take place during the experience itself, that is, before it becomes consolidated in memory. Directing attention to different features of somatic experience during memory retrieval (which does not include reexperiencing of the bodily stimulus) does not seem to have any effect on symptom recall (Chapter 6). In order to modify prior emotional memories, the old memories should be reactivated together with the engagement in new emotional experiences. This will enable a subsequent update of the reactivated memories through the process of reconsolidation. The approach focusing on the role of emotional arousal and memory reconsolidation in therapeutic change was recently reviewed by Lane et al. (2015). 8.3.2. Improving symptom assessment Intensity of somatic experiences seems poorly represented in memory. In all of our experimental studies, pronounced biases in intensity reports were demonstrated even after a very short delay (immediately after the end of the experience). As already mentioned in this 151

Chapter 8

chapter, memory-based ratings depend on retrospective integration processes which do not rely on simple mathematical averaging. Instead, the more memorable moments of the experience are given heavier weights in the final rating. Furthermore, the results of our studies suggest that an increase in the retention interval does not further increase bias in the ratings. More specifically, no differences were observed between the intensity ratings given immediately (the end of a trial), after a longer delay (the end of an experiment), and, in some cases, even after two or four weeks. This could suggest that all memory-based ratings follow similar retrospective integration processes regardless of the retention interval (what could change, however, is the weight given to different elements of the experience; e.g., Gedney et al., 2003). In addition, once a retrospective rating has been given, it may be encoded in memory as well, leading to stabilization of the memory. This finding has at least two implications for the assessment of symptoms. First, if the intensity of the actual experience is of primary interest, then it is important to collect the intensity ratings in real-time during the experience, and to refrain from using memory-based ratings. If feasible, real-time measures, such as ecological momentary assessment (Shiffman et al., 2008) or concurrent symptom ratings are recommended (Conner & Feldman Barrett, 2012). However, to ensure that the ratings capture the ongoing state at the moment of assessment, careful attention should be paid to the phrasing used in the prompt questions, even in the EMA research. Namely, the time indicator should refer to “right now/at this moment” rather than to “recently”, “in the last few minutes”, or “from the last measurement”. Second, if the retrospective rating is unavoidable or necessary, then a different outcome measure than the typically used average level of symptom intensity could be considered. It is possible that other outcome indices, such as the variability of symptoms or the peak intensity, would be more relevant for patients to report their symptoms. However, more research considering the optimization of self-reported symptoms is still needed. In addition, it is also crucial to mention that those two types of ratings should not be compared as better or worse, because they reflect different types of knowledge and tap from different functional selves (see Conner & Feldman Barrett, 2012, for an extensive description of this approach). According to this approach, momentary ratings reflect the “experiencing self”, while retrospective ones correspond to the “remembered self” (Kahneman & Riis, 2005). This distinction has important implications for psychosomatic research. The studies interested in the correspondence between physiological responses and self-reports should use real-time or state ratings, because the experiential self (momentary ratings) shows a closer connection with bodily processes than the remembering self (recalled or trait ratings). For 152

General discussion

example, autonomic activation was found to be more related to momentary states (stress, positive affect) than to trait characteristics (Kamarck et al., 2005; Steptoe, Leigh Gibson, Hamer, & Wardle, 2007). On the other hand, research concerning future health decisions like routine check-ups would profit from the retrospective symptom ratings, as patients’ somatic memories play a key role in prospective decision making (e.g., Redelmeier et al., 2003). 8.4. Strengths and limitations This doctoral project contributed to the understanding of memory processes in retrospective symptom reporting by exploring this issue through multiple studies using various methodologies (psychometric, experimental, and correlational). The multiple experimental studies based on two different standardized symptom induction paradigms (respiratory stimulus via rebreathing, and cold pain stimuli) in analogous participant samples and in clinical groups are the major strength of this project. This approach gave the possibility to replicate of a number of findings within one doctoral project as well as to directly compare the outcomes of different studies. The experimental approach facilitated the control over the bodily stimuli delivered to participants along with the measurement of both self-reported symptoms and psychophysiological responses to bodily stimuli during the experience. The average of momentary ratings given during the symptom induction could be used as a comparison standard for the retrospective, memory-based ratings regarding the same event. In addition, the retrospective ratings were longitudinally followed up in a within-subject longitudinal manner up to 2 and 4 weeks after the experience. Nevertheless, several limitations of the reported studies can be identified. A first set of limitations is related to the information processing focus – its manipulation, assessment, and the measurement of its effects. Even though the PF manipulation at encoding (Chapter 5) was considered successful, based on the observed changes in affective responses, the actual focus on different elements of the experience was not directly measured. This measurement was not incorporated due to the within-subject design of this study: As participants were undergoing the same symptom inductions twice, with the only difference being the manipulation of processing focus, the explicit assessment of focus could make the aims of the study too obvious for the participants. This limitation was taken into account in the subsequent study (Chapter 6), when the measurement of focus on sensory and affective aspects of experience was included after the PF manipulation at retrieval stage. The reported focus ratings did not differ between the PF conditions, even though the participants confirmed that they understood

153

Chapter 8

and followed the instructions. This suggests that the temporary PF manipulation at the retrieval stage was not strong enough to affect the spontaneous PF, which could be one of the explanations of the overall absence of PF-related effects on affective and symptom ratings. However, the lack of effects could be also related to the choice of the outcome variables. More specifically, the self-reports of experienced symptoms did not include the differentiation between the sensory (symptom intensity) and the affective components (symptom unpleasantness) of somatic experience. As shown in earlier studies, the manipulation of PF might exert a differential effect on the way the intensity and unpleasantness of pain/dyspnea are perceived and remembered (e.g., Loggia, Mogil, & Bushnell, 2008; Rainville, Carrier, Hofbauer, Bushnell, & Duncan, 1999; von Leupoldt et al., 2006; von Leupoldt, Taube, Schubert-Heukeshoven, Magnussen, & Dahme, 2007). Unfortunately, such a distinction could not be included in this dissertation due to practical reasons (continuous momentary ratings during the symptom induction could only include one dimension). To overcome the abovementioned limitations, future studies interested in the effects of PF on symptom reporting might consider including a manipulation check by means of the self-reported focus ratings, preferably in a more complex form than a single-item measure, as well as the separate ratings of the symptom intensity/unpleasantness as the outcome measures if feasible. Second, the investigation of retrospective memory for experienced symptoms was limited to a 2-week time frame, which might be perceived as a short retention interval. However, no consensus exists over the preferred or recommended time frame that should be applied in research focused on somatic memories. Consequently, the retention intervals used in the studies range from a typical week (e.g., Giske et al., 2010; Stone et al., 2004) to several months (e.g., Bąbel, Pieniążek, & Zarotyński, 2015; Lefebvre & Keefe, 2002) or even more than a year (e.g., Gedney et al., 2003). It is assumed that with longer time frames, the accessibility to episodic memory (the “remembered”) decreases, giving way to the increased influence of the semantic memory (the “known/believed”) on the retrospective ratings (Robinson & Clore, 2002a). Event-specific episodic knowledge presumably becomes less accessible within a week or two unless rehearsed (Burt, Watt, Mitchell, & Conway, 1998). As the main focus of our studies was placed on the “remembered” symptoms, we have followed Brodie and Niven (2000) and chose the 2-week time frame to limit the influence of semantic knowledge on the symptom ratings. Recently, Geng et al. (2013) demonstrated that the shift between a short and long retention interval occurs between 3 to 7 weeks after the experience. This could suggest that longer retention intervals could be also applied to study retrospective 154

General discussion

ratings without increased impact of semantic memory. In line with this approach, in Chapter 6 we have shown that no differences were observed between the retrospective symptom ratings provided after 2 or 4 weeks after the experience. Another limitation of the current project is related to the participant sample, which in the majority of laboratory studies consisted of young and healthy women, usually university students. Gender is a well-established factor resulting in the differences in symptom reporting (e.g., Gijsbers van Wijk, Huisman, & Kolk, 1999; Van Diest et al., 2005) and pain experience (e.g., Fillingim & Maixner, 1995; Riley, Robinson, Wise, Myers, & Fillingim, 1998). For this reason, the number of men and women should be comparable to be able to control for the effects of gender in the studies investigating bodily experiences. However, as HSR is more common among women than men, the studies investigating symptom reporting among the individuals characterized by HSR usually rely on female sample to avoid uneven gender distribution. The reliance on healthy female sample could speak against the generalizability of our findings. However, the results regarding symptom perception and memory in high HSR groups were similar to the findings in clinical populations (Bogaerts et al., 2012; Walentynowicz, Bogaerts, et al., 2016), suggesting that cognitive processes and biases may be comparable in those groups. Nevertheless, to improve the generalizability of the proposed mechanisms future research should also focus on male and patient populations. 8.5. Unresolved issues and further research perspectives The findings of this dissertation extended the science of self-report by highlighting, in line with other cognitive-psychological models of symptom reporting, the influential role of affective processes. Moreover, they also showed that retrospective symptom ratings depend on memory-based integration processes, which are not based on a simple mathematical averaging, but rather take into account the weight assigned to various features of the somatic experience. Nevertheless, a number of unresolved questions and issues emerging from the work described in this dissertation would profit from further elaboration in future studies. 8.5.1. The relevance/weight of specific moments in symptom evaluation One of the conclusions drawn from our findings is that the weight given to different elements of the bodily experience influences memory-based symptom intensity ratings, with elements weighted more heavily having greater impact on the ratings than the ones with a lower weight. Earlier studies have consistently shown that the moments of the highest stimulus intensity (peak) and the final moments (end) are weighted more heavily in the 155

Chapter 8

retrospective recall (e.g., Jensen et al., 2008; Kahneman et al., 1993; Redelmeier & Kahneman, 1996; Stone et al., 2000), while the duration of experiences is not well represented in memory and largely neglected. In addition, other features, such as affective responses to somatic experiences, may also contribute to ratings of past states, as shown throughout this thesis. As there are considerable individual differences in reliance on the peak and end effects (Schneider, Stone, Schwartz, & Broderick, 2011), a tendency to weight certain features of a bodily experience more in the global evaluation may be grounded in a relatively general and stable predisposition. In that respect, it is interesting to note that bias is not always characteristic of maladaptive processing. Indeed, giving different weights to various elements of the experience may be an adaptive response, possibly leading to a less aversive memory of the experience when aversive experiences eventually turn out better. This is what happened in non-clinical samples, while absence of this bias was characteristic of patients with MUS/SSD. To further explore this issue, future studies could examine the individual differences in weights given to different elements of a somatic experience and their effects on self-reports of symptom intensity. In a first step, the cognitive interviewing technique (Broderick et al., 2006) could be used to directly inquire about the strategies used to formulate a retrospective intensity rating and the importance given to various features of the event. Second, an experimental approach could be applied to explore whether increasing the weight of different aspects has a visible effect on symptom ratings. This could be achieved, for example, by the processing focus manipulations described in Chapters 5 and 6. Finally, to test which elements are (not) well represented in the memory, designs using temporal profiles of experienced symptoms could be proposed to measure the concordance between momentary and retrospective ratings of different features. In a similar fashion, continuous retrospective ratings of experienced arousal have been recently compared with ongoing physiological arousal measured during an initial experience, demonstrating high correspondence between those two measures (McCall, Hildebrandt, Bornemann, & Singer, 2015). In future studies that look into this issue, the assessment of symptom intensity could provide data to look at the memory representation of the sensory aspects of an experience, while the measurement of symptom unpleasantness could inform about the affective features. 8.5.2. Processing focus and memory consolidation Another set of findings offering solid grounds for future elaborations is related to the processing focus manipulation and its effect on the self-reported bodily symptoms during 156

General discussion

different stages of memory formation. In this dissertation, the focus was placed on the impact of PF during two of those stages, encoding and retrieval. The effects of processing focus on retrospective symptom reporting during the consolidation stage were unfortunately not explored in this doctoral project due to time constraints. While the role of PF during this phase has never been tested directly, the findings from studies concerning the other types of self-focused processing suggest that repeated or extended focus on affective-motivational aspects of the past somatic experience could lead to unfavorable consequences. Namely, the affective PF manipulation applied in this dissertation shows similarities to some unconstructive types of repetitive thinking, such as worry and abstract/analytical ruminative self-focus (e.g., Watkins, 2008). Ample evidence indicates that abstract ruminative processing, characterized by decontextualized and evaluative thinking about causes, meanings, and consequences of symptoms, is related to a number of negative outcomes including more overgeneral memories (e.g., Crane, Barnhofer, Visser, et al., 2007; Raes & Hermans, 2008; Watkins & Teasdale, 2001, 2004; Williams et al., 2007), increased negative generalization (Van Lier et al., 2016, 2014), poorer problem solving (Watkins & Moulds, 2005), more global self-judgments (Rimes & Watkins, 2005), and more negative mood related to prior failure (Watkins, 2004). Moreover, another form of repetitive thinking – (health) worry – is also assumed to play a crucial role in somatic health. According to the perseverative cognition hypothesis (Brosschot, Gerin, & Thayer, 2006), worry prolongs the cognitive representation of stressors and, consequently, stress-related physiological activation. Indeed, worry seems not only to be related to somatic complaints, but also to mediate the effects of stressful events on those complaints (Brosschot & van der Doef, 2006; Verkuil, Brosschot, Meerman, & Thayer, 2012; Verkuil, Brosschot, & Thayer, 2007). Interestingly, a simple worry intervention was able to decrease the frequency of worry episodes for six days, what subsequently predicted the reduction in self-reported health complaints (Brosschot & van der Doef, 2006). Finally, a recent study investigating the effects of experimentally induced worry on associative fear memory demonstrated that worry inducted after the initial fear acquisition can enhance the subsequent physiological fear responding (measured by fear-potentiated startle) and impair fear extinction (Gazendam & Kindt, 2012). Taking into account the negative consequences of unconstructive processing styles described above, it could be expected that recurrent focusing on the negative emotions elicited by the bodily experience (ruminative thinking) could maintain the affect associated with this event, and eventually could lead to biased

157

Chapter 8

retrospective symptom ratings. However, further research is needed to investigate this assumption. 8.5.3. Memory reconsolidation While the PF manipulation at encoding exerted effects on retrospective symptom ratings, probably due to the enhancement of negative affective responses to the stimulus, the manipulation at retrieval did not affect the recalled symptom ratings. Reconsolidation research could give a possible explanation for this observation. When a memory is reactivated, it may temporarily re-enter the unstable state during which it is vulnerable to change (Nader, 2003). In certain conditions, when the previously formed memory trace becomes destabilized, it can be updated and re-stabilized during a process called reconsolidation (e.g., Hupbach, Gomez, Hardt, & Nadel, 2007; Kindt, Soeter, & Vervliet, 2009; Nader, Schafe, & Le Doux, 2000; see Lee, 2009, for a review). The reconsolidation processes can modify episodic and autobiographical memories as well as fear memories (see Schwabe et al., 2014, for a review), and stressful experiences after memory reactivation can affect them. However, in contrast to a rather robust association between post-event stress and enhanced consolidation of the material (e.g., Cahill, Gorski, & Le, 2003), the direction of the effect on reconsolidation is not completely clear, with studies demonstrating both enhancement (Coccoz, Maldonado, & Delorenzi, 2011) and impairment of memory (Schwabe & Wolf, 2010) due to the post-retrieval stressor. Moreover, whether the reactivated memories will undergo the reconsolidation or not depends on the number of boundary conditions. For example, the memory update will not take place when no or little new information is present during memory retrieval (Sevenster et al., 2012; Wichert et al., 2013). Also, autobiographical memory performance was not affected when the reactivation took place without a post-recall stressor (Schwabe & Wolf, 2010). Because the temporary PF manipulation used in Chapter 6 during the retrieval stage did not lead to either increased distress or learning new information, which could update the memory for past somatic experience, it is probable that the reactivated memory trace was not subject to the reconsolidation processes. The number of studies investigating the effects of reconsolidation on human memory performance and therapeutic treatment effects has been rapidly increasing over the last 10 years. Future research on somatic memories and their modification could most certainly profit from a more intense exploration of this issue.

158

General discussion

8.5.4. The role of autobiographical memory specificity An additional issue that results from the findings of this dissertation is the role of autobiographical memory specificity in retrospective symptom reporting. In Chapter 7, we have demonstrated that patients with somatic symptom disorder recall fewer specific and more categoric memories in response to health-related cue words. Whereas this novel finding still requires replication in a larger sample, it has potentially important consequences for understanding how somatic memories are formed and retrieved in this patient group. rAMS, found across a broad range of psychopathological disorders (Williams et al., 2007) is not only a correlate of depressive symptoms, but is also considered a relatively stable marker of an unfavorable course of pathology, impacting severity of symptoms, illness duration, and treatment success (Sumner, 2012). While the prognostic role of rAMS for the course of affective disorders has already been established, future research could focus on investigating the extent to which rAMS can predict the development of somatic and depressive symptoms in SSD. Moreover, while it was assumed that health-related events are more self-relevant for the patients with SSD, and, consequently, would elicit more overgeneral memories (Barnhofer et al., 2007; Spinhoven et al., 2007), the perceived self-relevance of the used cue words was not rated. Future studies could further improve the health-related autobiographical memory test by assessing the self-relevance of the health-related cues in both patient and control groups. 8.5.5. Memory enhancement As a final note, attention should be brought to the terms used while reporting the findings of memory research. Studies focused on exploring the factors affecting memory enhancement often refer to an increase in memory performance/accuracy. However, as will be argued below, memory enhancement does not necessarily denote increased memory accuracy. The assessment of memory performance always depends on a reference point or a comparison standard. In the majority of memory studies, an external, measurable standard can be used (e.g., a word list consisting of 40 to-be-remembered items). In those cases, more accurate/better/enhanced memory will be reported when the number of correct items at time 2 will be higher than the number of correct items at time 1. However, in case of internal states such as emotions or bodily symptoms, an external comparison standard is not available and instead the comparisons are always performed within subject. In this sense, an accurate recall is understood as “the rating of a given person at time 2 did not differ from the rating given by this person at time 1”. Consequently, any deviation from the initial ratings will be perceived 159

Chapter 8

as less accurate or biased, leading to either retrospective over- or underreporting of symptoms. Consequently, the message conveyed under the term “accurate recall” in the research focused on internal states is completely different than in other memory research, where an external reference point is available. An interesting question that follows is what memory enhancement actually means in the context of symptoms. From an objective perspective, enhanced symptom memory should indicate a lack of change between the two symptom ratings over time. But memory enhancement

could

also

be

seen

as

an

increase

in

subjective

feeling

of

recollection/remembering. In this case, enhanced memory is perceived by the respondent as more vivid or easier to remember. However, this enhancement does not always correspond with the actual increase in accuracy. This is often reported in case of the enhancing effects of emotions on memories: Emotions can lead to more vivid memories, which are perceived as well-remembered, but not necessarily are correct (see Phelps & Sharot, 2008, for a review). In conclusion, researchers should be aware that the terms “memory enhancement” and “increased memory accuracy” should not be perceived as synonyms. Moreover, before focusing on the enhancement of somatic memories, it should be first specified what such enhancement denotes and how it could be measured.

160

References

9References

Ahles, T. A., Blanchard, E. B., & Leventhal, H. (1983). Cognitive control of pain: Attention to the sensory aspects of the cold pressor stimulus. Cognitive Therapy and Research, 7, 159–177. Aizawa, E., Sato, Y., Kochiyama, T., Saito, N., Izumiyama, M., Morishita, J., … Fukudo, S. (2012). Altered cognitive function of prefrontal cortex during error feedback in patients with irritable bowel syndrome, based on fMRI and dynamic causal modeling. Gastroenterology, 143, 1188–1198. American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed., text rev.). Washington, DC: Author. American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). Washington, DC: Author. Anderson, A. K., & Phelps, E. A. (2001). Lesions of the human amygdala impair enhanced perception of emotionally salient events. Nature, 411, 305–309. Aronson, K. R., Feldman Barrett, L., & Quigley, K. (2006). Emotional reactivity and the overreport of somatic symptoms: Somatic sensitivity or negative reporting style? Journal of Psychosomatic Research, 60, 521–530. Arrindell, W. A., & Ettema, J. H. M. (2003). Symptom Checklist (SCL-90). Handleiding bij een multidimensionele psychopathologie-indicator [SCL-90: Manual for a multidimensional psychopathology indicator]. Lisse, The Netherlands: Swets & Zeitlinger. Bąbel, P. (2015). Memory of pain and affect associated with migraine and non-migraine headaches. Memory, 23, 864–875. Bąbel, P. (2016). Memory of pain induced by physical exercise. Memory, 24, 548–559. Bąbel, P., Pieniążek, L., & Zarotyński, D. (2015). The effect of the type of pain on the accuracy of memory of pain and affect. European Journal of Pain, 19, 358–368. Bar-Haim, Y., Lamy, D., Pergamin, L., Bakermans-Kranenburg, M. J., & van Ijzendoorn, M. H. (2007). Threat-related attentional bias in anxious and nonanxious individuals: A meta-analytic study. Psychological Bulletin, 133, 1–24. Barnhofer, T., Crane, C., Spinhoven, P., & Williams, J. M. G. (2007). Failures to retrieve specific memories in previously depressed individuals: Random errors or contentrelated? Behaviour Research and Therapy, 45, 1859–1869. Barnhofer, T., Jong-Meyer, R., Kleinpaß, A., & Nikesch, S. (2002). Specificity of autobiographical memories in depression: An analysis of retrieval processes in a thinkaloud task. British Journal of Clinical Psychology, 41, 411–416. Baron, R. S., Logan, H., & Hoppe, S. (1993). Emotional and sensory focus as mediators of dental pain among patients differing in desired and felt dental control. Health Psychology, 12, 381–389. Barsky, A. J., Orav, E. J., & Bates, D. W. (2005). Somatization increases medical utilization and costs independent of psychiatric and medical comorbidity. Archives of General Psychiatry, 62, 903–910. Barsky, A. J., Peekna, H. M., & Borus, J. F. (2001). Somatic symptom reporting in women and men. Journal of General Internal Medicine, 16, 266–275. Barsky, A. J., & Wyshak, G. (1990). Hypochondriasis and somatosensory amplification. British Journal of Psychiatry, 157, 404–409. 161

References

Beauducel, A., & Wittmann, W. W. (2005). Simulation study on fit indexes in confirmatory factor analysis based on data with slightly distorted simple structure. Structural Equation Modeling, 12, 41–75. Bennett, A. V, Amtmann, D., Diehr, P., & Patrick, D. L. (2012). Comparison of 7-day recall and daily diary reports of COPD symptoms and impacts. Value in Health, 15, 466–474. Bennett, A. V, Patrick, D. L., Lymp, J. F., Edwards, T. C., & Goss, C. H. (2010). Comparison of 7-day and repeated 24-hour recall of symptoms of cystic fibrosis. Journal of Cystic Fibrosis, 9, 419–424. Bieg, M., Goetz, T., & Lipnevich, A. A. (2014). What students think they feel differs from what they really feel - Academic self-concept moderates the discrepancy between students’ trait and state emotional self-reports. PLoS ONE, 9, 1–9. Bogaerts, K., Janssens, T., De Peuter, S., Van Diest, I., & Van den Bergh, O. (2010). Negative affective pictures can elicit physical symptoms in high habitual symptom reporters. Psychology & Health, 25, 685–698. Bogaerts, K., Millen, A., Li, W., De Peuter, S., Van Diest, I., Vlemincx, E., … Van den Bergh, O. (2008). High symptom reporters are less interoceptively accurate in a symptom-related context. Journal of Psychosomatic Research, 65, 417–424. Bogaerts, K., Notebaert, K., Van Diest, I., Devriese, S., De Peuter, S., & Van den Bergh, O. (2005). Accuracy of respiratory symptom perception in different affective contexts. Journal of Psychosomatic Research, 58, 537–543. Bogaerts, K., Rayen, L., Lavrysen, A., Van Diest, I., Janssens, T., Schruers, K., & Van den Bergh, O. (2015). Unraveling the relationship between trait negative affectivity and habitual symptom reporting. PLoS ONE, 10, 1–15. Bogaerts, K., Van Eylen, L., Li, W., Bresseleers, J., Van Diest, I., De Peuter, S., … Van den Bergh, O. (2010). Distorted symptom perception in patients with medically unexplained symptoms. Journal of Abnormal Psychology, 119, 226–234. Bogaerts, K., Wan, L., Diest, I. Van, Stans, L., Decramer, M., & Van den Bergh, O. (2012). Peak-end memory bias in laboratory-induced dyspnea. Psychosomatic Medicine, 74, 974–981. Bohn, D., Bernardy, K., Wolfe, F., & Häuser, W. (2013). The association among childhood maltreatment, somatic symptom intensity, depression, and somatoform dissociative symptoms in patients with fibromyalgia syndrome: A single-center cohort study. Journal of Trauma & Dissociation, 14, 342–358. Bolton, J. E. (1999). Accuracy of recall of usual pain intensity in back pain patients. PAIN, 83, 533–539. Borg, G. (1982). Ratings of perceived exertion and heart rates during short-term cycle exercise and their use in a new cycling strength test. International Journal of Sports Medicine, 3, 153–158. Bradley, M. M., & Lang, P. J. (1994). Measuring emotion: The self-assessment manikin and the semantic differential. Journal of Behavior Therapy and Experimental Psychiatry, 25, 49–59. Brittlebank, A. D., Scott, J., Williams, J. M., & Ferrier, I. N. (1993). Autobiographical memory in depression: State or trait marker? The British Journal of Psychiatry, 162, 118–121. Broderick, J. E., Schwartz, J. E., Vikingstad, G., Pribbernow, M., Grossman, S., & Stone, A. A. (2008). The accuracy of pain and fatigue items across different reporting periods. PAIN, 139, 146–157. Broderick, J. E., Stone, A. A., Calvanese, P., Schwartz, J. E., & Turk, D. C. (2006). Recalled pain ratings: A complex and poorly defined task. The Journal of Pain, 7, 142–149. Brodie, E. E., & Niven, C. A. (2000). Remembering an everyday pain: The role of knowledge 162

References

and experience in the recall of the quality of dysmenorrhoea. PAIN, 84, 89–94. Brosschot, J. F., Gerin, W., & Thayer, J. F. (2006). The perseverative cognition hypothesis: A review of worry, prolonged stress-related physiological activation, and health. Journal of Psychosomatic Research, 60, 113–124. Brosschot, J. F., & van der Doef, M. (2006). Daily worrying and somatic health complaints: Testing the effectiveness of a simple worry reduction intervention. Psychology & Health, 21, 19–31. Brown, M. B., & Benedetti, J. K. (1977). On the mean and variance of the tetrachoric correlation coefficient. Psychometrika, 42, 347–355. Brown, R. J. (2004). Psychological mechanisms of medically unexplained symptoms: An integrative conceptual model. Psychological Bulletin, 130, 793–812. Brown, R. J. (2006). Medically unexplained symptoms: A new model. Psychiatry, 5, 43–47. Buchanan, T. W. (2007). Retrieval of emotional memories. Psychological Bulletin, 133, 761– 779. Burt, C. D. B., Watt, S. C., Mitchell, D. A., & Conway, M. A. (1998). Retrieving the sequence of autobiographical event components. Applied Cognitive Psychology, 12, 321–338. Cahill, L., Babinsky, R., Markowitsch, H. J., & McGaugh, J. L. (1995). The amygdala and emotional memory. Nature, 377, 295–296. Cahill, L., Gorski, L., & Le, K. (2003). Enhanced human memory consolidation with postlearning stress: Interaction with the degree of arousal at encoding. Learning & Memory, 10, 270–274. Cahill, L., & McGaugh, J. L. (1995). A novel demostration of enhanced memory associated with emotional arousal. Consciousness and Cognition, 4, 410–421. Cahill, L., & McGaugh, J. L. (1998). Mechanisms of emotional arousal and lasting declarative memory. Trends in Neurosciences, 21, 294–299. Carson, A. J., Ringbauer, B., Stone, J., McKenzie, L., Warlow, C., & Sharpe, M. (2000). Do medically unexplained symptoms matter? A prospective cohort study of 300 new referrals to neurology outpatient clinics. Journal of Neurology, Neurosurgery & Psychiatry, 68, 207–210. Chajut, E., Caspi, A., Chen, R., Hod, M., & Ariely, D. (2014). In pain thou shalt bring forth children: The peak-and-end rule in recall of labor pain. Psychological Science, 25, 2266– 2271. Christianson, S.-Å., & Loftus, E. F. (1991). Remembering emotional events: The fate of detailed information. Cognition & Emotion, 5, 81–108. Cioffi, D. (1991). Sensory awareness versus sensory impression: Affect and attention interact to produce somatic meaning. Cognition & Emotion, 5, 275–294. Coccoz, V., Maldonado, H., & Delorenzi, A. (2011). The enhancement of reconsolidation with a naturalistic mild stressor improves the expression of a declarative memory in humans. Neuroscience, 185, 61–72. Cohen, L. L., Blount, R. L., Cohen, R. J., Ball, C. M., McClellan, C. B., & Bernard, R. S. (2001). Children’s expectations and memories of acute distress: Short- and long-term efficacy of pain management interventions. Journal of Pediatric Psychology, 26, 367– 374. Conner, T., & Feldman Barrett, L. (2012). Trends in ambulatory self-report: The role of momentary experience in psychosomatic medicine. Psychosomatic Medicine, 74, 327– 337. Constantinou, E., Bogaerts, K., Van Diest, I., & Van den Bergh, O. (2013). Inducing symptoms in high symptom reporters via emotional pictures: The interactive effects of valence and arousal. Journal of Psychosomatic Research, 74, 191–196. 163

References

Constantinou, E., Bogaerts, K., Van Diest, I., & Van den Bergh, O. (2016). Influences of mood on information processing styles in high and low symptom reporters. Health Psychology Report, 3, 300–311. Conway, M. A. (2005). Memory and the self. Journal of Memory and Language, 53, 594– 628. Conway, M. A. (2009). Episodic memories. Neuropsychologia, 47, 2305–2313. Conway, M. A., & Pleydell-Pearce, C. W. (2000). The construction of autobiographical memories in the self-memory system. Psychological Review, 107, 261–288. Correa, Á., Miró, E., Martínez, M. P., Sánchez, A. I., & Lupiáñez, J. (2011). Temporal preparation and inhibitory deficit in fibromyalgia syndrome. Brain and Cognition, 75, 211–216. Craig, A. D. (2003). A new view of pain as a homeostatic emotion. Trends in Neurosciences, 26, 303–307. Crane, C., Barnhofer, T., Visser, C., Nightingale, H., & Williams, J. M. G. (2007). The effects of analytical and experiential rumination on autobiographical memory specificity in individuals with a history of major depression. Behaviour Research and Therapy, 45, 3077–3087. Crane, C., Barnhofer, T., & Williams, J. M. G. (2007). Cue self-relevance affects autobiographical memory specificity in individuals with a history of major depression. Memory, 15, 312–323. Crane, C., & Martin, M. (2003). Illness schema and level of reported gastrointestinal symptoms in irritable bowel syndrome. Cognitive Therapy and Research, 27, 185–203. Creed, F. H., Davies, I., Jackson, J., Littlewood, A., Chew-Graham, C., Tomenson, B., … McBeth, J. (2012). The epidemiology of multiple somatic symptoms. Journal of Psychosomatic Research, 72, 311–317. Dalgleish, T., Rolfe, J., Golden, A.-M., Dunn, B. D., & Barnard, P. J. (2008). Reduced autobiographical memory specificity and posttraumatic stress: Exploring the contributions of impaired executive control and affect regulation. Journal of Abnormal Psychology, 117, 236–241. Dalgleish, T., Tchanturia, K., Serpell, L., Hems, S., Yiend, J., de Silva, P., & Treasure, J. (2003). Self-reported parental abuse relates to autobiographical memory style in patients with eating disorders. Emotion, 3, 211–222. Dalgleish, T., Williams, J. M. G., Golden, A.-M. J., Perkins, N., Barrett, L. F., Barnard, P. J., … Watkins, E. (2007). Reduced specificity of autobiographical memory and depression: The role of executive control. Journal of Experimental Psychology: General, 136, 23– 42. De Clercq, A., Verschuere, B., De Vlieger, P., & Crombez, G. (2006). Psychophysiological Analysis (PSPHA): A modular script-based program for analyzing psychophysiological data. Behavior Research Methods, 38, 504–510. De Peuter, S., Lemaigre, V., Van Diest, I., & Van den Bergh, O. (2008). Illness-specific catastrophic thinking and overperception in asthma. Health Psychology : Official Journal of the Division of Health Psychology, American Psychological Association, 27, 93–9. De Peuter, S., Put, C., Lemaigre, V., Demedts, M., Verleden, G., & Van den Bergh, O. (2007). Context-evoked overperception in asthma. Psychology & Health, 22, 737–748. De Peuter, S., Van Diest, I., Lemaigre, V., Verleden, G., Demedts, M., & Van den Bergh, O. (2004). Dyspnea: The role of psychological processes. Clinical Psychology Review, 24, 557–581. de Waal, M. W. M., Arnold, I. A., Eekhof, J. A. H., & van Hemert, A. M. (2004). Somatoform disorders in general practice: Prevalence, functional impairment and comorbidity with anxiety and depressive disorders. The British Journal of Psychiatry, 164

References

184, 470–476. de Waal, M. W. M., Arnold, I. A., Spinhoven, P., Eekhof, J. A. H., & van Hemert, A. M. (2005). The reporting of specific physical symptoms for mental distress in general practice. Journal of Psychosomatic Research, 59, 89–95. de Wied, M., & Verbaten, M. N. (2001). Affective pictures processing, attention, and pain tolerance. PAIN, 90, 163–172. Debeer, E., Hermans, D., & Raes, F. (2009). Associations between components of rumination and autobiographical memory specificity as measured by a Minimal Instructions Autobiographical Memory Test. Memory, 17, 892–903. Debeer, E., Raes, F., Williams, J. M. G., & Hermans, D. (2011). Context-dependent activation of reduced autobiographical memory specificity as an avoidant coping style. Emotion, 11, 1500–1506. Dick, B. D., Verrier, M. J., Harker, K. T., & Rashiq, S. (2008). Disruption of cognitive function in fibromyalgia syndrome. PAIN, 139, 610–616. Dolan, R. J. (2002). Emotion, cognition, and behavior. Science, 298, 1191–1194. Dolcos, F., LaBar, K. S., & Cabeza, R. (2004). Interaction between the amygdala and the medial temporal lobe memory system predicts better memory for emotional events. Neuron, 42, 855–863. Dolcos, F., LaBar, K. S., & Cabeza, R. (2005). Remembering one year later: Role of the amygdala and the medial temporal lobe memory system in retrieving emotional memories. Proceedings of the National Academy of Sciences, 102, 2626–2631. Ehring, T., Zetsche, U., Weidacker, K., Wahl, K., Schönfeld, S., & Ehlers, A. (2011). The Perseverative Thinking Questionnaire (PTQ): Validation of a content-independent measure of repetitive negative thinking. Journal of Behavior Therapy and Experimental Psychiatry, 42, 225–232. Eich, E., Reeves, J. L., Jaeger, B., & Graff-Radford, S. B. (1985). Memory for pain: Relation between past and present pain intensity. PAIN, 23, 375–380. Eli, I., Baht, R., Kozlovsky, A., & Simon, H. (2000). Effect of gender on acute pain prediction and memory in periodontal surgery. European Journal of Oral Sciences, 108, 99–103. Engelen, U., De Peuter, S., Victoir, A., Van Diest, I., & Van den Bergh, O. (2006). Verdere validering van de Positive and Negative Affect Schedule (PANAS) en vergelijking van twee nederlandstalige versies. Gedrag & Gezondheid, 34, 89–102. Erskine, A., Morley, S., & Pearce, S. (1990). Memory for pain: A review. PAIN, 41, 255–265. Everts, B., Karlson, B., Währborg, P., Abdon, N.-J., Herlitz, J., & Hedner, T. (1999). Pain recollection after chest pain of cardiac origin. Cardiology, 92, 115–120. Falco, D. E., Peynircioglu, Z. F., & Hohman, T. J. (2015). Tendency to recall remote memories as a mediator of overgeneral recall in depression. Clinical Psychological Science, 3, 913–925. Feine, J. S., Lavigne, G. J., Thuan Dao, T., Morin, C., & Lund, J. P. (1998). Memories of chronic pain and perceptions of relief. PAIN, 77, 137–141. Fillingim, R. B., & Maixner, W. (1995). Gender differences in the responses to noxious stimuli. Pain Forum, 4, 209–221. Fink, P., Rosendal, M., & Olesen, F. (2005). Classification of somatization and functional somatic symptoms in primary care. Australian and New Zealand Journal of Psychiatry, 39, 772–781. Fredrickson, B. L. (2000). Extracting meaning from past affective experiences: The importance of peaks, ends, and specific emotions. Cognition & Emotion, 14, 577–606. Friedberg, F., & Sohl, S. (2008). Memory for fatigue in chronic fatigue syndrome: The relation between weekly recall and momentary ratings. International Journal of 165

References

Behavioral Medicine, 15, 29–33. Gasper, K., & Clore, G. L. (2002). Attending to the big picture: Mood and global versus local processing of visual information. Psychological Science, 13, 34–40. Gazendam, F. J., & Kindt, M. (2012). Worrying affects associative fear learning: A startle fear conditioning study. PLoS ONE, 7, 1–10. Gedney, J. J., & Logan, H. (2004). Memory for stress-associated acute pain. The Journal of Pain, 5, 83–91. Gedney, J. J., & Logan, H. (2006). Pain related recall predicts future pain report. PAIN, 121, 69–76. Gedney, J. J., Logan, H., & Baron, R. S. (2003). Predictors of short-term and long-term memory of sensory and affective dimensions of pain. The Journal of Pain, 4, 47–55. Geng, X., Chen, Z., Lam, W., & Zheng, Q. (2013). Hedonic evaluation over short and long retention intervals: The mechanism of the peak-end rule. Journal of Behavioral Decision Making, 26, 225–236. Gijsbers van Wijk, C. M. T., Huisman, H., & Kolk, A. M. (1999). Gender differences in physical symptoms and illness behavior. Social Science & Medicine, 49, 1061–1074. Gijsbers van Wijk, C. M. T., & Kolk, A. M. (1997). Sex differences in physical symptoms: The contribution of symptom perception theory. Social Science & Medicine, 45, 231– 246. Giske, L., Sandvik, L., & Røe, C. (2010). Comparison of daily and weekly retrospectively reported pain intensity in patients with localized and generalized musculoskeletal pain. European Journal of Pain, 14, 959–965. Glass, J. M., Park, D. C., Minear, M., & Crofford, L. J. (2005). Memory beliefs and function in fibromyalgia patients. Journal of Psychosomatic Research, 58, 263–269. Griffith, J. W., & Raes, F. (2015). Factor structure of the Ruminative Responses Scale. European Journal of Psychological Assessment, 31, 247–253. Griffith, J. W., Sumner, J. A., Debeer, E., Raes, F., Hermans, D., Mineka, S., … Craske, M. G. (2009). An item response theory/confirmatory factor analysis of the Autobiographical Memory Test. Memory, 17, 609–23. Hadjistavropoulos, H. D., Hadjistavropoulos, T., & Quine, A. (2000). Health anxiety moderates the effects of distraction versus attention to pain. Behaviour Research and Therapy, 38, 425–438. Haller, H., Cramer, H., Lauche, R., & Dobos, G. (2015). Somatoform disorders and medically unexplained symptoms in primary care. Deutsches Ärzteblatt International, 112, 279– 287. Hamann, S. B., Ely, T. D., Grafton, S. T., & Kilts, C. D. (1999). Amygdala activity related to enhanced memory for pleasant and aversive stimuli. Nature Neuroscience, 2, 289–293. Han, J., Zhu, Y., Li, S., Luo, D., Zheng, H., Van Diest, I., … Van den Bergh, O. (2004). Medically unexplained dyspnea: Psychophysiological characteristics and role of breathing therapy. Chinese Medical Journal, 117, 6–13. Han, J., Zhu, Y., Li, S., Zhang, J., Cheng, X., Van den Bergh, O., & Van de Woestijne, K. P. (2008). The language of medically unexplained dyspnea. Chest, 133, 961–968. Hariri, A. R., Bookheimer, S. Y., & Mazziotta, J. C. (2000). Modulating emotional responses. NeuroReport, 11, 43–48. Hashtroudi, S., Johnson, M. K., Vnek, N., & Ferguson, S. A. (1994). Aging and the effects of affective and factual focus on source monitoring and recall. Psychology and Aging, 9, 160–170. Hauer, B. J. A., Wessel, I., Geraerts, E., Merckelbach, H., & Dalgleish, T. (2008). Autobiographical memory specificity after manipulating retrieval cues in adults reporting childhood sexual abuse. Journal of Abnormal Psychology, 117, 444–453. 166

References

Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. New York: The Guilford Press. Haythornthwaite, J. A., Lawrence, J. W., & Fauerbach, J. A. (2001). Brief cognitive interventions for burn pain. Annals of Behavioral Medicine, 23, 42–49. Henningsen, P., Zimmermann, T., & Sattel, H. (2003). Medically unexplained physical symptoms, anxiety, and depression. Psychosomatic Medicine, 65, 528–533. Hermans, D., De Decker, A., De Peuter, S., Raes, F., Eelen, P., & Williams, J. M. G. (2008). Autobiographical memory specificity and affect regulation: Coping with a negative life event. Depression and Anxiety, 25, 787–792. Hermans, D., Defranc, A., Raes, F., Williams, J. M. G., & Eelen, P. (2005). Reduced autobiographical memory specificity as an avoidant coping style. British Journal of Clinical Psychology, 44, 583–589. Hermans, D., Van Den Broeck, K., Belis, G., Raes, F., Pieters, G., & Eelen, P. (2004). Trauma and autobiographical memory specificity in depressed inpatients. Behaviour Research and Therapy, 42, 775–789. Horn, J. (1965). A rationale and test for the number of factors in factor analysis. Psychometrica, 30, 179–185. Hotopf, M. (2002). Childhood experience of illness as a risk factor for medically unexplained symptoms. Scandinavian Journal of Psychology, 43, 139–146. Hotopf, M., Mayou, R., Wadsworth, M., & Wessely, S. (1999). Childhood risk factors for adults with medically unexplained symptoms: Results from a national birth cohort study. American Journal of Psychiatry, 156, 1796–1800. Houtveen, J. H., & Oei, N. Y. L. (2007). Recall bias in reporting medically unexplained symptoms comes from semantic memory. Journal of Psychosomatic Research, 62, 277– 282. Houtveen, J. H., & van Doornen, L. J. P. (2007). Medically unexplained symptoms and between-group differences in 24-h ambulatory recording of stress physiology. Biological Psychology, 76, 239–249. Howren, M. B., Suls, J., & Martin, R. (2009). Depressive symptomatology, rather than neuroticism, predicts inflated physical symptom reports in community-residing women. Psychosomatic Medicine, 71, 951–957. Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55. Hunter, M., Philips, C., & Rachman, S. (1979). Memory for pain. PAIN, 6, 35–46. Hupbach, A., Gomez, R., Hardt, O., & Nadel, L. (2007). Reconsolidation of episodic memories: A subtle reminder triggers integration of new information. Learning & Memory, 14, 47–53. Jamison, R. N., Raymond, S. A., Slawsby, E. A., McHugo, G. J., & Baird, J. C. (2006). Pain assessment in patients with low back pain: Comparison of weekly recall and momentary electronic data. The Journal of Pain, 7, 192–199. Jamison, R. N., Sbrocco, T., & Parris, W. C. V. (1989). The influence of physical and psychosocial factors on accuracy of memory for pain in chronic pain patients. PAIN, 37, 289–294. Janssens, T., Verleden, G., De Peuter, S., Van Diest, I., & Van den Bergh, O. (2009). Inaccurate perception of asthma symptoms: A cognitive-affective framework and implications for asthma treatment. Clinical Psychology Review, 29, 317–327. Jarrett, M., Burr, R., Cain, K., Hertig, V., Weisman, P., & Heitkemper, M. (2003). Anxiety and depression are related to autonomic nervous system function in women with irritable bowel syndrome. Digestive Diseases and Sciences, 48, 386–394. 167

References

Jasper, F., Egloff, B., Roalfe, A., & Witthöft, M. (2015). Latent structure of irritable bowel syndrome symptom severity. World Journal of Gastroenterology, 21, 292–300. Jasper, F., Hiller, W., Rist, F., Bailer, J., & Witthöft, M. (2012). Somatic symptom reporting has a dimensional latent structure: Results from taxometric analyses. Journal of Abnormal Psychology, 121, 725–738. Jensen, M. P., Mardekian, J., Lakshminarayanan, M., & Boye, M. E. (2008). Validity of 24-h recall ratings of pain severity: biasing effects of “Peak” and “End” pain. PAIN, 137, 422–427. Johnston, N. E., Atlas, L. Y., & Wager, T. D. (2012). Opposing effects of expectancy and somatic focus on pain. PLoS ONE, 7, 1–9. Joormann, J., Teachman, B. A., & Gotlib, I. H. (2009). Sadder and less accurate? False memory for negative material in depression. Journal of Abnormal Psychology, 118, 412– 417. Kahneman, D., Fredrickson, B. L., Schreiber, C. A., & Redelmeier, D. A. (1993). When more pain is preferred to less: Adding a better end. Psychological Science, 4, 401–405. Kahneman, D., Krueger, A. B., Schkade, D. A., Schwarz, N., & Stone, A. A. (2004). A survey method for characterizing daily life experience: The Day Reconstruction Method. Science, 306, 1776–1780. Kahneman, D., & Riis, J. (2005). Living, and thinking about it: Two perspectives on life. In F. Huppert, N. Baylis, & B. Keverne (Eds.), The science of well-being (pp. 285–304). Oxford: Oxford University Press. Kamarck, T. W., Schwartz, J. E., Shiffman, S., Muldoon, M. F., Sutton-Tyrrell, K., & Janicki, D. L. (2005). Psychosociol stress and cardiovascular risk: What is the role of daily experience? Journal of Personality, 73, 1749–1774. Katon, W., Lin, E., Von Korff, M., Russo, J., Lipscomb, P., & Bush, T. (1991). Somatization: A spectrum of severity. The American Journal of Psychiatry, 148, 34–40. Keefe, F. J., Rumble, M. E., Scipio, C. D., Giordano, L. A., & Perri, L. M. (2004). Psychological aspects of persistent pain: current state of the science. The Journal of Pain, 5, 195–211. Kensinger, E. A. (2009). Remembering the details: Effects of emotion. Emotion Review, 1, 99–113. Kensinger, E. A., Garoff-Eaton, R. J., & Schacter, D. L. (2006). Memory for specific visual details can be enhanced by negative arousing content. Journal of Memory and Language, 54, 99–112. Kensinger, E. A., Garoff-Eaton, R. J., & Schacter, D. L. (2007). How negative emotion enhances the visual specificity of a memory. Journal of Cognitive Neuroscience, 19, 1872–1887. Kent, G. (1985). Memory of dental pain. PAIN, 21, 187–194. Keogh, E., Hatton, K., & Ellery, D. (2000). Avoidance versus focused attention and the perception of pain: Differential effects for men and women. PAIN, 85, 225–230. Keogh, E., & Herdenfeldt, M. (2002). Gender, coping and the perception of pain. PAIN, 97, 195–201. Kerstner, T., Witthöft, M., Mier, D., Diener, C., Rist, F., & Bailer, J. (2015). A diary-based modification of symptom attributions in pathological health anxiety: Effects on symptom report and cognitive biases. Journal of Consulting and Clinical Psychology, 83, 578– 589. Khan, A. A., Khan, A., Harezlak, J., Tu, W., & Kroenke, K. (2003). Somatic symptoms in primary care: Etiology and outcome. Psychosomatics, 44, 471–478. Kindt, M., Soeter, M., & Vervliet, B. (2009). Beyond extinction: Erasing human fear responses and preventing the return of fear. Nature Neuroscience, 12, 256–258. 168

References

Kirmayer, L., Robbins, J., & Paris, J. (1994). Somatoform disorders: Personality and the social matrix of somatic distress. Journal of Abnormal Psychology, 103, 125–136. Konnopka, A., Schaefert, R., Heinrich, S., Kaufmann, C., Luppa, M., Herzog, W., & König, H.-H. (2012). Economics of medically unexplained symptoms: A systematic review of the literature. Psychotherapy and Psychosomatics, 81, 265–275. Koyama, Y., Koyama, T., Kroncke, A. P., & Coghill, R. C. (2004). Effects of stimulus duration on heat induced pain: The relationship between real-time and post-stimulus pain ratings. PAIN, 107, 256–266. Kroenke, K. (2003a). Patients presenting with somatic complaints: Epidemiology, psychiatric co-morbidity and management. International Journal of Methods in Psychiatric Research, 12, 34–43. Kroenke, K. (2003b). The interface between physical and psychological symptoms. Primary Care Companion to The Journal of Clinical Psychiatry, 5, 11–18. Kroenke, K., & Spitzer, R. L. (1998). Gender differences in the reporting of physical and somatoform symptoms. Psychosomatic Medicine, 60, 150–155. Kroenke, K., Spitzer, R. L., & Williams, J. B. W. (2002). The PHQ-15: Validity of a New Measure for Evaluating the Severity of Somatic Symptoms. Psychosomatic Medicine, 64, 258–266. Kuwert, P., Braehler, E., Freyberger, H. J., & Glaesmer, H. (2012). More than 60 years later. The Journal of Nervous and Mental Disease, 200, 911–914. Kuyken, W., & Brewin, C. R. (1995). Autobiographical memory functioning in depression and reports of early abuse. Journal of Abnormal Psychology, 104, 585–591. LaBar, K. S., & Cabeza, R. (2006). Cognitive neuroscience of emotional memory. Nature Reviews Neuroscience, 7, 54–64. LaBar, K. S., & Phelps, E. A. (1998). Arousal-mediated memory consolidation: Role of the medial temporal lobe in humans. Psychological Science, 9, 490–493. Lane, R. D., Ryan, L., Nadel, L., & Greenberg, L. (2015). Memory reconsolidation, emotional arousal, and the process of change in psychotherapy: New insights from brain science. Behavioral and Brain Sciences, 38, 1–64. Larsen, R. J. (1992). Neuroticism and selective encoding and recall of symptoms: Evidence from a combined concurrent-retrospective study. Journal of Personality and Social Psychology, 62, 480–488. Larson, R., & Csikszentmihalyi, M. (1983). The Experience Sampling Method. New Directions for Methodology of Social & Behavioral Science, 15, 41–56. Lee, J. L. C. (2009). Reconsolidation: Maintaining memory relevance. Trends in Neurosciences, 32, 413–420. Lefebvre, J. C., & Keefe, F. J. (2002). Memory for pain: The relationship of pain catastrophizing to the recall of daily rheumatoid arthritis pain. The Clinical Journal of Pain, 18, 56–63. Lethem, J., Slade, P. D., Troup, J. D. G., & Bentley, G. (1983). Outline of a fear-avoidance model of exaggerated pain perception-I. Behaviour Research and Therapy, 21, 401–408. Leventhal, H., Brown, D., Shacham, S., & Engquist, G. (1979). Effects of preparatory information about sensations, threat of pain, and attention on cold pressor distress. Journal of Personality and Social Psychology, 37, 688–714. Leventhal, H., & Everhart, D. (1979). Emotion, pain, and physical Illness. Emotions in Personality and Psychopathology, 263–302. Leventhal, H., Leventhal, E. A., & Contrada, R. J. (1998). Self-regulation, health, and behavior: A perceptual-cognitive approach. Psychology & Health, 13, 717–733. Levine, L. J., & Safer, M. A. (2002). Sources of bias in memory for emotions. Current Directions in Psychological Science, 11, 169–173. 169

References

Lim, S.-L., & Kim, J.-H. (2005). Cognitive processing of emotional information in depression, panic, and somatoform disorder. Journal of Abnormal Psychology, 114, 50– 61. Linton, S. J., & Melin, L. (1982). The accuracy of remembering chronic pain. PAIN, 13, 281– 285. Liu, X., Liu, Y., Li, L., Hu, Y., Wu, S., & Yao, S. (2014). Overgeneral autobiographical memory in patients with chronic pain. Pain Medicine, 15, 432–439. Logan, H. L., Baron, R. S., & Kohout, F. (1995). Sensory focus as therapeutic treatments for acute pain. Psychosomatic Medicine, 57, 475–484. Loggia, M. L., Mogil, J. S., & Bushnell, M. C. (2008). Experimentally induced mood changes preferentially affect pain unpleasantness. The Journal of Pain, 9, 784–791. Lonigan, C. J., & Vasey, M. W. (2009). Negative affectivity, effortful control, and attention to threat-relevant stimuli. Journal of Abnormal Child Psychology, 37, 387–399. Löwe, B., Spitzer, R. L., Williams, J. B. W., Mussell, M., Schellberg, D., & Kroenke, K. (2008). Depression, anxiety and somatization in primary care: Syndrome overlap and functional impairment. General Hospital Psychiatry, 30, 191–199. Martin, A., Buech, A., Schwenk, C., & Rief, W. (2007). Memory bias for health-related information in somatoform disorders. Journal of Psychosomatic Research, 63, 663–671. Mather, M., & Johnson, M. K. (2003). Affective review and schema reliance in memory in older and younger adults. American Journal of Psychology. Mather, M., & Sutherland, M. R. (2011). Arousal-biased competition in perception and memory. Perspectives on Psychological Science, 6, 114–133. Matsumoto, N., & Mochizuki, S. (2016). Effects of self-relevant cues and cue valence on autobiographical memory specificity in dysphoria. Cognition and Emotion, 1–9. McCall, C., Hildebrandt, L. K., Bornemann, B., & Singer, T. (2015). Physiophenomenology in retrospect: Memory reliably reflects physiological arousal during a prior threatening experience. Consciousness and Cognition, 38, 60–70. McGaugh, J. L. (2004). The amygdala modulates the consolidation of memories of emotionally arousing experiences. Annual Review of Neuroscience, 27, 1–28. McGaugh, J. L. (2006). Make mild moments memorable: Add a little arousal. Trends in Cognitive Sciences, 10, 345–347. McNally, R. J., Perlman, C. A., Ristuccia, C. S., & Clancy, S. A. (2006). Clinical characteristics of adults reporting repressed, recovered, or continuous memories of childhood sexual abuse. Journal of Consulting and Clinical Psychology, 74, 237–242. Meagher, M. W., Arnau, R. C., & Rhudy, J. L. (2001). Pain and emotion: Effects of affective picture modulation. Psychosomatic Medicine, 63, 79–90. Meek, P. M., Lareau, S. C., & Anderson, D. (2001). Memory for symptoms in COPD patients: How accurate are their reports? European Respiratory Journal, 18, 474–481. Michael, E. S., & Burns, J. W. (2004). Catastrophizing and pain sensitivity among chronic pain patients: Moderating effects of sensory and affect focus. Annals of Behavioral Medicine, 27, 185–194. Miron-Shatz, T., Stone, A., & Kahneman, D. (2009). Memories of yesterday’s emotions: Does the valence of experience affect the memory-experience gap? Emotion, 9, 885–891. Muthén, L. K., & Muthén, B. O. (2012). Mplus 7 [Software]. Los Angeles, CA: Muthén & Muthén. Nader, K. (2003). Memory traces unbound. Trends in Neurosciences, 26, 65–72. Nader, K., Schafe, G. E., & Le Doux, J. E. (2000). Fear memories require protein synthesis in the amygdala for reconsolidation after retrieval. Nature, 406, 722–726. Nelson, S., Baldwin, N., & Taylor, J. (2012). Mental health problems and medically unexplained physical symptoms in adult survivors of childhood sexual abuse: An 170

References

integrative literature review. Journal of Psychiatric and Mental Health Nursing, 19, 211–220. Niemi, P. M., Portin, R., Aalto, S., Hakala, M., & Karlsson, H. (2002). Cognitive functioning in severe somatization - A pilot study. Acta Psychiatrica Scandinavica, 106, 461–463. Nimnuan, C., Hotopf, M., & Wessely, S. (2001). Medically unexplained symptoms: An epidemiological study in seven specialities. Journal of Psychosomatic Research, 51, 361–367. Nimnuan, C., Rabe-Hesketh, S., Wessely, S., & Hotopf, M. (2001). How many functional somatic syndromes? Journal of Psychosomatic Research, 51, 549–557. Noel, M., Chambers, C. T., McGrath, P. J., Klein, R. M., & Stewart, S. H. (2012a). The influence of children’s pain memories on subsequent pain experience. PAIN, 153, 1563– 1572. Noel, M., Chambers, C. T., McGrath, P. J., Klein, R. M., & Stewart, S. H. (2012b). The role of state anxiety in children’s memories for pain. Journal of Pediatric Psychology, 37, 567–579. Noel, M., Rabbitts, J. A., Tai, G. G., & Palermo, T. M. (2015). Remembering pain after surgery: A longitudinal examination of the role of pain catastrophizing in children’s and parents' recall. PAIN, 156, 800–808. Nolen-Hoeksema, S. (1991). Responses to depression and their effects on the duration of depressive episodes. Journal of Abnormal Psychology, 100, 569–582. Page, L. A., & Wessely, S. (2003). Medically unexplained symptoms: Exacerbating factors in the doctor-patient encounter. Journal of the Royal Society of Medicine, 96, 223–227. Pennebaker, J. W. (1982). The psychology of physical symptoms. New York: Springer-Verlag. Perkin, G. D. (1989). An analysis of 7836 successive new outpatient referrals. Journal of Neurology, Neurosurgery, and Psychiatry, 52, 447–448. Petersen, S., van den Berg, R. A., Janssens, T., & Van den Bergh, O. (2011). Illness and symptom perception: A theoretical approach towards an integrative measurement model. Clinical Psychology Review, 31, 428–439. Petersen, S., Van Staeyen, K., Vögele, C., von Leupoldt, A., & Van den Bergh, O. (2015). Interoception and symptom reporting: Disentangling accuracy and bias. Frontiers in Psychology, 6, 1–8. Petersen, S., von Leupoldt, A., & Van den Bergh, O. (2015). Interoception and the uneasiness of the mind: Affect as perceptual style. Frontiers in Psychology, 6, 1–6. Phelps, E. A., & Sharot, T. (2008). How (and why) emotion enhances the subjective sense of recollection. Current Directions in Psychological Science, 17, 147–152. Phillips, M. L., Gregory, L. J., Cullen, S., Cohen, S., Ng, V., Andrew, C., … Aziz, Q. (2003). The effect of negative emotional context on neural and behavioural responses to oesophageal stimulation. Brain, 126, 669–684. Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40, 879–891. Price, D. D. (2000). Psychological and neural mechanisms of the affective dimension of pain. Science, 288, 1769–1772. Prins, B., Decuypere, A., & Van Damme, S. (2014). Effects of mindfulness and distraction on pain depend upon individual differences in pain catastrophizing: An experimental study. European Journal of Pain, 18, 1307–1315. Raes, F., & Hermans, D. (2008). On the mediating role of subtypes of rumination in the relationship between childhood emotional abuse and depressed mood: Brooding versus reflection. Depression and Anxiety, 25, 1067–1070. Raes, F., Hermans, D., de Decker, A., Eelen, P., & Williams, J. M. G. (2003). 171

References

Autobiographical memory specificity and affect regulation: An experimental approach. Emotion, 3, 201–206. Raes, F., Hermans, D., Williams, J. M. G., & Eelen, P. (2006). Reduced autobiographical memory specificity and affect regulation. Cognition and Emotion, 20, 402–429. Raes, F., Schoofs, H., Griffith, J. W., & Hermans, D. (2012). Rumination relates to reduced autobiographical memory specificity in formerly depressed patients following a selfdiscrepancy challenge: The case of autobiographical memory specificity reactivity. Journal of Behavior Therapy and Experimental Psychiatry, 43, 1002–1007. Raes, F., Schoofs, H., Hoes, D., Hermans, D., Van Den Eede, F., & Franck, E. (2009). “Reflection” en “brooding” als subtypes van rumineren: Een herziening van de Ruminative Response Scale [Reflection and brooding as subtypes of rumination: A revision of the ruminative response scale]. Gedragstherapie, 42, 205–214. Raes, F., Watkins, E., Williams, J. M. G., & Hermans, D. (2008). Non-ruminative processing reduces overgeneral autobiographical memory retrieval in students. Behaviour Research and Therapy, 46, 748–756. Rainville, P., Carrier, B., Hofbauer, R. K., Bushnell, M. C., & Duncan, G. H. (1999). Dissociation of sensory and affective dimensions of pain using hypnotic modulation. PAIN, 82, 159–171. Read, D. J. C. (1967). A clinical method for assessing the ventilatory response to carbon dioxide. Australasian Annals of Medicine, 16, 20–32. Redelmeier, D. A., & Kahneman, D. (1996). Patients’ memories of painful medical treatments: Real-time and retrospective evaluations of two minimally invasive procedures. PAIN, 66, 3–8. Redelmeier, D. A., Katz, J., & Kahneman, D. (2003). Memories of colonoscopy: A randomized trial. PAIN, 104, 187–194. Reid, S., Wessely, S., Crayford, T., & Hotopf, M. (2001). Medically unexplained symptoms in frequent attenders of secondary health care: Retrospective cohort study. BMJ, 322, 767–769. Renoult, L., Davidson, P. S. R., Palombo, D. J., Moscovitch, M., & Levine, B. (2012). Personal semantics: At the crossroads of semantic and episodic memory. Trends in Cognitive Sciences, 16, 550–558. Renoult, L., Tanguay, A., Beaudry, M., Tavakoli, P., Rabipour, S., Campbell, K., … Davidson, P. S. R. (2016). Personal semantics: Is it distinct from episodic and semantic memory? An electrophysiological study of memory for autobiographical facts and repeated events in honor of Shlomo Bentin. Neuropsychologia, 83, 242–256. Reyna, V. F., & Brainerd, C. J. (1995). Fuzzy-trace theory: An interim synthesis. Learning and Individual Differences, 7, 1–75. Reyna, V. F., & Brainerd, C. J. (1998). Fuzzy-trace theory and false memory: New frontiers. Journal of Experimental Child Psychology, 71, 194–209. Rief, W., & Barsky, A. J. (2005). Psychobiological perspectives on somatoform disorders. Psychoneuroendocrinology, 30, 996–1002. Rief, W., & Broadbent, E. (2007). Explaining medically unexplained symptoms-models and mechanisms. Clinical Psychology Review, 27, 821–841. Rief, W., & Martin, A. (2014). How to use the new DSM-5 somatic symptom disorder diagnosis in research and practice: A critical evaluation and a proposal for modifications. Annual Review of Clinical Psychology, 10, 339–367. Riley, J. L., Robinson, M. E., Wise, E. A., Myers, C. D., & Fillingim, R. B. (1998). Sex differences in the perception of noxious experimental stimuli: A meta-analysis. PAIN, 74, 181–187. Rimes, K. A., & Watkins, E. (2005). The effects of self-focused rumination on global 172

References

negative self-judgements in depression. Behaviour Research and Therapy, 43, 1673– 1681. Rimmele, U., Davachi, L., Petrov, R., Dougal, S., & Phelps, E. A. (2011). Emotion enhances the subjective feeling of remembering, despite lower accuracy for contextual details. Emotion, 11, 553–562. Robinson, M. D., & Barrett, L. F. (2010). Belief and feeling in self-reports of emotion: Evidence for semantic infusion based on self-esteem. Self and Identity, 9, 87–111. Robinson, M. D., & Clore, G. L. (2002a). Belief and feeling: Evidence for an accessibility model of emotional self-report. Psychological Bulletin, 128, 934–960. Robinson, M. D., & Clore, G. L. (2002b). Episodic and semantic knowledge in emotional self-report: Evidence for two judgment processes. Journal of Personality and Social Psychology, 83, 198–215. Roche, P. A., & Gijsbers, K. (1986). A comparison of memory for induced ischaemic pain and chronic rheumatoid pain. PAIN, 25, 337–343. Roelofs, J., Peters, M. L., van der Zijden, M., & Vlaeyen, J. W. S. (2004). Does fear of pain moderate the effects of sensory focusing and distraction on cold pressor pain in pain-free individuals? The Journal of Pain, 5, 250–256. Roelofs, K., & Spinhoven, P. (2007). Trauma and medically unexplained symptoms. Clinical Psychology Review, 27, 798–820. Safer, M. A., Levine, L. J., & Drapalski, A. L. (2002). Distortion in memory for emotions: The contributions of personality and post-event knowledge. Personality and Social Psychology Bulletin, 28, 1495–1507. Schaefer, M., Egloff, B., Gerlach, A. L., & Witthöft, M. (2014). Improving heartbeat perception in patients with medically unexplained symptoms reduces symptom distress. Biological Psychology, 101, 69–76. Schneider, S., Stone, A. A., Schwartz, J. E., & Broderick, J. E. (2011). Peak and end effects in patients’ daily recall of pain and fatigue: A within-subjects analysis. The Journal of Pain, 12, 228–235. Schoofs, H., Hermans, D., & Raes, F. (2010). Brooding and reflection as subtypes of rumination: Evidence from confirmatory factor analysis in nonclinical samples using the Dutch Ruminative Response Scale. Journal of Psychopathology and Behavioral Assessment, 32, 609–617. Schwabe, L., Nader, K., & Pruessner, J. C. (2014). Reconsolidation of human memory: Brain mechanisms and clinical relevance. Biological Psychiatry, 76, 274–280. Schwabe, L., & Wolf, O. T. (2010). Stress impairs the reconsolidation of autobiographical memories. Neurobiology of Learning and Memory, 94, 153–7. Sephton, S. E., Studts, J. L., Hoover, K., Weissbecker, I., Lynch, G., Ho, I., … Salmon, P. (2003). Biological and psychological factors associated with memory function in fibromyalgia syndrome. Health Psychology, 22, 592–597. Sevenster, D., Beckers, T., & Kindt, M. (2012). Retrieval per se is not sufficient to trigger reconsolidation of human fear memory. Neurobiology of Learning and Memory, 97, 338–345. Sharot, T., Delgado, M. R., & Phelps, E. A. (2004). How emotion enhances the feeling of remembering. Nature Neuroscience, 7, 1376–1380. Sharot, T., & Phelps, E. A. (2004). How arousal modulates memory: disentangling the effects of attention and retention. Cognitive, Affective & Behavioral Neuroscience, 4, 294–306. Shiffman, S., Stone, A. A., & Hufford, M. R. (2008). Ecological Momentary Assessment. Annual Review of Clinical Psychology, 4, 1–32. Smets, J., Griffith, J. W., Wessel, I., Walschaerts, D., & Raes, F. (2013). Depressive symptoms moderate the effects of a self-discrepancy induction on overgeneral 173

References

autobiographical memory. Memory, 21, 751–761. Smith, R. C., & Dwamena, F. C. (2007). Classification and diagnosis of patients with medically unexplained symptoms. Journal of General Internal Medicine, 22, 685–691. Smith, W. B., & Safer, M. A. (1993). Effects of present pain level on recall of chronic pain and medication use. PAIN, 55, 355–361. Smyth, J. M., & Stone, A. A. (2003). Ecological Momentary Assessment research in behavioral medicine. Journal of Happiness Studies, 4, 35–52. Sohl, S. J., & Friedberg, F. (2008). Memory for fatigue in chronic fatigue syndrome: Relationships to fatigue variability, catastrophizing, and negative affect. Behavioral Medicine, 34, 29–38. Solberg Nes, L., Roach, A. R., & Segerstrom, S. C. (2009). Executive functions, selfregulation, and chronic pain: A review. Annals of Behavioral Medicine, 37, 173–183. Spinhoven, P., Bockting, C. L. H., Kremers, I. P., Schene, A. H., & Williams, J. M. G. (2007). The endorsement of dysfunctional attitudes is associated with an impaired retrieval of specific autobiographical memories in response to matching cues. Memory, 15, 324–338. Squire, L. R. (2004). Memory systems of the brain: A brief history and current perspective. Neurobiology of Learning and Memory, 82, 171–177. Stegen, K., Neujens, A., Crombez, G., Hermans, D., Van de Woestijne, K. P., & Van den Bergh, O. (1998). Negative affect, respiratory reactivity, and somatic complaints in a CO2 enriched air inhalation paradigm. Biological Psychology, 49, 109–122. Stegen, K., Van Diest, I., Van de Woestijne, K. P., & Van den Bergh, O. (2000). Negative affectivity and bodily sensations induced by 5.5% CO2 enriched air inhalation: Is there a bias to interpret bodily sensations negatively in persons with negative affect? Psychology & Health, 15, 513–525. Stegen, K., Van Diest, I., Van de Woestijne, K. P., & Van den Bergh, O. (2001). Do persons with negative affect have an attentional bias to bodily sensations? Cognition and Emotion, 15, 813–829. Steinbrecher, N., Koerber, S., Frieser, D., & Hiller, W. (2011). The prevalence of medically unexplained symptoms in primary care. Psychosomatics, 52, 263–271. Steptoe, A., Leigh Gibson, E., Hamer, M., & Wardle, J. (2007). Neuroendocrine and cardiovascular correlates of positive affect measured by ecological momentary assessment and by questionnaire. Psychoneuroendocrinology, 32, 56–64. Stone, A. A., Broderick, J. E., Kaell, A. T., DelesPaul, P. A., & Porter, L. E. (2000). Does the peak-end phenomenon observed in laboratory pain studies apply to real-world pain in rheumatoid arthritics? The Journal of Pain, 1, 212–217. Stone, A. A., Broderick, J. E., Shiffman, S. S., & Schwartz, J. E. (2004). Understanding recall of weekly pain from a momentary assessment perspective: Absolute agreement, between- and within-person consistency, and judged change in weekly pain. PAIN, 107, 61–69. Stone, A. A., Schwartz, J. E., Broderick, J. E., & Shiffman, S. S. (2005). Variability of momentary pain predicts recall of weekly pain: A consequence of the peak (or salience) memory heuristic. Personality & Social Psychology Bulletin, 31, 1340–1346. Stone, A. A., Shiffman, S., Schwartz, J. E., Broderick, J. E., & Hufford, M. R. (2002). Patient non-compliance with paper diaries. BMJ, 324, 1193–1194. Storbeck, J. (2013). Negative affect promotes encoding of and memory for details at the expense of the gist: Affect, encoding, and false memories. Cognition and Emotion, 27, 800–819. Storbeck, J., & Clore, G. L. (2005). With sadness comes accuracy; with happiness, false memory: Mood and the false memory effect. Psychological Science, 16, 785–791. 174

References

Storbeck, J., & Clore, G. L. (2011). Affect influences false memories at encoding: Evidence from recognition data. Emotion, 11, 981–989. Suengas, A. G., & Johnson, M. K. (1988). Qualitative effects of rehearsal on memories for perceived and imagined complex events. Journal of Experimental Psychology: General, 117, 377–389. Suls, J., & Fletcher, B. (1985). The relative efficacy of avoidant and nonavoidant coping strategies: A meta-analysis. Health Psychology, 4, 249–288. Suls, J., & Howren, M. B. (2012). Understanding the physical-symptom experience: The distinctive contributions of anxiety and depression. Current Directions in Psychological Science, 21, 129–134. Sumner, J. A. (2012). The mechanisms underlying overgeneral autobiographical memory: An evaluative review of evidence for the CaR-FA-X model. Clinical Psychology Review, 32, 34–48. Sumner, J. A., Griffith, J. W., & Mineka, S. (2010). Overgeneral autobiographical memory as a predictor of the course of depression: A meta-analysis. Behaviour Research and Therapy, 48, 614–625. Sumner, J. A., Griffith, J. W., Mineka, S., Rekart, K. N., Zinbarg, R. E., & Craske, M. G. (2011). Overgeneral autobiographical memory and chronic interpersonal stress as predictors of the course of depression in adolescents. Cognition and Emotion, 25, 183– 192. Sumner, J. A., Mineka, S., Adam, E. K., Craske, M. G., Vrshek-Schallhorn, S., WolitzkyTaylor, K., & Zinbarg, R. E. (2014). Testing the CaR–FA–X model: Investigating the mechanisms underlying reduced autobiographical memory specificity in individuals with and without a history of depression. Journal of Abnormal Psychology, 123, 471–486. Talarico, J. M., & Rubin, D. C. (2003). Confidence, not consistency, characterizes flashbulb memories. Psychological Science, 14, 455–461. Talmi, D. (2013). Enhanced emotional memory: Cognitive and neural mechanisms. Current Directions in Psychological Science, 22, 430–436. Talmi, D., Schimmack, U., Paterson, T., & Moscovitch, M. (2007). The role of attention and relatedness in emotionally enhanced memory. Emotion, 7, 89–102. Tasmuth, T., Estlanderb, A.-M., & Kalso, E. (1996). Effect of present pain and mood on the memory of past postoperative pain in women treated surgically for breast cancer. PAIN, 68, 343–347. Terry, R. H., Brodie, E. E., & Niven, C. A. (2007). Exploring the phenomenology of memory for pain: Is previously experienced acute pain consciously remembered or simply known? Journal of Pain, 8, 467–475. Terry, R. H., Niven, C. A., Brodie, E. E., Jones, R. B., & Prowse, M. A. (2008). Memory for pain? A comparison of nonexperiential estimates and patients’ reports of the quality and intensity of postoperative pain. The Journal of Pain, 9, 342–349. Thomas, M. L., & Locke, D. E. C. (2010). Psychometric properties of the MMPI-2-RF Somatic Complaints (RC1) Scale. Psychological Assessment, 22, 492–503. Toffalini, E., Mirandola, C., Coli, T., & Cornoldi, C. (2015). High trait anxiety increases inferential false memories for negative (but not positive) emotional events. Personality and Individual Differences, 75, 201–204. Toussaint, A., Murray, A. M., Voigt, K., Herzog, A., Gierk, B., Kroenke, K., … Löwe, B. (2016). Development and validation of the somatic symptom disorder–B criteria scale (SSD-12). Psychosomatic Medicine, 78, 5–12. Treynor, W., Gonzalez, R., & Nolen-Hoeksema, S. (2003). Rumination reconsidered: A psychometric analysis. Cognitive Therapy and Research, 27, 247–259. Tulving, E. (1972). Episodic and semantic memory. In E. Tulving & W. Donaldson (Eds.), 175

References

Organization of Memory (pp. 381–403). New York: Academic Press. Tulving, E. (1983). The elements of episodic memory. New York: Oxford University Press. Tulving, E. (2002). Episodic memory: From mind to brain. Annual Review of Psychology, 53, 1–25. Van den Bergh, O., Bogaerts, K., & Van Diest, I. (2015). Symptom perception, awareness and interpretation. In J. D. Wright (Ed.), International encyclopedia of the social & behavioral sciences (2nd ed., pp. 866–872). Oxford, UK: Elsevier. Van den Bergh, O., Devriese, S., Winters, W., Veulemans, H., Nemery, B., Eelen, P., & Van de Woestijne, K. P. (2001). Acquiring symptoms in response to odors: a learning perspective on multiple chemical sensitivity. Annals of the New York Academy of Sciences, 933, 278–290. Van den Bergh, O., Stegen, K., & Van De Woestijne, K. P. (1997). Learning to have psychosomatic complaints: Conditioning of respiratory behavior and somatic complaints in psychosomatic patients. Psychosomatic Medicine, 59, 13–23. Van den Bergh, O., Stegen, K., & Van De Woestijne, K. P. (1998). Memory effects on symptom reporting in a respiratory learning paradigm. Health Psychology, 17, 241–248. Van den Bergh, O., Winters, W., Devriese, S., & Van Diest, I. (2002). Learning subjective health complaints. Scandinavian Journal of Psychology, 43, 147–152. Van den Bergh, O., Winters, W., Devriese, S., Van Diest, I., Vos, G., & de Peuter, S. (2004). Accuracy of respiratory symptom perception in persons with high and low negative affectivity. Psychology & Health, 19, 213–222. Van Den Brink, M., Bandell-Hoekstra, E. N. G., & Huijer Abu-Saad, H. (2001). The occurrence of recall bias in pediatric headache: A comparison of questionnaire and diary data. Headache, 41, 11–20. Van der Does, A. J. W. (2002). Handleiding bij de Nederlandse bewerking van de BDI-II [Manual of the Dutch version of the BDI-II]. San Antonio, Texas/Lisse, The Netherlands: The Psychological Corporation/Swets Test Publishers. Van Diest, I., De Peuter, S., Eertmans, A., Bogaerts, K., Victoir, A., & Van den Bergh, O. (2005). Negative affectivity and enhanced symptom reports: Differentiating between symptoms in men and women. Social Science & Medicine, 61, 1835–1845. van Eck van der Sluijs, J., ten Have, M., Rijnders, C., van Marwijk, H., de Graaf, R., & van der Feltz-Cornelis, C. (2015). Medically unexplained and explained physical symptoms in the general population: Association with prevalent and incident mental disorders. PLoS ONE, 10, 1–14. Van Lier, J., Vanbrabant, K., Vervliet, B., & Raes, F. (2016). Abstract “why” thoughts lead to more negative generalization: A replication and improved data-analytic approach. Manuscript submitted for publication. Van Lier, J., Vervliet, B., Vanbrabant, K., Lenaert, B., & Raes, F. (2014). Abstract thinking about negative events in dysphoric students leads to negative generalization. Journal of Experimental Psychopathology, 5, 314–328. Van Oudenhove, L., Jasper, F., Walentynowicz, M., Witthöft, M., Van den Bergh, O., & Tack, J. (in press). The latent structure of the functional dyspepsia symptom complex: A taxometric analysis. Neurogastroenterology & Motility. van Ravesteijn, H., Wittkampf, K., Lucassen, P., van de Lisdonk, E., van den Hoogen, H., van Weert, H., … Speckens, A. (2009). Detecting somatoform disorders in primary care with the PHQ-15. The Annals of Family Medicine, 7, 232–238. Verkuil, B., Brosschot, J. F., Meerman, E. E., & Thayer, J. F. (2012). Effects of momentary assessed stressful events and worry episodes on somatic health complaints. Psychology & Health, 27, 141–158. Verkuil, B., Brosschot, J. F., & Thayer, J. F. (2007). A sensitive body or a sensitive mind? 176

References

Associations among somatic sensitization, cognitive sensitization, health worry, and subjective health complaints. Journal of Psychosomatic Research, 63, 673–681. Vlaeyen, J. W. S., & Linton, S. J. (2000). Fear-avoidance and its consequences in chronic musculoskeletal pain: A state of the art. PAIN, 85, 317–332. Vlaeyen, J. W. S., & Linton, S. J. (2012). Fear-avoidance model of chronic musculoskeletal pain: 12 years on. PAIN, 153, 1144–1147. von Leupoldt, A., & Dahme, B. (2007a). Experimental comparison of dyspnea and pain. Behavior Research Methods, 39, 137–143. von Leupoldt, A., & Dahme, B. (2007b). Psychological aspects in the perception of dyspnea in obstructive pulmonary diseases. Respiratory Medicine, 101, 411–422. von Leupoldt, A., Mertz, C., Kegat, S., Burmester, S., & Dahme, B. (2006). The impact of emotions on the sensory and affective dimension of perceived dyspnea. Psychophysiology, 43, 382–386. von Leupoldt, A., Taube, K., Schubert-Heukeshoven, S., Magnussen, H., & Dahme, B. (2007). Distractive auditory stimuli reduce the unpleasantness of dyspnea during exercise in patients with COPD. Chest, 132, 1506–1512. Walentynowicz, M., Bogaerts, K., Stans, L., Van Diest, I., Raes, F., & Van den Bergh, O. (2016). Memory for symptoms in patients with somatic symptom disorder. Manuscript in preparation. Walentynowicz, M., Bogaerts, K., Van Diest, I., Raes, F., & Van den Bergh, O. (2015). Was it so bad? The role of retrospective memory in symptom reporting. Health Psychology, 34, 1166–1174. Walentynowicz, M., Raes, F., Van Diest, I., & Van den Bergh, O. (2016). Ways of encoding information and their effects on retrospective symptom reporting. Manuscript in preparation. Walentynowicz, M., Witthöft, M., Raes, F., Van Diest, I., & Van den Bergh, O. (2016). Sensory and affective components of symptom perception: A psychometric approach. Manuscript in preparation. Wan, L., Stans, L., Bogaerts, K., Decramer, M., & Van den Bergh, O. (2012). Sensitization in medically unexplained dyspnea: Differential effects on intensity and unpleasantness. Chest, 141, 989–995. Watkins, E. (2004). Adaptive and maladaptive ruminative self-focus during emotional processing. Behaviour Research and Therapy, 42, 1037–1052. Watkins, E. (2008). Constructive and unconstructive repetitive thought. Psychological Bulletin, 134, 163–206. Watkins, E., Moberly, N. J., & Moulds, M. L. (2008). Processing mode causally influences emotional reactivity: Distinct effects of abstract versus concrete construal on emotional response. Emotion, 8, 364–378. Watkins, E., & Moulds, M. (2005). Distinct modes of ruminative self-focus: Impact of abstract versus concrete rumination on problem solving in depression. Emotion, 5, 319– 328. Watkins, E., & Nolen-Hoeksema, S. (2014). A habit-goal framework of depressive rumination. Journal of Abnormal Psychology, 123, 24–34. Watkins, E., & Teasdale, J. D. (2001). Rumination and overgeneral memory in depression: Effects of self-focus and analytic thinking. Journal of Abnormal Psychology, 110, 333– 357. Watkins, E., & Teasdale, J. D. (2004). Adaptive and maladaptive self-focus in depression. Journal of Affective Disorders, 82, 1–8. Watkins, E., Teasdale, J. D., & Williams, R. (2000). Decentring and distraction reduce overgeneral autobiographical memory in depression. Psychological Medicine, 30, 911– 177

References

920. Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54, 1063–1070. Watson, D., & Pennebaker, J. W. (1989). Health complaints, stress, and distress: Exploring the central role of negative affectivity. Psychological Review, 96, 234–254. Weisenberg, M., Raz, T., & Hener, T. (1998). The influence of film-induced mood on pain perception. PAIN, 76, 365–375. Wessely, S., Nimnuan, C., & Sharpe, M. (1999). Functional somatic syndromes: one or many? The Lancet, 354, 936–939. Wheeler, M. A., Stuss, D. T., & Tulving, E. (1997). Toward a theory of episodic memory: The frontal lobes and autonoetic consciousness. Psychological Bulletin, 121, 331–354. Whitehead, B. R., & Bergeman, C. S. (2013). Self-reported health bias: The role of daily affective valence and arousal. Psychology and Health, 28, 784–799. Wichert, S., Wolf, O. T., & Schwabe, L. (2013). Updating of episodic memories depends on the strength of new learning after memory reactivation. Behavioral Neuroscience, 127, 331–338. Wientjes, C. J., & Grossman, P. (1994). Overreactivity of the psyche or the soma? Interindividual associations between psychosomatic symptoms, anxiety, heart rate, and end-tidal partial carbon dioxide pressure. Psychosomatic Medicine, 56, 533–540. Williams, J. M. G., Barnhofer, T., Crane, C., Herman, D., Raes, F., Watkins, E., & Dalgleish, T. (2007). Autobiographical memory specificity and emotional disorder. Psychological Bulletin, 133, 122–148. Williams, J. M. G., & Broadbent, K. (1986). Autobiographical memory in suicide attempters. Journal of Abnormal Psychology, 95, 144–149. Williams, J. M. G., & Dritschel, B. H. (1992). Categoric and extended autobiographical memories. In M. A. Conway, D. C. Rubin, H. Spinnler, & W. A. Wagenaar (Eds.), Theoretical perspectives on autobiographical memory (pp. 391–410). Dordrecht, The Netherlands: Kluwert. Wingenfeld, K., Terfehr, K., Meyer, B., Löwe, B., & Spitzer, C. (2013). Memory bias for emotional and illness-related words in patients with depression, anxiety and somatization disorders: An investigation with the directed forgetting task. Psychopathology, 46, 22– 27. Witthöft, M., Fischer, S., Jasper, F., Rist, F., & Nater, U. M. (2016). Clarifying the latent structure and correlates of somatic symptom distress: A bifactor model approach. Psychological Assessment, 28, 109–115. Witthöft, M., Gerlach, A. L., & Bailer, J. (2006). Selective attention, memory bias, and symptom perception in idiopathic environmental intolerance and somatoform disorders. Journal of Abnormal Psychology, 115, 397–407. Witthöft, M., Hiller, W., Loch, N., & Jasper, F. (2013). The latent structure of medically unexplained symptoms and its relation to functional somatic syndromes. International Journal of Behavioral Medicine, 20, 172–183. Witthöft, M., Loch, N., & Jasper, F. (2013). Somatoforme Beschwerden und Stile der Emotionsregulation. Verhaltenstherapie & Verhaltensmedizin, 34, 444–464. Yiend, J. (2010). The effects of emotion on attention: A review of attentional processing of emotional information. Cognition and Emotion, 24, 3–47. Zijlema, W. L., Stolk, R. P., Löwe, B., Rief, W., White, P. D., & Rosmalen, J. G. M. (2013). How to assess common somatic symptoms in large-scale studies: A systematic review of questionnaires. Journal of Psychosomatic Research, 74, 459–468.

178

Appendix A

Appendix A Validation of the Checklist for Symptoms in Daily Life (CSD) questionnaire (Chapter 2) The Checklist for Symptoms in Daily Life CSD was developed on the basis of another questionnaire, the Psychosomatic Symptom Checklist (Wientjes & Grossman, 1994). This supplemental material describes the analyses aimed to validate the CSD. To this end, data from multiple group testing sessions among first-year psychology students were collected in order to perform a principal component analysis of the CSD and to explore the convergent validity of the questionnaire. Methods Participants The data were collected during six consecutive group testing sessions (2009-2014) among all first-year psychology students from the University of Leuven, Belgium, in return for the course credit. All group testing sessions were approved by the Multidisciplinary Ethics Committee of the Faculty of Psychology and Educational Sciences of the University of Leuven. Due to the fact that the assessment procedure and questionnaires accompanying the symptom reporting questionnaire varied slightly between the testing sessions, three samples were formed to analyze the data. Sample 1 (S1, N = 1288, 82.5% women, Mage = 18.5, SDage = 1.39) completed the group testing in the years 2009 to 2011 in a paper-pencil format. Sample 2 (S2, N = 392 at Time 1, 79.3% women; N = 343 at Time 2, 81.9% women; Mage = 18.45, SDage = 1.31) completed the group testing session in 2012 twice over the period of 6 weeks. The responses in this latter sample were collected by means of a web-based survey. Sample 3 (S3, N = 684, 82% women, Mage = 18.76, SDage = 2.92) completed the group session in a web-based format in the years 2013 and 2014. Measures Checklist for Symptom in Daily Life (CSD; S1-3). The CSD is a self-report assessment of symptom reporting based on the Psychosomatic Symptom Checklist (Wientjes & Grossman, 1994). Apart from the original 35 symptoms, which mainly focus on the hyperventilation-related complaints, it includes four additional items: stuffy nose, low back pain, joint pain, and burning feeling in the eyes (see Table A-1 for the entire CSD). The occurrence of the 39 symptoms in the past year is assessed on a 5-point scale (never, seldom, sometimes, often, very often). Symptom Checklist-90 (SCL-90; S2). The Dutch version of the SCL-90 (Arrindell & Ettema, 2003) is a 90-item self-report inventory covering a broad range of psychological symptoms. The frequency of symptoms in a past week is rated on a 5-point scale (1 = not at all; 5 = extremely). In the present study only the somatization subscale (SCL-90 SOM) was used. Cronbach’s alpha in the present sample was .78.

179

Appendix A

Patient Health Questionnaire-15 (PHQ-15; S3). The PHQ-15 (Kroenke et al., 2002; van Ravesteijn et al., 2009) is a widely used self-report questionnaire measuring the distress related to 15 somatic symptoms over the previous 4 weeks. The severity of each symptom is rated on the 3-point scale (0 = not bothered at all, 1 = bothered a little, 2 = bothered a lot). Cronbach’s alpha in the present sample was .72. Data analysis Principal component analysis with varimax rotation was performed to examine the factorial structure of the CSD (S1). To determine the number of reliable factors, the parallel analysis method (Horn, 1965) was used. In this procedure, the size of the observed eigenvalues is compared with the eigenvalues obtained from random data. Internal consistency of the CSD and the subscales was measured with Cronbach’s alphas (S1). To investigate the convergent validity, correlations were calculated between CSD and scores on other well-known questionnaires measuring somatic symptoms (Zijlema et al., 2013), namely the SCL-90 SOM (S2) and the PHQ-15 (S3). Test-retest reliability of the CSD was measured over the period of 6 weeks (S2, Time 1 and 2). Results and discussion According to the parallel analysis (Horn, 1965), a five-factor solution emerged. Based on the questionnaire items loading at least >.40 on each factor (Table A-2), the factors were interpreted as Cardiorespiratory (Items 3, 5, 6, 7, 19, 24, 25, 31, and 36), Psychological (Items 1, 9, 10, 27, and 38), Neurological (Items 12, 18, 23, 30, 35, and 37), Flu/Cold (Items 4, 8, 11, 12, 13, 14, 15, 16, 21, and 29), and Cerebral (Items 2, 22, 26, and 32). Five items (20, 28, 33, 34, 39) could not be allocated to a factor as they loaded less than .40 on the extracted factors. The five factors accounted for 43.02% of variance. The CSD and its subscales were reliable. Cronbach’s α for the CSD was .91, and for the subscales of CSD it ranged from .67 to .85, as displayed in Table A-2. The scale demonstrated good convergent validity, as indicated by strong correlations between the CSD and the SCL-90 SOM (S2; r = .64, p < .001), as well as the PHQ-15 (S3; r = .70, p < .001). Finally, the total CSD score showed a moderate test-retest reliability for the time interval of 6 weeks, r = .62, p < .001. In summary, our findings on the internal consistency and the convergent validity show that the CSD can be perceived as a reliable and valid instrument to assess symptom reporting.

180

Appendix A Table A-1. The Checklist for Symptoms in Daily Life (CSD). This questionnaire consists of 39 symptoms, which you can experience in daily life. Please indicate how often have you experienced each of the symptoms in the last year on the following scale: Never Seldom Sometimes Often Very often 1. Tenseness ☐ ☐ ☐ ☐ ☐ 2. Dizziness ☐ ☐ ☐ ☐ ☐ 3. Faster/deeper breathing ☐ ☐ ☐ ☐ ☐ 4. Joint pain ☐ ☐ ☐ ☐ ☐ 5. Dyspnea ☐ ☐ ☐ ☐ ☐ 6. Pressure on chest ☐ ☐ ☐ ☐ ☐ 7. Pounding heart ☐ ☐ ☐ ☐ ☐ 8. Feeling sleepy ☐ ☐ ☐ ☐ ☐ 9. Feeling anxious ☐ ☐ ☐ ☐ ☐ 10. Feeling of unrest, panic ☐ ☐ ☐ ☐ ☐ 11. Feeling of heat ☐ ☐ ☐ ☐ ☐ 12. Low back pain ☐ ☐ ☐ ☐ ☐ 13. Shivering ☐ ☐ ☐ ☐ ☐ 14. Feeling of head warmth ☐ ☐ ☐ ☐ ☐ 15. Pressure or knot in throat ☐ ☐ ☐ ☐ ☐ 16. Stuffy nose ☐ ☐ ☐ ☐ ☐ 17. Tingling in arms ☐ ☐ ☐ ☐ ☐ 18. Tingling in face ☐ ☐ ☐ ☐ ☐ 19. Suffocating feeling ☐ ☐ ☐ ☐ ☐ 20. Burning feeling in the eyes ☐ ☐ ☐ ☐ ☐ 21. Headache ☐ ☐ ☐ ☐ ☐ 22. Nausea ☐ ☐ ☐ ☐ ☐ 23. Tingling in feet ☐ ☐ ☐ ☐ ☐ 24. Unable to breathe deeply enough ☐ ☐ ☐ ☐ ☐ 25. Faster heart rate ☐ ☐ ☐ ☐ ☐ 26. Blacking out ☐ ☐ ☐ ☐ ☐ 27. Confused, dreamlike feeling ☐ ☐ ☐ ☐ ☐ 28. Toe or leg cramps ☐ ☐ ☐ ☐ ☐ 29. Stomach cramps ☐ ☐ ☐ ☐ ☐ 30. Tingling in legs ☐ ☐ ☐ ☐ ☐ 31. Irregular heart rate ☐ ☐ ☐ ☐ ☐ 32. Fainting ☐ ☐ ☐ ☐ ☐ 33. Hands tremble ☐ ☐ ☐ ☐ ☐ 34. Stomach feels blown up ☐ ☐ ☐ ☐ ☐ 35. Tingling in fingers ☐ ☐ ☐ ☐ ☐ 36. Chest pain, around heart region ☐ ☐ ☐ ☐ ☐ 37. Stiffness in fingers and arms ☐ ☐ ☐ ☐ ☐ 38. Fits of crying ☐ ☐ ☐ ☐ ☐ 39. Cold hands or feet ☐ ☐ ☐ ☐ ☐

181

Appendix A

Table A-2. Results of the exploratory factor analysis with varimax rotation (N = 1288). Items of the Checklist for Symptoms in Daily Life are presented with factor loadings for the extracted factors. Cronbach’s alphas, eigenvalues and the percentage of explained variance for each extracted factor are included. Factor loadings Factor (% variance) items 1 2 3 4 5 (1) Cardiorespiratory factor (22.78%) Pressure on the chest .71 Faster heart rate .70 Faster/deeper breathing .68 Dyspnea/need for air .67 Irregular heart rate .65 Pounding heart .64 Unable to breathe deeply enough .61 Chest pain around heart region .53 Suffocating feeling .51 (2) Psychological factor (6.61%) Feeling of unrest/panic .83 Feeling anxious .80 Tenseness .73 Fits of crying .67 Confused/dreamlike feeling .48 (3) Neurological factor (2.22%) Tingling in arms .75 Tingling in feet .75 Tingling in legs .73 Tingling in fingers .68 Tingling in face .59 Stiffness in fingers or arms .48 (4) Cold/flu factor (4.31%) Feeling of warm .55 Feeling of head warm .52 Low back pain .51 Feeling sleepy .46 Stuffy nose .45 Joint pain .44 Pressure or knot in throat .42 Stomach cramps .42 Shivering .41 Headache .41 (5) Cerebral factor (3.63%) Dizziness .66 Blacking out .66 Fainting .64 Nausea .49 Cronbach’s α .85 .81 .81 .72 .67 Eigenvalue 8.88 2.58 2.22 1.68 1.41 Note. All items shown load > .40 onto the factors.

182

Appendix B

Appendix B The latent structure of somatic symptoms in the CSD and the PHQ-15: Gender differences (Chapter 2) The four models (general factor, correlated factor, hierarchical, and bifactor models) were tested separately for males and females in both the CSD (Table B-1) and the PHQ-15 (Table B-2). Based on the fit indices reported in Tables B-1 and B-2, the bifactor models outperformed the other models in all examined cases. The results for the male and the female samples were very similar with one exception observed in the bifactor model of the CSD. In the case of male sample (Figure B-1, left panel), the loadings in the cerebral factor were nonsignificant, while they were significant in the analysis of the female sample (Figure B-2, left panel). However, it is important to bear in mind that due to the small sample size of the male group (n = 184), the findings regarding this group should be interpreted with caution. Association between the bifactor models (CSD and PHQ-15) and NA Male sample In the CSD bifactor model in the male sample, NA was strongly related to the CSD general symptom factor, r = .59, p < .001, and the symptom-specific psychological factor, r = .55, p < .001 (see Figure B-1, right panel). A weaker but significant correlation with a neurological factor was also observed (r = -.19, p = .001). No significant associations between NA and other symptom-specific factors were found (all rs < .05, ps > .62). The model fit was good: χ²(521) = 668.52, p < .001; RMSEA = .039 (90% CI: .030-.048); CFI: .964. In the PHQ-15 bifactor model in the male sample, NA was only associated with the PHQ-15 general symptom factor, r = .58, p < .001 (see Figure B-3, top right panel). No significant correlation was observed between NA and any of the symptom-specific factors (all rs < .19, ps > .19). The model fit was good, χ²(64) = 74.56, p > .05; RMSEA = .030 (90% CI: .000-.055); CFI: .977. Female sample In the CSD bifactor model in the female sample, NA was strongly related to the CSD general symptom factor, r = .59, p < .001, and the symptom-specific psychological factor, r = .57, p < .001 (see Figure B-2, right panel). A weaker but significant correlation with a neuroloical factor was also observed (r = -.10, p = .005). No significant associations between NA and other symptom-specific factors were found (all rs < .06, ps > .22). The model fit was acceptable: χ²(521) = 1960.87, p < .001; RMSEA = .056 (90% CI: .054-.059); CFI: .925. In the PHQ-15 bifactor model in the female sample, NA was only associated with the PHQ-15 general symptom factor, r = .52, p < .001 (see Figure B-3, bottom right panel). No significant correlation was observed between NA and any of the symptom-specific factors (all rs < .12, ps > .12). The model fit was excellent, χ²(64) = 83.91, p = .048; RMSEA = .019 (90% CI: .002-.029); CFI: .991.

183

Appendix B Table B-1. Goodness of fit for the four different models tested with the CSD for males (n = 184) and females (n = 870). Model I Model II Model III Model IV General factor Correlated Hierarchical Bifactor model model factor model model Males (n = 184) χ 2 (df)a 1236.59 (527) 767.01 (517) 782.96 (522) 639.87 (493) CFI .826 .939 .936 .964 RMSEA .086 .051 .052 .040 90% CI RMSEA .079-.092 .043-.059 Females (n = 870) χ 2 (df)a 5644.64 (527) 2249.88 (517) CFI .732 .909 RMSEA .106 .062 90% CI RMSEA .103-.108 .059-.065 a 2 All χ values were highly significant, p < .001

.044-.060

.031-.049

2320.44 (522) .906 .063 .060-.066

1905.61 (493) .926 .057 .055-.060

Table B-2. Goodness of fit for the four different models tested with the PHQ-15 for males (n = 184) and females (n = 869). Model I Model II Model III Model IV General factor Correlated Hierarchical Bifactor model model factor model model Males (n = 184) χ 2 (df)a CFI RMSEA

125.72 (65) .845 .071

65.47 (59) .984 .024

102.26 (65) .905 .056

62.17 (56) .984 .024

90% CI RMSEA .052-.090 .000-.053 .034-.076 .000-.054 Females (n = 869) χ 2 (df)a 413.29 (65) 134.80 (59) 133.34 (61) 78.49 (56) CFI .828 .963 .964 .989 RMSEA .079 .038 .037 .021 90% CI RMSEA .071-.086 .030-.047 .028-.045 .008-.032 a All χ 2 values were significant, p < .05, except for models II and IV, males: p > .05

184

Appendix B

Figure B-1. Bifactor model of somatic symptoms in the CSD with standardized factor loadings (left) and with negative affectivity (NA; right) in male sample. Single headed arrows represent factor loadings; factor loading coefficients printed in bold are significant at p < .05. Double-headed arrows represent latent correlation coefficients; all correlation coefficients printed in bold are significant at p < .001; residual terms of manifest variables not shown.

185

Appendix B

Figure B-2. Bifactor model of somatic symptoms in the CSD with standardized factor loadings (left) and with negative affectivity (NA; right) in female sample. Single headed arrows represent factor loadings; factor loading coefficients printed in bold are significant at p < .05. Double-headed arrows represent latent correlation coefficients; all correlation coefficients printed in bold are significant at p < .05; residual terms of manifest variables not shown.

186

Appendix B

Figure B-3. Bifactor model of somatic symptoms in the PHQ-15 with standardized factor loadings (left) and with negative affectivity (NA; right) in male (top) and female samples (bottom). Single headed arrows represent factor loadings; factor loading coefficients printed in bold are significant at p < .05. Double-headed arrows represent latent correlation coefficients; all correlation coefficients printed in bold are significant at p < .001; residual terms of manifest variables not shown.

187

Appendix C

Appendix C The multiple mediator analyses (Chapter 3) This appendix includes multiple mediator analyses examining the mediating role of state NA experienced during the trials on retrospective (follow-up) symptom ratings in both groups (high and low symptom reporters). Data Analyses The multiple mediator analyses were performed on all four trials: short pain trial, long pain trial, short dyspnea trial, long dyspnea trial. In each of the analyses, we first examined whether group (high HSR coded 1, vs. low HSR coded 0) predicted follow-up symptom rating (path c), whether group predicted state NA during the trial (path a1), and whether state NA predicted follow-up symptom rating after controlling for group (path b1). As it is also possible that follow-up ratings were influenced by the actual symptom ratings given during the trials, separate analyses investigated whether group predicted concurrent symptom ratings during the trial (path a2), and concurrent symptom ratings predicted follow-up symptom rating after controlling for group (path b2). Finally, we investigated whether the relationship between the group and the retrospective (follow-up) symptom ratings was mediated by the effects of both state NA and concurrent symptom ratings. To this end, a multiple mediation model using the bootstrapping procedure (Preacher and Hayes, 2008) was tested. Such method, designed for small sample sizes, can estimate not only total effects, but also specific indirect effects (paths a1b1 and a2b2) of each mediator. The model coefficients, direct and indirect effects are reported in unstandardized form (Hayes, 2013). The data analysis was performed with the PROCESS Procedure for SPSS (Hayes, 2013). Results Short pain trial. Group predicted follow-up pain rating (c = 17.67, p = .02), group predicted state NA (a1 = 3.79, p = .01), but state NA did not predict pain rating after controlling for group (b1 = 1.20, p = .12). Moreover, group did not predict concurrent pain (a2 = 3.81, p = .51), but concurrent pain predicted pain rating after controlling for group (b2 = 1.02, p < .001). Due to the fact that for neither state NA nor concurrent symptom ratings paths a and b were significant, those variables could not be treated as candidate mediators. Long pain trial. Group predicted follow-up pain rating (c = 17.54, p = .01), group predicted state NA (a1 = 3.54, p = .01), and state NA predicted pain rating after controlling for group (b1 = 2.02, p = .003). Group did not predict concurrent pain (a2 =10.32, p = .10), but concurrent pain predicted pain rating after controlling for group (b2 = .61, p < .001). Due to the fact that for concurrent symptom ratings path a2 was not significant, only state NA was added to the model as a mediator (Figure C-1, panel a). The indirect effect of state NA on follow-up ratings (a1b1 = 7.15) was significant (95% CI [2.37, 14.22]). Short dyspnea trial. Group predicted follow-up dyspnea rating (c = 13.50, p = .05), group predicted state NA (a1 = 5.21, p = .003), and state NA predicted dyspnea rating after 188

Appendix C

controlling for group (b1 = 2.04, p < .001). Moreover, group predicted concurrent dyspnea (a2 = 10.82, p = .005), and concurrent dyspnea predicted follow-up dyspnea rating after controlling for group (b2 = 1.01, p < .001). State NA and concurrent symptom ratings could be treated as candidates for mediators (Figure C-1, panel b). Specific indirect effects for both state NA, a1b1 = 6.89 (95% CI [1.22, 15.75]) and concurrent symptom ratings, a2b2 = 8.20 (95% CI [1.93, 16.88]) on follow-up ratings were significant. The point estimate for the contrast between the two indirect effects, -1.31 (95% CI [-12.10, 8.45]), was not significantly different from zero, showing that the effects did not differ in size. Long dyspnea trial. Group marginally predicted follow-up dyspnea rating (c = 13.13, p = .07), group predicted state NA (a1 = 4.71, p = .003), and state NA predicted dyspnea rating after controlling for group (b1 = 2.40, p < .001). Furthermore, group did not predict concurrent dyspnea (a2 = 6.33, p = .13), but concurrent dyspnea predicted follow-up dyspnea rating after controlling for group (b2 = 1.08, p < .001). Due to the fact that for concurrent symptom ratings path a2 was not significant, only state NA was added as a mediator to the model (Figure C-1, panel c). The indirect effect of state NA on follow-up ratings (a1b1 =11.30) was significant (95% CI [4.30, 20.92]). From the abovementioned mediation analysis it can be seen that for three out of four trials (except for short pain trial) state NA experienced during the trial was a significant mediator of the association between the habitual symptom reporting and retrospective symptom ratings. This is in line with our explanation regarding affective bias of the memory for symptoms, as described in the discussion.

Figure C-1. Simple (panels a and c) and multiple (panel b) mediator models for long pain trial (panel a), short dyspnea trial (panel b), and long dyspnea trial (panel c). The panels show direct and indirect effects of a group (high/low HSR) on the retrospective symptom ratings, mediated by state NA (all panels) and concurrent symptom ratings (panel b). The model coefficients are reported in unstandardized form.

189

Appendix D

Appendix D Manipulation of processing focus: Instructions (Chapters 5 and 6) In everyday life we constantly experience all sorts of bodily sensations. They can vary in type (such as hunger, butterflies in the stomach, fatigue, or a stiff knee) and can differ in unpleasantness (pleasant, neutral, or unpleasant) and intensity (barely noticeable to attention demanding). Our attention is often externally directed and we do not notice many sensations from within our body, but we can also pay attention to our body and then the feelings such as butterflies in the stomach or being fatigued become more pronounced. We can also focus on the sensations in our body in various ways. We can focus, for example like a doctor, on the different sensations to know what they exactly are, where we can locate them, how intense they are, if they change, and so on. Or we can also focus on the emotional experience related to the sensations, for example how pleasant, annoying or bothersome they are, whether you become frightened or cheerful, in short, how do they make you feel. A comparable example is a piece of music on the radio. You can focus, for example like an expert or researcher, on the different instruments that you can distinguish, how intense the sounds are, which rhythm the music has, etc. But you can also get carried away emotionally with the music, immerse yourself in it and become aware of the emotions and feelings that music evokes in you, such as sadness or joy. Another example is how you can feel after a feast when you have eaten too much. You can focus, like an objective, neutral observer, on the expansion of and sensations in your stomach, or on possible hot flushes. But you can also focus on the annoying and unpleasant feeling in your body, maybe even with some guilt over that big belly, or on the lazy and satisfied feeling, enjoying the memory of a good meal. Sensory-perceptual processing focus Hopefully it is clear that you can focus on the sensations in your body in various ways. In the next trial you will experience some physical sensations. We ask you to pay attention (like an objective, neutral observer) to the different sensations that occur when the cold stimulus is delivered to the inside of your arm with the thermode/ you are breathing through the mouthpiece. In particular, concentrate on the changes in temperature and on the sensations that you experience on your skin/in the intensity of your breathing or in the breathing effort, or on the changes in sensations in your mouth, throat, airways, or lungs. But maybe you also feel different kinds of sensations elsewhere in your body or changes in your position and muscles. Focus on all bodily sensations and how they change during this somatic experience, no matter how weak or strong they are. Observe how they change from moment to moment in the course of time. Try to notice them in a neutral manner (like an objective, neutral observer) without assessing whether they are good or bad. Also, try to remember all of the sensations, because after this trial we will ask you to describe all of them in detail.

190

Appendix D

Affective-motivational processing focus Hopefully it is clear that you can focus on the sensations in your body in various ways. In the next trial you will experience some physical sensations. We ask you to pay attention to the different feelings that occur when the cold stimulus is delivered to the inside of your arm with the thermode/ you are breathing through the mouthpiece. We ask you to completely give in and immerse yourself in these feelings. Notice how pleasant or unpleasant they are, how intense they are, or if they make you feel relaxed, calm, scared, or angry, etc. But you may also feel some surprise, sadness, joy, cheerfulness or indifference during this trial. Focus on all emotions and feelings that you experience and how they change during this somatic experience, no matter how weak or strong they are. Try to notice how this cold stimulus/ the breathing through the mouthpiece makes you feel, how you react to these experiences, and how they change. Try to remember all of the emotional experiences, because after this trial we will ask you to describe all of them in detail.

191

Appendix E

Appendix E Supplementary material (Chapter 5) Results Retrospective peak symptom ratings. Retrospective peak pain ratings decreased over the course of two weeks in both groups, C4 for Time, F(1, 41) = 9.71, p = .003, !!! = .19; Time, F(1.55, 63.61) = 8.28, p = .002, !!! = .17 (Figure E-2). Peak dyspnea ratings remained at the same level throughout this time. However, peak dyspnea was retrospectively rated higher for the first session compared to the second session. This habituation effect was expressed by the interaction between the PF Manipulation and Order, F(1, 41) = 8.50, p = .006, !!! = .17. Retrospective end symptom ratings. The change in retrospective end pain ratings over time was influenced by both order of PF manipulations and group, Time × Group × Order, F(1.63, 66.81) = 16.13, p < .001, !!! = .28 (Figure E-3). This interaction revealed that a session order, in which the first experimental session included the affective PF manipulation, resulted in the largest discrepancy in immediate and delayed end pain ratings between high and low HSR. However, after 2 weeks the ratings of the two groups did not differ. With regard to the end dyspnea evaluations, the ratings increased over time to the same extent in both groups, F(2, 82) = 11.89, p < .001, !!! = .23. Also, an interaction effect between Group and PF manipulation at a trend level showed that while in the affective PF condition the groups did not differ from each other, in the sensory PF high HSR retrospectively rated end dyspnea higher than low HSR, F(1, 41) = 3.68, p = .062, !!! = .08.

Figure E-1. Mean values and standard errors of heart rate (beats per minute) for high and low HSR during the pain (left panel) and the dyspnea (right panel) induction trials. Data from two sessions are aggregated. To obtain a detailed picture of change over time, data are displayed per 30-s intervals. Pain induction: baseline phase: 0-60 s, induction phase: 60-125 s; recovery phase: 125-185 s; dyspnea induction: baseline phase: 0-60 s, induction phase: 60-210 s; recovery phase: 210-360 s.

192

Appendix E

Figure E-2. Mean retrospective pain ratings (0-100, left) and dyspnea ratings (0-100, right) for high and low habitual symptom reporters (HSR) in sensory and affective processing focus conditions. Top panels represent average ratings, middle panels – peak ratings, and bottom panels – end ratings. Whiskers denote standard errors.

193

Appendix E

Figure E-3. Mean retrospective end pain ratings (0-100) for high and low habitual symptom reporters (HSR) as a function of processing focus manipulation order (sensory-affective/affective-sensory). Whiskers denote standard errors.

194

Appendix F

Appendix F Supplementary materials (Chapter 6)

Table F-1. Means and standard deviations for the processing focus (PF) manipulation groups and F ratios of one-way ANOVAs for the trait characteristics and self-report variables for baseline (BL), pain induction trial (PT) and dyspnea induction trial (DT) at T0 (N = 90). Measure IndTrial PF manipulation group F(df) Sensory PF Affective PF Undirected PF (n = 31) (n = 31) (n = 28) Descriptives Age 20.35 (2.78) 20.48 (2.58) 19.96 (2.43) .31 (2, 87) BMI 22.02 (2.68) 21.70 (2.72) 21.07 (2.49) .99 (2, 87) CSD 89.48 (20.75) 86.03 (17.97) 84.68 (19.88) .48 (2, 87) NA 21.81 (6.75) 20.97 (6.82) 19.79 (6.86) .65 (2, 87) T0 ratings State NA BL 16.29 (3.30) 15.06 (3.58) 15.04 (3.72) 1.25 (2, 87) PT 17.16 (3.27) 16.45 (4.90) 16.93 (4.86) .21 (2, 87) DT 19.35 (4.54) 19.87 (6.65) 20.68 (6.84) .35 (2, 87) State symptoms BL 18.32 (2.48) 18.06 (2.79) 18.14 (3.61) .06 (2, 87) PT 26.00 (5.45) 24.42 (5.18) 25.82 (6.30) .72 (2, 87) DT 28.39 (7.12) 29.87 (7.77) 30.86 (7.99) .79 (2, 87) Valence BL 6.29 (1.23) 6.45 (1.06) 6.46 (1.32) .20 (2, 87) PT 4.13 (1.57) 4.39 (1.73) 4.61 (1.89) .57 (2, 87) DT 3.39 (1.54) 3.58 (1.86) 3.25 (1.40) .31 (2, 87) Arousal BL 4.29 (1.58) 3.97 (1.56) 3.79 (1.75) .74 (2, 87) PT 5.13 (1.46) 5.16 (2.02) 5.18 (1.66) .01 (2, 87) DT 5.55 (2.26) 5.68 (2.07) 5.71 (1.86) .05 (2, 87) Control BL 4.52 (1.69) 5.10 (1.62) 4.79 (1.77) .91 (2, 87) PT 3.50 (1.75) 4.31 (2.02) 3.39 (1.75) 2.11 (2, 83) DT 3.17 (1.80) 3.53 (2.30) 2.82 (1.68) .97 (2, 85) Threat PT 3.97 (1.94) 3.74 (2.14) 3.36 (1.62) .75 (2, 87) DT 5.19 (2.06) 5.23 (2.38) 5.00 (2.04) .09 (2, 87) T0 symptom ratings (0-100) Immediate pain 35.68 (20.14) 35.19 (17.99) 29.93 (17.36) .85 (2, 87) Delayed pain 37.13 (19.35) 32.61 (19.71) 33.57 (20.24) .45 (2, 87) Immediate dyspnea 37.77 (22.90) 43.74 (20.98) 35.43 (18.46) 1.25 (2, 87) Delayed dyspnea 28.55 (24.98) 46.16 (20.78) 36.93 (20.50) 1.48 (2, 87) T0 Focus Focus-sens PT 6.94 (2.07) 7.03 (1.22) 6.89 (1.77) .05 (2, 87) DT 6.29 (1.70) 7.03 (1.22) 7.32 (1.44) 3.93 (2, 87)* Focus-emo PT 4.13 (1.95) 5.03 (1.80) 4.64 (2.02) 1.72 (2, 87) DT 5.10 (2.36) 5.29 (1.94) 5.25 (2.32) .07 (2, 87) Focus-other PT 3.16 (2.46) 3.10 (2.01) 3.04 (2.15) .02 (2, 87) DT 3.87 (2.50) 4.03 (2.36) 3.32 (2.11) .74 (2, 87) Note. BMI = body mass index, CSD = Checklist for Symptoms in Daily Life, NA= negative affectivity, IndTrial = induction trial, Focus-sens = focus on sensations, Focus-emo = focus on emotions, Focus-other = focus on other aspects Due to the software failure, Ns for the control variable are lower than 90 (PT, n = 86; DT, n = 88). *p < .05, **p < .01, ***p < .001.

195

Appendix F

Figure F-1. The interaction of Time with the T0 (upper panels) and T1 ratings (lower panels) of pain/dyspnea (T0/T1 pain/dyspnea) scores for average symptom ratings for pain (left panels) and dyspnea (right panels) ratings.

196

Appendix G

Appendix G Autobiographical memory task (Chapter 7) Dutch version Instructies: Deze test gaat over herinneringen aan gebeurtenissen die je zelf hebt meegemaakt. Ik zal je enkele woorden voorlezen. Het is de bedoeling dat jij bij elk woord je een gebeurtenis probeert te herinneren waaraan dat woord je doet denken. Deze gebeurtenis kan verwijzen naar iets dat recent gebeurd is of heel lang geleden (10 of 15 jaar). Laat ons afspreken dat de gebeurtenis of het moment waaraan het woord je herinnert, moet ‘dateren’ van minstens zeven dagen geleden. Dus je mag geen gebeurtenissen noemen van de afgelopen zeven dagen. Maar het mag dus ook gaan om iets dat veel langer geleden gebeurd is. Het kan een belangrijke gebeurtenis zijn of iets triviaals, iets wat niet echt belangrijk was. Wat wel belangrijk is, is dat de herinnering die je vertelt moet verwijzen naar een specifieke gebeurtenis. Met specifiek wordt bedoeld dat de herinnering verwijst naar één welbepaalde gebeurtenis die op een bepaalde dag plaats vond (maar niet langer dan één dag geduurd heeft). Als ik bijvoorbeeld het woord ‘goed’ geef, zou je kunnen zeggen ‘ik voel me steeds goed op feestjes’. Dit antwoord is echter niet specifiek, het verwijst niet naar één welbepaalde gebeurtenis die op een bepaalde dag plaats vond. Als je zou zeggen ‘ik voelde me goed op het laatste feestje bij Veerle’ is dit een beter antwoord. Dit is een specifieke gebeurtenis. Je zou ook kunnen antwoorden ‘vorige zomer voelde ik me goed’, maar dit verwijst naar een gebeurtenis die langer dan één dag geduurd heeft. Een specifieke gebeurtenis daarentegen is iets dat één welbepaalde keer als dusdanig gebeurd is en korter geduurd heeft dan één dag. Het is ook belangrijk dat je bij elk woord steeds een andere herinnering of gebeurtenis noemt. Je mag dus niet tweemaal naar exact eenzelfde gebeurtenis of herinnering verwijzen. Voor we beginnen zal ik eerst drie oefenwoorden geven, om te kijken of alles duidelijk is. Bij elke cue: Kan je je één specifiek moment herinneren waar het woord ________ je aan doet denken? Oefenwoorden: ontspanning, dokter, actief Stimuluswoorden: ziekte, herstel, griep, gezondheid, koortsig, genezen, bacterie, vaccinatie, hoofdpijn, behandeling

197

Appendix G

English version Instructions: This test inquires about the memories of the events that you have experienced yourself. I will read out some words to you. For each word, I want you to think about an event of which this word makes you think. This event may refer to something that has happened recently or a long time ago (for example 10 or 15 years). Let us agree that the event or the moment of which this word reminds you must have happened at least seven days ago. Thus, you cannot mention the events from the past seven days. But you may also refer to something that has happened much longer ago. It could be an important event or something trivial, something that was not really important. What is important though is that the memory that you recall should refer to a specific event. It means that it should refer to one particular event that took place on a specific day, but lasted less than one day. For example, in response to a word “good”, you could say “I always feel good at parties”. However, this response is not specific, because it does not refer to a specific event that took place on a particular day. On the other had, a response “I felt good at the last party at Emma’s” would be a better answer, because it is a specific event. You could also answer "Last summer I felt good”, but this refers to an event that lasted longer than one day. In contrast, a specific event is something that happened at a particular time and place and that lasted less than one day. It is also important that you provide a different memory for each word. That means that you cannot refer to exactly the same memory or event twice. Before we begin, we will first practice with three practice words to see if everything is clear. For every cue: Can you recall a specific moment that the word __________ reminds you of? Practice words: relaxation, doctor, active Cue words: disease, recover, flu, health, feverish, cure, bacterium, vaccination, headache, treatment

198

Appendix H

Appendix H h-RRS scale construction (Chapter 7) Method Participants. Analyses were conducted on a sample of first-year psychology students from the University of Leuven, Belgium, who completed the questionnaire twice over a period of 6 weeks. There were 388 participants (79.9% women) at Time 1, and 341 (82.4% women) at Time 2. Materials and procedure. The modified version of the RRS (h-RRS) consisted of items describing self-focused and symptom-focused thoughts about possible causes, meanings, and consequences of bodily sensations and symptoms. Participants rated the frequency of thoughts on a 4-point rating scale (almost never, sometimes, often, almost always). Results and discussion A principal component analysis with oblimin (oblique) rotation was performed. Parallel analysis method (Horn, 1965), which compares the size of the observed eigenvalues with the ones taken from random data, indicated a two-factor structure. A factor loading cutoff of .40 was used, and all items were retained (all loadings >.58). The first factor included items 4, 5, 6, 10, 11, and 12, while the second one consisted of items 1, 2, 3, 7, 8, and 9. The inspection of the items revealed a structure resembling the original RRS factors – brooding and reflection (Treynor et al., 2003). In line with the previous findings, we interpret the first factor, which consists of the items reflecting a mental struggle against/non-acceptance of bodily sensations and complaints, as body brooding. The second factor includes items focused on the analysis of causes, meanings, and consequences, which was interpreted as body reflection. Both subscales were reliable. The coefficient alpha for the body brooding subscale was .89 at Time 1, while for the body reflection subscale it was .87. The test-retest correlation for brooding subscale was r = .57, while for the reflection r = .51.

199

Appendix H

h-RRS; Ruminative Response Scale to bodily sensations and symptoms. People think and do many different things when they feel bodily sensations (e.g., faster or deeper breathing while running up the stairs, faster heart rate after exercises, sensations in the stomach before or after eating, etc). Please read each of the items below and indicate whether you almost never, sometimes, often, or almost always think or do each one when you feel bodily sensations. Please indicate what you generally do, not what you think you should do.

1 almost never

2 sometimes

3 often

4 almost always

h-RRS1.

I think “What are the causes of these sensations?”

h-RRS2.

I think “What do these sensations mean?”

h-RRS3.

I think “What are the consequences of these sensations for my health?”

h-RRS4.

I think “Why does this happen to me again?”

h-RRS5.

I think “I wish I didn’t have these sensations!”

h-RRS6.

I think “I can only feel good when I don’t have these sensations anymore!”

People think and do many different things when they feel bodily symptoms (e.g., palpitations, stomach pain, dyspnea, muscle pain, headache, etc). Please read each of the items below and indicate whether you almost never, sometimes, often, or almost always think or do each one when you feel bodily symptoms. Please indicate what you generally do, not what you think you should do.

1 almost never 2 sometimes

3 often

4 almost always

h-RRS7.

I think “What are the causes of these symptoms?”

h-RRS8.

I think “What do these symptoms mean?”

h-RRS9.

I think “What are the consequences of these symptoms for my health?”

h-RRS10.

I think “Why does it happen to me again?”

h-RRS11.

I think “I wish I didn’t have these symptoms!”

h-RRS12.

I think “I can only feel good when I don’t have these symptoms anymore!”

200

Appendix I

Appendix I Supplementary material (Chapter 7) Table I-1. Pearson product-moment coefficients (r) between the indices of reduced autobiographical memory specificity (specific, categoric, and extended memories) and the main psychological variables (depression and rumination) for the whole sample (N = 54), and the SSD (n = 30) and control group (n = 24) separately. Variables 1 2 3 4 5 6 7 Whole sample (N = 54) 1. Specific memories

-

2. Categoric memories

-.79***

-

3. Extended memories

-.80***

.42**

-

4. BDI-II

-.34*

.19

.30*

-

5. RRS-brooding

-.16

.05

.11

.66***

-

6. h-RRS brooding

-.34*

.19

.19

.73***

.75***

-

7. h-RRS reflection

-.21

.14

.17

.57***

.49***

.61***

-

SSD patients (n = 30) 1. Specific memories

-


-.77***

-


-.76***

.46

-

4. BDI-II

-.07

-.07

.09

-

5. RRS-brooding

.02

-.16

-.02

.60***

-

6. h-RRS brooding

-.13

-.05

-.03

.57**

.77***

-

7. h-RRS reflection

-.02

-.01

.03

.50**

.58**

.69***

-

Control group (n = 24) 1. Specific memories

-


-.70***

-


-.73***

.14

-

4. BDI-II

-.02

-.02

.01

-

5. RRS-brooding

.01

.08

-.16

.51*

-

6. h-RRS brooding

.05

.09

-.14

.54**

.50*

-

7. h-RRS reflection

.02

-.02

-.07

.25

.00

-.03

-

Note. BDI-II = Beck Depression Inventory; RRS = Ruminative Response Scale; h-RRS = Ruminative Response Scale to bodily sensations and symptoms. *p < .05, ** p < .01, *** p < .001.

201