Behavioural and Cognitive Psychotherapy, 2014, 42, 117–123 First published online 19 July 2013 doi:10.1017/S1352465813000611
Hear Today, Not gone Tomorrow? An Exploratory Longitudinal Study of Auditory Verbal Hallucinations (Hearing Voices) Nicky Hartigan University of Surrey, Guildford, UK
Simon McCarthy-Jones Macquarie University, Sydney, Australia
Mark Hayward University of Sussex, and Sussex Partnership NHS Foundation Trust, Guildford, UK
Background: Despite an increasing volume of cross-sectional work on auditory verbal hallucinations (hearing voices), there remains a paucity of work on how the experience may change over time. Aims: The first aim of this study was to attempt replication of a previous finding that beliefs about voices are enduring and stable, irrespective of changes in the severity of voices, and do not change without a specific intervention. The second aim was to examine whether voice-hearers’ interrelations with their voices change over time, without a specific intervention. Method: A 12-month longitudinal examination of these aspects of voices was undertaken with hearers in routine clinical treatment (N = 18). Results: We found beliefs about voices’ omnipotence and malevolence were stable over a 12-month period, as were styles of interrelating between voice and hearer, despite trends towards reductions in voicerelated distress and disruption. However, there was a trend for beliefs about the benevolence of voices to decrease over time. Conclusions: Styles of interrelating between voice and hearer appear relatively stable and enduring, as are beliefs about the voices’ malevolent intent and power. Although there was some evidence that beliefs about benevolence may reduce over time, the reasons for this were not clear. Our exploratory study was limited by only being powered to detect large effect sizes. Implications for clinical practice and future research are discussed. Keywords: Voice-hearing, schizophrenia, psychosis.
Reprint requests to Simon McCarthy-Jones, ARC Centre of Excellence in Cognition and its Disorders, Department of Cognitive Science, Macquarie University, North Ryde, NSW 2109, Australia. E-mail:
[email protected] © British Association for Behavioural and Cognitive Psychotherapies 2013
118
N. Hartigan et al. Introduction
Auditory verbal hallucinations (AVHs: hearing voices) are often, but not always, distressing experiences (McCarthy-Jones, 2012). Whilst the beliefs a person holds about their voices impacts on the ensuing levels of distress they experience (Peters, Williams, Cooke and Kuipers, 2012), recent work has begun to explore how the relationships people have with their voices contribute to distress levels. One framework that has been used to analyse such relationships is Birtchnell’s theory of relating (1996). Using this framework, Vaughan and Fowler (2004) found that Voice Dominance (the voice relating to the person in a dominant, insulting manner) and Hearer Distance (the hearer avoiding communication with the voice) were associated with greater levels of distress in hearers, and argued that it is not a hearer’s beliefs about the voice’s power per se that leads to distress, but the way in which the voice is perceived to use its power, for example by being domineering and bullying. A later study by Sorrell, Hayward and Meddings (2010) found that Voice Dominance, Hearer Distance, and Voice Intrusiveness (the voice imposing itself on the hearer) were all significantly positively correlated with levels of distress, but that when levels of malevolence and omnipotence were controlled for, these correlations became non-significant. Sorrell et al. concluded that it is a combination of both relating styles and beliefs about voices (specifically malevolence and omnipotence) that determines distress, with beliefs about voices either moderating (influence the strength of the relationship) or mediating the association between relating styles and distress. To date, almost all such work has been cross-sectional, with only one study directly exploring how beliefs about voices change over time. This study by Csipke and Kinderman (2006) investigated change in beliefs about voices’ intent, for clinical hearers over a 6-month period (N = 16), finding that the frequency and severity of voices decreased significantly but not finding significant changes in beliefs about the malevolent or benevolent intent of voices. This offered support for the proposal that beliefs about voices are enduring and stable, irrespective of immediate frequency and severity of voices, and do not change without a specific intervention. The first aim of our study was to attempt replication of these findings in a 12-month longitudinal study, using a more psychometrically reliable measure of beliefs about voices’ benevolent or malevolent intent and including a measure of perceived voice-power. We hypothesized that beliefs about voices’ malevolence, benevolence and omnipotence, as assessed by the Beliefs about Voices Questionnaire-Revised (BAVQ-R; Chadwick, Lees and Birchwood, 2000), would remain stable over a 12-month period, irrespective of changes in the severity of voices, as assessed by the Psychotic Symptoms Rating ScaleAuditory Hallucinations Scale (PSYRATS-AH: Haddock, McCarron, Tarrier and Faragher, 1999). We also aimed to examine whether relationships with voices change over time. Previous research has found that over time hearers have increasing levels of intimacy with their voices and more detailed dialogues with them (Nayani and David, 1996), with Chin, Hayward and Drinnan (2009) finding developments in the relationship over time were evident for all participants in their study. These findings suggest that relating styles may change over time, with voice-hearers becoming closer to their voices. To investigate this we examined whether voice-hearers’ relations with their voices, as assessed by the Voice and You scale (VAY; Hayward, Denney, Vaughan and Fowler, 2008), changed over a 12-month period.
Change in voices over time
119
Method Participants Inclusion criteria were being over 18 years of age, in receipt of mental health services, and having heard voices for at least 6 months. Exclusion criteria were participants’ voice-hearing being a likely consequence of substance misuse or organic illness, currently experiencing acute psychosis or distress that meant they were not able to give informed consent, or lacking sufficient English language skills. As this was an exploratory study, we designed the study to be able to detect large (and hence clinically relevant) effect sizes. G∗ Power was used to calculate the necessary sample size to detect, with 80% power, a large effect size (Cohen’s d = 0.8) at an alpha of .05. This indicated for a repeated-measures t-test that a sample size of 15 would be required. Given that Csipke and Kinderman’s (2006) 6-month longitudinal study had an attrition rate of 65%, in order to attempt to achieve our desired sample size we assessed 32 participants (19 male) at baseline, recruited from local adult mental health services. All were currently taking antipsychotic medication. We were able to reassess 18 of these participants at 12-month follow-up. Reasons for attrition included participants withholding their consent to be recontacted, the researchers being unable to make contact with the participant, and participants being in hospital or unwell. The mean age of the 18 participants was 39.39 years (SD = 9.95); all were of White British ethnicity, and they had been hearing voices for a mean of 16.50 years (SD = 9.46). Nine participants reported their predominant voice was male, four female, and five unknown. Eight had voices that they personified as real people, seven as unknown people, and three as supernatural entities. Ethical approval was received for this study from the South West Surrey Local Research Ethics Committee. Procedure Participants were interviewed face-to-face. Follow-up interviews were conducted as close as possible to 12 months (range 10–14 months) following the original interview date. The following measures were completed at baseline and 12-month follow-up. During the course of the 12 months, 10 (56%) participants attended Hearing Voices Groups (peer-support groups for people who hear voices), 2 (11%) underwent formal individual psychotherapy, and 7 (39%) changed medication. Measures Voice and You (VAY; Hayward et al., 2008) is a 28-item, self-report measure of a person’s assessment of themselves in relation to their predominant voice, with four subscales: Voice Dominance (e.g. “My voice tries to get the better of me”); Voice Intrusiveness (e.g. “My voice does not let me have time to myself”); Hearer Dependence (e.g. “My voice helps me make up my mind”); and Hearer Distance (e.g. “I don’t wish to spend much time listening to my voice”). Psychotic Symptoms Rating Scale - Auditory Hallucinations Scale (PSYRATS-AH; Haddock et al., 1999) is an 11-item scale, administered by the researcher, that assesses characteristics of voices such as frequency, duration, severity, loudness, location, degree of negative content and controllability. It also assesses the impact of the voice upon the hearer in terms of distress caused, beliefs about origin of voices, and disruption caused to the hearer’s life.
120
N. Hartigan et al. Table 1. Changes in relevant variables over a 12-month period
Scale and item Beliefs about voices: BAVQ-R: Malevolence BAVQ-R: Benevolence BAVQ-R: Omnipotence Relationship to voices: VAY: Voice Dominance VAY: Voice Intrusiveness VAY: Hearer Dependence VAY: Hearer Distance PSYRATS-AH variables: Frequency Duration Loudnessb Negative content (amount) Negative content (degree) Distress (intensity) Distress (amount)c Controllabilityc Disruptiond
t, sig.
Effect size, da (95% CI)
Baseline
12m follow-up
11.67 (5.95) 5.00 (5.88) 12.00 (4.52)
11.39 (6.28) 3.33 (4.97) 11.11 (4.48)
14.83 (7.14) 9.44 (4.33) 9.33 (6.47) 13.89 (5.89)
15.28 (7.36) 9.67 (4.88) 8.50 (5.69) 13.22 (6.72)
− .55, p = .59 − .30, p = .76 .94, p = .36 .71, p = .49
− .19 (−3.49−3.21) − .11 (−2.11−2.15) .32 (−2.67−2.95) .24 (−2.48−3.35)
3.00 (1.14) 2.89 (0.90) 2.50 (1.10) 3.17 (1.04) 3.44 (1.10) 2.83 (1.20) 3.12 (1.11) 3.06 (1.20) 1.92 (0.90)
2.78 (.94) 2.44 (1.15) 2.38 (.96) 3.28 (1.27) 3.11 (1.23) 2.67 (1.33) 2.53 (1.46) 3.18 (1.33) 1.33 (0.78)
.78, p = .45 1.52, p = .13 .48, p = .63 − .49, p = .63 1.46, p = .16 .68, p = .51 1.98, p = .07∗ − .49, p = .63 2.55, p = .03∗
.26 (−.27−.69) .54 (.12−1.07) .17 (−.37−.64) − .17 (−.65−.42) .49 (−.02−1.05) .22 (−.34−.83) .70 (.17−1.39) − .17 (−.74−.46) 1.06 (.55−1.50)
.36, p = .72 2.54, p = .02∗ 1.47, p = .16
.12 (−2.63−3.02) .89 (−1.83−3.18) .49 (−1.59−2.56)
Notes: a Difference between means divided by the pooled standard deviation times the square root of 1 minus the correlation (r) between scores at baseline and at 12m follow-up; b Due to missing data, n = 16; c Due to missing data, n = 17; d Due to missing data, n = 12. ∗ Trend towards significance, as per paper’s statistical criterion, i.e. .015 < p < .10
Beliefs about Voices Questionnaire-Revised (BAVQ-R; Chadwick et al., 2000) is a 35-item questionnaire with a number of sub-scales; however, we only report here on the key subscales of beliefs about the voice’s malevolence, benevolence and omnipotence. Results We adjusted conventional significance levels to take into account the multiple tests performed. As Bonferroni corrections have been argued to be overly conservative (Narum, 2006), we employed the correction suggested by Benjamini and Yekutieli (2001). As a total of 16 tests were performed (Table 1), this corresponded to a corrected significance level of p = .015, with a trend towards significance for this study being defined as .015< p < .10. To test our first hypothesis, we examined whether beliefs about voices’ malevolence, benevolence and omnipotence changed over time. As can be seen from Table 1, at baseline assessment BAVQ-R scores indicated that participants mostly perceived their voice to have malevolent intent, to have omnipotence and low benevolence. As Table 1 also shows, at the corrected significance level there was no significant change in malevolence or omnipotence over time, despite there being a trend towards a reduction in both the amount of voice-related distress and voice-related disruption. However, there was a trend towards a reduction in beliefs about the voices’ benevolence.
Change in voices over time
121
To evaluate our second hypothesis, we tested whether scores on the VAY changed significantly over time. As indicated in Table 1, there were no significant changes in VAY scores. Descriptive statistics indicated that both at baseline and follow-up participants typically perceived their voice to relate to them primarily from a position of dominance, perceived themselves relating to their voice from a position of distance, with the voices being intrusive (with average scores on both these scales being closest to the “Quite often true” response). Participants had low levels of dependence on their voice (with average scores on this scale being “Sometimes true”), both at baseline and at follow-up. We also undertook two re-performances of our analyses. First, we re-analysed our data using the Wilcoxon Signed-Rank test, as some of our variables were non-normally distributed. Second, as two participants had undergone formal individual psychotherapy over the course of the 12-month follow-up period, we re-performed our analyses excluding these individuals. Neither of these re-analyses changed our original pattern of findings, with the same variables (benevolence, amount of voice-related distress and voice-related disruption) continuing to show trends towards significance (.015 < all p’s
