Alimentary Pharmacology & Therapeutics
Validation of a brief symptom questionnaire (ReQuest in Practice) for patients with gastro-oesophageal reflux disease G. RUBIN*, P. UEBEL , A. BRIMO-HAYEKà, K. H. HEY§, H. DOERFLER– & R. C. HEADING**
*University of Sunderland, Sunderland, UK; Ludwigshafen, Germany; àDortmund, Germany; §Paderborn, Germany; –Department of Gastroenterology, Nycomed GmbH, Konstanz, Germany; **Glasgow Royal Infirmary, Glasgow, UK Correspondence to: Prof G. Rubin, University of Sunderland, Priestman Building, Green Terrace, Sunderland, UK. E-mail:
[email protected]
Publication data Submitted 6 December 2007 First decision 2 January 2008 Resubmitted 5 February 2008 Accepted 5 February 2008 Epub OnlineAccepted 9 February 2008
SUMMARY Background A clinical need exists for a means of assessing symptom control in patients with gastro-oesophageal reflux disease. The ReQuest questionnaire has been extensively validated for symptom assessment in both erosive and non-erosive gastro-oesophageal reflux disease but was designed for research purposes. We derived a shorter version (ReQuest in Practice) that would be more convenient for clinical practice. Aim To validate ReQuest in Practice in patients suffering from gastrooesophageal reflux disease. Methods Multicentre, non-interventional, crossover comparison. Patients completed ReQuest in Practice followed by ReQuest or vice versa. Before and after a planned endoscopy, patients completed the health-related quality of life questionnaire GERDyzer. Internal consistency and the Intraclass Correlation Coefficient were calculated. Construct validity was evaluated by correlation with ReQuest and GERDyzer. Results There was high internal consistency of ReQuest in Practice (Cronbach’s alpha: 0.9) and a high Intraclass Correlation Coefficient of 0.99. The measurement error of ReQuest in Practice was 4.1. High correlation between ReQuest in Practice and ReQuest (Spearman correlation coefficient: 0.9) and GERDyzer (Spearman correlation coefficient: 0.8) demonstrated construct validity. Conclusions ReQuest in Practice was proven to be valid and reliable. Its close correlation with ReQuest makes it a promising tool to guide the clinical management of patients across the full spectrum of both erosive and nonerosive gastro-oesophageal reflux disease. Aliment Pharmacol Ther 27, 846–851
846
ª 2008 The Authors Journal compilation ª 2008 Blackwell Publishing Ltd doi:10.1111/j.1365-2036.2008.03641.x
V A L I D A T I O N O F R E Q U E S T I N P R A C T I C E 847
INTRODUCTION
Inclusion criteria
For many patients, gastro-oesophageal reflux disease (GERD) is a chronic disorder requiring long-term therapy and regular review.1 Assessment of the effectiveness of treatment is an essential part of the clinical management. Although outcomes in the early proton pump inhibitor trials were predominantly based on endoscopic findings, there has been a gradual recognition that symptom relief is the most relevant primary outcome measure, both for clinical trials and in practice.2 This is especially important with regard to patients suffering from non-erosive GERD where symptom relief is the only measure for treatment response.3–6 As a result, symptom questionnaires have been developed to assess patient symptom burden and as an aid to management. To be meaningful, a questionnaire should be sensitive to the condition, measure the frequency and intensity of all known dimensions of GERD, both typical and atypical, have proven psychometric properties such as validity, reliability and responsiveness to change and be easy to understand. To measure the full impact of the disease, questionnaires should also be designed for selfcompletion.7 ReQuest was developed as a self-completion GERD symptom questionnaire for use in clinical trials. It has been validated in patients with both erosive and nonerosive GERD.5, 8 To adapt ReQuest for use in clinical practice, we maintained the same structure, but simplified the method and frequency of assessment. We retained the two subscales of gastrointestinal (GI) symptom-related dimensions (‘acid complaints’, ‘upper abdominal ⁄ stomach complaints’, ‘lower abdominal ⁄ digestive complaints’ and ‘nausea’) and quality of life. The aim of this study was the validation and psychometric evaluation of ReQuest in Practice.
Patients who were known to have erosive oesophagitis previously and were suspected of having a relapse, for whom an endoscopy was planned on clinical grounds and who were at least 18 years of age were eligible for inclusion. They were required to have experienced at least moderate (clearly noticeable, but tolerable without immediate relief) heartburn at least twice during the previous 7 days.
Exclusion criteria Pregnant and nursing women, and those expected to be noncompliant, were excluded from the study.
TREATMENT AND CONDUCT Before gastroscopy, all patients completed ReQuest and ReQuest in Practice, together with a quality of life questionnaire (GERDyzer) (Fig. 1). The investigator classified the endoscopic appearance of the erosive oesophagitis according to the Los Angeles Classification and documented demographic data. The following questionnaires were used.
ReQuest in Practice ReQuest in Practice comprises six dimensions such as ‘acid complaints’, ‘upper abdominal ⁄ stomach com40 patients
40 patients
ReQuest in Practice
ReQuest
ReQuest in Practice
ReQuest
METHODS Design GERDyzer
This study was a non-interventional, crossover comparison in six centres in Germany.
Patient population A total of 120 patients with suspected erosive oesophagitis, who were planned to be endoscopically examined independent of this study, were enrolled. ª 2008 The Authors, Aliment Pharmacol Ther 27, 846–851 Journal compilation ª 2008 Blackwell Publishing Ltd
Endoscopy
GERDyzer
Figure 1. Flow chart of the conduct of the non-interventional study.
848 G . R U B I N et al.
plaints’, ‘lower abdominal ⁄ digestive complaints’, ‘nausea’, ‘general well-being’ and ‘sleep disturbances’. For each dimension (except ‘general well-being’), distress is evaluated by means of a 100 mm visual analogue scale (VAS) ranging from ‘not at all’ to ‘extremely severe’ (‘general well-being’: from ‘wonderful’ to ‘extremely poor’). Dimensions related to GI complaints (‘acid complaints’, ‘upper abdominal ⁄ stomach complaints’, ‘lower abdominal ⁄ digestive complaints’ and ‘nausea’) form the subscale ReQuest in Practice-GI. The subscale ReQuest in Practice-WS covers the dimensions ‘general well-being’ and ‘sleep disturbances’.
ReQuest ReQuest comprises seven dimensions such as ‘acid complaints’, ‘upper abdominal ⁄ stomach complaints’, ‘lower abdominal ⁄ digestive complaints’, ‘nausea’, ‘general well-being’, ‘sleep disturbances’ and ‘other complaints’. Each dimension (except ‘general well-being’) is assessed with respect to its frequency and intensity. Frequency is measured by means of a seven-point Likert scale ranging from ‘0’ to ‘more than 10 times ⁄ constant’. The intensity of each dimension (except ‘general well-being’) is assessed by means of a 100 mm VAS ranging from ‘not at all’ to ‘extremely severe’. The sum score for each dimension (except ‘general well-being’) is formed by the weighted product of its frequency and intensity. For the dimension ‘general well-being’, only the intensity is measured. ReQuest can be divided into two subscales, ReQuestGI (including ‘acid complaints’, ‘upper abdominal ⁄ stomach complaints’, ‘lower abdominal ⁄ digestive complaints’ and ‘nausea’) and ReQuest-WSO (encompassing ‘general well-being’, ‘sleep disturbances’ and ‘other complaints’).
STATISTICAL ANALYSES Psychometric analyses These were performed for the full ReQuest in Practice and ReQuest as well as for each subscale separately. The full GERDyzer and each of its dimensions were also psychometrically analysed.
Evaluation of Cronbach’s alpha The correlation coefficient Cronbach’s alpha9 measures internal consistency by determining the degree to which an instrument is free from random error. It was calculated for ReQuest in Practice as well as for its subscales GI and WS. Calculating Cronbach’s alpha makes it possible to determine the degree of association between the symptom dimensions within each subscale. A value for Cronbach’s alpha close to 1 implies a high internal consistency of the construct, whereas a Cronbach’s alpha close to 0 indicates that the construct only possesses little internal consistency.
Calculation of the measurement error The measurement errors of the total score, the GI score and the WS score of ReQuest in Practice were determined.
Calculation of the Intraclass Correlation Coefficient The Intraclass Correlation Coefficients (ICC) of ReQuest in Practice, its GI and WS subscales in combination with ReQuest and its subscales were calculated using exact F-statistics [analysis of variance (ANOVA)]. The following formula was used to determine the ICC: ICC ¼ ðF 1Þ=ðF þ k 1Þ;
GERDyzer The GERD Analyzer (GERDyzer) is a health-related quality of life questionnaire to evaluate the impact of GERD on the quality of life. GERDyzer assesses 10 dimensions of quality of life (i.e. general well-being, pain ⁄ discomfort, physical health, energy, daily activities, leisure activities, social life, diet ⁄ eating ⁄ drinking habits, mood and sleep). Each dimension is assessed using a 100 mm VAS ranging from ‘not at all’ to ‘very much’ (except ‘general well-being’: from ‘excellent to ‘unbearably bad’).
where k = (number of observations ) number of ratings) ⁄ (number of subjects ) 1). An ICC close to 1 implies that there is little variation between the questionnaires, whereas a value close to 0 indicates a high variation.
Calculation of Spearman and Pearson’s correlation coefficients Construct validity measures correlation with existing scales. Spearman and Pearson correlation coefficients ª 2008 The Authors, Aliment Pharmacol Ther 27, 846–851 Journal compilation ª 2008 Blackwell Publishing Ltd
V A L I D A T I O N O F R E Q U E S T I N P R A C T I C E 849
were calculated for the total ReQuest in Practice and ReQuest and for each subscale separately, as well as for the total GERDyzer and each dimension. Pearson or Spearman correlation coefficients close to 1 or )1 indicate similarity between the established questionnaire and the new questionnaire, i.e. they measure the same information. Coefficients near 0 imply little or no correlation at all, i.e. the scales measure different things. In contrast, values close to 0.5 or )0.5 point to a relevant correlation.10
RESULTS Patient population A total of 120 patients were enrolled in the study. 40 patients did not complete ReQuest or ReQuest in Practice according to instructions. Demographic data and patient characteristics are shown in Table 1.
errors of the GI and WS subscales were 3.3 and 2.6 respectively. The ICC with regard to all combinations of ReQuest in Practice vs. ReQuest (Table 2) was very high (0.99). This also held true for both subscales ReQuest in Practice-GI (ICC = 0.98) and ReQuest in Practice-WS (ICC = 0.98). ReQuest in Practice showed a high correlation (Table 3) with ReQuest (Spearman correlation coefficient: 0.95). Pearson and Spearman correlation coefficients were similar; so, only the latter are shown. Correlation between GI and WSO subscales for ReQuest in Practice and ReQuest were similarly high. The subscale ReQuest in Practice-WS correlated well (Table 4) with the total score of the healthrelated quality of life questionnaire GERDyzer (Spearman correlation coefficient about 0.80). As Pearson and Spearman correlation coefficients were
Table 2. Calculation of the ICC between ReQuest and ReQuest in Practice (per-protocol population, n = 80)
Psychometric properties n
ReQuest in Practice possessed high internal consistency, indicated by a Cronbach’s alpha coefficient of 0.92. Both the GI and WS subscales of ReQuest in Practice also showed high internal consistency as indicated by Cronbach’s alpha coefficients of 0.90 and 0.72 respectively. The measurement error of the ReQuest in Practice total score was calculated as 4.1. The measurement
F
k ICC
ReQuest vs. ReQuest in Practice 79* 142 2 0.99 ReQuest-GI vs. ReQuest in Practice-GI 80 104 2 0.98 ReQuest-WS vs. ReQuest in Practice-WS 79* 92.6 2 0.98 * One patient was missing. Replacement of missing entries in ReQuest in Practice or ReQuest was not possible. F, F-value derived from exact F-statistics (analysis of variance) for the scores within the repeated measurement model; k = (number of observations ) number of ratings) ⁄ (number of subjects ) 1); ICC, Intraclass Correlation Coefficient.
Table 1. Demographic data and patient characteristics (PP) PP population (n = 80) Parameter Age (years; mean s.d.) Gender, n (%) Missing Male Female GERD grade, n (%) Missing Grade A Grade B Grade C Grade D
49.0 15.4 1 (1) 39 (49) 40 (50) 4 (5) 37 (46) 30 (38) 7 (9) 2 (3)
PP, per-protocol; n, number of patients; s.d., standard deviation; GERD, gastro-oesophageal reflux disease.
ª 2008 The Authors, Aliment Pharmacol Ther 27, 846–851 Journal compilation ª 2008 Blackwell Publishing Ltd
Table 3. Spearman correlation coefficients of the total ReQuest in Practice and its subscales GI and WS vs. the total ReQuest and its subscales GI and WSO (per-protocol population, n = 80) ReQuest ReQuest ReQuest in Practice in Practice-GI in PracticeWS score sum score score ReQuest sum score 0.95* ReQuest-GI score 0.94 ReQuest-WSO score 0.86*
0.93* 0.94 0.82*
0.92* 0.89 0.88*
* n = 79 patients (replacement of missing entries in ReQuest in Practice or ReQuest was not possible). All correlations were shown to be significant (P < 0.0001).
850 G . R U B I N et al.
GERDyzer General feeling Effect on pain ⁄ discomfort Effect on physical health Effect on energy level Interference with everyday activities Interference with leisure activities Interference with social life Interference with eating and drinking habits Effect on mood Effect on sleep GERDyzer total score
ReQuest in Practice
ReQuest in Practice-WS
ReQuest in Practice ‘general well-being’
0.76 0.69 0.76 0.66 0.74 0.78 0.78 0.56
0.70 0.64 0.74 0.65 0.68 0.72 0.67 0.54
0.59 0.58 0.66 0.67 0.77 0.69 0.71 0.50
0.58 0.65 0.84
0.55 0.76 0.80
0.68 0.48 0.76
Table 4. Spearman correlation coefficients of the total ReQuest in Practice, its WS subscale and ‘general wellbeing’ score vs. GERDyzer total score and its dimension scores (per-protocol population)
All correlations were shown to be significant (P < 0.0001). ReQuest in Practice-WS = subscale of ReQuest in Practice (comprising ‘general well-being’ and ‘sleep disturbances’).
quite similar, only the latter are shown. ReQuest in Practice-WS also correlated with the different dimensions of GERDyzer (Spearman correlation coefficients above 0.5).
DISCUSSION ReQuest has been shown to be valid and reliable for the daily assessment of symptoms in clinical studies in patients suffering from erosive as well as non-erosive GERD.5, 8 ReQuest in Practice is a short symptom questionnaire, derived from ReQuest and designed to provide an instrument for monitoring GERD-related symptoms in clinical practice. This study demonstrates that ReQuest in Practice has psychometric properties that are very close to those of the fully validated parent questionnaire, which was developed for use as an outcome criterion in clinical studies. ReQuest covers the full spectrum of typical and atypical GERD symptoms and offers the possibility to record symptom changes on a daily basis in the course of clinical trials. Psychometric evaluation has shown high internal consistency, high test–retest reliability and high construct validity.5 Both the full and the shortened versions are suitable for self-completion, but ReQuest in Practice requires only six responses on a VAS, from which a sum score can be directly derived, making it convenient to use immediately prior to or during the patient’s consulta-
tion with the doctor. It takes 1 min to complete and score. The full ReQuest requires responses on a seven-point scale (frequency) as well as a VAS (intensity) for each of seven dimensions. A score is then calculated as the weighted product for the two scales for each dimension. It takes 3 min to complete and score, and results in a level of detail that is more suited to research purposes. Validation of ReQuest in Practice resulted in high internal consistency as expressed by high values of Cronbach’s alpha, indicating that the questionnaire and each subscale measure consistently what they claim to measure. The high scores of the GI and WS subscale further showed that combining dimensions into the two subscales was appropriate. The measurement errors of the total score (4.1), the GI score (3.3) and the WS score (2.6) of ReQuest in Practice were higher than for the ReQuest total score (4), GI score (3.1) and WSO score (1.7) respectively (unpublished results). Reduction of symptom frequency and intensity to distress within ReQuest in Practice results in a higher measurement error in comparison with ReQuest. Thus, the ReQuest questionnaire, which is used in clinical trials, is a more accurate measurement instrument, while its simplified version ReQuest in Practice provides sufficient accuracy and is more convenient for day-to-day practice. Calculation of the ICC of the ReQuest in Practice total score and its GI and WS score in comparison to ª 2008 The Authors, Aliment Pharmacol Ther 27, 846–851 Journal compilation ª 2008 Blackwell Publishing Ltd
V A L I D A T I O N O F R E Q U E S T I N P R A C T I C E 851
ReQuest and its subscores yielded values close to 1.0, indicating little variance between ReQuest in Practice and the parental scale ReQuest. Construct validity of ReQuest in Practice was measured by correlation with its precursor ReQuest. The correlation between the total score, the GI and WS scores of ReQuest in Practice and the respective scores of ReQuest resulted in Spearman coefficients close to 1.0, implying high similarity between measurements of the two instruments. The complete score of ReQuest in Practice and the ‘general well-being’ score correlated well with the total score and the single dimensions of the health-related quality of life questionnaire GERDyzer. This indicates that it is appropriate to combine aspects that affect the patient’s quality of life into dimension ‘general wellbeing’ of ReQuest in Practice. ReQuest in Practice was developed for use in everyday clinical practice and this validation study was undertaken in a pragmatic clinical setting. This was a cross-sectional study and we did not seek to assess responsiveness to change. Furthermore, patients in this study were known to have previously had erosive oesophagitis. However, validity in erosive and non-erosive GERD as well as responsiveness to change have been previously demonstrated for ReQuest5, 8 and the close correlation between ReQuest and ReQuest in
REFERENCES 1 National Institute for Clinical Excellence. Dyspepsia – Management of dyspepsia in adults in primary care. Clinical Guideline 17, National Institute for Clinical Excellence, London, 2004. 2 Vakil N, van Santen SV, Kahrilas P, Dent J, Jones R, the Global Consensus Group. The Montreal Definition and Classification of Gastroesophageal Reflux Disease: a Global Evidence-Based Consensus. Am J Gastroenterol 2006; 101: 1900–20. 3 Dent J, Brun J, Fendrick AM, et al. An evidence-based appraisal of reflux disease management – the Genval Workshop report. Gut 1999; 44: 1–16. 4 Tack J, Fass R. Review article: approaches to endoscopic-negative
Practice allows us to infer that these properties are shared between the two.11 This means that Request in Practice is suitable for use across the spectrum of GERD, including those patients seen in primary care, where endoscopic status may be unknown. Further research is now needed to evaluate its utility in clinical management and clinical decision making in GERD. In general, ReQuest in Practice and its two subscales were proven to be valid and showed very good psychometric properties: high internal consistency, high ICC and high construct validity. It has potential as an instrument for use in clinical practice to monitor treatment-induced symptom changes in GERD patients.
ACKNOWLEDGEMENTS Declaration of personal interests: In the past 5 years, G. Rubin and R. Heading have received consultancy fees and conference attendance support from Nycomed GmbH and Reckitt Benckiser. H. Doerfler is Medical Director (Gastroenterology), Clinical Research and Development, for Nycomed. Declaration of funding interests: This study was funded by Nycomed. Initial data analysis and writing support was provided by Waldemar Braun and Ilona Vogt-Humberg, who are employees of IFE, Institute for Research and Development, Witten, Germany.
reflux disease: part of the GERD spectrum or a unique acid-related disorder? Aliment Pharmacol Ther 2004; 19 (Suppl. 1): 28–34. 5 Bardhan KD, Stanghellini V, Armstrong D, Berghoefer P, Gatz G, Moennikes H. International validation of ReQuestTM in patients with endoscopy-negative gastro-oesophageal reflux disease. Aliment Pharmacol Ther 2004; 20: 891–8. 6 Martinez SD, Malagon IB, Garewal HS, Cui H, Fass R. Non-erosive reflux-disease (NERD) – acid reflux and symptom patterns. Aliment Pharmacol Ther 2003; 17: 537–45. 7 McColl E, Junghard O, Wiklund I, Revicki DA. Assessing symptoms in gastroesophageal reflux disease: how well do clinicians’ assessments agree with
ª 2008 The Authors, Aliment Pharmacol Ther 27, 846–851 Journal compilation ª 2008 Blackwell Publishing Ltd
8
9 10
11
those of their patients? Am J Gastroenterol 2005; 100: 11–8. Bardhan KD, Stanghellini V, Armstrong D, Bergho¨fer P, Gatz G, Mo¨nnikes H. Evaluation of GERD symptoms during therapy. Part I: Development of the new GERD questionnaire ReQuest. Digestion 2004; 69: 229–37. Bland JM, Altmann DG. Cronbach’s alpha. BMJ 1997; 314: 572. Nunally JC, Bernstein IH. Psychometric Theory, 3rd edn. New York, NY: McGraw Hill, Inc., 1994. Moennikes H, Bardhan KD, Stanghellini V, Berghoefer P, Bethke TD, Armstrong D. Evaluation of GERD symptoms during therapy. Digestion 2007; 75 (Suppl. 1): 41–7.
