Responsiveness to change and English language validation of the WPAI-GERD questionnaire - results from a Canadian study Wahlqvist P 1, Guyatt G 2, Armstrong D 2, Austin P 2, Barkun A 3, Chiba N 2, degl'Innocenti A 1, El-Dika S 4, Fallone C 3, Heels-Ansdell D 2, Tanser L 5, Veldhuyzen van Zanten S 6, Wiklund I 1, Schünemann H 4 1 AstraZeneca R&D Mölndal, Sweden; 2 McMaster University, Hamilton, ON, Canada; 3 McGill University Health Centre, Montreal, QC, Canada; 4 University at Buffalo, Buffalo, NY, USA; 5 AstraZeneca Canada, Mississauga, ON, Canada; 6 Dalhousie University, Halifax, NS, Canada Results from the effect size calculations are shown in Table 4. Results indicate moderate responsiveness of each productivity variable, except for GERD-related absence from work where the effect size indicates a small responsiveness. Further, the large and statistically significant changes after treatment reported in Table 1 are in themselves indicators of responsive- ness of all relevant productivity variables. C ONCLUSIONS Cross-sectional construct validity of the English WPAI-GERD version was confirmed and results indicated that the WPAI-GERD is responsive to change. Although these results also indicated poor longitudinal construct validity, the overall findings suggest that further study of the instrument remains warranted. O BJECTIVES Cross-sectional construct validity of the Work Productivity and Activity Impairment questionnaire for Gastro-Esophageal Reflux Disease (WPAI- GERD) has previously been established in a study of Swedish patients consulting a general practitioner because of GERD symptoms. 1 The purpose of this study was to assess cross-sectional and longitudinal construct validity of the English language version, as well as responsiveness to change of the WPAI-GERD. M ETHODS The WPAI-GERD was used in a clinical study in Canadian GERD patients with moderate or severe symptoms treated with esomeprazole 40 mg once daily for 4 weeks. 2 Productivity variables obtained included GERD-specific hours absence from work, percent of reduced productivity while at work, and percent of reduced productivity while carrying out regular daily activities other than work during the preceding week. Before starting treatment, patients completed the Medical Outcomes Short Form 36 (SF-36) questionnaire 3 for assessing their health status during the past week. The SF-36 covers 8 dimensions: Physical Functioning, Role-Physical, Bodily Pain, General Health, Vitality, Social Functioning, Role- Emotional and Mental Health. Before (at baseline) and after treatment (at follow-up), patients also completed the following questionnaires, all using a one-week recall period: Quality of Life in Reflux and Dyspepsia (QOLRAD), 4 a GERD-specific health-related quality of life questionnaire with 25 items that combine into 5 dimensions: Sleep disturbance, Food and drink problems, Emotional distress, Vitality, and Physical/social functioning. A questionnaire for assessing severity of four symptoms (Heartburn, Acid reflux, Stomach ache, Belching) using a 7-point Likert scale. A questionnaire for assessing severity of overall stomach problems using a 7-point Likert scale. Cross-sectional validity refers to a validity assessment made at one measured time point, and longitudinal validity refers to a validity assessment made with regard to a change between two measurements. Evidence of construct validity was in the present study defined as: A statistically significant (p<0.05, adjusted for multiplicity) and sufficiently high correlation coefficient (preferably ≥ 0.30 for absence from work and ≥ 0.40 for other productivity variables) between each productivity variable and other variables where expected. 1 Responsiveness to change refers to the extent to which an instrument can accurately measure change in a subject’s condition over time. Responsiveness was assessed by calculating the effect size (mean change divided by the standard deviation at baseline) for each productivity variable. An effect size between was defined as small, between as moderate, and >0.8 as large. 5 R ESULTS The analysis included 217 patients, of whom 71% (n=153) were employed. Results obtained from the WPAI-GERD during the study period are shown in Table 1. The improvement (difference from start of treatment) in productivity per patient and week was 0.6 hours for absence from work, 11.0% units for reduced work productivity (=4.7 hours equivalent) and 16.1% units for reduced productivity in daily activities other than work due to GERD symptoms. This improvement translates into an avoided loss of work productivity of 5.3 hours in total ( ) on a weekly basis per patient employed. Table 1. WPAI-GERD results at baseline, at follow-up and change between baseline and follow-up. N BaselineFollow-upChange Mean95% CIMean95% CIMeanp-value Hours absence from work due to GERD (0.4, 1.3)0.3(0.1, 0.5) Hours actually worked (37.5, 42.0)40.0(37.6, 42.5) Percent reduced productivity, work (11.0, 17.0)3.0(1.8, 4.2)-11.0<0.001 Hours lost, reduced productivity, work*1265.8(4.5, 7.1)1.1(0.6, 1.6)-4.7<0.001 Percent reduced productivity, activities (17.9, 24.0)4.9(3.4, 6.3)-16.1<0.001 * Hours lost due to reduced productivity while at work = number of hours actually worked multiplied by percent of reduced productivity. Abbreviations: CI = Confidence Interval. A matrix of baseline, follow-up and change score correlations between productivity variables and other study variables relevant for validation purposes are shown in Table 2. Cross-sectional correlations (i.e. at baseline or follow-up) were all in the right direction and in accordance with expectations in general. 1 For example, the highest correlation with productivity variables would be expected to be with the QOLRAD dimension “Physical/social functioning”, which contains specific questions on the impact of GERD symptoms on work and daily activities. Overall, results support cross-sectional construct validity of the English WPAI-GERD. Change score correlations with QOLRAD dimensions were also according to expectations in general, with a high correlation between productivity and the “Physical/social functioning” dimension. However, change score correlations between productivity variables and symptom assessments were low, and these results do not support longitudinal construct validity of the WPAI-GERD. Table 2. Correlations between productivity variables and other study variables at baseline, at follow-up and change between baseline and follow-up, Pearson correlation coefficients. Hours absence from work, Reduced productivity, Hours lost, reduced productivity, Reduced productivity, GERD (N=139)work (N=130)work (N=126)activities (N=215) BaselineFollow-upChangeBaselineFollow-upChangeBaselineFollow-upChangeBaselineFollow-upChange SF-36 dimensions: Physical Functioning *--0.38*--0.49*-- Role-Physical0.40*--0.54*--0.49*--0.56*-- Bodily Pain *--0.49*--0.54*-- General Health * *-- Vitality *--0.36*--0.47*-- Social Functioning0.38*--0.54*--0.50*--0.59*-- Role-Emotional0.34*--0.43*--0.39*--0.52*-- Mental Health *--0.32*--0.43*-- QOLRAD dimensions: Emotional distress0.33* * 0.47*0.62*0.34* 0.43*0.57*0.32* 0.49*0.51*0.39* Sleep disturbance0.35* *0.56* * *0.46*0.28* Food/drink problem0.33*0.30* *0.50* *0.46* *0.45*0.34* Physical/social functioning0.38* * 0.52*0.62*0.46* 0.50*0.55*0.45* 0.63*0.53*0.56* Vitality0.43* * 0.52*0.65*0.39* 0.47*0.59*0.37* 0.57*0.54*0.42* Symptom severity: Stomach pain *0.43* * *0.44*0.20 Heartburn * * *0.29*0.21 Belching * * *0.11 Acid reflux * * *0.41*0.23* Overall symptoms * * *0.59*0.17 * Statistically significant (  =0.05), p-values adjusted for multiplicity (Bonferroni) at each measurement time point: p< at baseline, p< at follow-up and for change. Changes in each productivity variable between baseline and follow-up by change in heartburn severity are displayed in Table 3. In patients with a large or moderate change in symptoms, mean values indicate that a change in productivity is related to the magnitude of change in symptom severity, which in turn would indicate longitudinal construct validity. However, results also suggest that the insufficiently low change score correlations found between productivity and symptoms are mainly caused by; (1) high variability in productivity changes between individual responses, and (2) patients with a small or no change in symptoms also reporting an improvement in productivity. Table 3. WPAI-GERD results, change between baseline and follow-up by change in heartburn symptom severity. Improvement in heartburn Change in hours absence Change in percent reduced Change in hours lost, Change in percent reduced symptom severity # from work, GERD productivity, work reduced productivity, work productivity, activities NMean (SD)NMean (SD)NMean (SD)NMean (SD) +5 or (4.5) (24.2) (9.3) (23.5) (3.0) (15.7) (9.3) (20.7) (0.6) (14.7) (5.6) (19.9) (1.9) (7.6) (3.3) (17.6) +1 or no improvement (2.0) (17.3) (6.4) (19.9) # Number of units improvement between baseline and follow-up in the 7-graded heartburn symptom severity scale. Abbreviations: SD = Standard Deviation. Table 4. Effect size calculations. NMean95% CI Hours absence from work, GERD (0.05, 0.38) Percent reduced productivity, work (0.47, 0.81) Hours lost, reduced productivity, work (0.47, 0.81) Percent reduced productivity, activities (0.59, 0.84) Abbreviations: CI = Confidence Interval. R EFERENCES 1. Wahlqvist P, Carlsson J, St å lhammar N-O, Wiklund I. Validity of a Work Productivity and Activity Impairment Questionnaire for Patients with Symptoms of Gastro- Esophageal Reflux Disease (WPAI-GERD) - Results From a Cross-Sectional Study. Value in Health 2002;5: Schunemann HJ. A randomized multi-center trial to compare utility elicitation techniques with and without hypothetical health states. Quality of Life Research 2003;12(7):A Ware JE, Snow KK, Kosinski MA. SF-36 Health Survey Manual and Interpretation Guide. Boston, Massachusetts: New England Medical Centre, Talley NJ, Fullerton S, Junghard O, Wiklund I. Quality of life in patients with endoscopy-negative heartburn: reliability and sensitivity of disease-specific instruments. American Journal of Gastroenterology 2001; 96: Cohen J. Statistical power analysis for the behavioral sciences. New York: Academy Press, 1977.