Minimal difference in pain associated with change in quality of life in children with rheumatic disease.код для вставкиСкачать
Arthritis & Rheumatism (Arthritis Care & Research) Vol. 47, No. 5, October 15, 2002, pp 501–505 DOI 10.1002/art.10661 © 2002, American College of Rheumatology ORIGINAL ARTICLE Minimal Difference in Pain Associated With Change in Quality of Life in Children With Rheumatic Disease SONNY DHANANI,1 JOSEE QUENNEVILLE,1 MEGAN PERRON,2 MOHAMED ABDOLELL,1 3 AND BRIAN M. FELDMAN Objective. To establish the minimal change on a pain visual analog scale (VAS) associated with change in self-reported quality of life in pediatric rheumatology patients. Methods. Subjects were a cohort of 533 pediatric rheumatology patients in Toronto. Pain and perceived quality of life were measured at 2 consecutive visits to a clinic. Results. Among patients who rated quality of life at the second visit as “a little better” and “much better,” the mean change in pain score on a 10-cm VAS was reduced by 0.82 and 1.45 cm, respectively. For those whose quality of life changed to “a little worse” and “much worse,” the pain scores increased by 1.90 and 3.69 cm. Conclusion. Our results suggest that future studies of the assessment and treatment of pain in this population should aim for a minimum reduction in pain score of 0.82 cm on a 10-cm VAS to achieve clinical improvement in quality of life. KEY WORDS. Minimal difference; Pain; Quality of life; Pediatrics; Rheumatic disease. INTRODUCTION Rheumatic diseases include both common and rare pediatric chronic diseases, such as juvenile rheumatoid arthritis, systemic lupus erythematosus, dermatomyositis, and the spondylarthropathies. These diseases can affect many aspects of life, such as pain and physical function, which in turn can have profound effects on a child’s daily living and psychosocial well being. Thus, the clinical assessment of children with rheumatic diseases, such as juvenile rheumatoid arthritis, is commonly based on a diverse set of measures (1,2). In particular, pain is an important but poorly underDr. Feldman’s work is supported by a Career Scientist Award from the Ontario Ministry of Health. 1 Sonny Dhanani, MD, BSc, Josee Quenneville, MSc, Mohamed Abdolell, MSc: The Hospital for Sick Children and the University of Toronto, Toronto, Ontario, Canada; 2 Megan Perron, RN: The Bloorview-MacMillan Centre, Toronto, Ontario, Canada; 3Brian M. Feldman, MD, MSc, FRCPC: The Hospital for Sick Children, the University of Toronto, and The Bloorview-MacMillan Centre, Toronto, Ontario, Canada. Address correspondence to Brian M. Feldman, MD, MSc, FRCPC, Associate Professor, Pediatrics and Public Health Sciences, The Hospital for Sick Children, University of Toronto, 555 University Avenue, Toronto, Ontario, M5G 1X8, Canada. E-mail: brian.Feldman@sickkids.on.ca. Submitted for publication August 8, 2000; accepted in revised form February 3, 2002. stood aspect of rheumatic disease in children (3), perhaps because its measurement is also not yet well understood. Varni and Thompson were the first to develop a standardized pain assessment tool for children: the Pediatric Pain Questionnaire (PPQ), which established the use of a visual analog scale (VAS) to measure pain intensity (4,5). This scale is now accepted along with other aspects of the PPQ as a valid and reliable method of assessing pain in the pediatric population (6,7), and has been used as part of several comprehensive assessment tools, such as the Childhood Health Assessment Questionnaire (CHAQ) (2,8). Although the importance of pain has long been accepted, methods for assessing it in the pediatric population are only now being devised. Similarly, quality of life has been recognized as one of the most important clinical outcomes in rheumatology, but it too is not well understood (2,9,10). Some studies have suggested that pain may correlate with clinical disease activity and functional status (6,11–13), whereas others have examined the relevance of pain to the general health and quality of life of both children and adults (8,11,14,15). Although a relationship between pain and quality of life has been established, studies have not been done to determine the minimal clinically important change in pain, as measured by VAS, that will lead to change in quality of life. Our objective was to determine the minimal change in the pediatric pain VAS that is associated with change in 501 502 Dhanani et al scale their pain level in the past week. The number of centimeters marked is recorded as a score from 0 –10, with 0 ⫽ no pain and 10 ⫽ very severe pain. The categorical change in quality of life score was obtained from the Quality of My Life scale (10), a 5-category scale that asks “Since my last visit my life is. . . .” Possible answers are “much worse,” “a little worse,” “the same,” “a little better,” or “much better.” Patients were asked to circle the category that best described their quality of life compared with the previous visit. If patients were too young to complete the questionnaires themselves, parents helped complete these scales. The pain VAS and the scales in the CHAQ have been developed for use in children as young as 1 year old. Figure 1. Outcome measures. self-reported quality of life in pediatric rheumatology patients, using a retrospective cohort design. Establishment of this minimal important change in the perception of pain may enhance our understanding of its impact in rheumatic disease and help define a standard for studies dealing with the treatment of pain. PATIENTS AND METHODS Patients. Subjects were pediatric patients attending the rheumatology clinic at The Hospital for Sick Children and the Bloorview-MacMillan Centre in Toronto, Ontario, Canada from August 1995 to February 1999. These facilities are tertiary care centers serving metropolitan Toronto as well as central and northern Ontario. All patients, regardless of type of rheumatic disease, age, race, or sex, are routinely asked to answer a standard battery of questions. These questionnaires are completed by either the patient or parent on repeated visits to the clinic. Parents are asked to complete or help complete the questionnaires if patients are too young to respond adequately themselves. Our study included all those who completed pain scales and change in quality of life assessments on 2 consecutive visits. Patients were excluded if they made only 1 visit to the clinic or if they returned for a second visit but did not complete pain scales or quality of life assessments. Measurements. The main instruments used for this study were a pain VAS and a categorical scale measuring change in quality of life (Figure 1). These measures formed part of a larger standard battery of questionnaires that included scales for quality of life, coping, global disease activity, global disease impact, health-related quality of life, and the CHAQ (2). These scales were also used in our study and have been previously validated in various pediatric rheumatic diseases (2,8,10). The pain VAS used in the CHAQ is a double-anchored 10-cm analog scale with anchors of “no pain” and “very severe pain.” Patients were asked to indicate on the linear Data collection and statistical analysis. Patients completed questionnaires at each clinic visit between August 1995 and February 1999. Data from the pain VAS and the quality of life scale were compiled, entered into a computer database, and analyzed using Datadesk 6 (Data Description Inc., Ithaca, NY). The mean change in pain score associated with each category of quality of life assessment between 2 consecutive visits was established with 95% confidence intervals (95% CI). We calculated change in pain scores by subtracting the pain score at the second visit from the pain score at the first visit. Many patients had more than 2 visits. For the initial analysis, the first 2 visits for each subject were used to see whether there were noticeable changes in quality of life associated with change in pain. To increase the precision, we performed a subsequent analysis that calculated the mean change in pain score for all patients who at any 2 consecutive visits scored the change in quality of life for the second as “a little better.” Separate analyses were likewise done for each of the other 4 possible responses to the change in quality of life scale. For example, if patient A from the first to second visit was “much better,” then his change in pain score was used in the calculation of mean pain change for the category of “much better.” Subsequently, if patient A from the third to fourth visit was “much worse,” then this change in pain score was used in calculating the mean pain change for the category “much worse.” But, if patient A was again “much better” from the third to fourth visit, then the data from this visit was not used. In this way, some subjects contributed data to the mean change in pain score for 1 or more categories of change in quality of life. However, the calculation of mean pain change associated with any one category of change in quality of life did not use any subject more than once. Analysis of variance (ANOVA) or chi-square tests were used where appropriate to compare the demographic and clinical profiles of subjects who chose each category of change in quality of life. P values ⬍ 0.05 were considered statistically significant. Subsequent pair-wise comparisons were corrected for multiple statistical testing by the least significant difference (LSD) test. General linear models were also used to determine whether such factors as sex, age, parent or child responder, type of rheumatic disease, coping, disease severity, and overall disability confounded the relationship of change in pain and change in quality of Minimal Difference in Pain Needed to Change Quality of Life life. For these models, change in quality of life was recoded as a number from –2 to ⫹2. Because change in quality of life was an ordinal variable in its original form, we used standard regression diagnostics to ensure that the assumptions underlying multiple regression were met. Regression analysis was carried out to examine the contribution of change in pain to perceptions of change in quality of life; the R2 from the ANOVA model was used to estimate the amount of variability accounted for by change in pain. Ordinal logistic regression was used to model the probability of “worse” quality of life as a function of pain change. A proportional odds model was employed. The proportional odds assumption was investigated using plots of the cumulative logits to confirm that no crossover effects were indicated. The effect of the predictor variable on the odds of being in a “worse” quality of life category is reported as an odds ratio with 95% CI. Values of the change in pain that are associated with a 50% chance of reporting that they are “much worse,” “much worse or a little worse” and “much worse or a little worse or the same” were obtained from the ordinal logistic regression model. RESULTS Data from a total of 533 patients were available for analysis between August 1995 and February 1999. The sex distribution was 69.6% female and 30.4% male. The mean age was 11.3 years (range 1.3–19.2 years). The median time between clinic visits was 87 days. The most common diagnosis was juvenile rheumatoid arthritis (41.4%); however, a full spectrum of illnesses was encountered, including systemic lupus erythematosus (14.3%), nonarticular rheumatism (8.5%), and spondylarthritis (4.7%). Other diagnoses (31.1%) included pain related to myositis-, orthopedic-, psoriatic-, infectious-, vasculitic-, and inflammatory bowel-related illnesses. The age and sex statistics of the study group were representative of the rheumatology subspecialty population in general. Questionnaires were completed by the patient 81.8% of the time (parents assisted in 11.5%) and by parents in 18.2% of cases. We had some concern that questionnaires completed by parents might differ systematically. However, when the data from parent-completed forms were removed from the analysis, the results were unchanged. Thus, further analysis of data included all the questionnaires. We calculated the average change in pain score associated with each category in quality of life between the first 2 visits. For patients whose quality of life was rated as “the same” between the first 2 visits, the mean difference in pain score was 0.07 cm (95% CI – 0.39, 0.24; n ⫽ 156). Change in quality of life to “a little better” showed that the mean change in pain score was reduced by 0.89 cm (95% CI –1.29, – 0.49; n ⫽ 156). In the group of patients whose quality of life changed to “a little worse,” average pain scores increased by 2.15 cm (95% CI 1.33, 2.98; n ⫽ 50). The difference in pain scores observed for changes in quality of life to “much better” and “much worse” between the first 2 visits were –1.22 cm (95% CI –1.68, – 0.75; n ⫽ 503 151) and ⫹3.29 cm (95% CI 1.77, 4.82; n ⫽ 12), respectively. There was a significant difference in the amount of change in pain among the patients in different categories of change in quality of life (n ⫽ 533; F ⫽ 24.99, degrees of freedom ⫽ 4, P ⱕ 0.0001). Pair-wise comparisons revealed that patients who rated their quality of life as “a little better” had a significantly different change in pain as compared with those who rated their quality of life as “the same” (P ⫽ 0.005, LSD). Likewise, those whose quality of life was “a little worse” had a significantly different change in pain compared with those whose quality of life was “the same” (P ⬍ 0.0001). After analyzing data from the first 2 visits alone, we calculated the average change in pain score associated with each category of change in quality of life between any 2 consecutive visits to increase the power of our estimation. For patients whose quality of life was rated as “the same” between any 2 visits, the mean difference in pain score was 0.06 cm (95% CI – 0.34, 0.22; n ⫽ 228). This change in pain score was significantly different from patients who rated quality of life as “a little better:” In this category, the mean change in pain score was reduced by 0.82 cm (95% CI –1.15, – 0.49; n ⫽ 234). In the group of patients whose quality of life changed to “a little worse,” average pain scores increased by 1.90 cm (95% CI 1.30, 2.51; n ⫽ 92). The difference in pain scores observed for changes in quality of life to “much better” and “much worse” were –1.45 cm (95% CI –1.83, –1.07; n ⫽ 210) and ⫹3.69 cm (95% CI 2.59, 4.80; n ⫽ 26), respectively. Table 1 shows the characteristics of the subjects in each category of change in quality of life. The group that scored quality of life as “much better” was significantly younger than the rest. There were minor, but statistically significant differences in sex as well. Fewer patients with juvenile rheumatoid arthritis, but more with systemic lupus erythematosus, rated their quality of life as the same between 2 visits. Patients who scored their quality of life as “a little worse” or “much worse” had significantly worse global disease severity scores and higher scores on measures of global disease impact (2). The R2 from the ANOVA model was used to estimate the amount of variability accounted for by change in pain. The result yielded an R2 of 0.159 (P ⬍ 0.0001), suggesting that change in pain alone explained about 16% of the variance in quality of life. Furthermore, ordinal logistic regression was used to estimate the probability of worse quality of life for any pain change. This suggested that a 1-cm increase in change in pain on a 10-cm scale increased the odds of a worse quality of life by a factor of 1.3 with a 95% CI of 1.22, 1.38. Another way to determine the minimal clinically important change in pain is to determine the threshold value at which 60% of the subjects reported that their life was “a little better” or “much better.” From the ordinal logistic regression model, this threshold value was – 0.7 cm. We constructed general linear models to examine the influence of other disease and demographic factors on change in quality of life measured at the second visit. The best predictor remained change in pain scores, though other factors were also significant (Table 2). 504 Dhanani et al Table 1. Characteristics of the 533 study subjects by change in quality of life at second clinic visit* Variable Much better Subjects Age, years (SD) Female/male Diagnosis, n (%) JRA Lupus Nonarticular rheumatism Seronegative spondylarthropathy Other Days since previous visit (SD) Global disease severity score from CHAQ (SD) Global disease impact score from CHAQ (SD) 151 9.9 (4.4) 107/44 Little better Same Little worse Much worse P 156 11.5 (4.2) 92/64 164 12.1 (4.4) 126/38 50 11.9 (4.3) 39/11 12 11.9 (4.4) 7/5 ⱕ0.0001 (ANOVA) ⱕ0.005 (2) 76 (50.3) 12 (7.9) 9 (6.0) 3 (2.0) 51 (33.8) 96 (200) 0.3 (0.5) 67 (42.9) 16 (10.3) 14 (9.0) 13 (8.3) 46 (29.5) 83 (260) 0.6 (0.6) 62 (37.8) 40 (24.4) 15 (9.1) 5 (3.0) 42 (25.6) 86 (79) 0.5 (0.6) 23 (46.0) 6 (12.0) 5 (10.0) 3 (6.0) 13 (26.0) 83 (54) 1.2 (0.8) 5 (41.7) 1 (8.3) 1 (8.3) 1 (8.3) 4 (33.3) 76 (82) 1.7 (0.8) 0.4 (0.5) 0.7 (0.7) 0.6 (0.7) 1.4 (0.7) 2.0 (0.7) ⱕ0.007 (2) 0.57 (ANOVA) ⱕ0.0001 (ANOVA) ⱕ0.0001 (ANOVA) * CHAQ ⫽ Childhood Health Assessment Questionnaire; JRA ⫽ juvenile rheumatoid arthritis; ANOVA ⫽ analysis of variance; 2 ⫽ chi-squared test. DISCUSSION In addition to reemphasizing that pain alters quality of life, this study identified the minimal clinically important difference leading to significant improvement in quality of life in pediatric rheumatology patients. This difference can be defined as the smallest difference in the domain of interest (level of pain) that patients perceive as beneficial; it is thus influential in their management (16). In our study, this average reduction of pain score associated with a minimal improvement in quality of life was 0.82 cm on a 10-cm VAS. Defining the minimal score may set a standard for future studies that involve the treatment of pain; in other words, a reduction of 0.82 cm may be used in other studies in this population to show significant change in clinical care. Studies in adults have suggested that pain has a profound effect on a person’s quality of life (14,15). Our results show that changes in pain scores are associated with changes in quality of life. We showed that every 1-cm increase in pain score increased the odds of a worse quality of life by a factor of 1.3. Furthermore, the threshold value of clinically important change in pain associated with a 60% chance of patients reporting they are “much better” or “a little better” was – 0.7 cm. This was similar to the mean pain change of – 0.82 cm associated with “a little better” quality of life. Patients who judged their quality of Table 2. General linear model (multiple linear regression) in which change in quality of life is the dependent variable Independent variable Pain change (10-cm scale) Global disease severity (scale 0–3) Age (years) Sex Health-related quality of life (scale 0–3) Quality of life (scale 0–3) Regression (␤) t coefficient statistic P 0.116 0.272 7.77 ⱕ0.0001 3.62 ⱕ0.0003 0.029 0.102 ⫺0.173 3.17 ⱕ0.0016 2.35 ⱕ0.0192 ⫺2.19 ⱕ0.0290 ⫺0.228 ⫺2.71 ⱕ0.0069 life as “the same” showed little or no difference in pain scores—a result that suggests that no change in pain is associated with none in quality of life. Furthermore, lower pain scores were associated with improved quality of life and increased pain scores with worsening quality. Although pain is an important factor, it is only one aspect of quality of life. This likely accounts for the range of results seen in our study, in which change in pain accounted for about 16% of the variability in quality of life. Other factors, such as global disease severity, also had a significant influence on change in quality of life. Previous studies established minimally important differences with other disorders, such as asthma and chronic lung disease (17,18). Although these studies used a 7-point scale as opposed to a VAS, they were able to consistently demonstrate a minimal important difference in diseasespecific quality of life assessments. Using a quality of life questionnaire for asthma and a global rating of change, they found that the minimal important difference of quality of life score per item was close to 0.5. These authors suggest that this minimal difference may be generalized to many areas of health-related quality of life assessments (17,19). If such a broad applicability can be demonstrated, these results could help clinicians interpret quality of life assessments, which are used in many areas of medicine. As a result, the same standards may eventually be used to assess the magnitude of changes across a wide variety of instruments. Determining the minimal important difference may also help clinicians judge the benefit when comparing 2 treatments, calculating sample size, and assessing therapeutic interventions (20,21). Interpreting values of minimal clinically important differences is subject to several limitations. First, the term “a little better” used in our scale may not actually represent the smallest difference important to that patient’s quality of life (19); a patient may report change in quality of life in smaller increments than can be measured with conventional scales. Second, most studies assessing the minimal important difference depend on patients’ judgments when they are stable and cooperative. However, results may be influenced by other factors that may alter interpretation of Minimal Difference in Pain Needed to Change Quality of Life quality of life (18). When patients are unstable or more seriously ill, they may be more heavily influenced by factors such as mood. A third limitation may be that in our study we used quality of life as a standard measure. However, smaller changes in pain may still be important to patients but may not affect quality of life. Another, more specific factor might have been more sensitive to smaller changes in pain. A point of statistical methodology in our study should be addressed. In our secondary assessment, to increase the precision of estimation of the mean change in pain scores for each quality of life category, we used multiple measurements from some subjects. In other words, we used data from not only the first and second visits but in some cases also from subsequent visits. Because the perception of pain and quality of life are highly personal, it is conceivable that some people would consistently associate only large changes in pain scores with a small change in quality of life and vice versa. Theoretically this might bias our results, especially if these patients were repeatedly used in the data sets. However, in our primary analysis we used data from only the first 2 visits, i.e., each patient was used only once. The results were very similar to that of the subsequent larger data set. For example, change in quality of life to “a little better” showed a mean change in pain of – 0.89 cm as compared with – 0.82 cm. Thus, we felt that there was adequate independence of the responses. A few other limitations and assumptions of our study in particular need to be addressed. First, the study was retrospective. However, our patients were selected from a population in a clinic that kept careful records and ensured that questionnaires were completed in full. Second, we must be aware of errors in causal inference, assuming that change in pain score led to change in quality of life. Changes in quality of life may affect pain perception itself. Thus, we cannot assume a cause-and-effect relationship and can only report an association between pain change and quality of life. Third, there may be confounding factors that may affect both pain and quality of life independently. However, we were unable to find other factors that statistically confounded the relationship between pain and quality of life. In summary, our study showed that pain is an important factor associated with alteration in quality of life in pediatric rheumatology patients. Reduced pain scores over time were associated with improved quality of life, and increased pain scores with worsening quality of life. Furthermore, we were able to define the minimal clinically important difference associated with significant improvement in quality of life in this population. Establishment of this minimal important change in the perception of pain may enhance our understanding of its impact in rheumatic disease and help define a standard for studies dealing with the treatment of pain. Our results suggest that future studies of the assessment and treatment of pain in this population should aim for a minimum reduction in pain score of 0.82 cm on a 10-cm VAS. ACKNOWLEDGMENTS We would like to thank the Department of Rheumatology at The Hospital for Sick Children in Toronto for their 505 cooperation and resources. Special thanks to Mr. Frank Quinlan of the Department of Medical Editing at The Hospital for Sick Children for his time, effort, and invaluable suggestions. REFERENCES 1. Murray KJ, Passo MH. Functional measures in children with rheumatic diseases. Pediatr Clin North Am 1995;42:1127–53. 2. Singh G, Athreya BH, Fries JF, Goldsmith DP. Measurement of health status in children with juvenile rheumatoid arthritis. Arthritis Rheum 1994;37:1761–9. 3. Lovell DJ, Walco GA. Pain associated with juvenile rheumatoid arthritis. Pediatr Clin North Am 1989;36:1015–26. 4. Varni JW, Thompson KL, Hanson V. The Varni/Thompson pediatric pain questionnaire. I. Chronic musculoskeletal pain in juvenile rheumatoid arthritis. Pain 1987;28:27–38. 5. Varni JW, Bernstein BH. Evaluation and management of pain in children with rheumatic diseases. Rheum Dis Clin North Am 1991;17:985–99. 6. Gragg RA, Rapoff MA, Danovsky MB, Lindsley CB, Varni JW, Waldron SA, et al. Assessing chronic musculoskeletal pain associated with rheumatic disease: further validation of the pediatric pain questionnaire. J Pediatr Psychol 1996;21:237– 50. 7. Walco GA, Varni JW, Ilowite NT. Cognitive-behavioral pain management in children with juvenile rheumatoid arthritis. Pediatrics 1992;89:1075–9. 8. Ward MM. Clinical measures in rheumatoid arthritis: which are most useful in assessing patients? J Rheumatol 1994;21: 17–27. 9. Gill TM, Feinstein AR. A critical appraisal of the quality of quality-of-life measurements. JAMA 1994;272:619 –26. 10. Feldman BM, Grundland B, McCullough L, Wright V. Distinction of quality of life, health-related quality of life, and health status in children referred for rheumatologic care. J Rheumatol 2000;27:226 –33. 11. Abu-Saad HH, Uiterwuk M. Pain in children with juvenile rheumatoid arthritis: a descriptive study. Pediatr Res 1995; 38:194 –7. 12. Jaworski TM. Juvenile rheumatoid arthritis: pain-related and psychosocial aspects and their relevance for assessment and treatment. Arthritis Care Res 1993;6:187–96. 13. Varni JW, Thompson-Wilcox K, Hanson V, Brik R. Chronic musculoskeletal pain and functional status in juvenile rheumatoid arthritis, an empirical model. Pain 1988;32:1–7. 14. Skevington SM. Investigating the relationship between pain and discomfort and quality of life, using the WHOQOL. Pain 1998;76:395– 406. 15. Fries JF, Spitz PW, Young DY. The dimension of health outcomes: the health assessment questionnaire, disability and pain scales. J Rheumatol 1982;9:789 –93. 16. Jaeschke R, Singer J, Guyatt GH. Measurement of health status: ascertaining the minimal clinically important difference. Control Clin Trials 1989;10:407–15. 17. Juniper EF, Guyatt GH, Willan A, Griffith LE. Determining a minimal important change in disease-specific quality of life questionnaire. J Clin Epidemiol 1994;47:81–7. 18. Guyatt GH, Berman LB, Townsend M, Pugsley SO, Chambers LW. A measure of quality of life for clinical trials in chronic lung disease. Thorax 1987;42:773– 8. 19. Redelmeier DA, Guyatt GH, Goldstein RS. Assessing the minimal important difference in symptoms: a comparison of two techniques. J Clin Epidemiol 1996;49:1215–9. 20. Wright JG. The minimal important difference: who’s to say what is important? J Clin Epidemiol 1996;49:1221–2. 21. Guyatt GH, Juniper EF, Walter SD, Griffith LE, Goldstein RS. Interpreting treatment effects in randomised trials. BMJ 1998; 316:690 –3.