Article type
Year
Abstract
Background: Investigators performing systematic reviews and meta-analyses are typically faced with data derived from different assessment tools to measure a common concept such as pain. A common method is to combine data regardless of instrument using standardised mean differences (SMD). However, it is unclear whether the estimated SMD and statistical heterogeneity will depend on the choice of instrument used.
Objectives: To determine whether the two most frequently used instruments to assess knee or hip pain in osteoarthritis trials, the WOMAC Pain subscale (WOMAC) and Pain Overall measured on a visual analogue scale (VAS), result in different pooled estimates of treatment effects and statistical heterogeneity between trials.
Methods: Meta-epidemiological study of meta-analyses of large-scale osteoarthritis trials comparing active treatment with placebo, sham, or non-intervention, reporting at least one follow-up for both, WOMAC and VAS, with at least 100 patients per group. We pooled SMDs for each instrument using an inverse-variance random-effects model within each meta-analysis and subsequently combined pooled estimates of treatment effect and statistical heterogeneity across meta-analyses using a random-effects model. Estimates were compared between instruments using a paired sign-rank test. Negative SMDs indicated a benefit of experimental interventions as compared with control.
Results: We identified 5 meta-analyses including 24 large-scale trials with 13400 patients that reported on both WOMAC and VAS. The average SMD was 0.22 (95% CI 0.34 to 0.09) according to WOMAC and 0.32 (95% CI 0.46 to 0.18) according to VAS (pfor difference = 0.038, see Table1). Corresponding I2 estimates were 53% (95% CI 0% to 80.7%) and 85% (95% CI 66.6% to 93.4%), respectively (p = 0.043).
Conclusions: Differences in estimated SMDs between WOMAC and VAS were small on average, but WOMAC resulted in considerably lower heterogeneity between trials. Therefore, WOMAC should be given precedence over VAS when extracting data for meta-analysis of osteoarthritis trials using SMDs.
Objectives: To determine whether the two most frequently used instruments to assess knee or hip pain in osteoarthritis trials, the WOMAC Pain subscale (WOMAC) and Pain Overall measured on a visual analogue scale (VAS), result in different pooled estimates of treatment effects and statistical heterogeneity between trials.
Methods: Meta-epidemiological study of meta-analyses of large-scale osteoarthritis trials comparing active treatment with placebo, sham, or non-intervention, reporting at least one follow-up for both, WOMAC and VAS, with at least 100 patients per group. We pooled SMDs for each instrument using an inverse-variance random-effects model within each meta-analysis and subsequently combined pooled estimates of treatment effect and statistical heterogeneity across meta-analyses using a random-effects model. Estimates were compared between instruments using a paired sign-rank test. Negative SMDs indicated a benefit of experimental interventions as compared with control.
Results: We identified 5 meta-analyses including 24 large-scale trials with 13400 patients that reported on both WOMAC and VAS. The average SMD was 0.22 (95% CI 0.34 to 0.09) according to WOMAC and 0.32 (95% CI 0.46 to 0.18) according to VAS (pfor difference = 0.038, see Table1). Corresponding I2 estimates were 53% (95% CI 0% to 80.7%) and 85% (95% CI 66.6% to 93.4%), respectively (p = 0.043).
Conclusions: Differences in estimated SMDs between WOMAC and VAS were small on average, but WOMAC resulted in considerably lower heterogeneity between trials. Therefore, WOMAC should be given precedence over VAS when extracting data for meta-analysis of osteoarthritis trials using SMDs.