Influence of flare design on symptomatic efficacy of non-steroidal anti-inflammatory drugs in osteoarthritis: a meta-analysis of randomized placebo-controlled trials

Summary Introduction Non-steroidal anti-inflammatory drugs' (NSAIDs) symptomatic efficacy in osteoarthritis (OA) is often assessed in trials with a "flare design", i.e., including only patients with an increase in their pain after stopping their usual treatment (NSAIDs or analgesic). Objective To evaluate the influence of the "flare design" on NSAIDs apparent symptomatic efficacy in OA. Design Search strategy : a systematic literature research was performed in Medline, EMBASE and The Cochrane Register up to March 2009. All randomized controlled trials comparing NSAIDs vs placebo symptomatic efficacy in hip, knee, or digital OA were included. Data collection and analysis : efficacy was evaluated on pain (visual analog scale), and on function (Western Ontario and McMaster Universities OA index or Lequesne index). The magnitude of the treatment effect was evaluated by calculating Cohen's effect size (ES). Meta-analysis of ES according to flare design yes/no was performed. Results Among the 343 identified studies, 33 (20,915 patients) were included: 27 (18,667 patients) vs 6 (2248 patients) respectively in the group with vs without "flare design". Populations were comparable in each group. ESs were, for pain, −0.66 (95% confidence interval, −0.71 to −0.61), vs −0.45 (−0.54 to −0.36) in the flare design vs "no flare design" group, and for function, −0.50 (−0.55 to −0.44) vs −0.25 (−0.36 to −0.14) respectively. Conclusion Our study suggests that the flare design used in clinical trials evaluating NSAIDs results in a treatment effect of higher magnitude. These results should be considered when designing a trial and/or interpreting the results of a trial.