Objective: To identify sources of heterogeneity (statistical, methodological, and clinical) in studies evaluating non-steroidal anti-inflammatory drugs (NSAIDs) vs acetaminophen in patients with knee and hip osteoarthritis (OA) to elucidate variations in outcomes. Method: A database search (1966 to January 2010) was made for (randomized) controlled trials ((R)CTs) comparing NSAIDs vs acetaminophen in knee and hip OA. Extracted data included baseline demographic/clinical characteristics, outcomes at follow-up, and characteristics of study design. Heterogeneity was examined with subgroup analyses by exploring changes in effect size and with I2of Higgins. Pain measures were expressed as standardized mean differences. Results: 15 RCTs, including 21 comparisons of NSAIDs and acetaminophen were included. Statistical heterogeneity was absent (Cochran's Q-test=14.11; I2=0; P=0.78). Moderate clinical heterogeneity was found for comparisons which included both hip and knee OA vs knee OA only (I2=51; P=0.09). NSAIDs seemed slightly more effective than acetaminophen if more patients with hip OA were included. However, the pooled effect sizes of comparisons with knee OA vs both knee and hip OA are equal. Low clinical heterogeneity was found for comparisons with low dosage of acetaminophen, normal dosage of NSAIDs, and moderate pain intensity at baseline. Low methodological heterogeneity was found for comparisons with a short duration. Conclusion: Future trials should present the results of hip and knee OA separately, as moderate clinical heterogeneity was found. There might be differences in effectiveness of NSAIDs vs acetaminophen in patients with hip vs knee OA. No significant methodological and statistical heterogeneity was found in studies evaluating NSAIDs vs acetaminophen.

, , , ,
doi.org/10.1016/j.joca.2011.04.013, hdl.handle.net/1765/26632
Osteoarthritis and Cartilage
Erasmus MC: University Medical Center Rotterdam