Outcome measures used in arthroplasty trials : systematic review of the 2008 and 2013 literature

Background/objective: Previously published literature assessing the reporting of outcome measures used in joint replacement randomised controlled trials (RCTs) has revealed disappointing results. It remains unknown as to whether international initiatives have led to any improvement in the quality of reporting and/or a reduction in the heterogeneity of outcome measures used. Our objective was to systematically assess and compare primary outcome measures and risk of bias of joint replacement RCTs published in 2008 and 2013. Methods: We searched MEDLINE, EMBASE and CENTRAL for RCTs investigating adult patients undergoing joint replacement surgery. Two authors independently identified eligible trials, extracted data and assessed risk of bias using the Cochrane tool. Results: Seventy RCTs (30 in 2008; 40 in 2013) met the eligibility criteria. There was no significant difference in the number of trials judged to be at low overall risk of bias (N=6, 20%) in 2008 compared with six (15%) in 2013 (χ2 =0.302, P=0.75). Significantly more trials published in 2008 did not specify a primary outcome measure (N=25, 83%) compared with 18 (45%) trials in 2013, χ2 = 10.6316, P=0.001). When specified, there was significant heterogeneity in the measures used to assess primary outcomes. Conclusion: While less than a quarter of trials published in both 2008 and 2013 were judged to be at low overall risk of bias, significantly more trials published in 2013 specified a primary outcome. Although this might represent a temporal trend towards improvement, the overall frequency of primary outcome reporting and the wide heterogeneity in primary outcomes reported remain suboptimal