Context: Noninferiority and equivalence trials aim to show that the experimental treatment is not clinically worse than (noninferior) or clinically similar to (equivalent) a control active treatment. These study objectives imply particular planning and analysis.
Objective: To assess the methodologic quality of reports of randomized controlled trials of noninferiority and equivalence.
Design: We searched MEDLINE and the Cochrane Central Register of Controlled Trials for reports of randomized controlled trials of noninferiority and equivalence hypotheses published between January 1, 2003, and December 31, 2004.
Main outcome measures: Data extracted by use of a standardized form involved assessment of choice of noninferiority or equivalence margins, sample size calculation, sets of patients analyzed, method of statistical testing and reporting results, and conclusions.
Results: A total of 162 reports were included in the analysis (116 reports of noninferiority and 46 of equivalence). The margin defining noninferiority or equivalence was described in most reports (156 [96.3%]), with justification of the margin in only 33 (20.4%). Almost one quarter of the reports (35 [21.6%]) did not describe a sample size calculation, and an additional 11 (6.8%) did not take into account a prespecified noninferiority or equivalence margin. Less than half of the reports (69 [42.6%]) described both an intent-to-treat (ITT; all randomized patients are included in the analysis) or modified ITT (patients who never received treatment are excluded) and per-protocol (patients who did not complete the treatment are excluded) analysis, and only about half of those (39 [56.5%]) described both types of results. Results were displayed with confidence intervals in 136 reports (84.0%). Only 33 articles (20.3%) fulfilled reporting requirements specific to noninferiority and equivalence trials, 4 of them (12.1%) with misleading conclusions.
Conclusions: Reporting of noninferiority and equivalence trials has important deficiencies: absence of noninferiority or equivalence margin, only an ITT (or a per-protocol) analysis performed, and results not adequately reported. Moreover, even for articles fulfilling these requirements, conclusions are sometimes misleading.