Interim analysis in clinical trials: A meta-epidemiological study

Abstract

Objective: The ethical responsibility of the researcher requires that the safety of participants be monitored throughout the study. Thus, a data monitoring committee is required, whose main function is the interim analysis, which monitors variables such as dramatic benefits, adverse effects, mortality and futility that lead to the early termination of a study. The objective of this study is to determine the most frequently used statistical methods in interim analysis in randomized clinical trials published during 2016 in a high impact factor medical journal.

Methodology: Meta-epidemiological descriptive study, consisting of randomized clinical trials published in The New England Journal of Medicine from January 7 to November 10 of 2016.

Results: 104 articles were analyzed, where the mean sample was 5,531 ± 762 patients and and average follow-up time was 31 ± 45 months. Of the studies evaluated, 88 (84.61%) performed interim analysis, while 16 (15.38%) did not. Regarding statistical methods used, O’Brien-Fleming was used in 30 (34.09%) studies, Haybittle-Peto in 13 (14.77%), other methods in two (2.27%), and the method was not specified in 43 (48.86%).

Conclusions: most of the randomized clinical trials reviewed had interim analysis and described it in their protocol. Approximately half of the studies did not specify the statistical method; however, the most commonly described method was O’Brien-Fleming, followed by Haybittle-Peto; no study used the Pocock method.