Undisclosed changes in outcomes in randomized controlled trials: an observational study.

PURPOSE

We wanted to investigate the frequency of undisclosed changes in the outcomes of randomized controlled trials (RCTs) between trial registration and publication.

METHODS

Using a retrospective, nonrandom, cross-sectional study design, we investigated RCTs published in consecutive issues of 5 major medical journals during a 6-month period and their associated trials registry entries. Articles were excluded if they did not have an available trial registry entry, did not have analyzable outcomes, or were secondary publications. The primary outcome was the proportion of publications in which the primary outcome of the trial was, without disclosure, changed between that recorded in the trial registry and that reported in the final publication. The secondary outcome was the proportion of publications in which the secondary outcome was changed without disclosure.

RESULTS

We reviewed 158 reports of RCTs and included 110 in the analysis. In 34 (31%), a primary outcome had been changed, and in 77 (70%), a secondary outcome had been changed.

CONCLUSIONS

There are substantial and important undisclosed changes made to the outcomes of published RCTs between trial registration and publication. This finding has important implications for the interpretation of trial results. Disclosure and discussion of changes would improve transparency in the performance and reporting of trials.