Incomplete and Unreported: Failure to Report Results of Incomplete Studies Results in Publication Bias

Abstract

Background: Clinical research that is incomplete and unreported can lead to bias in health care decisions. Selective reporting of outcomes in clinical research overestimates the benefits and safety of treatments and threatens the integrity of systematic reviews of health care interventions. The Common Rule does not require that results be made available from clinical studies to research subjects and the scientific community. Methods: We analyzed all closed studies (i.e., no longer recruiting) registered in the online database ClinicalTrials.gov as of May 2010 to identify the number, length, and characteristics of complete and incomplete (terminated, suspended, or withdrawn) studies. Random samples of terminated studies, which were defined as studies in which the recruitment or enrollment of participants has been halted prematurely and will not resume, were analyzed to identify the reasons for termination and availability of the results in ClinicalTrials.gov or in published articles. Results: We identified 3,554 (6.2%) terminated studies among 57,233 closed studies. Reasons for termination were unknown in 34% and safety-related in 11% of terminated studies. Industry-funded studies omitted termination reasons less often than National Institutes of Health (NIH)-funded studies (OR 0.3, 95%CI: 0.1, 0.9). Drug studies omitted termination reasons more often than all other intervention studies. (OR 2; 95% CI: 1.1, 3.7) and also cited safety as the termination reason (OR 8.6; 95%CI: 1.9, 36.8) more often than other intervention studies. Terminated studies lasted on average 2.2 ± 1.9 years, and results were available for 5% in ClinicalTrials.gov and for 8% in peer-reviewed journals. Conclusions: Existing clinical research policies should mandate registration of all clinical trials in ClinicalTrials.gov as a condition of studies approval by institutional review boards. Mandatory posting of the results in ClinicalTrials.gov for all terminated studies, along with the exact reasons for termination irrespective of study funding and market or publication status, would enhance comprehensive evidence analysis by clinicians and policymakers, as well as public trust in clinical research.

Keywords:

• clinical trials as topic
• National Institutes of Health (U.S.)
• National Library of Medicine (U.S.)
• publishing
• registries
• research peer review
• research support as topic

Notes

See WHO Trial Registration Data Set (Version 1.2.1) at http://www.who.int/ictrp/network/trds/en for a list of the full criteria.