Comparison of registered and published primary outcomes in randomized controlled trials

Sylvain Mathieu ¹, Isabelle Boutron , David Moher

⁰INSERM, U738, Assistance Publique des Hôpitaux de Paris, Hôpital Bichat-Claude Bernard, Département d'Epidémiologie Biostatistique et Recherche Clinique, Paris, France.

Jama +

|

September 3, 2009

View abstract on PubMed

Summary

Selective outcome reporting bias is prevalent in clinical trials. Many registered trials show discrepancies between planned and published outcomes, favoring significant results.

Area Of Science

Clinical Research
Medical Journal Publishing
Biostatistics

Background

International guidelines mandate trial registration before participant enrollment.
This policy aims to enhance transparency and prevent selective reporting of trial findings.

Purpose Of The Study

Assess publication rates of registered trials in high-impact journals.
Compare primary outcomes in registries versus publications.
Identify bias favoring significant outcomes in reporting.

Main Methods

Searched MEDLINE/PubMed for randomized controlled trials (RCTs) in cardiology, rheumatology, and gastroenterology indexed in 2008.
Included trials from top 10 general and specialty medical journals.
Extracted trial registration data using a standardized form.

Main Results

Only 45.5% of 323 trials were adequately registered (timely, clear primary outcome).
Significant discrepancies (31%) between registered and published outcomes were found in adequately registered trials.
In cases where discrepancies could be assessed, statistically significant results were favored 82.6% of the time.

Conclusions

Selective outcome reporting is a prevalent issue in clinical trial publications.
Discrepancies suggest bias in how trial results are presented post-registration.

Related Concept Videos