Twenty percent of secondary publications of randomized controlled trials of drugs did not provide new results relative to the primary publication

Abstract

Objective: The objective of the study was to estimate the proportion of secondary publications of randomized controlled trials (RCTs) that provide new results relative to the primary publication.

Study design and setting: We searched for RCTs published in 2014 in the five medical journals with the highest impact factors. Secondary publications for each primary publication were then identified by their registration number. The main outcome measure was the proportion of secondary publications providing results already reported in the primary publication and/or nonprespecified analyses and/or a meta-analysis pooling results of studies not identified by systematic review.

Results: A total of 144 primary publications were identified; 94 (65%) had at least one secondary publication within 30 months after a primary publication. Of the secondary publications, 20% reported only results present in the primary publication, and 35% reported results not prespecified or pooled analyses not based on a systematic review. Factors associated with having at least one secondary publication were a large number of randomized trial participants (odds ratio [95% confidence interval]: 3.2 [1.1-9.3] for trials with >1,000 vs. ≤500 participants), investigating a biologic product (4.8 [1.4‒16.3] vs. a nonbiologic product) and cardiologic field vs. other fields (7.6 [1.46-39.8]).

Conclusion: Most drug RCTs with results published in high-impact-factor journals had secondary publications. More than half of these secondary publications provided results already reported in the primary publication or results of nonprespecified analyses.

Keywords: Medical writing; Multiplicity; Outcome; Randomized controlled trial; Secondary findings; Secondary publications.