Research Letter
February 17, 2025
Heterogeneity of Primary Outcomes in Large Atherosclerotic Cardiovascular Disease Trials Published in Prominent Medical Journals
Daniel Shepshelovich, Dafna Yahav, Danielle R. Rome, et al
JAMA Intern Med. Published online February 17, 2025. doi:10.1001/jamainternmed.2024.7694
The extent of heterogeneity in end points reported by randomized clinical trials (RCTs) of atherosclerotic cardiovascular disease (ASCVD) is unknown. We aimed to describe the heterogeneity of primary end points reported in contemporary ASCVD RCT publications and to compare them with the original end points at RCT initiation.
Methods
This cohort study was considered nonhuman participants research by the Columbia University institutional review board and did not require informed consent. The study followed the STROBE reporting guideline.
All ASCVD RCTs published between January 1, 2019, and December 31, 2023, in the New England Journal of Medicine, The Lancet, JAMA, the European Heart Journal, Circulation, and the Journal of the American College of Cardiology were reviewed. RCTs of revascularizations and acute care, which often include different end points, were excluded. Secondary analyses, extended follow-up publications, and reports of surrogate end points were also excluded. Published end points were compared with the original end points at RCT initiation, extracted from ClinicalTrials.gov, and the effect of end point modifications on published results was assessed. Because most RCTs used time-to-event analysis and the original data were generally unavailable, an accurate comparison of trial results using different composite end points was not feasible. Rather, the Fisher exact test was used to compare the cumulative incidence of different composite end points reported in both study groups. Statistical significance was defined as 2-sided P < .05. Analysis was performed with SPSS, version 28 (IBM SPSS).
Results
This study included 50 RCTs (Table). Nearly all RCTs (49 [98%]) used composite end points, the most common (14 [28%]) being the 3-point major adverse cardiovascular events (MACE), a composite end point of nonfatal myocardial infarction, nonfatal stroke, and cardiovascular death. The remaining 36 RCTs (72%) included 29 different end point combinations (Figure). Of 22 positive RCTs reporting end points other than 3-point MACE, the results of 2 trials1,2were driven by end point components not included in the 3-point MACE.
The published and original primary end points were different in 23 RCTs (46%). In 15 trials (30%), components of the original composite end points were edited after the initiation of patient accrual. These changes were reported by 11 publications (73%); the most common rationale (6 [40%]) was a need to bolster statistical power. Eleven trials (73%) were estimated to report similar results using the original primary outcome; 4 (27%) did not report adequate data for estimation. In 8 trials (16%), the initial ClinicalTrials.gov entry reported an undefined original end point (eg, cardiovascular disease, MACE), edited later to various detailed definitions.
Discussion
Our findings suggest that end point heterogeneity in contemporary ASCVD RCTs published in prominent journals might influence reported results3 and may complicate comparing the results of different studies of similar interventions or collecting data for meta-analyses.4 Finally, the reported heterogeneity might lead to lower barriers to expanding trial end points rather than increasing sample size or follow-up time when the event rate is lower than expected.
End point modifications between study initiation and publication compromise the trial’s scientific validity.5 This is underlined by RCTs that reported positive results driven by end point components not included in the 3-point MACE.
This study has several limitations. First, our results might not extend to trials not included in our sample; however, we selected top-tier journals and included major trials testing primary and secondary ASCVD prevention interventions. Second, the estimation of results using alternative composite end points might be inaccurate, given the lack of access to the primary data. Third, exploration of predictors of end point change was not feasible owing to the small number of trials.
In conclusion, the findings of this cohort study suggest that large ASCVD RCTs published in prominent journals use heterogeneous end points that are often changed from trial inception to publication. Regulators and journal editors should generally require that ASCVD RCTs use uniform end points (eg, 3-point MACE) and report the original protocol end points to generate higher-quality data and improve informed decision-making for patients with ASCVD.
