Study design, result posting, and publication of late-stage cardiovascular trials.

Abstract

Pre-registration of study protocols in accessible databases is required for publication of study results in high-impact medical journals. Nonetheless, data on characteristics of clinical trials registered in these databases and their outcome, in terms of result reporting and publication are limited. We searched for interventional, late-phase cardiovascular disease (CVD) studies in adults registered in Clinicaltrials.gov. first posted after 1 January 2013 and completed up to 31 December 2018. Data on study design, result reporting, and publication were collected, and potential associations with a pre-defined set of explanatory factors were examined. In total, 250 CVD trials were included in the analysis. Of these, 193 (77.2%) were randomized studies, 99 (39.6%) open label designs, and 126 (50.4%) had industry as main sponsor. One hundred and seventy-nine trials (71.6%) evaluated the effect of drugs and 27 (10.8%) evaluated devices. The most common primary outcomes were non-clinical endpoints (76.0%), with only 17% of studies evaluating clinical endpoints. Industry-funded trials focused on patent-protected drugs and devices more often than non-industry-funded trials (72.0% vs. 30.6%, P < 0.001 and 55.0% vs. 26.3%, P = 0.033, respectively). Sixty-three studies (25.2%) had results posted on clinicaltrials.gov, and 116 (46.4%) had results published in the scientific literature. In multivariate analysis, industry sponsorship was statistically significantly associated with results posting [odds ratio (OR): 3.38; 95% confidence interval (CI): 1.56-7.30, P = 0.002] and publication (OR: 0.41; 95% CI: 0.23-0.75, P = 0.004). Among late-stage cardiovascular trials only one-fourth had results posted on clinicaltrials.gov and <50% had results published. Industry sponsors were more likely to invest in research on patent-protected drugs and devices than were non-industry sponsors. Industry-sponsored studies were more likely to have their results posted, but less likely to have their results published in the scientific literature.