Evolution of Age and Female Representation in the Most-Cited Randomized Controlled Trials of Cardiology of the Last 20 Years

BACKGROUND: Older adults and women have historically been underrepresented in randomized controlled trials of cardiology. Recent temporal evolution and factors influencing representation are incompletely investigated. We aimed to contrast age and female representation in the most influential randomized controlled trials in cardiology of the last 20 years to population prevalence and to assess the study factors affecting representation. METHODS AND RESULTS: Using Web of Science, we selected the 25 most-cited cardiology articles each year between 1996 and 2015, and extracted mean age, percentage of women, funding source, sample size, disease condition, intervention type, and exclusion criteria. The outcomes were the evolution of the mean age and the percentage of women over time. Protocol design elements and year of publication were assessed as predictors of outcomes in multivariable regressions. A total of 500 studies were analyzed, where the mean age was 62.6 +/- 7.4 years and the median percentage of women was 28.6% (22.2-40.5). Compared with population prevalence derived from National Health and Nutrition Examination Survey 2015-2016, gaps in representation were apparent and more pronounced for coronary artery disease (-5.0 years; -27.2% women) and heart failure (-6.0 years; -25.4% women). The mean age (0.15 year per year; 95% confidence interval, 0.04-0.26) and percentage of women (+0.29% per year; 95% confidence interval, 0.07-0.50), slightly but significantly increased over time. Private funding, small sample size, and exclusions pertaining to maximal age, atrial fibrillation, and diabetes mellitus were associated with a decreased mean age in multivariable linear regressions. Age and life expectancy exclusions were associated with lower female percentage. CONCLUSIONS: Although age and female representation increased over time, the modest trends are unlikely to resolve the persistently wide gaps with actual populational prevalence, especially for coronary artery disease and HF. Representation is modulated by the cardiovascular condition studied and some modifiable protocol elements.