Dual-Outcome Intention-to-Treat Analyses in the Women’s Health Initiative Randomized Controlled Hormone Therapy Trials

Abstract

Abstract Dual-outcome intention-to-treat hazard rate analyses have potential to complement single-outcome analyses for the evaluation of treatments or exposures in relation to multivariate time-to-response outcomes. Here we consider pairs formed from important clinical outcomes to obtain further insight into influences of menopausal hormone therapy on chronic disease. As part of the Women’s Health Initiative, randomized, placebo-controlled hormone therapy trials of conjugated equine estrogens (CEE) among posthysterectomy participants and of these same estrogens plus medroxyprogesterone acetate (MPA) among participants with an intact uterus were carried out at 40 US clinical centers (1993–2016). These data provide the context for analyses covering the trial intervention periods and a nearly 20-year (median) cumulative duration of follow-up. The rates of multiple outcome pairs were significantly influenced by hormone therapy, especially over cumulative follow-up, providing potential clinical and mechanistic insights. For example, among women randomized to either regimen, hazard ratios for pairs defined by fracture during intervention followed by death from any cause were reduced and hazard ratios for pairs defined by gallbladder disease followed by death were increased, though these findings may primarily reflect single-outcome associations. In comparison, hazard ratios for diabetes followed by death were reduced with CEE but not with CEE + MPA, and those for hypertension followed by death were increased with CEE + MPA but not with CEE.