Abstract ED01-02: Methodologic considerations for large-scale prevention trials: Lessons from the Women's Health Initiative

Abstract

Abstract The Women's Health Initiative was one of the premiere primary prevention trials of its era. The core component was a partial factorial randomize trial designed to test three chronic disease prevention hypotheses in over 68,000 postmenopausal women: hormone therapy (HT) for the prevention of coronary heart disease; a low fat diet for the prevention of breast cancer and colorectal cancer, and calcium/vitamin D supplements for hip fracture risk reduction. This large diverse program provides useful lessons for many of the issues that arise in the design, implementation, analysis and interpretation of large-scale public health trials. The greatest public health impact of WHI has come from the two HT trials. The estrogen plus progestin trial in women with an intact uterus was stopped early, in 2002, based on an increased risk in breast cancer and overall health risks exceeding benefits. In 2004, the estrogen alone trial in women with prior hysterectomy, was stopped for an increased risk in stroke. These results were immediately translated into a public health message that had a rapid impact on HT use, followed shortly by a reduction in national breast cancer incidence rates. The HT trial results were considered highly controversial, however, because of their failure to show benefit for coronary heart disease, a hypothesis supported by numerous observational studies, animal studies, and randomized trials with intermediate outcomes. These contradictory findings fueled a persistent debate about the interpretation and integrity of the trial. Many of the purported flaws in the trial were easily refuted, owing to the strength of the trial's design, implementation and analyses. The design required excellent power for conservative projections of intervention effects. Key implementation factors included multicenter recruitment directly from the general population of postmenopausal women using broadly inclusive eligibility criteria, well-established, commonly used interventions, double-blinded trial conduct, ongoing support for adherence to interventions, broad-spectrum but prioritized outcomes ascertainment processes, standard of care safety monitoring including requirements for regular mammography, and a written trial monitoring plan to address the multiple testing for the numerous trial outcomes and interim analyses. To address these discrepancies between observational and randomized trial data directly, joint analyses of the WHI Observational Study (OS) and the HT trials were conducted. While the usual analysis of HT effects in the OS led to inferences similar to the preceding observational studies, analyses that carefully controlled for multiple confounders and took into account time from initiation of therapy and for breast cancer, time between menopause and first use of HT, yielded good alignment between the WHI HT trial and observational studies. These analyses yielded opposite trends with time for CHD and breast cancer, that is early initiators tended to have lower CHD risk but greater breast cancer risk than later initiators, providing some support for the so-called “timing hypothesis” yet confirming the limited utility of these agents for chronic disease prevention. Citation Format: Garnet L. Anderson. Methodologic considerations for large-scale prevention trials: Lessons from the Women's Health Initiative. [abstract]. In: Proceedings of the Eleventh Annual AACR International Conference on Frontiers in Cancer Prevention Research; 2012 Oct 16-19; Anaheim, CA. Philadelphia (PA): AACR; Cancer Prev Res 2012;5(11 Suppl):Abstract nr ED01-02.