Estimating the Effect of Low-Dose Aspirin on Fetal Loss in a Trial with Non-Compliance

项目总结/文摘

Project Summary/Abstract Miscarriage is the most common pregnancy outcome, and can have lasting psychological implications that persist even after the birth of a healthy child. The few treatment options that exist are costly, potentially invasive, undermine quality of life, or are associated with adverse sequelae. Aspirin is a well-tolerated over- the-counter drug, but little is known about its effects on promoting live birth in women at high risk of recurrent miscarriage. The Effects of Aspirin in Gestation and Reproduction (EAGeR) study was a randomized trial of the effect of low-dose aspirin on preventing miscarriage and promoting live birth. While the study found a small effect of aspirin on live birth, interpreting these results is complicated by several challenges that routinely affect randomized trials. These include failure of study participants to comply with the assigned treatment, mis- measurement of exposure to aspirin, and studying aspirin effects in a sample that is not representative of the population of interest. Each of these creates the potential for incorrect inferences regarding treatment effects in more general populations. We will reanalyze data from this prospective cohort study using tools that properly account for each of these complications. Specifically, we will (Aim 1) estimate the effect of taking daily low dose aspirin on miscarriage and successful live birth; (Aim 2) correct these effects estimates for potential measurement error; and (Aim 3) generalize these effect estimates to clinically representative populations. Successful completion of this work will fill critical gaps in knowledge about the benefits of aspirin in preventing miscarriage and promoting live birth. Our cost-effective, efficient, and methodologically innovative project will help guide clinicians and public health practitioners in making informed decisions about this simple and inexpensive preventive strategy. Furthermore, by developing and disseminating our project's methods and software programs, we will facilitate the widespread use of our state-of-the-art analytic techniques to account for common problems in the analysis of randomized trials. Consequently, the impact of this project will be wide, and will enable researchers in any substantive area to fully address critical issues that plague nearly all trials in the biomedical and social sciences.