Abstract WP232: The Importance Of Time-varying Confounding And Selection Bias In Stroke Research: The Example Of Anticoagulation And Health Related Quality Of Life

Abstract

Background: Guidelines support anticoagulation for primary stroke prevention in those with atrial fibrillation and additional risk factors. Primary endpoints in longitudinal studies such as mortality and morbidity over time are impacted by various forms of selection bias such as differential attrition and competing risks, as well as time-varying covariates such as medical comorbidities. We aim to demonstrate how different modelling strategies accounting for time-varying covariates and selection bias influence the association estimates between warfarin use and health-related quality of life over time in those with atrial fibrillation. Methods: We performed a secondary analysis of the AFFIRM trial quality of life substudy (N=716). Dichotomized warfarin use was ascertained at study baseline, 2 months later, and annually for up to 6 years. Generic health related quality of life was measured at every time point using a self-reported ordinal 5-point Likert-scale (lower score represents better health-related quality of life). Static and time-varying covariates were ascertained throughout the study period. Confounder-adjusted generalized linear mixed model regressions were used to demonstrate the traditional association between anticoagulation and health related quality of life. Inverse probability of treatment and censorship weights were used to ascertain the influence of time-varying confounding and selection bias respectively. Stratified analysis by age (age≥70 years) evaluated for effect modification. Results: The association of warfarin use with better health-related quality of life over time in those with atrial fibrillation was strengthened when accounting for time-varying confounding and selection bias (OR 0.33, 95% CI: 0.16-0.68) compared to traditional regression analyses (OR 0.61, 95% CI: 0.0.38-0.98). The association was most pronounced in those >70 years upon stratified analysis. Conclusion: Anticoagulation is associated with improved quality of life over time in patients with atrial fibrillation. While constrained by the secondary nature of the analysis, we demonstrate how time-varying confounding and selection bias may influence longitudinal effect estimates in stroke prevention research.