A meta-learner-based framework to analyze treatment heterogeneity in survival outcomes: application to pediatric asthma care under COVID-19 disruption

Oregon Convention Center 

Precision medicine aims to estimate heterogeneous treatment effects (HTE) that vary among individuals. However, a notable research gap exists in the estimation of HTE in the observational studies with survival outcomes. This paper proposes a comprehensive methodology for analyzing HTE in survival data, including a pseudo-outcome framework that generates six meta-learners to estimate HTE, a variable importance metric to identify predictive variables, and a data-adaptive procedure to select beneficial subgroups. Finite sample performance is evaluated in various observational study settings. We further analyzed the heterogeneous effects of a written asthma action plan (WAAP) on time-to-ED (Emergency Department) return for asthma exacerbation, using a large EHR dataset spanning pre- to post-COVID-19 pandemic. We identified vulnerable subgroups of patients but with enhanced benefit from WAAP during the pandemic and depicted patient profiles. Our study offers valuable insights for healthcare policymakers and providers in advocating influenza vaccination and strategic WAAP distribution to particularly vulnerable groups during disruptive events, ultimately enhancing pediatric asthma care.