Online Program

Return to main conference page
Tuesday, January 7
Tue, Jan 7, 11:00 AM - 12:45 PM
East Coast Ballroom
Causal Inference: Matching and Beyond

Estimating the effects of medical interventions: a framework for the design and analysis of longitudinal studies with treatment by indication (307878)

Presentation

Mark Glickman, Harvard University 
*Reagan Mozer, Bentley University 

Keywords: observational studies, time of treatment, causal inference

In a medical setting, observational studies commonly involve patients who initiate treatment and others who do not, and the goal is often to infer the effects on a time-to-event outcome. A difficulty with such studies is that the notion of a treatment initiation time does not exist in the control group. We propose an approach to infer the causal effects of a medical intervention in longitudinal studies when the time of treatment is observed only for treated units and where treatment is given by indication. We present a framework for conceptualizing an underlying randomized experiment in this setting based on separating the process that governs the time of indication for treatment from the assignment mechanism. Our approach involves inferring the missing indication times and estimating treatment effects that incorporate uncertainty about those times, which induces uncertainty in both the selection of the control group and the measurement of time-to-event outcomes. We demonstrate our approach to study the effects on mortality of inappropriately prescribing phosphodiesterase type 5 inhibitors (PDE5Is) using administrative data from the Veterans Affairs (VA) health care system.