Estimation of Optimal Treatment Regimens with Competing Outcomes

• Activity Number: 46
• Type: Invited
• Date/Time: Sunday, August 9, 2015 : 4:00 PM to 5:50 PM
• Sponsor: ASA
• Author(s): Eric Laber*
• Companies: North Carolina State University

Abstract

Delivering the best patient care requires consideration of multiple, potentially competing, outcomes. The canonical example of competing outcomes is the trade-off between treatment efficacy and side-effects though other potentially important factors include cost, availability, patient preference, and patient burden. We review four approaches for estimating an optimal regimen with competing outcomes: (i) forming a clinician-guided composite outcome to balance competing outcomes; (ii) using preference elicitation to obtain a patient-specific distribution over composite outcomes; (ii) set-valued treatment regimes that screen-out completely dominated treatments; and (iv) constrained treatment regimes that maximize efficacy subject to constraints on treatment harm. Our review focuses on practical differences associated with each of these methods. Each method is illustrated using data from a randomized clinical trial.