Online Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 169 - Estimating Heterogeneity in Treatment Effects in Complex Real World Settings
Type: Topic Contributed
Date/Time: Tuesday, August 4, 2020 : 10:00 AM to 11:50 AM
Sponsor: Health Policy Statistics Section
Abstract #313791
Title: WITHDRAWN The Value of Personalized Medicine: Treatment Effect Heterogeneity in the Treatment of Advanced Heart Failure
Author(s): Jeffrey Scott McCullough and Sriram Somanchi
Companies: University of Michigan and University of Notre Dame
Keywords: Health economics ; Machine learning; Heterogeneous treatment effects; Generalized random forests

Personalized medicine holds the potential to improve quality and control costs by tailoring treatments to patients. We develop a model that separates quality into clinical judgment – matching patients to surgical vs medical treatment – and provider skill. This distinction is important as judgment errors may be prevented through algorithmic decision support.

We apply this framework to the treatment of advanced heart failure (AHF). This is a high-mortality condition with an average one-year survival of 60%. Ventricular Assist Devices (VAD) increase one-year survival by as much as 40 percentage points in clinical trials, but real world AHF populations differ substantially from trial populations. We extend generalized random forests to implement nonlinear instrumental variables. Models are estimated using Medicare claims data from 2008 – 2015.

Preliminary results suggest that some patients realize treatment effects of 30 to 40 percentage point increases in one-year survival, consistent with clinical trial results. Average treatment on the treated, however, is about 15 percentage points. This is because many patients are poorly matched to the treatment.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2020 program