Primary Submission Category: Multilevel Causal Inference
Targeted Quality Measurement of Health Care Providers
Authors: Yige Li, José Zubizarreta, Nancy Keating, Mary Beth Landrum,
Presenting Author: Yige Li*
Assessing the quality of cancer care across US healthcare providers poses significant challenges, mainly due to small practice sizes, large number of practices, and diversity of patient populations. Patients vary widely in cancer type and other critical factors, making comparisons across practices complex. In this paper, we propose an approach to adjust for such patient diversity. Our framework follows recent advancements in health outcomes research, framing quality measurement as a causal inference problem. Using covariate profiles to describe patient populations, our approach combines a weighting step and a regression step, accounting for heterogeneous effects when practice-covariate interactions are present in the data. Through extensive simulations, we compare the performance of several methods in terms of point estimates and rankings of hundreds of practices. The results show our approach produces stable and robust estimates, as well as reliable rankings. For the case study, we provide the results of 600 practices for a couple of patient covariate profiles. The proposed approach is helpful for public reporting and has the potential to help individual patient make decisions.