Patient Heterogeneity and Treatment Choices: Evidence from the Dialysis Industry

with Paul Eliason, Ben Heebsh, Ryan McDevitt, and Jimmy Roberts
Revise and Resubmit at American Economic Review
Recent draft available here. BibTeX citation available here.
Note that this paper previously circulated as “The Effect of Bundled Payments on Provider Behavior and Patient Outcomes: Evidence from the Dialysis Industry.”

Abstract: We study the use and effectiveness of epoetin alfa (EPO), historically one of Medicare’s largest drug expenses. Using a novel instrumental variable based on elevation and a large payment reform, we estimate a model of providers’ treatment decisions that incorporates the heterogeneous effect EPO has on patients. We find that increasing doses by 1% reduces anemia by 0.1%, on average, but increases heart attacks by 0.3% and mortality by 0.4%, with patients exhibiting substantial heterogeneity along these dimensions. Our counterfactual analysis shows that patients’ well-being would decline considerably if providers disregarded their heterogeneous responses as mandated by some Medicare guidelines.