Leila Agha

Published Research

    with David Molitor
    Forthcoming at Review of Economics and Statistics

The Determinants of Productivity in Medical Testing: Intensity and Allocation of Care
    with Jason Abaluck, and Christopher Kabrhel, Ali Raja, Arjun Venkatesh
    2016, American Economic Review, 106(12): 3730-64.
    NIHCM Research Award Honorable Mention

    Previous title: "Negative Tests and the Efficiency of Medical Care: What Determines         Heterogeneity in Imaging Behavior?"

Big Names or Big Ideas: Do Peer-Review Panels Select Top Science Proposals?
   with Danielle Li
   2015, Science, 348(6233):434-438.

   2014, Journal of Health Economics, 34: p.19-30.  

Current Research

    with Brigham Frandsen and James Rebitzer
    Revision requested at the Journal of Public Economics

Team Formation and Performance: Evidence from Healthcare Referral Networks
    with Keith M. Ericson, Kimberly Geissler, and James Rebitzer

Personalized Medicine and Patient Selection: Discretion vs. Guidelines
    with Jason Abaluck and David Chan
Efficient care requires that care be administered to patients who stand to benefit most. Clinical risk scores are an increasingly common tool to achieve this. However, scores typically summarize baseline risk absent treatment while clinicians would like to know treatment effects. We make use of a large database of detailed clinical records from the Veterans Health Administration to study anticoagulation treatment choices and outcomes for patients with atrial fibrillation. We estimate how anticoagulation affects stroke and hemorrhage across patients with different characteristics using variation generated by quasi-random assignment of patients to physicians with different propensities to prescribe anticoagulation. Our model allows heterogeneous treatment effects to vary flexibly with observable characteristics and allows for the possibility  that physicians are selecting patients into treatment based on unobservable treatment  effect heterogeneity. We use the model to compare patient outcomes under the status quo, under strict adherence to standard and optimal treatment rules, and under  discretionary adherence to standard and optimal treatment rules. Optimal guidelines  have the potential to reduce stroke incidence by two thirds relative to current practice,  without increasing the rate of adverse side effects.