Author ORCID Identifier

Defense Date


Document Type


Degree Name

Doctor of Philosophy


Healthcare Policy & Research

First Advisor

Peter Cunningham

Second Advisor

Andrew Barnes

Third Advisor

Marianne Baernholdt

Fourth Advisor

Askar Chukmaitov


Provider incentives are a commonly used policy tool to mold provider behaviors.1 However, while we frequently measure the change in patient outcomes, failure to consistently produce changes in outcomes does not mean that providers are not changing their behavior. This paper focuses on two programs with null or inconsistent quality outcomes to try to identify why such inconsistency occurs. The two programs, both ratified in the Affordable Care Act, are 1) patient-centered medical homes (PCMHs), and 2) the Medicare Hospital Value-Based Purchasing (HVBP) program.

Chapter 1: Using data from the Medical Expenditure Panel survey (MEPS), I match provider characteristic surveys to member experience with care in order to evaluate characteristics key to patient-centered medical homes. I find that patient-perceived patient-centeredness of a practice is not related to the number of PCMH attributes a practice reports. However, some characteristics do play specific and significant roles in patient perception and outcomes. For instance, case management is not only associated with increased patient perception of after-hours access to care, but overall costs were reduced. Interestingly, having after hours clinic hours was more common with practices highly consistent with PCMH criteria, but these hours did not result in decreased emergency department use or cost of care.

Chapter 2: The second provider incentive studied is the Medicare Hospital Value-Based Purchasing Program (HVBP). This program assigns payment adjustments based on performance on a series of rotating quality metrics. To date, changes in patient outcomes cannot be attributed to the program; however, it should not be concluded that hospitals are not responding at all. I identify changes in staffing by provider type as an early indicator of hospital response to payment incentives. Data come from the Virginia Health Information (VHI) Hospital Cost Report, 2010-2017. Using a generalized linear model, I find that when receiving a penalty, hospitals reduce staffing among the most and least expensive personnel (physicians and nursing aides). Hospitals increase nursing and administrative staff following a bonus. These findings are consistent with hospitals responding to incentives both by aiming to improve efficient use of resources and maintain or improve quality of care.

Chapter 3: Finally, I assess potential unintended consequences of the HVBP program, specifically the provision of charity care. Using the VHI cost reports for year 2013 to 2017 with a regression discontinuity model, I find that hospitals receiving a bonus decrease their charity care among the lowest income patients (under 100% federal poverty level (FPL)). Hospitals receiving a penalty tend to reduce charity care among higher income patients (100%-200% FPL). These findings are consistent with two separate responses to the incentives. Hospitals receiving bonuses appear to be cream-skimming healthier, wealthier individuals while hospitals receiving penalties appear to be shifting the focus of their charity care to the most needy, likely in an effort to reduce cost of care levels overall while maintaining their community benefit programs, potentially as a result of goal gradient cognitive bias.


© The Author

Is Part Of

VCU University Archives

Is Part Of

VCU Theses and Dissertations

Date of Submission