Author ORCID Identifier


Defense Date


Document Type


Degree Name

Doctor of Philosophy


Health Related Sciences

First Advisor

Carolyn Watts, PhD


The purpose of this study is to explore differences in quality between Medicare Accountable Care Organizations (ACO) and Health Maintenance Organizations (HMO). Three outcomes measures reported by these plans use different methodologies but possess enough alignment to permit comparison: percent of diabetic patients with last HbA1c > 9.0%, colon cancer screening rate and ER visits per 1,000. These outcomes are the dependent variables (DV). A secondary purpose is to explore differences in quality based on the size of the beneficiary population served, using the same measures.

As the Medicare program faces threats to its solvency in coming decades, with 10,000 baby boomers becoming eligible every day, and the ongoing national conversation about healthcare more generally, approaches to Value-Based Purchasing (VBP) are becoming more common. Organizations seeking to identify the types of VBP arrangements in which they should enter have precious little information on the comparative performance of VBP approaches relative to outcomes measures. Different structures create different incentives through the plan design and risk/reward. The convergence or dissipation of the plan incentives at the level of the provider, particularly in primary care, may be a source of variance.

This study is retrospective, non-experimental, and uses publicly available data on the performance of Medicare ACO and HMO plans in calendar year 2015, for the identified measures. Using the Donabedian Structure-Process-Outcome framework, this study explores the impact of structure by type of plan and size of population served, relative to the outcomes. Race, average Hierarchical Condition Category (HCC) risk score and duration of operations are control variables. The analysis uses multiple hierarchical regression to better understand the relationship between the independent variables (IV) and DVs, after the impact of the control variables (CV).

After controls, the IVs did offer some explanation of variation in outcomes. The ACO plans fared better on HbA1c control, while HMO plans had fewer ER visits per 1,000. No discernable difference existed between the HMO and ACO plans with regard to colon cancer screening rate. Serving larger populations led to better performance on all three measures. In general performance was worse on each measure in both models when the percent of not-White patients or average HCC risk score increased. A longer duration of operations also associated to better performance on the outcome measures.


© The Author

Is Part Of

VCU University Archives

Is Part Of

VCU Theses and Dissertations

Date of Submission