Spencer Luster
Dissertation Title: "Understanding the Health Care Delivery System: Empirical Investigations of Insurer Site of Care Policies and Physician Competencies"
In the United States, the path from having a medical condition to receiving medical care for that condition is a multistep process involving multiple stakeholders and professionals. In particular, for the vast majority of the age under-65 population in the United States, patients will need to interact with both a physician and a commercial health insurer. The role of the physician has always been central in determining the care that a patient will receive. What is less clear, however, is how variation in competency among physicians is related to the care that patients receive, and from whom patients receive it. Conversely, in the last few decades the role of health insurers has shifted from one of primarily being aggregators of risk and ex-post indemnifiers to having a more direct impact on medical care through utilization management policies.
In Chapter 1, I, along with collaborators Alton Lu, Yanchun Xu, and Russ Michael, investigate a particular type of utilization management policy among commercial health insurers. These policies set out to shift the site of care of administration of physician administered drugs from the hospital outpatient department (HOPD) to other, less expensive sites of care (Site of Care policies). We are able to extract the timing of these policies for specific drugs from insurers’ publicly-available medical policies, and then link exposure to these policies to individual insurance claims with a novel data set from Blue Health Intelligence. Using a staggered difference in difference design, we show that Site of Care policies lead to a shift of approximately 4.4 percentage points, or 21%, of patients from the HOPD to other Sites of Care by two years after policy implementation. We also estimate that these policies lead to an approximately 6.4% relative reduction in spending by two years after policy implementation, and that this point estimate may be best interpreted as a lower bound on (the point-estimate for) savings.
In Chapter 2, I, along with Kenji Yamazaki, Sean Hogan, Eric Holmboe, J. Michael McWilliams, and Michael Chernew, investigate different ways of identifying low-performing internal medicine physicians using the Accreditation Council for Graduate Medical Education Internal Medicine Milestone Scores. We first show that identifying low-performing doctors is a necessary first step to understanding the relationship between utilization and physician ability due to the program-specific nature of Milestone evaluation and the likely non-random sorting of physicians to residency program and independent practice setting. Using the universe of residents graduating between 2015-2021 (excluding 2020 due to COVID-19), we show that methods of determining low performance that rely on taking the Milestone scores at face value produce different partitions of “low-performing” and “not low-performing” doctors than measures that take into account variation in mean scores across programs.
In Chapter 3, I, again with Kenji Yamazaki, Sean Hogan, Eric Holmboe, J. Michael McWilliams, and Michael Chernew, examine the relationship between comparative advantage and fellowship choice in Orthopedic Surgery. By using the uniquely specific ACGME Milestones for Orthopedic Surgery, we show that comparative advantage in the final two years of residency consistently predicts sub-specialty choice, with suggestive evidence as early as the middle of the third year. We discuss multiple interpretations of this result, and the potential implications for policy makers.