The Effect of Physician Group Mergers on the Health Outcomes of Medicare Beneficiaries

Wednesday, June 26, 2019: 12:00 PM
Wilson B - Mezzanine Level (Marriott Wardman Park Hotel)

Presenter: Thomas Koch

Co-Authors: Nathan Wilson; Brett Wendling

Discussant: Elena Prager

Previous research has documented substantial consolidation across several parts of the health care sector. In some instances, this consolidation is associated with higher prices (e.g., general acute-care hospitals in Cooper, et al., 2018) and lower quality (e.g., cardiology in Koch, Wendling and Wilson, 2018). While these patterns provide interesting cross-sectional or panel comparisons across different parts of the concentration spectrum, they do not necessarily capture the causal effects of mergers and acquisitions.

We measure the effect of physician mergers on the health outcomes of Medicare FFS beneficiaries. First, we identify physician group mergers using the MD-PPAS, which reports the primary and secondary TAX ID (i.e., group identifier) for each provider. In the spirit of Walden (2016), we are able to follow providers and groups over time to identify mergers. These acquisitions may be within specialty (i.e., horizontal) or across specialty (vertical or, potentially, unrelated). We also check this method of identification against lists of mergers identified from independent data sources.

Second, we measure the effect of the merger on a variety of health outcomes. We study traditional health outcomes, such as mortality and hospitalization. We also measure the effect of mergers on ambulatory health conditions, such as diabetes and hypertension. In related work, we have shown these measures to have more variation and improved power, relative to the more traditional measures. They are also more closely linked to outpatient, ambulatory care that is provided by many of the providers we study (e.g., general practitioners or family medicine doctors).

With a large number of mergers identified off changes in TAX ID, we can stratify these effects across specialties and by the initial concentration level of the area prior to the merger. I.e., we can assess whether the effects of the merger are uniform across the concentration spectrum or are particularly harmful or beneficial when the market is already highly concentration.