class: center, middle, inverse, title-slide # The Effect of Physician Group Mergers on the Health Outcomes of Medicare Beneficiaries ## Presenter: Brett Wendling ### Discussant: Ian McCarthy ### SHESG Annual Meeting, October 8, 2021 --- <!-- Adjust some CSS code for font size, maintain R code font size --> <style type="text/css"> .remark-slide-content { font-size: 30px; padding: 1em 2em 1em 2em; } .remark-code, .remark-inline-code { font-size: 20px; } </style> <!-- Set R options for how code chunks are displayed and load packages --> # Changing physician landscape .center[ <img src="shesg-2021-wendling_files/figure-html/unnamed-chunk-2-1.png" style="display: block; margin: auto;" /> ] Data from AMA Physician Practice Benchmark Survey, 2012-2018 --- count: false # Changing physician landscape .center[ <img src="shesg-2021-wendling_files/figure-html/unnamed-chunk-3-1.png" style="display: block; margin: auto;" /> ] Data from AMA Physician Practice Benchmark Survey, 2012-2018 --- count: false # Changing physician landscape .center[ <img src="shesg-2021-wendling_files/figure-html/unnamed-chunk-4-1.png" style="display: block; margin: auto;" /> ] Data from AMA Physician Practice Benchmark Survey, 2012-2018 --- count: false # Changing Physician Landscape Physicians are now: - less likely to own their own practice - more likely to be employed by a hospital or part of practice owned by hospital - more likely to be part of a very large practice -- **Relatively little research on practice structure** --- class: inverse, center, middle name: review # What does this paper do? <html><div style='float:left'></div><hr color='#EB811B' size=1px width=1055px></html> --- # In pictures .pull-left[ **Practice at time** `\(t\)` <br> <br> <br> <br> <br> <br> <br> ] .pull-right[ **Practice at time** `\(t+1\)` <br> <br> <br> <br> <br> <br> <br> ] -- .pull-left[ at least half of physicians leave practice ("merger-to") ] .pull-right[ at least 10% new physicians ("merger-from") ] --- # Identification and Estimation - Outcomes: Mortality and progression of disease - Identification: DD(ish) - Estimation: OLS with indicators for different merger types and controls for patient characteristics --- class: inverse, center, middle name: questions # Some thoughts and questions <html><div style='float:left'></div><hr color='#EB811B' size=1px width=1055px></html> --- # Capturing mergers in the data - What happens if 50% of physicians leave a practice, and the practice subsequently (or concurrently) gains more than 10% of new physicians? - What happens if physicians move practices more than once? --- # Are these mergers? .pull-left[ **Sure** - Bulk of a practice dissolves (50%) - Existing practice expands by some meaningful amount (10%) ] .pull-right[ **Not really** - "merger-to" practices still exist - very hard to distinguish horizontal vs vertical ] --- # If not mergers, what? - Disruption in care - Information transmission within practices - Variations in practice styles due to peers --- # Identification - Is this a DD? Didn't see an "ever treated" dummy or person fixed effect... - Sample selection? How often does a person leave their physician after the physician changes practices? - Block estimation predicts treatment using patient characteristics, but nothing about physician practices. What source of endogeneity is this intended to adjust for? - Physicians may change practices due to their current patient profile or restrictions on how they can practice - Are physician characteristics in the data? --- # Estimation - What is the estimand of interest? - Not an ATT without strong homogeneity assumptions (across people and over time) - Staggered treatment, etc. --- class: clear .center[ ![:scale 55%](https://media.giphy.com/media/1yMcjnQoYejDQ8AwZN/giphy.gif) ] --- # Final ideas - Thinking about movement across practices... - Could exploit variation in non-compete agreement enforcement (Hausman and Lavetti, 2019)