class: center, middle, inverse, title-slide .title[ # Healthy Reviews! Online Physician Ratings Reduce Healthcare Interruptions ] .subtitle[ ## Presenter: Danny R. Hughes, Georgia Tech ] .author[ ### Discussant: Ian McCarthy, Emory University and NBER ] .date[ ### ASHEcon Annual Meeting, June 29, 2022 ] --- <!-- 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 --> # Consumer search - Highly relevant and underappreciated structure for studying health care decisions - Basic idea: - Patients incur a cost to accessing/assimilating information about physicians - Patients have some beliefs about their utilities over all possible options - Patients stop searching when costs meet or exceed expected benefits --- # Context - Physicians retire and patients must find a new physician - Online reviews lower the cost of this search and increase probability of a more recent visit to a new physician (not sure that the first implies the second, but more on this later) --- # Looking for New PCP .center[ <img src="ashecon-2022-hughes_files/figure-html/unnamed-chunk-2-1.png" style="display: block; margin: auto;" /> ] MedPAC *Report to Congress*, March 2012 and March 2020 --- # Finding a PCP among Medicare and MA .center[ <img src="ashecon-2022-hughes_files/figure-html/unnamed-chunk-3-1.png" style="display: block; margin: auto;" /> ] MedPAC *Report to Congress*, March 2012 and March 2020 --- # Finding a PCP among Commercial Insurance .center[ <img src="ashecon-2022-hughes_files/figure-html/unnamed-chunk-4-1.png" style="display: block; margin: auto;" /> ] MedPAC *Report to Congress*, March 2012 and March 2020 --- # Approach - Research design: DD - Sample: Patients of retired physicians in two periods - Pre period: 2007-2010 - Post period: 2015-2018 - Control (Treatment): 18 (16) top 100 population cities, based on cities with fewest (most) Yelp reviews per capita and nonmissing reviews in pre and post period --- class: inverse, center, middle name: questions # Some thoughts and questions <html><div style='float:left'></div><hr color='#EB811B' size=1px width=1055px></html> --- # Practical motivation - Do people use Yelp to find new PCPs? - You write, "...a relatively small proportion of patients rely on these platforms for choosing their physicians." - But some work **does** suggest that patients value online reviews in health care --- # Theoretical motivation - Really like the idea, but... - Paper considers search as decision to find a physician today or not - Are patients "searching" over physicians, or are they just deciding whether to take an action on a given day? - Is `\(v\)` independent of `\(E[Days]\)`? How? - Searching for a physician on a given day is not the same as visiting the physician that day - Search could *increase* time to visit because patients are drawn to the most capacity-constrained physicians - How should I think of waiting time here? --- # General concern - Treated cities very different than control cities (e.g., Laredo vs Anaheim), probably not just in levels - Could relate to practice sizes and consolidation, thereby affecting within-practice referrals - Null results (using days) when including retirement year/quarter FEs --- # Identification and estimation - Is 2x2 DD appropriate here? - Treatment/control is continuous (availability of online reviews) - IV using other Yelp reviews - Does this satisfy exclusion restriction? Only conditional on overall use of Yelp - Not if early adoption of Yelp (in general) is correlated with unobservables or if Yelp causes additional retirements - Are cities the right level of treatment/control? - Laredo is isolated, Anaheim is right next to LA --- # Some potential considerations... 1. More (or fewer) details on model 2. Treatment at county or HRR-level instead of city/zip 3. Measure visits at practice-level instead of physician-level 4. Visualizations of physician markets over time between treatment/control (total retirements, number of physicians, number of practices, size of practices, etc.) Bonus question: Why do physician's retire and where do they retire most?