Hospital and physician prices and treatment choice in labor and delivery

We study the effect of changing the price differential for cesarean versus vaginal deliveries paid by commercial (private) insurers to hospitals and physicians on cesarean rates. Using eight years of claims data containing negotiated prices, we exploit within-hospital-physician-group-insurer price variation arising from contract renegotiations between providers and insurers over time. We model the choice of vaginal versus cesarean delivery as a function of the difference in expected physician prices for cesarean versus vaginal delivery, the difference in expected hospital prices for cesarean versus vaginal, patient characteristics, and a time-invariant preference parameter specific to each hospital, physician group and insurer combination. Crucially, by using these hospital-physician group-insurer fixed effects, we are able to account for unobserved quality, marginal cost, or preference heterogeneity across different provider subsets defined by a particular hospital, physician group and insurer for each treatment.

We find that increasing the physician price differential by $100 yields a 0.55 percentage point (1.9%) increase. This is equivalent to an average price-elasticity of approximately 0.08. We note this corresponds to a price-elasticity that is similar in magnitude to (though slightly larger than) previously reported elasticities for changes in the difference in physician payments across the two treatments.

Although we find that hospital payments do not have a significant effect on treatment decisions when we study the full sample, they do have a large and significant impact for births delivered by physician groups who only deliver at a single hospital: a $1000 increase in the hospital’s payment for a cesarean delivery compared to its payment for a vaginal delivery leads to a 1.1 percentage point (3.7%) increase in the cesarean rate. This is equivalent to an average price-elasticity of approximately 0.20. Since hospitals and physicians typically receive separate payments under both public and private insurance, our findings are consistent with the notion that hospitals transmit their incentives for choosing one treatment over the other to individuals present at a particular birth, including (but not limited to) the physician. Insofar as this suggests that hospitals are able to overcome the principal-agent problem that they face with physicians when those physicians are exclusive with that hospital and cannot steer patients to other hospitals, it also implies that hospitals and physicians may be able to coordinate their actions within accountable care organizations (ACOs). Our findings have implications for understanding hospital-physician principal-agent problems and the future of accountable care organizations.