Venue: The Fuqua School of Business, Duke University, 1 Towerview Drive, Durham, NC 27708-0120
Presentation
Entry by General Hospital Cardiac Units and Differential Distance to Medical Providers: How Certificate of Need Laws Affect Access to Specialty Services
We examine how new general hospital open heart surgery units affect the probability a patient gets invasive versus medical-only treatment for an acute myocardial infarction. Entry of a new unit will affect a patient's differential distance between medical and open heart facilities. In doing so, it will affect the probability of patient will be treated invasively. We measure the extent that entry affects the differential distance between each type of facility and then calculate how treatment intensity changes due to such entry. Furthermore, we examine how CON laws have affected the geographic distribution of open heart surgery units and thereby restricted access to specialty services. We then quantify the effect of CON on the probability a patient receives invasive treatment and the appropriateness of patients for invasive treatment among patients who are treated medically. The definition of appropriateness follows Chandra and Staiger (2007). Our analysis uses HCUP-SID discharge data from 8 states for the period 1995-2004. The discharge data is linked to American Hospital Association data, Census data and Medicare Cost Reports.
We first regress the zip code level risk-adjusted propensity for invasive treatment on differential distance, zip code demographic and hospital market controls, and zip code fixed effects. The coefficient on differential distance measures the effect of changes in differential distance on the propensity for invasive treatment, assuming changes are based solely of fixed zip code characteristics. Next, we measure marginal and average treatment effects using instruments that are correlated with access to care (e.g. the appropriateness for invasive treatment among those treated medically and the percent of patients transferred from a hospital). The first-stage is a discrete-time hazard model that predicts the probability a hospital opens a unit using hospital catchment-area demographics, patient/competitor characteristics and instruments. The predicted probability of entry is then used to calculate predicted differential distance. If CON laws are more effective at limiting entry to only access-related entry then the effect of changes in differential distance controlling for access-related endogeneity would be less in CON states than other states. In other words, patients in other states would experience an increase in invasive treatment for reasons other than changes in access. Finally we simulate how much entry would occur in absence of CON and the resulting change in invasive surgeries.
The preliminary results suggest that entry into open-heart surgery by general hospitals is access-related in all states, regardless of CON. However, residents of CON states are less likely to be treated invasively due to the restrictions on entry (i.e. distance to a specialty facility is larger). In other words, while entry-related treatment patterns are similar, there are fewer entrants in CON states and thus CON laws restrict access-related entry. We discuss the results in the context of recent research (Cutler, 2007) which measures the value of access to specialty services related to differential distance.