Variations in Treatment Patterns and Patient Outcomes across Medical Facilities in Japan: Micro-level Evidence from Patients with Acute Myocardial Infarction treated by PTCA (Percutaneous Transluminal Coronary Angioplasty)(in Japanese)

The main object of this study is to investigate how treatment patterns and patient outcomes vary among medical facilities in Japan. We take advantage of a unique survey on 3,502 patients with acute myocardial infarction treated by PTCA (Percutaneous Transluminal Coronary Angioplasty) in 2003. The data include 50 medical facilities in collaboration with the CAMPAIGN Study as well as National Cost Database on Patients Treated by PTCA, collected by the Japanese Society of Interventional Cardiology (JSIC). These facilities are high-volume hospitals that hold both physical and human resources rich enough to perform high-tech treatments. We adopt the GMM (General Method of Moment) approach, using hospital-level variation in treatment intensity as an instrumental variable. Our estimates suggest that there is substantial variation in both treatment patterns and patient outcomes such as mortality rates, medical expenditures, and length of stay, even though medical facilities in the data are relatively homogeneous from structural points of view, such as physical and human capital. First, hospitals that use one drug treatment on admission aggressively tend to be aggressive on other drug treatments as well, both on admission and at discharge. Second, hospitals that perform PTCA aggressively tend to use fewer drug treatments. Third, the longer the time period after hospitalization, the larger the standard deviation in mortality rates across hospitals tends to be. Our results, however, show that the variation in treatment patterns across medical facilities does not influence mortality rates and medical expenditure, though stent use tends to shorten patient length of stay.