Physicians often practice in multiple hospitals, a behavior believed to be associated with less familiarity and prestige over essential hospital resources. Little is known about what impact this behavior has on procedural outcomes despite a growing recognition that improvements in process of care can lead to better quality. This proposal investigates if practicing in multiple hospitals is related to variation in patient outcoms following Percutaneous Coronary Intervention (PCI). PCI is a common hospital procedure that widens clogged coronary arteries, a condition that prevents oxygen from getting to the heart and can lead to heart failure if left untreated. The patient outcomes I evaluate in this proposal include readmission, mortality and repeat PCI. I will use medical claims from a 100% sample of Fee-For-Service (FFS) Medicare beneficiaries who received PCI over the period 2001 through 2004 to empirically observe this relationship. In order to measure the behavior of practicing in multiple hospitals, I will develop a set of index measures on how concentrated a physician's PCI volume is across their surrounding hospitals. There are three kinds of concentration that might matter: (a) the physician's concentration of PCI across different hospitals, (b) a given hospital's PCI volume spread across different physicians, and (c) the concentration levels of a hospital's PCI physician workforce. I will join the Medicare claim records to physician characteristics from the American Medical Association's (AMA) Masterfile, hospital characteristics from the American Hospital Association (AHA) Annual Survey, and geographic-market level characteristics from the Area Health Resource File (AHRF). I will document variation in each of the proposed concentration measures in multiple dimensions (among physicians, hospitals, and geographic regions) using the characteristics I incorporate from these additional data sources. I will estimate their association with patient outcomes such as mortality, re-admission and repeat PCI while controlling for patient, physician, hospital, and market characteristics. This research will improve on previous efforts to model this relationship, both expanding its generalizability and dealing with a likely selection bias. The physicians'choice of hospital is likely influenced b the quality of the hospital, which is not entirely observable to the researcher. This unobserved quality influences concentration at a hospital, as well as patient outcomes at a hospital. I overcome this problem by: (i) modeling physician PCI volume share at any given hospital using factors exogenous to patient outcomes at that hospital, (ii) predicting physician PCI volume share at each hospital and constructing physician-hospital concentration measures using these predicted shares, and (iii) modeling the relationship of predicted physician-hospital concentration with patient outcomes. This proposal addresses an important health policy question on patient safety and advances methodologies for estimating levels of physician-hospital concentration in an increasingly integrated physician-hospital landscape.
According to the American Heart Association, coronary heart disease (CHD) is the leading cause of death in the United States and is a major market for the medical industry, totaling almost $100 billion dollars in direct costs each year. Percutaneous coronary intervention (PCI) is a prevalent procedure that relieves constriction of the flow of oxygen to the heart. In 2010, over 490 thousand PCIs were performed in the United States. Procedure related mortality averages 1.27%, meaning that over 6 thousand patients die each year from complications related to the PCI. Physicians delivering these procedures often practice in multiple hospitals. This proposal's objective is to estimate the relationship between practicing in multiple hospitals and variation in patient outcomes (e.g., a repeat PCI, a readmission to the hospital, or mortality) following PCI.
