MEDICAL OUTCOMES IN THE PRICING OF HOSPITAL PROCEDURES

点击翻译按钮获取中文摘要

This proposal utilizes economic models of pricing behavior to assess the effects of hospital quality on the prices of major medical hospital procedures. Hospital quality is defined in terms of medical outcomes, i.e., less-than-expected in-hospital mortality rates. We will focus on cardiac procedures such as CABG and PTCA. These are relatively expensive but common procedures that are also characterized by significant in-hospital mortality. The empirical specification of the price equation will be derived from a Nash-bargaining model that describes the hospital-insurer interaction. While this model had been recently used to describe appendectomy pricing, we undertake further extension to incorporate product quality. The empirical strategy will be based on instrumental variable (IV) techniques, subject to appropriate specification tests. Other than quality, prices will depend on variables representing hospital bargaining power (Herfindahl index, ownership, system affiliation), insurer bargaining power (HMO penetration, market share, type of plan) and patient and clinical traits that account for technical variations within a given procedure. Statistical identification of the model will be attained by instrumenting quality on hospital size (beds, days, etc.), procedure specific volume, and propensity to perform these procedures (teaching intensity, acute and intensive care days). Following earlier literature, hospital volume is expected to improve outcomes. Given our focus on price-informed institutional players, the main hypothesis to be tested states that higher prices reflect higher quality, holding market structure constant. We will further examine whether differences between managed care plans, such as PPOs, EPOs, and point-of-service HMOs, also lead to varying degrees of price discounting. Since emergency care is more likely to occur outside an insurer s network we will distinguish between procedure done on an emergent or elective basis. Finally, regional price variation will also be examined. The core data will be drawn from the inpatient component of the MarketScan claims database for 1995-1996, together with a complementary file with matching hospital level and patient level mortality rates. These data will be merged with data elements from MEDPAR, the AHA Annual Survey of Hospitals, and the Area Resource File. Data prior to 1995 will not be used because MarketScan had not yet included a detailed description of insurance plans. However, based on 1994 frequencies, final samples are expected to range from 900 to 2,400 per year, depending on the procedure and elective/emergency setting.