A partial test of a hospital behavioral model.

The influence of hospital and community characteristics on the behavior of five dimensions of hospital output is examined in this article. These dimensions are the level of emergency stand-by capacity, total admissions, the diagnosis-mix of admissions and the hospital's 'style of practice' with regard to ancillary services and length of stay. A simultaneous equations model is estimated with data from a sample of 63 New England short-term general hospitals for 1970. The findings suggest that various types of short-term general hospitals have distinctive preferences for emergency capacity, volume, case mix and style of practice, and that style of practice may be more appropriately viewed as a rate of resource use per day. Specific findings of interest include the positive interdependence between protection against running out of emergency beds and length of stay, and between length of stay and ancillary service use. Hospitals that admit greater numbers of patients tend to treat more severely ill patients, and sicker patients tend to go to larger hospitals. Hospitals that provide more ancillary services tend to attract the more acutely ill patients. Relationships among other elements of the hospital's utility function represent trade-offs, i.e. substitution, in a constrained world. Among the exogenous factors, patient preferences and ability to pay have strong associations with the types of care provided by hospitals. Highly educated, high income communities, for example, tend to prefer risk averse, service intensive hospital output. Teaching hospitals are shown to prefer higher protection levels, service-intensive patterns of care, and higher admissions levels. Self-paying patients tend to be admitted for more discretionary types of diagnoses and to receive longer diagnosis-specific lengths of stay. A relatively greater supply of physician specialists in the market area is associated with increased use of ancillary services in the hospital. If replicated, these results have significant policy implications for reimbursing teaching hospitals; for defining accessibility of hospital care for the uninsured; for identifying the practice of 'skimming' by proprietary hospitals; and for specifying the role of community preferences in determining hospital performance, especially with respect to quality of care and level of emergency stand-by capacity.