General medical care in the U.S. was traditionally provided by physicians caring for patients in both ambulatory and hospital settings, but increasingly is divided between physicians specializing in hospital care (‘hospitalists’) and primary care physicians who practice only in ambulatory settings (‘ambulatory-based PCPs’). We develop and test a theory of the division of labor in general medical care where the choice between hospitalist and traditional models of care depends on the cost of coordinating care across settings compared to the cost of physicians switching between settings.
We use data on general internists, family and general practice physicians from the 2004-2005 Community Tracking Study Physician Survey to test the predictions of our model. Using fractional logit models, we regressed the percent of a physician’s hospitalized patients cared for by hospitalists on measures of coordination and switching costs (whether a physician’s practice was located among the 25 most congested MSAs). To study the effects of total work hours, we examined the effects of average hours worked during the week and physician gender. To study the effect of probability of admission, we measured the fraction of a physician’s patients above age 65.
The data were consistent with our model predictions. Hospitalist use increased as travels costs rose, coordination costs decreased, work hours decreased and female physician gender, and decreased with the fraction of patients above age 65. A number of previously unappreciated forces, including travel and communication costs, physician work hours, and the growing rate of ambulatory relative to hospital care, likely contributed to the increased use of hospitalists in the U.S.
Our findings suggest the potential for new models of care to optimize the use of the hospitalist model and traditional models by combining ambulatory and hospital care under the direction of a single physician. More generally, our research provides support for the value of theoretical constructs emphasizing the importance of coordination and switching costs in understanding the division of labor.
