Linking patients to hospitals. Defining urban hospital service populations.

Small area analysis is a widely used approach for identifying variations in the delivery of health services across geographically defined populations. There is growing consensus that, to a significant degree, these variations reflect the practice style of different physicians at area hospitals. However, to date, small area analysis has not been appropriate for studying the practice style of physicians at individual urban hospitals. This occurs, at least in part, because urban residents have easy geographic access to several hospitals, and no clear method of assigning small geographic areas to a single hospital is available. This study addresses this issue by taking a new approach to defining urban hospital service areas. As gate keepers to hospitals, physicians tend to admit their patients to one or two institutions. Therefore, urban hospital service areas are defined by ignoring geographic boundaries and linking patients to a single hospital on the basis of the admitting patterns of the physicians the patient contacts. The sensitivity of the assignment rules is tested; the results of the proposed technique are compared with those using a traditional geographic approach. The findings of the study reported here suggest that this is a feasible method, which yields stable sensitivity results and is generalizable to a variety of urban settings.