Predicting hospital market shares.

Distance, hospital characteristics, physician availability, and collateral opportunities for trade and entertainment all affect hospitals' market shares. The model presented here differs from most previous health facility research in that it is a multiattribute model that can be estimated by an efficient least squares technique that appropriately assures probability estimates confined to the zero-one interval. The model focuses on intercity hospital trade for a predominantly rural area, in contrast to previous intraurban studies. Special problems posed by physician influence and hospital case mix are discussed, and topics for further research are suggested.