ESTIMATING THE UTILIZATION IMPACTS OF HOSPITAL CLOSURES THROUGH HOSPITAL CHOICE MODELS - A COMPARISON OF DISAGGREGATE AND AGGREGATE MODELS

The potential for hospital closures, especially in rural areas, will a lmost certainly intensify under current market conditions. Hospital ch oice models, both aggregate and disaggregate, can be used to simulate and assess the likely impacts of such events. While these two types of models prevail in the literature, surprisingly little attention has b een given to their relative performance in meeting the needs of policy planners. This study estimated both types of models using the same da ta and time period to assess the direction and magnitude of difference s in outcomes simulated by each. Data on over 12,000 Medicare discharg es in a rural market in Northwestern Minnesota were used. Outcomes wer e measured in terms of patient flows to hospital types and average dis tances traveled post-closure. While the two models produced fairly sim ilar predictions regarding distance traveled, the aggregate model syst ematically predicted more patient flows to urban hospitals than did th e disaggregate model. This difference appears to reflect the bias in e stimated aggregate models due to the omission of individual characteri stics that negatively relate to the choice of urban hospitals by elder ly rural residents. Findings indicate that models based on aggregate d ata should thus be used with caution when evaluating policy and planni ng options. Copyright (C) 1996 Elsevier Science Ltd