Mapping the Predicted Need for Hospital Beds

There is a long history of approaches for predicting the need for hospital beds at specific locations. Past methods have tended to regard as key variables the ser vice area population, hospital use rate per population unit, average length of stay, and occupancy rates. The usual client for a bed-need study is an individual hos pital. There is no constraint on the forecasting meth odology to ensure that if the approach used for one hospital were to be applied to many hospitals, the sum mation of the forecasts for a region would yield re sults that are consistent with macro-regional forecasts of bed need. The methods used in this study are de signed to produce such consistent system-wide forecasts when aggregated from individual-level forecasts. That is, a system of allocations of patients to hospitals in individual places is made with a sufficient accounting of flows of patients from each area to all hospitals that hospital-use rates for individual areas prevail irrespective of the particular hospitals patronized. Such a system-wide simulation model corrects for the major flaw in most existing procedures in which the system-wide effects of adapting