The Use of Length of Stay Distributions to Predict Hospital Discharges

    loading  Checking for direct PDF access through Ovid

Abstract

Many hospital admissions scheduling or admissions control systems reported in the literature rely on estimates of future discharges to help control the variance in daily patient census. One of the two most frequently reported methods of estimating discharges attempts to explain the variance in historical length of stay (LOS) data. This paper explores the relationship between LOS variance explanation and the ability to predict discharges and concludes that even a large improvement in the ability to explain LOS variance will only marginally reduce the errors in the associated discharge predictions. In drawing this conclusion, a general discharge prediction model is developed and a more relevant statistic than per cent variance explained is introduced.

Related Topics

    loading  Loading Related Articles