Predicting length of hospital stay in a pediatric ward

The purpose of this study was to develop a statistical model for predicting hospital length of stay for pediatric patients. A sample of n = 589 patient records, representing a two year time period, was selected from a large, acute care hospital. The variables identified as possible determinants of length of stay and investigated for their effect were: age, sex, financial class, isolation status, number of operative procedures, presence of nosocomial infections, number of complications, resulting death and a classification variable identifying the final diagnosis as one of 24 mutually exclusive major diagnostic categories. An eight variable model was derived (p less than .0001) with a coefficient of multiple determination of 0.1612.