Prediction of length of stay in an inpatient dual diagnosis unit.

The institution of prospective payment systems, in which flat fees are paid per discharge, raised the concern that hospitals might preferentially admit patients expected to have a short length of stay (LOS). This concern presupposes that intake workers could accurately predict psychiatric hospitalization LOS, but this does not appear to have been empirically demonstrated. Accordingly, we examined the ability of two psychiatrists heading separate treatment teams on an inpatient, dual-diagnosis unit and a program coordinator who worked with both teams to predict LOS for 94 patients consecutively admitted to one or the other of these teams. Predictions were highly consistent across the raters and were significantly correlated with actual LOS (r = 0.25, 0.35, and 0.45 for the three raters). However, we found that the psychiatrists were accurate predictors only for patients for whom they were the attending psychiatrist. The program coordinator, who was involved in the treatment of all patients, was an accurate predictor for the patients of either psychiatrist. We concluded that the relationships found between predicted and actual LOS held true only when the rater also influenced treatment management and discharge. Our results do not support the proposition that specialized intake workers independent of those providing care would be able to predict LOS accurately.