A statistical model for length of psychiatric in‐patient treatment and an analysis of contributing factors

OBJECTIVE: Direct illness costs in psychiatry are strongly related to the length of in-patient stay (LOS). Prior studies have shown that LOS depends upon many factors; however, there is no systematic work on their interrelation and relative contribution. METHOD: A detailed statistical analysis of the factors explaining LOS for n = 4,706 consecutive admissions (1994-97) to the psychiatric hospital of the University of Tübingen is presented. RESULTS: The distribution of LOS follows an exponential decay function, suggesting a hazard-based process. Cox regression indicates that the incidence of discharge and hence LOS is modulated by a number of illness-related and other factors, and their relationship is explored. CONCLUSION: While a linear model is commonly assumed, LOS in psychiatry is governed by a hazard-based process. As a tool in quality management, LOS data for psychiatric hospitals might be routinely analyzed and the effects of non-illness-related factors minimized.