Factors associated with length of stay and the risk of readmission in an acute psychiatric inpatient facility: a retrospective study

OBJECTIVE This study was to investigate factors influencing the length of stay and predictors for the risk of readmission at an acute psychiatric inpatient unit. METHOD Two comparative studies were embedded in a retrospective cross-sectional clinical file audit. A randomly selected 226 episodes of admissions including 178 patients during a twelve-month period were reviewed. A total of 286 variables were collected and analysed. A case control study was employed in the study of length of stay. A retrospective cohort study was used to investigate the predictors for the risk of readmission. RESULTS Logistic regression analyses showed that 10 variables were associated with length of stay. Seclusion during the index admission, accommodation problems and living in an area lacking community services predicted longer stay. During the follow-up period 82 patients (46%) were readmitted. Cox regression analyses showed 9 variables were related to the risk of readmission. Six of these variables increased the risk of readmission, including history of previous frequent admission, risk to others at the time of the index admission and alcohol intoxication. More active and assertive treatment in the community post-discharge decreased the risk of readmission. CONCLUSIONS Length of stay is multifactorially determined. Behavioural manifestations of illness and lack of social support structures predicted prolonged length of stay. Good clinical practice did not necessarily translate to a shorter length of stay. Therefore, length of stay is predictable, but not readily modifiable within the clinical domain. Good clinical practice within the community following discharge likely reduces the risk of readmission. Quality of inpatient care does not influence the risk of readmission, which therefore raises a question about the validity of using the rate of readmission as an outcome measure of psychiatric inpatient care.