Objective: To examine the distribution of length of stay (LOS) in acute psychiatric units.
Method: Analysis of one year of LOS data from twelve acute psychiatric units in New South Wales, Australia.
Main findings: LOS had a complex distribution, with a ‘head’ of discharges within a day, a ‘body’ of discharges that approximated an exponential distribution and a ‘tail’ made up of a small group of patients discharged after very long admissions. There were substantial differences between units in the proportions of discharges classified as being in the head and the tail of the LOS distribution. With regards to the traditional measures of central tendency, the median LOS was affected by the proportion of patients discharged within a day, and the mean LOS was strongly influenced by small numbers of very long stay patients.
Conclusions: A graphic representation of logarithmically transformed LOS values might assist in understanding the LOS of individual acute psychiatric units. A modified median LOS, calculated after the removal of patients staying less than a day, was found to be a simple and reliable measure that can potentially provide a more meaningful measure of LOS than the mean or the median LOS.