Analyses of hospital functioning are frequently reported with patient outcomes as the independent variable. However, the main focus in this paper is the length of stay in hospital not just in terms of "short stay" or "long stay", but ranked in tenths so that ten-point distributions of patient use of hospital resources are available for study. From the analysis of these distributions three patterns emerge: (a) a "standard procedure" model with best fitting "curves" given by a linear function, eg routine surgery; (b) a "finite disease" model (exponential series) eg internal medicine and, (c) a "person in hospital care" model (power series) eg psychiatry.
