Patient and hospital treatment clusters.

Overview

Patient admission data and treatment data collected for 536 cases from 11 collaborating private psychiatric hospitals were subjected to cluster analysis. From the patient admission data, four large clusters and three small clusters emerged. Two very large and general clusters and four very small clusters were obtained from the treatment data. Both sets of clusters were cross-classified against diagnosis, length of hospital stay, and attainment of therapeutic goals. There were several significant but small relationships between clusters and these criteria. There was little overlap between diagnosis and the patient data or treatment data clusters. Length of stay related only to the treatment clusters which included a number of variables measuring different treatment amounts. The method seems potentially useful because it groups individuals along dimensions of interest, but further effort must be spent on more specific treatment descriptions and better outcome variables.