The patient outcome research teams (PORT) are faced with many problems in the conduct and analyses of their studies. Analyses of the Medis Group data bases for the pneumonia PORT have highlighted statistical issues that need to be addressed for this analysis and for the analysis of the cohort study which is currently underway. The following issues are of greatest importance: appropriate model -choice for the analysis of length of hospital stay, models to account for variability introduced by hospitals and hospital related variables in an analysis of length of stay, and the treatment of deaths in a length of stay analysis. The focus is on the development and implementation of modeling techniques that will be useful for analyzing length of stay data. Currently existing models will be explored for use in these data sets and results of analyses will be written up for the health services literature. Several new statistical methods will be developed for analyzing these data sets and will be presented in both the statistical and health services literature. Results from all of the above approaches will be summarized and optimal modeling strategies will be presented in the health services literature.
