The analysis of readmissions has important implications in psychiatric research, particularly in the areas of emergency-room administration, diagnostic classification, treatment and policy planning. Although many studies haw attempted to evaluate possible predictors of readmission, commonly employed statistical techniques generally ignore the fit of probability distributions of readmissions to the data. One type of probability distribution - the negative binomial - has previously been successfully fitted to readmission data and has been found to give a closer fit than a commonly employed distribution, the Poisson. Employing a form of generalized linear model - negative binomial regression - we extend these findings to another set of readmission data, using readily available statistical software.
|Number of pages||6|
|Journal||International Journal of Methods in Psychiatric Research|
|Publication status||Published - 1998|