Exploring predictors of readmission for pediatric psychiatric inpatients

Hospital readmission of pediatric populations with mental or psychiatric illness who have received a prior inpatient experience is a public health concern. Unfortunately, there are many researchers, practitioners, or policy makers who do not recognize this readmission issue as a public health concern. The nonrecognition maybe primarily because little research has been done about pediatric psychiatric readmission and especially predictors of such readmissions. There is a need to identify predictors of these readmissions to assist practitioners and researchers to design interventions that could be used to minimize these readmissions. In this study, we use data from an academic psychiatric hospital, namely the University of Utah Neuropsychiatric Institute (UNI), to explore predictors of pediatric readmissions. These data are for those children and adolescents who were inpatients between 1999 and 2010. We begin with a description of pediatric inpatients seen at an academic psychiatric hospital. The description includes variables such as: the most common diagnoses of the admitted patients, sex and age break down, the hospital length of stay, readmission patterns, discharge dispositions, and the most common ways that patients' care is paid for. The description is necessary to identify the independent variables that are used in the major part of the study of identifying the predictors of readmission. Logistic regression models are used in the second part of the study with readmission as the dependent variable. The third part of the study includes use of iv Receiver Operating Characteristic (ROC) curves and Statistical Control Processes to further investigate the association between hospital length of stay and readmission. The major findings of the study are that the diagnosis is the major predictor of the readmissions. The influence of other independent variables significantly decreases with the introduction of the specific diagnosis. For the majority of the diagnoses, increase or decrease of length of stay does not predict readmissions. To understand psychiatric illness and readmission among pediatric populations, diagnoses should be viewed individually and not all categorized as if they were one disease.