Markov models and the utilization of mental health services: a study of endogenously depressed patients

Patients in the mental health care system typically make more or less irregularly spaced visits to psychiatrists, both within and between episodes of a given illness. A Markov model is constructed which can predict the utilization of psychiatric services for such patients. Unlike previous Markov models of utilization, the current model takes as its starting point a model of an actual disease, specifically, endogenous depression. It is shown how one can estimate the parameters of both the model of utilization and the model of depression using data which were collected for clinical research purposes. The models provide reasonable fits to the data. Several applications of the models are worked out. In addition to predicting the utilization of mental health services, the models can be used to predict incidence, prevalence and recovery rates and to predict the changes in utilization which parallel changes in treatment regimens.