Service Utilization

• Assessing Service Utilization by Ellen Dornelas, 5/2/97
  
  

I will attempt to reconstruct my last message, which somehow got lost after i completed it. I am very interested in your topic, since I am more and more convinced that assessing outcomes of individual programs is meaningless without including service utilzation in the model. This allows the focus to be on client outcome related to treatment.

A model that I am familiar with in measureing service utilization is a logistic regression model in which the dependent (or outcome) variable of interest is dichotomouse. The independent variables can be the service utilization, measurement variables (or dummy varaibles representing categorical data -- e.g. attended work program, yes/no). The goal of the model is to either predict the dependent variable, such as an event (the client achieved independent living, or the client is now working), or could be to gain an understanding of the association between the dependent (outcome event) and the independent variables (the constellation of services used). The measurement is based on odds/ratio modelling.

I am not sure if the model that you spoke of is the same, since sometimes models are named differently, but that is how I would use service utilization information. One thing is certain: we need to focus more on the assessment of systems of services, especially since clients receive a continuum of services by design in our mental health systems. As noted in Bickman's Fort Bragg Demonstration Project, we need to do more work investigating services systems, thus, more work needs to be done in relation to service utilization.

I will be following up with an e-mail, as offered.

• Ordered logit by Ellen Dornelas, 5/14/97