predicting outcome from service utilization

• Assessing Service Utilization by Ellen Dornelas, 5/2/97
  • Question about Service Utilization Assessment by Bob Talbot, 12/13/97
    
    

Much of this work is in systems assessment. A good description of some interesting work in predicting outcomes from services is found in Roth,et al., "Services in systems: impact on client outcomes", found in New Research in Mental Health, vol. 12, pps 388-401, published by the Ohio Department of Mental Health (where you can ask for a copy via email: rothd@mhmail.mh.state.oh.us) actually, Roth et al. used cluster analytic methods in the analysis, and some of this is described in the paper. I am currently working with my colleague, Mike Nagy, on a work in which we are going to compare service utilization patterns between two mental health systems. this is somewhat different (eg, we are not looking at prediction of outcome per se). I believe that we will end up using cluster analytic techniques, as well.

In the work that I am doing with the agency in Western Mass, I will be using (hopefully, if the data permits) a logistic regression model to assess the probability of outcome (or status at discharge from the adult residential program) given the set of services used by clients during treatment tenure. i do not have enough n yet to do this, so we will see if it ever comes to reality. what I am interested in learning more about is what combination of services - over time - lead to what types of outcomes for clients in residential care: does everyone receive similar services? do some combination of services lead to greater levels of independence? does amount of contact with different types of services (providers) have a bearing on status at discharge? or does amount of contact combine with duration of contact to produce certain outcome status? so, I am interested in the probability of outcome status given certain service combinations. (probability of y, given x). the model that I am looking at allows this analysis.

No Replies