Models of case mix adjustment for Ohio Mental Health Consumer Outcomes among children and adolescents

Outcome data collection systems provide the opportunity for comparing agencies, programs, and therapists. However, the political stakes of comparisons raise important issues regarding their perceived validity. The goal of case-mix adjustment is to eliminate potential biasing factors that might render comparisons as inappropriate or irrelevant. In this paper, case-mix adjustment variables are identified using regression analysis followed by developing models of case mix adjustment using a sample of youth participants in the Ohio Mental Health Consumer Outcomes System. Results indicate that certain client level variables do predict outcomes. The total variance accounted for by variables other than baseline ratings of the outcome measure was small. When comparing agencies using adjusted and unadjusted models, some agencies did vary relative to other agencies. As with previous studies, however, the rank order of agencies was largely unchanged and the correlation between adjusted and unadjusted outcomes among agencies was significant.