In this study, I intend to explore the production pattern of the publicly funded drug treatment programs in New Jersey and assess their performance. Production and performance are measured quantitatively by the number and value of treatment slots and qualitatively by treatment effectiveness. Cost, clients' pretreatment conditions, and program management choices are studied for their effects on productive efficiency and effectiveness. The main hypotheses include: (1) there may be some association between the quantity and quality production functions of 65 drug treatment programs; (2) clinic inputs relate to quantity and quality of treatment services; (3) the client's characteristics affect treatment outcome at different treatment modalities. To assess 65 drug treatment programs, both single and multiple treatment modalities or multiple clinic sites, that serviced over 26,000 clients during 1990, the clients' pretreatment conditions were controlled in terms of severity of drug problems, support and motivation for treatment, and the treatment modality a client was in. Logistic models were fitted to the individual client's treatment outcome and pretreatment condition. The parameters estimated were shown to have 77% predictive power for each client's outcome and were used to calculate the expected treatment effectiveness of each treatment program given its constituent clients' pretreatment condition. Performance rankings were compared by quantitative production output, actual treatment effectiveness and relative effectiveness. The relative effectiveness was defined as the ratio of actual effectiveness to expected effectiveness, expressed in percentage. A Cobb-Douglas production function fits better than a translog form for explaining the quantitative production behavior of the treatment programs. Decreasing returns to scale was observed when input and output were aggregated by various ways, but not significantly different from constant returns to scale when the standard error was quite large. Study results of ordinary least squares, seemingly unrelated regressions, and two-stage least squares estimation methods indicate a slight negative association between the quantity and quality of the treatment program production behaviors. The large cost variations make cost-minimizing behavior unlikely at all treatment programs. It was an important finding that higher per-slot total staff number or wage expenditure did produce better treatment quality.
Thesis (Ph.D. in Health Care Systems) -- University of Pennsylvania, 1996. Source: Dissertation Abstracts International, Volume: 57-11, Section: B, page: 6879. Supervisor: Mark Pauly.