Scale, efficiency and organization in Norwegian psychiatric outpatient clinics for children
Background It is generally believed that 5 percent of the population under 18 years is in need of specialist psychiatric care. In 1998, however, services were delivered to only 2.1 percent of the Norwegian population. Access to services can be improved by increasing capacity, but also by increasing the utilization of existing capacity. Changing financial incentives has so far not been considered. Based on a relatively low number of registered consultations per therapist (1.1 per therapist day) the ministry has stipulated that productivity should increase by as much as 50 percent. Measuring productivity in psychiatric care is difficult, but we believe that studies of productivity should be an important input in policy making. The aim of this paper is to provide such an analysis of the productive efficiency of psychiatric outpatient clinics for children and youths, and in particular to focus on three issues: 1) is an increase in productivity of 50 percent a realistic goal, 2) are there economies of scale in the sector, and 3) to what extent can differences in productivity be explained by differences in staff-mix and patient-mix? We utilize an approach termed Data Envelopment Analysis (DEA) to estimate a best-practice production frontier. The potential for efficiency improvement is measured as the difference between actual and best-practice performance, while allowing for trade-offs between different staff groups and different mixes of service production. The DEA method gives estimates of efficiency and productivity for each clinic without the need for prices, and thus avoids the pitfalls of partial productivity ratios. The Kolmogorov-Smirnov statistic is used to compare efficiency distributions, providing tests of variable specification and scale properties. Based on 135 observations for the years 1997 to 1999, the tests lead to a model with two inputs, two outputs and variable returns to scale. The outputs are number of hours spent on direct and indirect interventions, while neither the number of interventions nor the number of patients was found to be significant. The inputs are the number of university-educated staff and other staff, but disaggregation of the latter group was not significant. The average of estimated clinic efficiencies is 71%. The mean productivity is 64%, but many large clinics have considerably lower performance due mainly to scale inefficiency. There seems to be considerable room for improved performance in these clinics. It is interesting that the potential is not that far from the officially stipulated goal of 50% increased productivity. Staff composition does matter for clinic performance, but the different groups do not have significantly different marginal productivities, indicating a lack of ability to utilize specialized skills. It should be noted that these results to some extent depend on the assumptions that medical practice is efficient, and that the available data accurately captures the activities of the clinics. More appropriate outcome measures, e.g. global assessment of functioning scores (GAF), will soon be available and will improve the policy value of this type of analysis, as will a more refined data set with information about the number of personnel in training positions. The analyses in this paper indicate that a lack of consensus on the issues of who should be treated, how they should be treated and by whom results in large variations in productive efficiency. These issues are being debated in Norway, and it should be interesting to see whether this in itself leads to higher efficiency or whether a change in the incentive structure will be needed.
Halsteinli, Vidar, Jon Magnussen, Sverre A.C. Kittelsen
C61, D24, I12
Health Care, Productivity, Data Envelopment Analysis
Prosjekt:4101 - Økonomi og helse
Research Council of Norway to HERO