Measuring hospital efficiency in Austria--a DEA approach

We investigate the evolution of efficiency and productivity in the hospital sector of an Austrian province for the time period 1994-1996. We use panel data to design non-parametric frontier models (Data Envelopment Analysis) and compare efficiency scores and time patterns of efficiency across medical fields. As health outcomes hardly can be measured in a direct way we make use of two different approaches for output measurement: In a first approach, we employ the number of case mix-adjusted discharges and of inpatient days, in a second we use credit points, which are calculated in course of the newly introduced diagnosis related group-type financing system. We calculate and compare individual efficiency scores for hospital wards as decision making units (DMU) in specified medical fields. To our knowledge the calculation of ward-specific efficiency scores has not up till now been the unit of non-parametric efficiency analysis. Our two models find different results: Model 1 with conservative output measurement calculates an average efficiency level of 96%, while model 2 with credit points for output measurement puts average efficiency at 70%. Whereas average efficiency in model 1 hardly changes and in model 2 increases modestly in the period 1994-1996, a closer look at single hospitals displays a variety of different efficiency developments over time.