The temporal variation of cost-efficiency in Switzerland's hospitals: an application of mixed models

This paper uses a mixed effects model to examine the temporal variation of cost efficiency in Switzerland's general hospitals. The variations in total costs, the number of empty beds and the length of hospital stays are analyzed using financial data from a sample of 168 hospitals operating from 1998 to 2003, as well as hospitalization records disaggregated to Diagnosis Related Groups. Individual intercepts and random coefficients are used to account for the unobserved time–invariant heterogeneity and the differences in temporal patterns across hospitals and DRG categories. The analysis illustrates the usefulness of mixed models to account for unobserved factors such as quality, with a relatively weak assumption that their temporal variations, rather than their initial levels, be uncorrelated with efficiency changes. The results indicate that hospitals have adopted measures to curtail hospitalizations and reduce empty beds. The extent and effectiveness of these measures vary significantly across individual hospitals. However, there is no evidence in favor of a particular ownership type or subsidization regime. While the link between reduction rates of empty beds and gains in cost–efficiency is statistically significant, the expected association between shortening hospital stays and cost–efficiency cannot be clearly established in the data.