Mean total costs, mean costs for pharmaceutical products, personnel and infrastructure costs
as well as the length of stay were analyzed as the sum of daily costs over the duration of treatment
(equaling average daily cost x average days). The distribution of individual mean values
within these categories fitted to a normal distribution according to the Kolmogorov-SmirnovTest
as implemented in SPSS 22.0. Therefore, the differences in the four variables between the
two study groups were assessed using ANCOVA as implemented in SPSS 22.0. Age and gender
were included as nuisance variables. In line with the recommendations of Barber et al.andto derive robust estimates of confidence intervals for estimated group differences of mean of
total costs, medication costs, personals costs and infrastructure costs, we also performed paired
samples t-test with bootstrapping as implemented in SPSS 22.0 (10,000 samples).