Also it has been used like clinical performance in- dicators the length of stay of the three more frequently DRGs of Spanish National Health System. These ones are: number 373: birth without complications; number 39: surgical operation on crystalline and number 127: hearth failure. The results the analysis of variance (ANOVA) and the Kruskal–Wallis test that were run in order to de- tect significant differences in performance variables between the various groups resulting from the auto- matic classification are summarised in Tables 8 and 9. Significant differences emerge among the average values for the following: number of beds, length of stay, ratio of primary to subsequent consultations and the occupancy rate. It is useful to ascertain which groups differ sig- nificantly in their averages. Therefore, in the case of the variables that had been used to perform the Kruskal–Wallis test, the differences in the averages were again compared taking the groups two by two, this time applying the Mann–Whitney U test, since the variable under consideration has two modalities (it either pertains to one group or the other). In the case of the occupancy rate and ratio of pri- mary to subsequent consultations variables, the proce-
Table 10 Results of multiple comparisons
Variables Groups 1 2 3 4 Staya 1 – 30 5 36 2 – – 14 ∗ 92.5 3 – – – 6 ∗∗ 4 – – – – No. bedsa 1 – 32 5∗ 44 2 – – 30.5∗ 127 3 – – – 14.5∗∗ 4 – – – – Occupancy ratiob 1– −0.70 −3.90 −10.30∗∗∗ 2 – – −3.20 −9.6∗∗∗ 3 – – – −6.4 4 – – – – First visit/latterb 1– −0.04 1.28 −0.37 2 – – 1.32 −0.33 3 – – – −1.66∗∗∗ 4 – – – – a U Mann–Whitney. b Scheff´e test. ∗ P