Should policy-makers and managers trust PSI? An empirical validation study of five patient safety indicators in a national health service

Table 1 PSI adjusted-incidence and variation across hospitals

	Mortality in Low- mortality DRGs		Decubitus ulcer		Catheter-related infection		Postoperative PE or DVT		Postoperative sepsis
Cases	683		18,738		5,375		9,727		10,602
Patients at risk	1,255,647		2,190,633		2,954,018		1,949,434		612,590
Adjusted-incidence ^* (range P₅P₉₅)	0.54	0 to 1.41	7.69	6.00 to 12.63	1.82	1.59 to 2.20	4.99	4.20 to 6.10	17.3	15.30 to 20.26
Statistics of variation
RV _95-5(CI95%)	12.88	9.36-14.98	2.03	1.79-2.32	1.37	1.31-1.44	1.39	1.33-1.58	1.31	1.24-1.35
RV _75-25(CI95%)	7.06	4.77-7.88	1.40	1.33-1.44	1.15	1.11-1.17	1.15	1.13-1.18	1.12	1.10-1.14
EB (CI95%)	0.32	0.19-0.51	0.34	0.25-0.45	1.14.	0.85-1.51	0.19	0.12-0.28	0.30	0.19-0.45

* Adjusted by: age, gender and Elixhauser's comorbidities except in PSI#2 where crude incidence is shown. Empirical Bayes statistic is a measure of systematic variation -variation beyond chance. A value different to 0 would represent systematic variation; as for its magnitude, the higher the value the more the systematic variation

ISSN: 1471-2288