The step from a voluntary to a mandatory national nosocomial infection surveillance system: the influence on infection rates and surveillance effect

Table 3 Results of univariable and multivariable analysis comparing primary BSI rates the 3 ^rd and 1 ^st year of participation in NEO-KISS

HCAI infection rate	Old/voluntary participants (N = 26, starting in NEO-KISS January 2000- December 2002)		New/mandatory participants (N = 95, starting in NEO-KISS January-December 2006)
	univariable analysis pooled data 3^rdvs. 1^st year	multivariable analysis adjusted effect measures	univariable analysis pooled data 3^rdvs. 1^st year	multivariable analysis adjusted effect measures
BSI Incidence density (per 1000 patient days)	RR = 0.79; (CI95 0.68-0.91); p = 0.001	IRR^a = 0.78; (CI95 0.66-0.93); p = 0.005	RR =0.79; (CI95 0.69-0.90); p < 0.001	IRR^a = 0.81; (CI95 0.68-0.97); p = 0.019

CI95, 95% confidence interval; RR, relative risk; IRR, adjusted incidence rate ratio; p, p-value;
^a Poisson regression models with the outcome number BSI and log number patient days as offset parameter calculated by generalized estimating equation (GEE) model which account for clustering effect within a single NICU, consider the following parameters: birth weight (5 categories, 250gram steps), gestational age (4 categories, <27/27-28/29-30/>30 weeks), sex, mode of delivery (sectio) and surveillance endpoint (3 categories: 1800 g/transfer/died).

ISSN: 2047-2994