Individual units rather than entire hospital as the basis for improvement: the example of two Methicillin resistant Staphylococcus aureus cohort studies

BACKGROUND: Two MRSA surveillance components exist within the German national nosocomial infection surveillance system KISS: one for the whole hospital (i.e. only hospital based data and no rates for individual units) and one for ICU-based data (rates for each individual ICU). The objective of this study was to analyze which surveillance system (a hospital based or a unit based) leads to a greater decrease in incidence density of nosocomial MRSA METHODS: Two cohort studies of surveillance data were used: Data from a total of 224 hospitals and 359 ICUs in the period from 2004 to 2009. Development over time was described first for both surveillance systems. In a second step only data were analyzed from those hospitals/ICUs with continuous participation for at least four years. Incidence rate ratios (IRR) with 95% confidence intervals were calculated to compare incidence densities between different time intervals. RESULTS: In the baseline year the mean MRSA incidence density of hospital acquired MRSA cases was 0.25 and the mean incidence density of ICU-acquired MRSA was 1.25 per 1000 patient days. No decrease in hospital-acquired MRSA rates was found in a total of 111 hospitals with continuous participation in the hospital- based system. However, in 159 ICUs with continuous participation in the unit-based system, a significant decrease of 29% in ICU-acquired MRSA was identified. CONCLUSIONS: A unit-based approach of surveillance and feedback seems to be more successful in decreasing nosocomial MRSA rates, compared to a hospital-based approach. Therefore each surveillance system should provide unit-based data to stimulate activities on the unit level.