Introduction: As the number of nosocomial methicillin-resistant Staphylococcus aureus (MRSA) infections in German intensive care units increases, the problem of MRSA infection as such is becoming ever more serious. The aim of this study was to investigate whether mortality rates from nosocomial MRSA pneumonia and primary bloodstream infections (BSI) differ significantly from those of nosocomial pneumonia and primary BSI caused by methicillin-susceptible S. aureus (MSSA).
Methods: For the analysis data from the ICU component of the German nosocomial infection surveillance system (KISS) were used (January 1997 to June 2002). To identify mortality risk factors a logistic regression analysis with step-wise variable selection was conducted including all cases of nosocomial S. aureus pneumonia and primary BSI. The possible risk factors that were evaluated were age > median, male gender, time in the ICU before infection > median, type of ICU, type and size of hospital, intubation, CVC use, total parenteral nutrition, year of investigation, infection caused by MRSA.
Results: Data from 274 ICUs and 505,487 ICU patients were recorded and a total of 6,888 cases of nosocomial pneumonia and 2,357 cases of primary BSI identified, of which 1,851 cases of S. aureus pneumonia and 378 cases of S. aureus primary BSI were considered for analysis. 59 of the 349 patients with MRSA pneumonia (16.9%) and 105 of the 1,502 patients with MSSA pneumonia (7.0%) died. 16 of the 95 patients with primary MRSA BSI (16.8%) and 17 of the 283 patients with primary MSSA BSI died (6.0%). Four factors were significantly associated with mortality from S. aureus pneumonia, one of them being pneumonia caused by MRSA (OR = 2.62; CI95 1.69-4.02). Only MRSA was significantly associated with death from S. aureus primary BSI (OR = 3.84; CI95 1.51-10.2).
Conclusion: Nosocomial pneumonia and primary BSI from MRSA may be associated with death, but the cause-effect relationship of severity of illness and MRSA remains to be determined due to the limitations of surveillance data.