Transmission-associated nosocomial infections: Prolongation of intensive care unit stay and risk factor analysis using multistate models
published online 31 January 2008.
Background
Almost all studies investigating prolongation of stay because of nosocomial infections (NI) took into account all cases of NI, regardless whether they were associated with transmission of nosocomial pathogens (and therefore preventable) or not. We investigated the prolongation of intensive care unit (ICU) length of stay (LOS) because of transmission-associated NI (TANI) in a prospective study on 5 ICUs with normal NI rates over an 18-month period.
Methods
All clinical isolates and nose swabs were collected at admission. Pulsed-field gel electrophoresis and arbitrary primed polymerase length polymorphism methods were used for identifying transmissions. A NI was considered as TANI if indistinguishable pathogens were found in patients treated in temporal proximity and in the same ICU. Statistically, the temporal dynamics of the data were described by a multistate model.
Results
One thousand eight hundred seventy-six patients were observed for development of NI using the Centers for Disease Control and Prevention definitions; 341 patients acquired at least 1 NI (15.1 NI per 1000 patient-days), and 30 of these (8.8%) were considered to be infected with TANI. The influence of all NI as a time-dependent covariate in a proportional hazards model was significant (P < .0001) with an extra LOS of 5.3 days (±standard error, 1.6), as was the case for TANI alone (P = .02) with an extra LOS of 11.4 days (±7.3). However, TANI showed no significant effect compared with other NI (P = .23). The multivariate risk factor analysis showed that colostomy significantly increased the TANI hazard ratio (HR, 3.8; 95% CI: 1.0-14.3; P = .047) but did not significantly alter the HR for discharge or death without prior NI or for other NI.
Conclusion
TANI occur in particular in patients with many manipulations and TANI significantly prolong ICU stay.
aInstitute of Medical Biometry and Medical Informatics, University Medical Center, and Freiburg Centre for Data Analysis and Modelling, University of Freiburg, Freiburg, Germany
bInstitute of Medical Microbiology and Hospital Epidemiology, Hannover Medical School, Hannover, Germany
cNational Institute for Public Health and Environment, Bilthoven, The Netherlands
dDepartment of Medical Microbiology, Groningen University Medical Center, Groningen, The Netherlands
eInstitute of Hygiene and Environmental Medicine, Charité, Ai University of Medicine, Berlin, Germany
Address correspondence to Jan Beyersmann, PhD, Institute of Medical Biometry and Informatics, University Medical Center Freiburg, Stefan-Meier-Strasse 26, D-79104 Freiburg, Germany.
ABSTRACT