OBJECTIVESTo determine the incidence of sepsis in Japanese intensive care units (ICUs) and to evaluate the impact of sepsis on mortality and length of stay (LOS).

METHODSUsing the JANIS database for the period between June 2002 and June 2004, 21,895 eligible patients aged ≥16 years, hospitalized in 28 participating ICUs for ≥24 hours, were monitored until ICU discharge. Adjusted hazard ratio (HR) with 95% confidence interval (CI) for the incidence of sepsis was calculated using Cox's proportional hazard model. Standardized mortality ratio (SMR) was calculated on the basis of the crude mortality in patients without nosocomial infection (NI) for respective APACHE II categories. Mean LOS for survivors was assessed by two-way analysis of variance with adjustment for APACHE II.

RESULTSSepsis was diagnosed in 450 patients (2.1%), with 228 meeting the definition on ICU admission and 222 during the ICU stay. The overall incidence of sepsis was 1.02/100 admissions or 2.00/1000 patient-days. A significantly higher HR for the incidence of sepsis was found in men (1.54, 95% CI: 1.14-2.07), APACHE II ≥21 (2.92, 95% CI: 1.92-4.44), ventilator use (3.30, 95% CI: 1.98-5.49), and central venous catheter use (3.45, 95% CI: 1.90-6.28). SMR was determined to be 1.18 (95% CI: 0.82-1.21) in NI patients without sepsis and 2.43 (95% CI: 1.88-3.09) in NI patients with sepsis. Mean LOS for survivors was calculated to be 11.8 days (95% CI: 11.3-12.4) in NI patients without sepsis and 15.0 days (95% CI: 13.3-17.0) in NI patients with sepsis compared with 3.8 days (95% CI: 3.8-3.9) in patients without NI.

CONCLUSIONSSepsis is not very common in Japanese ICUs, but its development leads to further increases in mortality and LOS in patients with NI.