|| Checking for direct PDF access through Ovid
To determine the location of acquisition, timing, and outcomes associated with severe sepsis in community and teaching hospital critical care units.Prospective, observational study.Twelve Canadian community and teaching hospital critical care units.All patients admitted between March 17, 2003, and November 30, 2004 to the study critical care units with at least a 24-hr length of stay or severe sepsis identified during the first 24 hrs.Daily monitoring for severe sepsis.We recorded data describing characteristics of patients, infections, systemic responses, and organ dysfunction. Severe sepsis occurred in 1238 patients (overall rate, 19.0%; range, 8.2%–35.3%). Hospital mortality was 38.1% (95% confidence interval [CI]: 35.4–40.8). Median intensive care unit length of stay was 10.3 days (interquartile range: 5.5, 17.9). Variables associated with mortality in multivariable analysis included age (odds ratio [OR] by decade 1.50; 95% CI: 1.36–1.65), acquisition location of severe sepsis (with community as the reference—hospital [OR: 1.69; CI: 1.16–2.46], early intensive care unit [OR: 2.15; CI: 1.42–3.25], late intensive care unit [OR: 2.65; CI: 1.82–3.87]), late intensive care unit (OR: 2.65; CI: 1.82–3.87), any comorbidity (OR: 1.42; CI: 1.04–1.93), chronic renal failure (OR: 2.03; CI: 1.10–3.76), oliguria (OR: 1.34; CI: 1.02–1.76), thrombocytopenia (OR: 2.12; CI: 1.43–3.13), metabolic acidosis (OR: 1.54; CI: 1.13–2.10), Multiple Organ Dysfunction Score (OR: 1.15; CI: 1.09–1.21) and Acute Physiology and Chronic Health Evaluation II predicted risk (OR: 3.75; CI: 2.08–6.76).These data confirm that sepsis is common and has high mortality in general intensive care unit populations. Our results can inform healthcare system planning and clinical study designs. Modifiable variables associated with worse outcomes, such as nosocomial infection (hospital acquisition), and metabolic acidosis indicate potential targets for quality improvement initiatives that could decrease mortality and morbidity.