—To examine the incidence, risk factors, and outcome of severe sepsis in intensive care unit (ICU) patients.
Design and Setting.
—Inception cohort study from a 2-month prospective survey of 11 828 consecutive admissions to 170 adult ICUs of public hospitals in France.
—Patients meeting clinical criteria for severe sepsis were included and classified as having documented infection (ie, documented severe sepsis, n=742), or a clinical diagnosis of infection without microbiological documentation (ie, culture-negative severe sepsis, n=310).
Main Outcome Measures.
—Hospital and 28-day mortality after severe sepsis.
—Clinically suspected sepsis and confirmed severe sepsis occurred in 9.0 (95% confidence interval [CI], 8.5 to 9.5) and 6.3 (95% CI, 5.8 to 6.7) of 100 ICU admissions, respectively. The 28-day mortality was 56% (95% CI, 52% to 60%) in patients with severe sepsis, and 60% (95% CI, 55% to 66%) in those with culture-negative severe sepsis. Major determinants of both early (<3 days) and secondary deaths in the whole cohort were the Simplified Acute Physiology Score (SAPS) II and the number of acute organ system failures. Other risk factors for early death included a low arterial blood pH (<7.33) (P<.001) and shock (P=.03), whereas secondary deaths were associated with the admission category (P<.001), a rapidly or ultimately fatal underlying disease (P<.001), a preexisting liver (P=.01) or cardiovascular (P=.002) insufficiency, hypothermia (P=.02), thrombocytopenia (P=.01), and multiple sources of infection (P=.02). In patients with documented sepsis, bacteremia was associated with early mortality (P=.03).
—Only three of four patients presenting with clinically suspected severe sepsis have documented infection. However, patients with clinically suspected sepsis but without microbiological documentation and patients with documented infection share common risk factors and are at similarly high risk of death. In addition to the severity of illness score, acute organ failures and the characteristics of underlying diseases should be accounted for in stratification of patients and outcome analyses.(JAMA. 1995;274:968-974)