Current understanding of sepsis.

Sepsis and septic shock represent the thirteenth leading cause of death in the United States. It has been estimated that there are 500,000 new episodes each year, with an associated crude mortality of 35% [1]. Over the last 4 decades the age-adjusted mortality has climbed steadily from 0.5 to 7 per 100,000 episodes [2]. Approximately one-third to one-half of patients who are septic have culture-positive blood, and much of our understanding of this clinical syndrome derives from studies ofnosocomial bacteremia and candidemia. Several approaches have been used to study the direct impact of bloodstream infections-and therefore of sepsis-apart from the impact of patients' underlying diseases. The first includes the use of models such as logistic regression to adjust for underlying illness and to provide estimates of relative contribution to mortality from an individual-patient viewpoint. Risk ratios or odds ratios are provided in such analyses. A risk ratio of three among patients with bloodstream infections suggests that their risk of dying is three times greater than that expected from the underlying diseases alone. The second approach is to examine from a population viewpoint the absolute impact of the infection by subtracting the crude (overall) mortality among tightly matched controls without bloodstream infection from the crude mortality for cases of bloodstream infection. The difference is called the attributable mortality. For example, ifmortality among patients with bloodstream infections is 35% and that among matched controls is 10%, the estimated direct or attributable mortality is 25%. The implication is that five-sevenths of the deaths (25 of 35) are due to the infection and two-sevenths are due to the underlying disease. Previous studies of bloodstream infections have identified Pseudomonas aeruginosa infection, Candida species infection, and polymicrobial infection as independent predictors of death, with estimated risk ratios of approximately three. More recent studies have confirmed these data and suggested that the occur-