Can emergency physicians identify a high mortality subgroup of patients with sepsis: role of procalcitonin.

To assess the potential role of procalcitonin and tumor necrosis factor-alpha, interleukin-6 and interleukin-8, in the prognosis of patients with sepsis. Prospective study. The emergency unit of a teaching hospital. We included 131 patients with sepsis: 15 (12%) with septic shock, 20 (15%) with severe sepsis and 96 (73%) with sepsis. Out of the 131 patients, 112 (85.5%) survived and 19 (14.5%) died. These two groups of patients differed with regard to simplified acute physiology score II, severity of infectious disease and underlying disease, bacteremia and type of microorganisms. The mean serum levels of tumor necrosis factor, interleukin-6, interleukin-8, procalcitonin and lactates at study entry were higher in nonsurvivors than in survivors. Multivariate regression analysis showed the most significant of these variables to be serum procalcitonin level (P=0.0007), simplified acute physiology score II (P=0.03) and serum lactate level (P=0.03). Using a model incorporating these three variables, with a cut-off value corresponding to a 15% probability of predicting mortality, death could be correctly predicted in 99.5% of cases and survival in 95%. This cut-off value allowed us to maximize the prediction of death. When serum procalcitonin levels were not taken into account, the best model included simplified acute physiology score II and serum lactate and interleukin-6 levels, but the rate of correct prediction of death then dropped to 84%. Stepwise multivariate logistic regression analysis showed serum procalcitonin level to be a valuable marker of sepsis severity, compared with the 15 other clinical, biochemical and bacteriologic variables tested.