Serum Cortisol: A Predictor of Mortality in Sepsis?

To evaluate the pattern and predictive value of serum cortisol levels in severe septic shock, we studied 20 consecutive critically ill patients with severe sepsis. Serum cortisol levels were measured at time of diagnosis, at 12 and 24 hours, and daily for 5 days. Underlying diagnoses included pneumonia (n = 9), urosepsis (n = 5), peritonitis (n = 4), and cellulitis (n = 2). Two patterns of cortisol secretion were identified. In Group A, serum cortisol levels were greater than 400 nmol/L and gradually decreased (n = 14). In Group B, levels were less than 400 nmol/L and remained below normal (n = 6). In Group A, 2 patients died (14.3%), whereas in Group B, 5 died (83.3%) (p < 0.05). The Simplified Acute Physiology Score (SAPS) was higher, intitial serum cortisol levels were lower, and serum lactate levels were higher in the nonsurvivors. In a logistical regression model, SAPS was the only independent predictor of death; when this variable was removed from the model, decreased initial serum cortisol was also predictive. Increased serum lactate was strongly associated with mortality. There appears to be an association between decreased initial serum cortisol and mortality in patients with severe septic shock.