ObjectiveShock in the course of fever is likely caused by septic shock. Because septic shock carries a high mortality rate, early recognition could benefit the patient. We tried to predict the development of shock in medical patients with fever and a clinical infection, on the basis of clinical and microbiological information, and to evaluate the role therein of systemic inflammatory response syndrome (SIRS) criteria: abnormal body temperature, tachycardia, tachypnea, and abnormal white blood cell counts.DesignProspective observational study.SettingDepartment of Internal Medicine at a university hospital.PatientsPatients were 212 consecutive medical patients with newly onset fever (temperature, >38.0°C axillary or >38.3°C rectally) and a clinical source of infection.Measurements and Main ResultsOf the 212 patients enrolled, 14 developed shock (i.e., a decrease in systolic arterial blood pressure of >40 mm Hg) during a maximum follow-up period of 7 days after inclusion. In univariate analyses, advanced age, prior urogenital disease, an abdominal source, nosocomial infections, and bacteremia predisposed patients to shock (p< .05). For clinical variables, obtained daily for 2 days after inclusion, a low performance (p< .001), the peak respiratory rate (p< .05), the peak heart rate (p< .05), the nadir score on the Glasgow Coma Scale (p< .005), the peak and nadir white blood cell counts (p< .005), and the nadir albumin (p< .01) and peak creatinine concentrations in blood (p< .001) predicted shock development. In multivariate analysis, the presence of bacteremia, the peak respiratory rate, the nadir Glasgow Coma Scale score, and the peak white blood cell count positively and the peak erythrocyte sedimentation rate negatively contributed to prediction of shock development. In contrast, SIRS had less predictive value, mainly because of lack of predictive value of peak heart rate and temperature in multivariate models.ConclusionIn febrile medical patients with a clinical infection, the development of shock involves an interaction between circulating microbial products and the host response, which can be recognized clinically by variables easily obtained at the bedside and partly different from the set used to define SIRS.