Prediction of mortality in febrile medical patients: How useful are systemic inflammatory response syndrome and sepsis criteria?

OBJECTIVE

The aim was to evaluate demographic, clinical, and laboratory variables in febrile patients, with or without a microbiologically confirmed infection, for prediction of death, in comparison to the systemic inflammatory response syndrome (SIRS) and its criteria, such as abnormal temperature, tachycardia, tachypnea, and abnormal WBC count, and to sepsis, that includes SIRS and an infection.

METHODS

A prospective cohort study.

METHODS

Department of internal medicine at a university hospital.

METHODS

In 300 consecutive, hospitalized medical patients with new onset of fever, demographic, clinical, and laboratory variables were obtained during the 2 days after inclusion, while microbiological results for a follow-up period of 7 days were collected. Patients were followed up for survival or death, up to a maximum of 28 days after inclusion.

RESULTS

Of all patients, 95% had SIRS, 44% had sepsis with a microbiologically confirmed infection, and 9% died. A model with a set of variables all significantly (p<0.01) contributing to the prediction of mortality was derived. The set included the presence of hospital-acquired fever, the peak respiratory rate, the nadir score on the Glasgow coma scale, and the nadir albumin plasma level within the first 2 days after inclusion. This set of variables predicted mortality for febrile patients with microbiologically confirmed infection even better. The predictive values for mortality of SIRS and sepsis were less than that of our set of variables.

CONCLUSIONS

In comparison to SIRS and sepsis, the new set of variables predicted mortality better for all patients with fever and also for those with microbiologically confirmed infection only. This type of effort may help in refining definitions of SIRS and sepsis, based on prognostically important demographic, clinical, and laboratory variables that are easily obtainable at the bedside.