Toward an epidemiology and natural history of SIRS (systemic inflammatory response syndrome)

BACKGROUND: New definitions for sepsis and the systemic inflammatory response syndrome (SIRS) have been established. Comparatively little is known, however, about the types of patients who will be included within these new definitions.
OBJECTIVES: To determine what is--and what is not--known about the epidemiology and natural history of severe sepsis and SIRS.
DESIGN: A comparative analysis of patient characteristics in the Methylprednisolone, Veterans Administration Systemic Sepsis, HA-1A, and E5 studies.
RESULTS: At least 15% of patients in these studies had no documented infection; the proportion of all patients with severe SIRS and no documented infection is probably higher. Even among patients with presumed infection, less than half had bacteremia, and only about half had gram-negative infection or shock. The difference in the mean mortality rate of the combined studies at 14 days was 26%, while at 1 month it was 42%. Gram-negative sepsis and gram-positive sepsis seem to have similar mortality rates. Whether shock increases 30-day mortality is unclear.
CONCLUSIONS: Patients with severe SIRS should not be assumed to have gram-negative infection; furthermore, data derived from studies of patients with gram-negative infection should be applied cautiously to all patients with SIRS. Studies of patients with sepsis or SIRS should include at least a 1-month follow-up if mortality is an end point. More consistent definitions of these disorders should permit more effective comparisons across studies.