Sepsis and meningitis in hospitalized children: performance of clinical signs and their prediction rules in a case-control study.

OBJECTIVE: Feverish illness is a common presentation to acute pediatric services. Clinical staff faces the challenge of differentiating the few children with meningitis or sepsis from the majority with self-limiting illness. We aimed to determine the diagnostic value of clinical features and their prediction rules (CPR) for identifying children with sepsis or meningitis among those children admitted to a District General Hospital with acute febrile illness.

METHODS: Acutely ill children admitted to a District General Hospital in England were included in this case-control study between 2000 and 2005. We examined the diagnostic accuracy of individual clinical signs and 6 CPRs, including the National Institute for Clinical Excellence "traffic light" system, to determine clinical utility in identifying children with a diagnosis of sepsis or meningitis.

RESULTS: Loss of consciousness, prolonged capillary refill, decreased alertness, respiratory effort, and the physician's illness assessment had high positive likelihood ratios (9-114), although with wide confidence intervals, to rule in sepsis or meningitis. The National Institute for Clinical Excellence traffic light system, the modified Yale Observation Scale, and the Pediatric Advanced Warning Score performed poorly with positive likelihood ratios ranging from 1 to 3.

CONCLUSIONS: The pediatrician's overall illness assessment was the most useful feature to rule in sepsis or meningitis in these hospitalized children. Clinical prediction rules did not effectively rule in sepsis or meningitis. The modified Yale Observation Scale should be used with caution. Single clinical signs could complement these scores to rule in sepsis or meningitis. Further research is needed to validate these CPRs.