Tool limits unnecessary hospitalizations, aggressive antibiotic therapy

Photo of Dr. Kuppermann with pediatric patient
Research by pediatric emergency medicine specialist Nathan Kuppermann is improving the care of patients at UC Davis and at emergency rooms throughout the nation.

A simple algorithm developed at Children's Hospital Boston and the University of California, Davis, Health System can help doctors rapidly distinguish infants and children with viral meningitis from those with bacterial meningitis — potentially reducing unnecessary hospital admissions and prolonged antibiotic treatment.

The finding, published in the January 3, 2007, issue of the Journal of the American Medical Association, is based on the review of medical charts from thousands of children diagnosed with meningitis at 20 academic medical centers across the United States from January 2001 to June 2004.

“Using a multi-center network of U.S. hospitals that care for acutely ill children, we showed our algorithm effectively identified children with meningitis who are at very low risk of having bacterial meningitis,” says Lise Nigrovic, an emergency medicine specialist at Children's Hospital Boston and principal investigator of the study. “The finding is important because it gives emergency room physicians nationwide a tool to guide their decision-making when caring for children with suspected meningitis, a serious and potentially life-threatening infection.”

Meningitis is an inflammation of the membranes (meninges) surrounding the brain and spinal cord, and is usually initially recognized by a higher number of white blood cells in the spinal fluid than normal. While viral infections cause most meningitis cases, about one in 25 cases is caused by bacterial or fungal infections, which yield the most severe illness. Although the best course of treatment depends on identifying the type of infection, definitive test results can take 24 to 72 hours to process, and children are often admitted to hospitals and started on antibiotics while physicians wait for these results.

“Viral or 'aseptic' meningitis beyond young infancy is usually a mild disease, while meningitis caused by bacteria can cause serious illness and death,” says  Nathan Kuppermann, professor and chair of emergency medicine at UC Davis Medical Center who developed the algorithm in 2002 with Nigrovic and Richard Malley, also of Children's Hospital Boston.

“Even though most patients will turn out to have viral meningitis, pediatric emergency department physicians usually hospitalize any child with meningitis to receive broad-spectrum antibiotics while waiting two to three days for the bacterial culture results. The ability to identify those children who are at very low risk of bacterial meningitis and can be considered for management on an outpatient basis will avoid unnecessary hospitalization and aggressive antibiotic therapy,” said Kuppermann.

Known as the Bacterial Meningitis Score, the algorithm uses easily obtained results from routine tests of blood and fluid, including spinal fluid, Gram stain results, neutrophil count and protein concentration, as well as bloodstream neutrophil count. It also takes into account other factors, such as whether the child has had a seizure during the current illness.

“The Bacterial Meningitis Score accurately identified patients at very low risk of bacterial meningitis, misclassifying only 0.1 percent of patients categorized as very low risk,” said Nigrovic. “To our knowledge, this is the first bacterial meningitis prediction model to be both externally validated and studied at multiple centers in the era of widespread conjugate pneumococcal immunization.”

Pneumococcus is the most important cause of bacterial meningitis. Since 2000, the heptavalent pneumococcal conjugate vaccine, which protects against this bacterium, has been recommended for all children under age 2 who are at highest risk for serious disease from pneumococcal infection. In the United States, the vaccine's introduction has greatly reduced the incidence of the disease.

“Any major change in the epidemiology of a disease has the potential to affect the performance of a prediction rule,” she said. “We were pleased to find that our prediction tool remains valid now that most infants and toddlers have received the pneumococcal vaccine.”

The researchers caution physicians against using the assessment tool for infants younger than 2 months of age, who are at greater risk of bacterial meningitis, and for whom the algorithm was slightly less accurate. They also caution that the tool should not be used to guide decision-making for children who have already received pretreatment with antibiotics before the spinal tap. Nevertheless, widespread implementation of this algorithm could result in a substantial decrease in unnecessary hospitalizations of children who are at very low risk of bacterial meningitis. The researchers plan to further study the best implementation of the rule.

The study was sponsored by the Pediatric Emergency Medicine Collaborative Research Committee of the American Academy of Pediatrics and supported by the Ambulatory Pediatric Association Young Investigator Grant and the National Research Service Award.