[Epidemiologic surveillance and evaluation of meningitis hospital care]

To analyze the clinical-epidemiological profile and in-hospital death predictors of infectious meningitis patients admitted to a public hospital.

There were studied 694 cases of meningitis notified and investigated by a public hospital's epidemiology service from 1986 to 2002, using the National Information System of Notifiable Diseases (SINAN) as part of the local routine of epidemiologic surveillance. Statistics analysis included multivariate logistic regression.

The most frequent etiologies were: cryptococcal (12.3%; case-fatality =37.7%); meningococcal (8.7%; fatality =13.3%); pneumococcal (7.2%; fatality =46%); tuberculous (6.1%; fatality =40.5%); staphylococcal (5.2%; fatality =38.9%), viral (5.5%; fatality =7.9%); Haemophilus (2.9%; fatality =20%). The proportion of cases of non-specified etiology was 38.8% (fatality =36%) and 17.3% were associated to HIV infection. It was found that 27.1% were nosocomial meningitis and 9.2% of the surviving cases had sequelae. The logistic regression model identified the following death predictors of infectious meningitis: etiology (reference: viral category) -- tuberculous, cryptococcal, staphylococcal, meningococcal, non-specified, other Gram-negative, Candida and pneumococcal; HIV co-infection; coma. Fever, vomiting and neck stiffness were associated to a lower odds of death.

The high proportion of non-specified etiology and high case-fatality may reflect problems in the hospital care process and/or case selection. The epidemiologic surveillance system operating at the hospital level was able to feedback the services with clinical indicators. The use of SINAN at the local level was considered useful and pertinent.