No difference between short-course and long-course antibiotics for bacterial meningitis in children, but available evidence limited.

Until the early twentieth century the prognosis for patients with acute bacterial meningitis was dismal. Since then, improvements in public health, the discovery of effective antimicrobial agents, the implementation of childhood vaccination programmes and the development of highly specialised acute care facilities have had a dramatic impact on the mortality and morbidity associated with this disease.1 The duration of antibiotic therapy for patients with bacterial meningitis has often been based more on tradition than on data.2 Potentially, shorter courses of antibiotic therapy would decrease the duration of hospitalisation, the number of adverse events and healthcare costs and could conserve the limited resources available in low-income countries. Karageorgopoulos and colleagues performed a metaanalysis that included randomised controlled trials on antibiotic therapy duration in community-acquired bacterial meningitis.3 Studies were included if the randomisation arms had identical antimicrobial treatment, daily antibiotic dosage and mode of administration. Therapy was considered to be of short duration if it lasted for 2–7 days and of long duration if it exceeded 7 days. The difference between short-course and long-course regimens was set at a minimum of 2 days. Five randomised open-label trials were included, with a total 426 patients. All included patients were children, and all were treated with ceftriaxone. Inclusion and exclusion criteria for randomisation varied considerably between studies: two studies included only patients with an ill-defi ned “initial rapid recovery”, one trial excluded patients with a positive cerebrospinal fl uid culture from repeated sampling 15 h or more after initiation of therapy, and one trial excluded patients with concomitant adjunctive therapies. Data on how many patients were evaluated for inclusion and how many were excluded were not presented. Three trials did not report the use of adjunctive therapy. In one trial adjunctive dexamethasone was administered to a substantial number of patients, but its use was unevenly distributed between treatment groups. The initial outcome evaluation was performed on the day antibiotic therapy ended, and therefore patients in the short-course group were evaluated earlier in the disease course. Late follow-up was performed on the day of discharge to 6 months after discharge. The meta-analysis showed no differences between treatment groups in terms of clinical success at the end of therapy, mortality, persistence of cerebrospinal fl uid abnormalities, adverse events, secondary nosocomial infections, hearing impairment or long-term neurological complications. Duration of hospitalisation was evaluated in two trials, and short-course therapy was associated with a decrease in hospital stay of 2 days. This meta-analysis will not change current practice. Pooled available data did not show a difference in clinical effectiveness and safety between short-term and long-term therapy, but several biases may have diminished the reliability of the results. The fi rst confounding factor is selection bias. The studies included had exceptionally low case-fatality rates (0–1%). Mortality for childhood bacterial meningitis in previously reported studies ranged from 8% to 20%.4 Inclusion of patients with a less severe illness will probably underestimate a potential difference between therapies. Second, many patients were withdrawn after the randomisation process, often for unknown reasons. Finally, the external validity of the meta-analysis is poor because of the changed epidemiology of meningitis. A large proportion of included patients had meningitis due to Haemophilus infl uenzae type b, a bacterium that has been virtually eradicated after routine vaccination. Another large proportion of included patients had pneumococcal meningitis. Over recent years, drug-resistant pneumococci have become an emerging problem worldwide.5 The external validity is further reduced by the recent introduction of adjunctive dexamethasone therapy, which may impede antimicrobial penetration into the cerebrospinal fl uid. The fi ndings of this meta-analysis support further research on shortened antimicrobial therapy in selected patients with bacterial meningitis. In Diederik van de Beek and Matthijs C Brouwer