Plain language summary pending.
Routine phenobarbitone in cerebral malaria is associated with fewer convulsions but possibly more deaths. Further trials with adequate design, more participants, and different doses of anticonvulsant drugs are needed.
Cerebral malaria is a common complication of Plasmodium falciparum infection, and kills over a million people every year. People with cerebral malaria become unconscious, and often have protracted convulsions. It is unclear whether giving anticonvulsant drugs routinely to people with cerebral malaria will improve the outcome of treatment and prevent death.
To evaluate the effect of routine anticonvulsant drugs in people with cerebral malaria.
We searched the Cochrane Infectious Diseases Group Specialized Register (May 2004), CENTRAL (The Cochrane Library Issue 2, 2004), MEDLINE (1966 to May 2004), EMBASE (1988 to May 2004), LILACS (1982 to May 200), Science Citation Index (May 2004), African Index Medicus (1999), reference lists of articles, and research organizations. We also contacted the authors for additional information.
Randomized and quasi-randomized controlled trials of people with cerebral malaria. The trials compared anticonvulsant drugs started on admission to hospital with no anticonvulsant drug or placebo.
Two reviewers independently extracted data from those trials eligible for inclusion. We assessed the risk of bias in the included trials by considering allocation sequence, concealment of allocation, blinding, and inclusion of all randomized participants. We used Review Manager (version 4.1) for the meta-analysis and also explored possible sources of heterogeneity.
Three trials with a total of 573 participants met the inclusion criteria. These trials all compared phenobarbitone with placebo or no treatment. In the two trials with adequate allocation concealment, death was more common in the anticonvulsant group (Risk Ratio 2.0; 95% confidence interval 1.20 to 3.33; fixed effect model). In all three trials, phenobarbitone compared with placebo or no treatment was associated with fewer convulsions (Risk Ratio 0.30; 95% confidence interval 0.19 to 0.45; fixed effect model).