Objective: To assess whether antibiotics should be given to all children with measles in communities with a high case fatality rate.
Design: Meta-analysis of randomised controlled trials that compared routine antibiotic prophylaxis with no antibiotic treatment or selective treatment of pneumonia or sepsis.
Subjects: Six trials of children admitted to hospital with measles: five in Glasgow, London, or New York between 1939 and 1954; and one in India in 1967.
Main outcome measures: Incidence of pneumonia or sepsis, and mortality.
Results: All but one of the trials were unblinded, and randomisation was either not described or was by alternate allocation. In four studies, the incidence of pneumonia or sepsis in the control group was similar to that in the antibiotic prophylaxis group; in the other two studies, the incidence of pneumonia or sepsis was unusually high in the control group so these children had a higher complication rate than the antibiotic group. Four of the 764 children given antibiotics died compared with one of the 637 controls (exact odds ratio 4.0, mid-P corrected 95% confidence interval 0.5 to 101.6).
Conclusion: The quality of the trials reviewed was poor, and they provide weak evidence for giving antibiotics to all children with measles. Available evidence suggests that, when mortality from measles is high, all children with measles should be treated with vitamin A but antibiotics should be given only if a child has clinical signs of pneumonia or other evidence of sepsis.