The survival benefit of measles immunization may not be explained entirely by the prevention of measles disease: a community study from rural Bangladesh.

OBJECTIVE To examine whether the reduction in childhood mortality after immunization can be explained by the prevention of measles and its long-term effects. METHODS and Data We re-analysed an existing data set from Matlab, Bangladesh. During 1982-1985, measles immunization was used from 9 months of age in half of the study area, and the other half was used as an unvaccinated control area. A total of 8134 immunized children had been matched by age with 8134 non-immunized children; 578 children died during the follow-up period of 3 years. Using these data, we calculated the vaccine effectiveness against death (VED) controlling for significant factors in a matched analysis. In the absence of measles, there should be no difference in mortality between immunized, uninfected children and non-immunized, uninfected children. We therefore calculated VED after the exclusion of all measles cases in the survival analysis. To assess the long-term effects of measles, we compared survival of unvaccinated children after measles disease with children who had not yet contracted measles. RESULTS Prior to immunization and again after 1985, childhood mortality rates were 10% lower in the area that had received immunization. Though measles deaths only constituted 12.4% of the non-accidental deaths, the VED controlling for significant factors was 49% (95% CI: 38-58%). The vaccine was protective against measles death throughout the study, but it also had a marked effect against other causes of death, particularly diarrhoea and oedema. This effect may have been particularly strong in the first 6 months after immunization (VED = 74, 95% CI: 57-84%). The VED was only reduced from 49% to 43% (95% CI: 31-54%) when measles cases were excluded in the survival analysis. Controlling for background factors, mortality among measles cases was increased during the acute phase (0-45 days) (mortality ratio [MR] = 17.35, 95% CI: 11.9-25.3) and in the following 1(1/2) months (MR = 2.35, 95% CI: 0.95-5.84). However, post-measles cases had significantly lower mortality than uninfected, non-immunized children in the following 9 months (MR = 0.40, 95% CI: 0.16, 0.98). CONCLUSIONS The non-randomized character of the original study and the possibility of uncontrolled confounding between the two areas prevent a precise estimate of the effectiveness of measles vaccine, but it is likely to have been substantial. Though there may have been some underreporting of cases of measles, the prevention of measles infection can only explain a limited part of the observed impact of measles immunization in Bangladesh. Furthermore, mortality may be reduced after the acute phase of measles infection. The observations from Bangladesh are consistent with recent research from Africa suggesting that measles immunization may have non-specific beneficial effects on survival.