Non-specific effects of vaccines in developing countriesBMJ 2000; 321 doi: https://doi.org/10.1136/bmj.321.7274.1423 (Published 09 December 2000) Cite this as: BMJ 2000;321:1423
We need evidence about the effect of vaccines on mortality from all causes
- Frank Shann, director ()
Papers p 1435
Immunisation has led to spectacular reductions in mortality in both developed and developing countries.1 However, we know too little about the overall effect of vaccines. We have taken vaccines and schedules that are effective in developed countries with low levels of childhood mortality and used them in populations with high death rates without studying their effect on total mortality.
In this week's BMJ Kristensen and colleagues report some startling results from a prospective cohort study in Guinea-Bissau, which was performed in remarkably difficult circumstances (p 1435).2 Their findings show that both BCG and measles vaccines halved child mortality—a spectacular effect that should be exploited to reduce substantially the number of children dying in developing countries. However, the combination of diphtheria, pertussis, and tetanus (DPT) and polio vaccines seemed to increase mortality (mortality ratio 1.84 (95% confidence interval 1.10 to 3.10)).
The increase in mortality from the DPT and polio vaccines is worrying, but these are preliminary findings. For ethical reasons, this was a follow up study and not a controlled trial and, understandably, the response rate was low (66%). In addition, the effect of this immunisation on mortality only just reached statistical significance, with a lower 95% confidence limit of 1.10 (where 1.00 is no effect). Mortality was 4.8% in 1295 children who received DPT and polio vaccines and 4.0% in 1822 controls. The overall effect of giving BCG in addition to DPT, and polio vaccines was to lower mortality, with a mortality ratio of 0.74 (0.53 to 1.03).
It is important to stress that the Guinea-Bissau study did not compare mortality in a region where DPT vaccine was given with mortality in a region where it was not—it compared the outcome of immunisation that began in early infancy with the outcome when immunisation was delayed. Most children in the study area were eventually given DPT vaccine, and mortality from pertussis is negligible in regions with widespread immunisation, probably because of herd immunity.3 This study provides no information about the consequences of stopping DPT immunisation in a high mortality area: that would be a disaster, as the incidence of pertussis, and child mortality, would increase dramatically.3 Similarly, this study does not cast doubt on the use of DPT vaccine in developed countries, where child mortality is low and acellular pertussis vaccine is used rather than the old whole cell vaccine.
Assuming that it is a reliable estimate, what does the mortality ratio of 1.84 for DPT and polio vaccines mean? It does not mean that giving the vaccines increases mortality by 84%. Rather, it implies that if we compared an area where every child was immunised (and herd immunity was induced) with an area where no child was immunised, the reduction in mortality in the immunised children would be less than expected—the large decrease in the number of deaths from diphtheria, pertussis, tetanus, and polio would be partly offset by increased mortality from other causes. On the other hand, the reduction in mortality from BCG and measles immunisation would be greater than the number of deaths caused by tuberculosis and measles.
There is a clear message from this study: in areas of high mortality, vaccines may have substantial effects on mortality from all causes through non-specific effects on deaths from diseases other than those targeted by the vaccines. There is strong evidence that measles immunisation lowers mortality through non-specific effects, and this association is unlikely to be due to confounding. 4 5 In addition to the data from Guinea-Bissau, other studies suggest that non-specific effects may cause BCG to lower mortality from all causes 6 7 and cause DPT vaccines to have less beneficial effects than expected in high mortality areas. 4 6
The expanded programme on immunisation began in 1974. It was called expanded because polio and measles vaccines were added to BCG and DPT vaccines and because the proportion of children in developing countries who were immunised was substantially increased (from less than 5%). It is imperative that we obtain more evidence about the effects of these vaccines on mortality from all causes in developing countries. It would be unethical to do controlled trials of the current vaccines used in the programme, so we will have to use cohort and case-control studies. In addition, new vaccines must not be introduced into developing countries without controlled trials of their effect on mortality from all causes. In areas where mortality is high, it should no longer be acceptable to use surrogate measures such as disease specific mortality, disease specific morbidity, or, worst of all, indirect evidence such as antibody responses. Lack of evidence about the effect of vaccines on mortality from all causes can lead to serious errors: examples include the use of high titre measles vaccine (which resulted in a higher mortality than standard titre vaccine in girls),4 failure to investigate the role of polysaccharide pneumococcal vaccine in children in high mortality areas (because of poor antibody responses and poor protection against otitis media in children in developed countries),8 and failure to appreciate the possibility of substantial non-specific effects from measles, BCG, and DPT vaccines.
The Guinea-Bissau investigators speculate that BCG and measles vaccines may have beneficial non-specific effects because they stimulate Th1 immunity and that DPT vaccine may have adverse non-specific effects because the aluminium adjuvant stimulates Th2 immunity.2 It has been suggested that reduced exposure to BCG and other microbes combined with increased exposure to aluminium, DPT vaccine, and other Th2 adjuvants may have contributed to the apparent increase in allergic disease in developed countries.9 10
Some will argue that the Guinea-Bissau data should not have been published, because publication might damage immunisation programmes. However, it would be inappropriate to suppress this evidence—just as it would be inappropriate to withhold DPT vaccine on the basis of these preliminary results. The appropriate response is to extract as much information as we can from this important study, to perform similar cohort studies in other high mortality areas, and to ensure that new vaccines are introduced into developing countries only after randomised trials of their effect on mortality from all causes. We need accurate information about the effect of the vaccines in the Expanded Programme on Immunisation on mortality from all causes in children in developing countries, and we need it soon.