Re: Association of BCG, DTP, and measles containing vaccines with childhood mortality: systematic review
In their systematic review, Higgins and colleagues stress that the evidence about whole-cell diphtheria-tetanus-pertussis vaccine (DTP) comes from observational studies with a high risk of bias, but they do not emphasise that all the major biases (frailty bias, survival bias and classification bias) lead to under-estimation of harm from DTP. The available evidence suggests that neither starting surveillance after vaccination nor censoring for measles vaccination after DTP causes over-estimation of harm from DTP. Importantly, bias cannot explain why mortality is reduced by BCG, then increased by DTP, then reduced again by measles vaccine (MV) within individual studies where there are the same biases for all three vaccines (review Figure 8).
The reviewers estimate that DTP is associated with a 38% (95% CI -8% to 108%) increase in all-cause mortality. Any suggestion of increased mortality is very worrying, especially as the actual increase is almost certainly larger because of frailty bias and survival bias. An ingenious recent analysis of observational studies of DTP suggests using a “bias index” defined as mortality in children classified as never vaccinated divided by mortality in children classified as having received any vaccine; a ratio greater than 2.0 suggests severe frailty bias or survival bias or both. The review included 10 studies in its analysis of DTP; if we exclude the three studies with a bias index greater than 2.0 (Papua New Guinea, Bangladesh and Burkina Faso), DTP is associated with a 91% (46% to 150%) increase in all-cause mortality (Figure). The Papua New Guinea study had an extraordinarily high mortality rate of 233 per 1000 person-years in unvaccinated children compared to only 31 per 1000 person-years in vaccinated children, and a bias index of 7.5; the Bangladesh data come from an unpublished analysis which the authors emphasise should not be used to compare DTP-after-BCG to BCG, and the bias index is 3.4; the Burkina Faso study has survival bias with a bias index of 2.3, and a substantial proportion of children received BCG with DTP or received MV after DTP.
In high-mortality countries, WHO recommends administration of BCG at birth, DTP at 6, 10 and 14 weeks, MV at 9 months, and a booster dose of DTP at 18 months. The reviewers suggest these recommendations should not change, yet their Figure 5 shows that mortality increases by 92% (25% to 194%) when DTP is given after BCG (as recommended by WHO) compared to giving DTP with BCG, and that mortality increases by 116% (25% to 274%) when DTP is given after MV (as recommended by WHO for the booster dose).
At the very least, these findings suggest that DTP should no longer be given with or after MV until randomised trials have shown this to be safe, and the benefit from giving a second dose of BCG with the last priming dose of DTP at 14 weeks should be tested urgently in randomised trials.
WHO recommends DTP as the most recent vaccine for 84% of the time between 1 month and 5 years of age, and there were 1.34 million post-neonatal under 5 deaths in the least developed countries in 2015. If all children follow the recommended schedule and DTP increases mortality by 38% (91% in studies with a bias index < 2.0), DTP is responsible for 310,000 (or 540,000) deaths annually in these countries alone. It is 2½ years since the systematic review’s findings were presented to WHO, and there have been no changes to the recommendations and no trials started.
1 Higgins J, Soares-Weiser K, López-López J, et al. Association of BCG, DTP, and measles containing vaccines with childhood mortality: systematic review. BMJ 2016;355:i5170.
2 Aaby P, Ravn H, Benn CS. The WHO Review of the Possible Non-Specific Effects of Diphtheria-Tetanus-Pertussis Vaccine. Pediatr Infect Dis J 2016;35:1247–57.
3 UNICEF. The state of the world’s children 2016: a fair chance every child. New York: UNICEF 2016. http://www.unicef.org/publications/files/UNICEF_SOWC_2016.pdf (accessed 16 Oct2016). Statistical Table 1.
4 Meeting of the Strategic Advisory Group of Experts on immunization, April 2014 – conclusions and recommendations. Wkly Epidemiol Rec 2014;89:233–5.
Figure. Random effects forest plot for DTP and all cause mortality for observational studies with a bias index < 2.0 and a bias index > 2.0 (ES = effect size, BI = bias index, CC = case control study).
Competing interests: No competing interests