Intended for healthcare professionals

Letters

Routine vaccination and child survival in Guinea-Bissau

BMJ 2001; 322 doi: https://doi.org/10.1136/bmj.322.7282.360 (Published 10 February 2001) Cite this as: BMJ 2001;322:360

Authors' reply to commentary

  1. Peter Aaby, anthropologist (psb{at}sol.gtelecom.gw),
  2. Henrik Jensen, statistician
  1. Bandim Health Project, Danish Epidemiology Science Centre, Apartado 861, Bissau, Guinea-Bissau
  2. Danish Epidemiology Science Centre, Statens Serum Institut, Copenhagen, Denmark
  3. Department of Paediatrics, Royal Children's Hospital, Parkville, Victoria 3051, Australia
  4. Instituto de Saude Coletiva, Federal University of Bahia, Salvador-Bahia, Brazil
  5. Department of Pharmacology, University of Cape Town Health Sciences Faculty, Observatory 7925, South Africa
  6. Bandim Health Project, Danish Epidemiology Science Centre, Apartado 861, Bissau, Guinea-Bissau
  7. Danish Epidemiology Science Centre, Statens Serum Institut, Copenhagen, Denmark

    EDITOR—In his commentary on our paper Fine claims that our observations lack the classic attributes of causality: gradient, strength, and coherence.1

    Firstly, the estimates only just reach conventional significance. However, the important finding was not the significance of the estimates for the individual vaccines but the fact that BCG and diphtheria, pertussis, and tetanus vaccines had opposite effects on mortality—the difference being strongly significant (P=0.005).

    Secondly, the association of diphtheria, pertussis, and tetanus vaccine with increased mortality could be due to higher vaccination coverage and higher mortality among children of young mothers. However, as we stated in our paper, adjustment for background factors had minimal effect on mortality estimates. Given the low prevalence of young mothers and their slightly higher vaccine coverage (table 2), adjustment for maternal age had little effect. Mortality ratios changed from 1.84 (1.10 to 3.10) to 1.82 (1.08 to 3.07) with one dose of diphtheria, pertussis, and tetanus vaccine and from 1.38 (0.73 to 2.61) to 1.37 (0.72 to 2.59) with 2-3 doses.

    Thirdly, our results may be anomalous because mortality declines with age for children who had not been vaccinated with BCG but increases with age among those who were (table 3). These findings are to be expected if diphtheria, pertussis, and tetanus vaccine increases mortality. Exposure to diphtheria, pertussis, and tetanus vaccine increased with age for children vaccinated with BCG (table below), whereas there was little change for those who were not so vaccinated. Presumably children who are not vaccinated with BCG have less contact with immunisation services and are therefore less likely to receive additional doses of diphtheria, pertussis, and tetanus vaccine during follow up.

    Exposure to diphtheria, tetanus, and pertussis (DTP) vaccine by age and vaccination with BCG

    View this table:

    Fourthly, the mortality ratio of 1.84 (1.10 to 3.10) for one dose of diphtheria, pertussis, and tetanus vaccine declined to 1.38 (0.73 to 2.61) for 2-3 doses. This difference is not significant (P=0.250) and we did not discuss it, but we mentioned in the discussion that the estimates were conservative because they were based on initial vaccination status and had to ignore subsequent vaccinations during follow up. Children who had received 2-3 doses of diphtheria, pertussis, and tetanus vaccine were more likely to receive measles vaccine during follow up (51%) than children who initially had received only one (28%) or no dose (13%). When the analysis was censored at 9 months of age to minimise interference from measles vaccinations, the mortality ratios for one dose (1.81 (1.00 to 3.26)) and 2-3 doses (1.75 (0.86 to 3.59)) were similar. We concluded that the vaccines had important non-specific effects on mortality, but we did not claim that the individual vaccines altered mortality. Moreover, the non-specific effects of diphtheria, pertussis, and tetanus vaccine may well have had gradient, strength, and coherence in our study.

    A more detailed version of our response to Fine's commentary is available on bmj.com.2

    References

    1. 1.
    2. 2.

    Lessons can be learnt from this study

    1. Kim Mulholland (mulhollk{at}cryptic.rch.unimelb.edu.au), professor,
    2. Mauricio L Barreto (mauricio{at}ufba.br), professor
    1. Bandim Health Project, Danish Epidemiology Science Centre, Apartado 861, Bissau, Guinea-Bissau
    2. Danish Epidemiology Science Centre, Statens Serum Institut, Copenhagen, Denmark
    3. Department of Paediatrics, Royal Children's Hospital, Parkville, Victoria 3051, Australia
    4. Instituto de Saude Coletiva, Federal University of Bahia, Salvador-Bahia, Brazil
    5. Department of Pharmacology, University of Cape Town Health Sciences Faculty, Observatory 7925, South Africa
    6. Bandim Health Project, Danish Epidemiology Science Centre, Apartado 861, Bissau, Guinea-Bissau
    7. Danish Epidemiology Science Centre, Statens Serum Institut, Copenhagen, Denmark

      EDITOR—The paper by Kristensen et al describes improved survival in infants in Guinea-Bissau who received measles or BCG vaccines but increased mortality in those who received diphtheria, tetanus, and pertussis vaccine.1 At the request of the World Health Organization we visited Guinea-Bissau durng 8-15 October 2000 to review this study.

      We found the methods to be as described in the paper. As data collection is continuing, we were able to accompany the field staff on several visits. We reviewed the data management procedures and found them to be in order. While several potential sources of bias could be hypothesised we were not able to identify any that could invalidate the study. This is a single study, which was not originally designed to address this issue, and the methodological weaknesses have been highlighted in the accompanying editorial and commentary. 1 2 By its characteristics and its surprising findings this study should provoke substantial further investigation of the subject. Although these results should not be used as a basis for changing national or global vaccination policy, they demand an immediate response. The first response from the WHO was to seek out data from other developing countries that may be used to address this question, and this is happening.

      Other studies may not show the same findings in relation to diphtheria, tetanus, and pertussis vaccine, in which case the particular circumstances in Guinea-Bissau should be considered. At the time of the study 25-30% of all children died before their 5th birthday, most from malaria or pneumonia, although this has not been well studied. Data from other places with high mortality and intense malaria transmission will therefore be particularly interesting.

      Regardless of the outcome of the ongoing investigations, the work in Guinea-Bissau highlights the importance of considering the overall impact of vaccines on children's health. Thus the finding that measles and BCG vaccines seem to improve survival beyond what can be attributed to the prevention of the specific diseases is particularly good news that should stimulate similar studies in other areas.

      Broader lessons can be learnt from this study. Public health research to support public health interventions is not a luxury but a necessity. Up till now, the overall impact of routine childhood vaccines on survival in places with high mortality has not yet been evaluated systematically. We should learn from this omission and ensure that all public health interventions are underpinned by the appropriate public health research, both experimental and observational, to ensure that they are both safe and efficacious.

      References

      1. 1.
      2. 2.

      WHO responds to Guinea-Bissau report

      1. Peter I Folb (pfolb{at}uctgsh1.uct.ac.za), professor
      1. Bandim Health Project, Danish Epidemiology Science Centre, Apartado 861, Bissau, Guinea-Bissau
      2. Danish Epidemiology Science Centre, Statens Serum Institut, Copenhagen, Denmark
      3. Department of Paediatrics, Royal Children's Hospital, Parkville, Victoria 3051, Australia
      4. Instituto de Saude Coletiva, Federal University of Bahia, Salvador-Bahia, Brazil
      5. Department of Pharmacology, University of Cape Town Health Sciences Faculty, Observatory 7925, South Africa
      6. Bandim Health Project, Danish Epidemiology Science Centre, Apartado 861, Bissau, Guinea-Bissau
      7. Danish Epidemiology Science Centre, Statens Serum Institut, Copenhagen, Denmark

        EDITOR—The paper by Kristensen et al challenges the safety of diphtheria, tetanus, and pertussis vaccine, one of the principal vaccines used by national immunisation programmes.1 The authors shared their results with the World Health Organization before publication, and they cooperated in allowing the organisation to review the field sites in Guinea-Bissau and the data bank in Copenhagen.

        The Global Advisory Committee on Vaccine Safety of the WHO, an independent group of experts in drug safety, vaccine science, and epidemiology that advises the Department of Vaccines and Biologicals of the organisation, has closely considered the reported findings and conclusions of the paper. It has found that numerous and serious deficiencies in the paper did not allow it to reach the same definitive conclusions reached by the authors. In particular, it found that the reported observations are incomplete and do not tally, no systematic effort has been made to address the likelihood of bias introduced by the method of data collection, and categorical inferences have been drawn from data that are either not significant or critically dependent on a very small number of results that might equally be explained by chance. In addition, the probability of the results being distorted by confounding factors has not been adequately addressed. The analysis was data driven and not based on a priori generation of a hypothesis, which makes interpretation of significance values and confidence limits problematic. The conclusions of this paper need to be scrutinised to the same extent as adverse events previously mistakenly attributed to diphtheria, tetanus, and pertussis vaccine.

        It is important that the safety and impact of current immunisation schedules should be studied, particularly in high risk situations. WHO is committed to giving support to investigators, including Kristensen et al, to conduct such work in a robust manner. We emphasise that no change is warranted in current policy with regard to immunisation practices, including BCG, diphtheria, tetanus, and pertussis, oral polio, and measles vaccines in national immunisation schedules.

        Footnotes

        • On behalf of the WHO Global Advisory Committee on Vaccine Safety.

        References

        1. 1.

        Authors' reply

        1. Peter Aaby, anthropologist (psb{at}sol.gtelecom.gw),
        2. Henrik Jensen, statistician
        1. Bandim Health Project, Danish Epidemiology Science Centre, Apartado 861, Bissau, Guinea-Bissau
        2. Danish Epidemiology Science Centre, Statens Serum Institut, Copenhagen, Denmark
        3. Department of Paediatrics, Royal Children's Hospital, Parkville, Victoria 3051, Australia
        4. Instituto de Saude Coletiva, Federal University of Bahia, Salvador-Bahia, Brazil
        5. Department of Pharmacology, University of Cape Town Health Sciences Faculty, Observatory 7925, South Africa
        6. Bandim Health Project, Danish Epidemiology Science Centre, Apartado 861, Bissau, Guinea-Bissau
        7. Danish Epidemiology Science Centre, Statens Serum Institut, Copenhagen, Denmark

          EDITOR—Our observations on the non-specific effects of vaccinations may have major implications for immunisation policies, so we invited the World Health Organization to collaborate in examining these effects. A WHO epidemiologist reviewed analyses in Copenhagen and three experts visited Guinea-Bissau. As reported above, they did not identify problems with the data or potential bias that could invalidate our study.

          Folb's letter for the Global Advisory Committee on Vaccine Safety and the WHO's website make several unsubstantiated criticisms.1

          Firstly, the criticism that our observations are incomplete and do not tally is undocumented.

          Secondly, that we have not systematically addressed the likelihood of bias is not true as the results and discussion sections deal extensively with potential sources of bias.

          Thirdly, we are said to have made categorical inferences and definitive conclusions; however, we suggest only that non-specific effects could be important as different routine vaccinations are associated with opposite trends in the survival analyses.

          Fourthly, the committee claims that we have not adequately addressed the probability that our results are distorted by confounding factors without even suggesting which confounding factors could produce opposite trends for different vaccines.

          Finally, the suggestion that our study is produced by data driven analysis is contradicted by the results of our previous vaccine studies.24 The importance of non-specific effects of vaccines has been evident since the early 1990s, when girls given high doses of measles vaccine were found to have a higher mortality than girls given the standard dose. 2 3 Further evidence that standard measles vaccine reduces mortality from diseases other than measles was published in 1995.4

          The observations of non-specific effects of routine vaccinations were unexpected in spite of previous work on high titre vaccines. 2 3 When an unexpected observation cannot be tested in a randomised trial, there are two important considerations: Is the observation consistent with existing data? Can it be replicated? For high titre measles vaccines, the results from three studies from Guinea-Bissau2 and Senegal3 were consistent when we reported our studies, and the finding was later replicated in Haiti. Our current observations are consistent with previous studies of the non-specific effects of vaccines, and Folb provides no information that contradicts them.24 We welcome Folb's assurance that the WHO will support further studies of immunisation in high risk populations as this will help clarify the extent to which the observations on non-specific effects are replicable. There is an urgent need for such research on vaccinations and child survival.

          References

          1. 1.
          2. 2.
          3. 3.
          4. 4.
          View Abstract