The aim of the thesis was to investigate the feasibility and effect of simultaneous high-dose vitamin A supplementation (V AS) and Bacille Calmette-Guérin (BCG) vaccination at first health care contact in low-income countries.
VAS is associated with 23-30% reduction in all cause mortality in 6 months to 5 years old children in low-income settings. So far, no strategy of supplementation in infants below 6 months of age has been consistently proven to reduce mortality. The remarkably beneficial effect of VAS in children above 6 months of age and the heavy burden of disease in younger infants lead us to believe that the potential improvements in health would be of high impact if a feasible and effective strategy of VAS targeting young infants could be identified.
To increase cost-effectiveness, it would be beneficial to provide VAS at existing contacts between children and the health care system. Such strategies oblige us to examine possible interactions with other interventions provided simultaneously.
Effect must be evaluated by measuring changes in morbidity and mortality in randomized trials. Two studies from Asia had reported a promising effect of supplementation at birth, particularly in boys, when our study was initiated. Later, a study from Zimbabwe showed no overall effect. The variability in the effect of VAS may be due to differences in baseline nutritional status or in level and type of pathogen exposure. It has also been hypothesized that VAS may interact with routine childhood immunizations.
With the primary aim of testing the hypothesis that VAS given with Bacille Calmette-Guérin vaccine (BCG) would be beneficial and lead to reduced mortality in the first year of life, a randomized study was carried out in Guinea-Bissau. The present thesis is based on four papers examining the feasibility and effects of the intervention within this cohort. Feasibility was examined as the rate of adverse effects and the effect on the immune response to BCG vaccination by simultaneously administered VAS. The effect of the intervention was studied as the impact on measles, rotavirus and diarrheal morbidity. We furthermore investigated the background factors associated with a beneficial effect of VAS.
From 2002-2005, infants were enrolled in a randomized study of 50,000 IU vitamin A versus placebo administered with BCG vaccination at first health care contact, most often immediately birth. The trial included 4,345 children.
Adverse effects were monitored through daily visits day on 1-3 after supplementation by a medical doctor, and through morbidity registration performed by a trained assistant performing a weekly structured interview during the first month post supplementation. To evaluate the effect on BCG-specific response ex vivo cytokine production to purified protein derivative of Mycobacterium tuberculosis (PPD) stimulation was measured in full blood drawn from 607 children at 6 weeks of age, and delayed type hypersensitivity (DTH) to PPD and sear reaction was measured in 2,709 infants at 2 and 6 months of age.
All children were followed by the demographic surveillance system of the Bandim Health Project (BHP), surveillance at the national hospital Simão Mendes (HN SM) and dedicated follow-up visits. Measles morbidity was registered for the entire cohort through the surveillance system. Rotavirus morbidity and diarrhea were examined in detail in a subgroup of 287 children. These children were visited repeatedly during the rotavirus season of 2005, and each Week a structured morbidity interview was performed and a fecal sample was analyzed for rotavirus.
Various statistical methods were used guided by the nature of the data. Cross-sectional observations, such as response/no response to PPD, were compared as prevalence data using a Poisson regression with robust variance estimates. Outcomes with known time at risk were compared in a Cox proportional hazard model. The importance of background factors and sex was examined by stratified analysis and interaction analysis.
No adverse effects were identified in the study. The rate of health care contacts and various symptoms during the first month were comparable between the groups. In boys, VAS was associated with a larger local reaction to BCG vaccination. Ex vivo VAS recipients responded with more IFN-y upon PPD stimulation. Boys from the VAS group had decreased prevalence of PPD-response in viva at 2 months of age. The PPD response in vivo at 6 months and the scar reaction was not affected by the intervention. Thus, VAS did not appear to affect long-term immunity to tuberculosis.
Measles incidence was affected in a sex-differentia1rnanner. In the first 6 months of life the hazard ratio in boys was 0.54 (0.25-1.15) and 1.63 (0.83-3.18) in girls (test of interaction, p=0.03).
Rotavirus colonization and rotavirus diarrhea was increased by VAS in the first 6 months of life. Non-rotavirus diarrhea tended to be reduced in boys and increased in girls, the interaction between sex and VAS again being significant (p=0.03).
Overall, our results did not document any beneficial or negative effect of VAS on morbidity, but we consistently found that boys and girls responded differently to VAS. Likewise, the result for mortality, not part of the present thesis, showed no overall effect but a tendency for a sex-differential effect: In males VAS Was associated with a tendency of decreased mortality, whereas it was associated with a tendency of increased mortality in girls (test of interaction, 1 3).
The studies indicate that administering VAS with BCG vaccination at first health care contact is feasible. However, given the current evidence concerning the effect on mortality and morbidity, a global recommendation of VAS at birth is unwarranted.
The study of adverse events and BCG response documented that the tested dose was acceptable as treatment for infants — it is below the limit causing toxic reactions and it did not hamper immune response to the simultaneously administered BCG vaccine. This suggests that with regard to vaccine response, co-administration of VAS and BCG can be undertaken, probably also in children above 6 months of age. It also implies that in terms of adverse effects this particular combination given to infants can be considered safe.
It seems plausible that VAS used as an “immune stimulant” in infants may have a very narrow therapeutic range. This could explain the strong interactions with different background factors and the lack of reproducible results between populations and within populations over time. The consequences of changing the purpose of VAS from therapy to prophylaxis might have been underestimated. In the prophylactic approach the aim is to provide the child with optimal resistance to infection, which may result in only a narrow range of post- - supplementation vitamin A status outcomes being acceptable. If the optimal range is very narrow, and possibly even narrower following vaccines or under immunologic stress, it is conceivable that the effect of different doses or regimes of VAS will continue to vary drastically in and between populations, due to the constant and rapid change in background factors determining vitamin A status.
VAS at birth was found to be beneficial for boys, with the strongest effect during the first 6 months following supplementation, and to have no effect in girls, possibly a negative effect after DTP vaccination. Our findings were consistent across outcomes and with earlier studies.
Based on the results from our studies, the next candidate approach for infant supplementation could be sex-specific, season-specific, or involve lower, perhaps repeated, doses or could be designed to fit a modified immunization program. It is important to keep trying to understand the mechanisms behind the variability in effect, but it is even more important to extend the paradigm of evidence-based medicine to include mass interventions in low-income settings.