The results of this review should be of interest to clinicians and policy makers alike. We found that, despite early observational studies suggesting an association between vitamin A deficiency and decreased risk of MTCT [4–9], RCTs show no such effect and actually raise the possibility of increased risk. Similarly, the single trial examining supplementation with multivitamins did not decrease MTCT. Supplementation with vitamin A or multivitamins was not associated with a reduction in childhood mortality. While one trial suggested multivitamins might decrease pre-term delivery, the results are not consistent.
There are several limitations to consider in this review. Due to the small number of studies included in each separate analysis, available methods for exploring the likelihood of publication bias are uninformative. We attempted to reduce this potential impact by systematically searching the databases, contacting authors, and searching for unpublished studies through registries. A further limitation is the impact that multiple childbirths from the same mother had on the results of our analyses . This information was not provided consistently across studies or through contact with authors and, although systematic evaluations of this have shown it does not significantly confound meta-analyses, could theoretically affect the estimates of effect . Finally, all of the studies compared vitamin supplementation vs. placebo. It is possible that a trial examining ARVs plus micronutrients vs ARVs alone would yield results generalizable to the current desired situation.
A strength of our meta-analysis is that we used a random effects model as this assumes a different underlying effect for each study and takes between-study variability into consideration as an additional source of variation. These effects are assumed to be randomly distributed and the central point of the distribution is the focus of the combined effect estimate. Thus, the random effects model gives greater weight to smaller studies than does the fixed effects model, and results in wider confidence intervals and a more conservative estimate of effect than the fixed effects model. This is especially warranted in this study as we identified significant heterogeneity in our pooled analysis of vitamin A on MTCT (I2 = 75%, heterogeneity P = 0.01) and pre-term delivery (I2 = 77%, heterogeneity P = 0.03).
We found large and unexplained heterogeneity between studies in the meta-analysis of vitamin A for prevention of MTCT. We were unable to explain this heterogeneity using our a priori determined explanations of heterogeneity. However, biological evidence may best explain this occurrence. MTCT is most likely to occur during the process of vaginal birth. Thus, it is important that a further investigation of the trial by Fawzi et al. demonstrated that vaginal HIV-1 viral shedding actually increased in women who were given Vitamin A supplementation but not in the case of other micronutrient supplementation (74.8% vs. 65.1%, P = 0.04) . This supports the plausibility that vitamin A may contribute to an increased risk of transmission. There does not appear to be evidence demonstrating the same risk with the use of other vitamins and multivitamins may still provide some level of protection for women living with HIV [25, 26].
There is also another explanation for the difference between the trial by Fawzi et al (2002)  and the trial in South Africa and Malawi [17, 18]. In the trials that found no effect, the supplements were given during the antenatal period only, whereas in the Tanzania trial supplementation continued during the antenatal and breastfeeding periods. It may be that a longer period of supplementation on a larger n resulted in greater power to detect effects. Indeed, earlier analysis of this sample by Fawzi (2000) did not reveal this effect . It is additionally possible that geographical differences exist from between Tanzania and the other countries (South Africa and Malawi). It is possible that nutritional status regarding important nutritional supplementation associated with HIV progression, such as selenium [27–29], is different in Tanzania.
Conducting trials to assess the impact of interventions on MTCT is an ethically challenging, yet politically eye-opening area. Section 29 of the Helsinki Declaration ethical principles for conducting research on human subjects states that "the benefits, risks, burdens and effectiveness of a new method should be tested against those of the best current prophylactic, diagnostic, and therapeutic methods. This does not exclude the use of placebo, or no treatment, in studies where no proven prophylactic, diagnostic or therapeutic method exists " However, in many impoverished nations, supplying antiretrovirals would also result in inducement to participate, a factor that is largely considered unethical to recruitment. Were antiretroviral treatment provided to these developing nation populations, vitamins would have to be tested in the presence of antiretroviral treatments, such as single-dose nevirapine or short-dose zidovudine . However, access to antiretroviral treatments in developing nations is extremely limited and although the Global Fund for AIDS, Malaria and Tuberculosis is making great strides at providing access to antiretrovirals for impoverished nations, the likelihood of effective treatment even in pregnancy is not guaranteed. The investigators of the trials reviewed here have provided evidence in a pragmatic fashion as they provide results from the population with which we would aim to generalize.
More than 95% of HIV-1-infected children acquired their infection from their mother . Mother-to-child transmission is largely preventable with interventions that are accessible to resource-poor countries: prevention of sexual transmission of HIV-1 through education for women of childbearing age, especially very young women; access to HIV-1 testing and reduction of unwanted pregnancies by HIV infected women informed of their serostatus; and ARV-based prevention of mother-to-child transmission. Prevention of mother-to-child transmission is the most cost-effective antiretroviral method and one of the most attractive interventions for prevention of HIV-1. A rapid scaling-up of implementation is crucial to allow programs to prevent mother-to-child transmission to affect the burden of paediatric HIV/AIDS. Such national initiatives should build a comprehensive continuum of care, including access to ARVs, for all members of affected families.
Using vitamins as a therapy to prevent MTCT seems inadvisable given the current state of evidence indicating a lack of consistent effect in prevention of vertical transmission [2, 3, 31]. However, in settings where poverty and social circumstances prevent adequate nutrition, the implementation of nutritional programs for pregnant women may play a role in preventing other harmful pregnancy outcomes. Future trials assessing the impact of effective nutrition on pregnant women living with HIV are not only an important effort in stemming the epidemic and improving the quality of life of patients, but also a human rights imperative . Specific trials aimed at women with low nutritional status may provide an additional armament in the fight against HIV/AIDS.
In summary, the findings from our systematic review and meta-analysis do not support the use of vitamin A supplementation as an aid in reducing the risk of mother-to-child transmission of HIV-1, and may in fact increase the risk. With respect to protection against pre-term delivery, vitamin A supplementation demonstrated a non-statistically significant protective trend. No role was found for maternal vitamin A supplementation in reducing childhood mortality at 1 year. We also found that multivitamin supplementation showed no effect on mother-to-child transmission, childhood mortality at 1 year, or prevention of pre-term delivery.