Summary of findings
The global AIDS epidemic is far more extensive than was predicted when HIV emerged three decades ago. UNAIDS and WHO estimate that 30.8 million adults were living with HIV worldwide in 2007, half of whom were women of childbearing age (UNAIDS 2009). Mother-to-child transmission (MTCT) of HIV is the dominant mode of acquisition of HIV infection for children, currently resulting in more than 1000 new HIV infections each day worldwide, most of which occur in sub-Saharan Africa.
MTCT of HIV can occur during pregnancy, delivery and breastfeeding. The increasing number of infected women of childbearing age makes the prevention of MTCT of HIV a public health priority in the hardest-hit developing countries with high antenatal HIV seroprevalence (UNAIDS 2009). Current strategies to reduce the risk of transmission include a short course of antiretroviral therapy (Volmink 2007), elective Caesarean section delivery (Read 2005), and avoidance of breastfeeding (Newell 2000). Despite their benefits, these interventions are impractical in many resource-limited developing countries because they require the determination of the HIV status of pregnant women and are costly, complex, and require skilled personnel. Simple, inexpensive, and effective interventions that could potentially be implemented in the absence of prenatal HIV testing programmes would be valuable.
Vitamin A supplementation during pregnancy is one low-cost intervention that has been suggested (Newell 2000). Vitamin A is the generic name for a group of fat-soluble compounds, which have the biological activity of the primary alcohol, retinol. The vitamin is involved in the regulation and promotion of growth and differentiation of many cells and maintains the integrity of the epithelial cells of the respiratory and digestive tracts. Vitamin A is necessary for formation of the photosensitive visual pigment in the retina, and reproductive functions. In the 1920s the vitamin was considered to be an anti-infective agent (Green 1928) and there is increasing evidence that it is essential for normal immune function (Ross 1996; Semba 1998).
Vitamin A deficiency is most prevalent in areas where diets lack preformed vitamin A, such as in South and Southeast Asia, and Sahelian and sub-Saharan Africa (West 2001). An estimated seven million pregnant women are deficient in vitamin A worldwide (West 2002). Vitamin A deficiency in pregnant women is associated with night blindness, severe anaemia, wasting, and malnutrition, and reproductive and infectious morbidity (Christian 1998a), and increased risk of mortality 1-2 years following delivery (Christian 2000).
Observational studies in sub-Saharan Africa have shown low serum vitamin A levels in HIV-infected women to be associated with significantly increased rates of MTCT of HIV (Semba 1994, Dushimimana 1992) and infant mortality (Dushimimana 1992, Semba 1995). However, four observational studies in the United States provided conflicting results: low serum vitamin A was associated with a higher risk of MTCT of HIV in two (Greenberg 1997, Landesmand 1996), but not the other two (Burger 1997, Burns 1999). These studies used different definitions for vitamin A deficiency (serum retinol levels <30mg/dl used by Semba et al [Burns 1999, Greenberg 1997] and <20mg/dl used by Burger et al [Burger 1997]) and had small sample sizes (e.g. in the Burger study, only 4/95 [4.2%] of women had serum retinol levels <20mg/dl).
Given the magnitude of the paediatric HIV epidemic in resource-limited countries (UNAIDS 2009) and the simplicity and low cost of vitamin A supplementation (Bell 1997), clarification of the role of vitamin A supplementation in MTCT of HIV is of considerable importance. This review aims to combine all high quality randomised trials conducted to date, to estimate the effect of vitamin A supplementation on MTCT of HIV, infant and maternal mortality and morbidity, as well as the tolerability of vitamin A supplementation. The ultimate goal is to determine whether vitamin A supplementation of HIV-infected women could be recommended as a public health policy to reduce MTCT of HIV. We have thus considered overall MTCT of HIV whether occurring during, before or after delivery.
The primary objective of this review was to estimate the effect of vitamin A supplementation during pregnancy and or the postpartum period on the risk of mother-to-child transmission of HIV infection.
The secondary objectives are to estimate the effect of vitamin A supplementation of pregnant or breastfeeding HIV positive women on infant and maternal mortality and morbidity and to describe any side effects for the mother and the neonate.
Criteria for considering studies for this review
Types of studies
Randomised controlled trials
Types of participants
Known HIV-infected pregnant or breastfeeding women (as diagnosed by an antibody test); any age, any clinical stage of HIV disease, and any setting.
Types of interventions
Vitamin A supplementation compared with placebo.
Types of outcome measures
- HIV infection status of the child (as defined by the authors)
Secondary outcome measures:
1. Infant death
3. Neonatal sepsis (as defined by the authors)
4. Neonatal admission to neonatal unit
5. Death by 24 months of age (including still births)
6. Side effects in the child (as defined by the authors)
7. Preterm delivery, less than 37 weeks of gestation
8. Very preterm delivery, less than 34 weeks of gestation
9. Birth weight
10. Low birth weight, less than 2500g
11. Very low birthweight, less than 2000g
12. Long term side effects in survivors
1. Maternal death
2. Postpartum infection (as defined by the authors)
3. Side effects in the mother (as defined by the authors)
4. Cost of the intervention
5. Acceptability of the intervention.
Search methods for identification of studies
See: Cochrane HIV/AIDS Group methods used in reviews.
See: HIV/AIDS Collaborative Review Group search strategy.
We formulated an exhaustive search strategy in an attempt to identify all relevant studies regardless of language or publication status (published, unpublished, in press, and in progress). Full details of the Cochrane HIV/AIDS Review Group methods and the journals hand-searched are published in The Cochrane Library in the section on Collaborative Review Groups (http://www.mrw.interscience.wiley.com/cochrane/clabout/articles/HIV/frame.html).
In June 2010 we searched PubMed [ Table 1], EMBASE [ Table 2], the Cochrane Central Register of Controlled Trials (CENTRAL: Table 3), AIDS Education Global Information System (AEGIS: www.aegis.com), and WHO International Clinical Trials Registry Platform (http://www.who.int/ictrp/search/en/) for randomised controlled trials of vitamin A supplementation in HIV-infected women, using terms specific to vitamin A and mother-to-child transmission of HIV in combination with standardised methodological filters for randomised controlled trials developed by the Cochrane Collaboration (Higgins 2008).
The search conducted in June 2010 was an update of one conducted in February 2008 for an earlier version of this review (Wiysonge 2005). In the latter, in addition to an exhaustive search of electronic databases and conference proceedings [ Table 4; Table 5; Table 6; Table 7], we also contacted investigators of identified trials and other content experts, agencies, organisations and academic centres in an attempt to locate any further trials (completed or ongoing, published or not). Relevant reviews and other publications were also scrutinised for any additional relevant studies or unpublished data.
Data collection and analysis
Titles and abstracts identified by the searches conducted in 2002, 2005, and 2008 (for previous versions of this review) were scrutinised independently for eligibility by at least two of four authors (CSW and MSS, EJK, or PB). The Cochrane Editorial Unit conducted the screening and initial selection of articles from the June 2010 search, and two authors (CSW, MS) took the final decision on study eligibility for inclusion in the review. Studies were included in the review if they were controlled trials (study design), comparing vitamin A supplementation with placebo (intervention) in HIV-infected women during pregnancy or the postpartum period (participants), and information was available on any of the specified outcome variables. For each included study, two authors (CSW and MSS, EJK, or PB) independently assessed whether the method of assigning participants to interventions was adequate, unclear, or inadequate (Higgins 2008), and extracted the data. We used pre-designed forms for data extraction and for requesting additional information from the investigators. On data abstraction forms we noted the review title, study reference and publication status, date of extraction, and reviewer's initials. Data were extracted under the following subheadings in the form: methods (method used to generate the randomisation sequence; method used to conceal treatment allocation; blinding of participants, care providers, and outcome assessors, losses to follow-up and how they were handled; other potential sources of bias), participants (setting; number of women randomised, baseline vitamin A level, gestational age [or hours after delivery] at initiation of supplementation), interventions (supplementation dose, length of supplementation, type of control group, co-interventions), outcomes (children infected with HIV, infant death, stillbirth, neonatal sepsis, neonatal admission to neonatal unit, death of child by 24 months of age, side effects in the child, preterm delivery, very preterm delivery, birth weight, low birth weight, very low birth weight, long term side effects in survivors, maternal death, maternal postpartum infection, side effects in the mother, cost of the intervention, number of complaints), and other notes. If data were available on MTCT of HIV at two or more periods, the more complete or later one was taken into account. The data extracted for dichotomous variables were the number of affected participants and the number of participants in the comparison group; and for continuous variables, the mean, the standard deviation and the number of participants in the comparison group.
Disagreement between authors about the eligibility or quality of a study, data extracted, or other aspects of the review was resolved by discussion and consensus. When more information was needed, we wrote to the principal investigator of the trial. Review authors assessing study eligibility and quality were not blinded to the names of the authors, their institutions, journals of publication, and results of the study.
We conducted meta-analysis using RevMan 5; analysing trial participants in groups to which they were randomised regardless of which or how much treatment they actually received. We expressed each study result as a relative risk (RR) for dichotomous data, and a mean difference (MD) for continuous data, with its 95% confidence intervals (CI). Heterogeneity between studies was assessed by graphical inspection of results and, more formally followed by, the chi-square test of homogeneity. In the absence of significant statistical heterogeneity between studies (P>0.1), we pooled the results using a fixed-effects method. When there was significant heterogeneity between study results, we used the random-effects method. When data were available we conducted a subgroup analysis based on the timing of vitamin A supplementation i.e. antenatal or postpartum. In addition, we used the GRADE approach (Guyatt 2008) to assess the quality of the evidence for the effect of vitamin A supplementation on each key outcome.
Description of studies
The search for trials published between 1 January 2007 and 8 June 2010 and those which were ongoing or completed but not yet published by 8 June 2010 yielded 117 records in AEGIS, 11 in CENTRAL, 11 in EMBASE, 14 in PubMed, and 34 in the WHO International Clinical Trials Registry Platform. The 34 hits from the WHO International Clinical Trials Registry Platform search portal did not reveal any additional ongoing trial. When we removed duplicates from the combined search results for AEGIS, CENTRAL,EMBASE and PubMed, 93 references were available to scrutinise. After screening the abstracts, 66 were excluded and 27 were left for further scrutiny. However, a review of the full text of the 27 articles did not reveal any additional trial over and above the ones already identified in the previous version of the review (Wiysonge 2005).
We have provided a detailed description of the characteristics of included studies in the table of included studies [Characteristics of included studies] and provide a summary below. Studies are referred to by the name of the principal investigator and the year of the major publication for the study.
It is not stated how selected women were allocated to this trial, which is described as double-blind and randomised. The participants comprised 728 HIV-infected women at 17-39 weeks' gestation in KwaZulu-Natal Province of South Africa; 30.6% of whom had serum retinol levels < 20 µg/dl. These women were assigned to daily oral vitamin A (5000 IU retinyl palmitate and 30 mg beta-carotene) or placebo. At delivery, women in the vitamin A group received a dose of 200,000 IU of retinyl palmitate while the placebo arm received an identical placebo. No woman received any antiretroviral therapy and it is not clear from the trial reports whether the women also received iron, folic acid, and/or chloroquine. Mother-infant pairs lost to follow-up were excluded from the analysis.
This is a randomised, placebo-controlled, double-blind trial. Participants consisted of 1,075 pregnant HIV-infected women enrolled at 12-27 weeks' gestation resident in Dar es Salam, Tanzania. The prevalence of vitamin A deficiency at baseline is not stated. The women were randomly assigned to a daily oral dose of one of: vitamin A (30mg beta carotene + 5000 IU retinyl palmitate) alone, vitamin A plus multivitamins ( 20mg B1, 20mg B2, 25mg B6, 100mg niacin, 50microg B12, 500mg C, 30 mg E, and 0.8 mg folic), multivitamins alone, or placebo. Of 1085 women initially randomised, 10 were eventually excluded for either being HIV-negative (n=7) or nonpregnant (n=3). At delivery, women receiving any vitamin A were given an additional 200,000 IU oral dose of vitamin A while the others received an extra dose of an identical placebo. It is not mentioned in this trial whether any woman received antiretroviral therapy. All women were given daily oral doses of iron and folic acid, and weekly doses of chloroquine. Mother-infant pairs lost to follow-up were excluded from the analysis.
This is a randomised non-placebo controlled trial. The participants comprised 697 pregnant HIV-infected women enrolled at 18-28 week's gestation in Blantyre, Malawi. The prevalence of vitamin A deficiency (< 0.70 µmol/L) was 51% during the second trimester. All women received orally administered daily doses of iron (30mg of elemental iron) and folate (400 µg) from enrolment until delivery. One half of the women were randomised to receive daily doses of orally administered vitamin A (10,000 IU). In addition, all women received oral vitamin A (100,000 IU) at 6 weeks postpartum, as per policy of the Malawi Ministry of Health. Mother-infant pairs lost to follow-up were excluded from the analysis. Twins were also excluded from the birth weight and mortality analyses because twins are known to have lower birth weights and higher mortality rates.
This is a randomised, placebo-controlled trial. The participants were 1,669 pregnant women (533 HIV-infected) enrolled at 22-35 weeks' gestation in Harare, Zimbabwe. The women were randomly assigned to a daily tablet containing either vitamin A (3000 micrograms retinol equivalents and 3.5 mg beta-carotene) and the recommended daily allowance of 12 micronutrients, or placebo. All women received iron and folic acid as part of routine antenatal care. Mother-infant pairs lost to follow-up were excluded from the analysis. All singletons and the first born of multiple pregnancies were included in the analysis. The authors did not report data on MTCT of HIV, due to a high loss to follow-up (32.5%).
This trial of postpartum vitamin A supplementation conducted in Zimbabwe randomly allocated 4495 HIV-positive women and their newborns within 96 hours of delivery - in a 2 x 2 factorial design - to a large single dose of vitamin A for the mother (400 000 IU) only, the newborn (50 000 IU) only, both the mother and the newborn, or neither of them. All women received iron and folic acid as part of routine antenatal care. Mother-infant pairs lost to follow-up were excluded from the analysis. All but 4 mothers initiated breastfeeding, no information on ART or cotrimoxazole prophylaxis.
Risk of bias in included studies
|Figure 1. Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.|
|Figure 2. Risk of bias summary: review authors' judgements about each risk of bias item for each included study.|
The trial is reported as "double-blind" and "randomised", but the methods used to generate the allocation sequence and to conceal the allocation are not reported. Fifty seven (7.8%) of 728 women for whom outcomes were not available were excluded from the analysis.
Randomisation sequence generation and allocation concealment were adequate. Randomisation was done in blocks of 20. At enrolment, each eligible woman was assigned the next numbered bottle of study drug. Fifty four (5.0%) women were lost to follow-up, and excluded from the analysis.
Randomisation sequence generation and allocation concealment were adequate. Participants were assigned to treatment using computer-generated random numbers, and treatment was concealed by pre-packing study supplements in sequentially numbered series assigned to study identification numbers. Nine percent of women were lost to follow-up before delivery and excluded from the analyses. Twins were excluded from the birth weight and mortality analyses because they are known to have lower birth weights and higher mortality rates.
Randomisation sequence generation and allocation concealment were adequate. Randomisation was done by simple blocked randomisation using a computer-generated random number sequence. Containers with 110 supplements or placebo tablets, which were coded A or B, respectively, were delivered by the manufacturer together with the code in 2 sealed envelopes. Duplicate containers, which corresponded to the random sequence, were consecutively numbered from 1 to 1800. The study participants were numbered consecutively at recruitment. One hundred and seventy-three (32.5%) HIV-infected women were lost to follow-up and excluded from the analysis.
Randomisation sequence generation and allocation concealment were adequate. Randomisation was done using computer-generated blocks of 12. Treatment assignment was concealed by pre-packing study supplements in sequentially numbered series assigned to study identification numbers. Mothers were assigned an original study identification number at enrolment and were given the next sequentially numbered opaque bottle with supplements. One hundred and forty-three (3.2%) mother-infant pairs were lost to follow-up and excluded from the analysis.
Effects of interventions
We identified six potentially eligible trials (Coutsoudis 1999;Chikobvu 2000; Fawzi 2002; Semba 2002; Friis 2004;Humphrey 2006). Of these, we have included five trials: four trials of antenatal vitamin A supplementation (Coutsoudis 1999; Fawzi 2002; Semba 2002; Friis 2004) that enrolled a total of 3033 HIV-infected pregnant women and one trial of postpartum supplementation (Humphrey 2006) that enrolled 4495 HIV-infected women at 96 hours or less after delivery. The remaining trial is still ongoing (Chikobvu 2000).
The combined relative risk for MTCT of HIV, comparing Vitamin A supplementation with placebo (4 trials, 6517 women) is RR 1.04, 95% CI 0.87 to 1.24, I
|Figure 3. Forest plot: HIV infection in child.|
There was evidence of heterogeneity among the trials with antenatal supplementation (I
Vitamin A supplementation of HIV-infected pregnant women significantly improved birth weight (3 trials, 1809 women: MD 89.78, 95%CI 84.73 to 94.83; heterogeneity P = 0.22, I
|Figure 4. Forest plot: Incidence of stillbirth.|
|Figure 5. Forest plot: Preterm (less than 37 weeks).|
|Figure 6. Forest plot: Low birth weight (less than 2500g).|
|Figure 7. Forest plot: Birthweight (in grammes).|
Vitamin A was postulated to reduce MTCT of HIV by affecting several maternal, fetal, and or child risk factors for transmission, including the clinical, immunological, or viral stage of HIV disease among pregnant women; the integrity of the epithelial lining of the placenta, maternal lower genital tract, or breast; the occurrence of prematurity and low birth weight; and the status of the systemic and digestive mucosal immune systems of the fetus and the child (Fawzi 1998, Fawzi 2000).
A synthesis of the currently available data does not show evidence of an effect of antenatal or postpartum vitamin A supplementation on the risk of MTCT of HIV. Due to the inconsistency in results across studies included in the meta-analysis, we reduced the quality of this evidence to moderate. The GRADE Working Group classifies research evidence as being of moderate quality if the true effect of the intervention is likely to be close to what was found in the research but there is a possibility that it is substantially different. Therefore, this review does not completely exclude the possibility of a beneficial or harmful effect of vitamin A supplementation on the risk of MTCT of HIV. This systematic review shows antenatal vitamin A supplementation of HIV-infected women to significantly improve birth weight, consistent with findings of previous observational studies (Semba 1995, Burns 1999). Overall, there was no significant effect of vitamin A supplementation on the frequency of preterm deliveries; although one trial which compared vitamin A supplementation to placebo in 728 pregnant women (Coutsoudis 1999) suggests significant reductions in the frequency of preterm birth (RR 0.64, 95% CI 0.364 to 94 for all preterm births and RR 0.37, 95% CI 0.15 to 0.94 for very preterm births). The review did not find evidence of an effect of vitamin A supplementation of HIV-infected pregnant women on the incidence of stillbirths and maternal mortality. However, a large, high-quality randomised placebo-controlled trial in women of unknown HIV status in Nepal (West 1999) showed a significant 40% relative reduction in maternal mortality with vitamin A supplementation. The quality of the evidence on the effects of antenatal or postpartum vitamin A supplementation was high for birth weight; moderate for MTCT of HIV, stillbirths, preterm deliveries, and child mortality; and very low for maternal deaths.
Vitamin A is a known teratogen, with a suggested dose-response effect. Although the doses of vitamin A used in the trials were within the currently recommended safe low doses (IVACG 1998; WHO 1998), none of the trials reported information on the potential adverse effects of vitamin A.
In high-income countries the incorporation of the three-pronged approach of (a) antiretroviral prophylaxis (Volmink 2007), (b) elective Caesarean section delivery and (c) formula feeding into clinical practice coupled with increased prenatal HIV-1 counselling and testing has resulted in MTCT of HIV rates of less than 1% (Mofenson 2000; Navér 2006; Warszawski 2008). However, due to their complexity, cost and other constraints many low- and middle-income countries are having difficulties implementing these interventions (Wiysonge 2001; McIntyre 2002); thus there is a need for simpler and less costly options.
Vitamin A is easily provided through existing health services (Wiysonge 2006), and has been shown to be effective in reducing child mortality by approximately 30% (Fawzi 1993; Glasziou 1993). Using data from Tanzania, Bell and Sacks (Bell 1997) estimated that the savings in health care costs and lost productivity associated with HIV exceed the vitamin A programme expenditures if a 3% reduction in MTCT of HIV is achieved. The cost-effectiveness was based on, but not sensitive to, an antenatal seroprevalence of 12.5%, a compliance of 60%, and baseline transmission of 32%. Therefore, given the cost-effectiveness of a vitamin A supplementation programme for reducing MTCT of HIV (Bell 1997), even substantially lesser degrees of benefit on MTCT of HIV and pregnancy outcomes than those achieved for childhood mortality may be acceptable to low- and middle-income countries. Vitamin A deficiency is common in most populations of low- and middle-income countries (WHO 1995) and preventing the deficiency through natural dietary means is unrealistic in most of these populations. It was thus necessary to clarify the effect of vitamin A supplementation on MTCT of HIV and infant and maternal mortality. Such clarification is important to programme managers and policy-makers considering affordable options for reducing MTCT of HIV and improving reproductive health care in resource-limited settings.
Currently available randomised controlled trial data do not show that vitamin A supplementation of HIV-infected pregnant or breastfeeding women has significant beneficial effects on the risk on MTCT of HIV , and have not excluded the possibility that this intervention could be harmful. The data raise the possibility that the association between low serum vitamin A levels and increased risk of MTCT of HIV seen in observational studies could have other explanations; for example, low serum vitamin A levels may be a marker of advanced HIV-1 infection and not causally related to MTCT of HIV.
Implications for practice
Currently available evidence does not support the use of vitamin A supplementation of HIV-infected pregnant or breastfeeding women to reduce mother-to-child transmission of HIV.
Implications for research
We will update this review when data from the ongoing Bloemfontein vitamin A trial (Chikobvu 2000) are published. Given that the currently available randomised controlled trial data do not exclude the possibility that vitamin A supplementation could be beneficial or harmful, there may be need for an appropriately powered randomised controlled trial to assess the additive effect of the intervention on the risk of MTCT of HIV in antiretroviral-treated women.
We are grateful to Prof Aubrey Sheiham, Sir Iain Chalmers, Dr Phil Alderson, Prof Alfred Sommer, Prof Justus Hofmeyr, the Cochrane Collaboration Secretariat, the UK Cochrane Centre, the Cochrane HIV/AIDS Group, the South African Cochrane Centre, and the Cochrane Editorial Unit for their support in preparing or updating this review. We thank Dr Lisa Butler (and anonymous peer reviewers) who critically reviewed an earlier version of this review and Profs Henrik Friis who shared his unpublished trial data with us.
Data and analyses
- Top of page
- Summary of findings [Explanations]
- Authors' conclusions
- Data and analyses
- What's new
- Contributions of authors
- Declarations of interest
- Sources of support
- Index terms
Last assessed as up-to-date: 13 September 2010.
Protocol first published: Issue 2, 2002
Review first published: Issue 2, 1995
Contributions of authors
CSW, PB, and JS wrote the first review published in 2002; CSW, MSS, JS and PB conducted the first update published in 2005; and CSW, MSS, and EJK conducted the current review update and the one of 2008.
Declarations of interest
Sources of support
- Department of Social Medicine, University of Bristol (JS), UK.
- Ministry of Public Health (CSW), Cameroon.
- University of Yaounde I (MSS), Cameroon.
- University of Cape Town (CSW, MSS), South Africa.
- Medical Research Council of South Africa (CSW), South Africa.
- Aubrey Sheiham Cochrane Collaboration Scholarship (CSW), UK.
- Department of Health (PB), UK.
- World Health Organization, Switzerland.The World Health Organization's Department of Nutrition for Health and Development provided funding for the preparation of this updated review.
Medical Subject Headings (MeSH)
*Pregnancy Complications, Infectious; HIV Infections [prevention & control; *transmission]; Infant, Newborn; Infectious Disease Transmission, Vertical [*prevention & control]; Randomized Controlled Trials as Topic; Vitamin A [*administration & dosage]; Vitamin A Deficiency [*complications; drug therapy]; Vitamins [*administration & dosage]
MeSH check words
Female; Humans; Pregnancy
* Indicates the major publication for the study