Summary of findings
Description of the condition
Vitamin A deficiency is considered to be a major public health problem in the developing countries (WHO 2000; WHO 2009). Globally, 9.8 million pregnant women are affected by night blindness, with more than 19 million having low serum retinol concentrations (< 0.70 µmol/L). Night blindness affects 5.2 million preschool children and an estimated 190 million have low serum retinol concentrations. The prevalence of low serum retinol concentrations in pregnant women is highest in South-East Asia (17.3%) followed by Africa (13.5%), whereas the prevalence of night blindness is approximately the same in the two regions (9.9% in South-East Asia versus 9.8% in Africa) (WHO 2009).
Deficiency of vitamin A may be secondary to decreased ingestion, defective absorption and altered metabolism; or increased requirements. Factors such as low dietary fat intake or intestinal infections may also interfere with the absorption of vitamin A. Vitamin A deficiency is the most important cause of childhood blindness and contributes significantly to morbidity and mortality from common childhood infections. It is a significant contributing factor in the 2.2 million diarrhoea deaths each year among children under five years of age, and in the nearly one million measles deaths (SOWC 1998).
Description of the intervention
Vitamin A is an essential micronutrient that is required for the maintenance of normal functioning of the human body. It was the first fat soluble vitamin to be discovered and has been known to be an important dietary constituent for nearly a century (Hopkins 1912; McCollum 1915). Vitamin A is part of a family of compounds called retinoids; the naturally occurring retinoids are retinol, retinal and retinoic acid. For human physiology, retinol is the predominant form and 11-cis-retinol is the active form. The inactive retinoids, also known as provitamin A, are produced as plant pigments and are called carotenoids. Although many carotenoids occur in foods, approximately only 50% can be metabolized into the active retinoid forms. Beta-carotene, a retinol dimer, has the most significant provitamin A activity. Vitamin A is stored in the liver as retinyl esters and, when needed, is transported into blood where it is carried by retinol binding protein (RBP) for delivery to other tissues (Shenai 1993).
Vitamin A is important for the normal functioning of the visual system, immune response, gene expression, reproduction, embryogenesis and hematopoiesis (Sommer 1996). It is essential for the maintenance of normal epithelial tissues throughout the body (Wolbach 1925). Preformed vitamin A is found only in animal foods such as liver, fish and dairy products (such as milk, cheese and butter); it constitutes 65% to 75% of the dietary vitamin A intake. The remaining dietary vitamin A comes from carotenoids present in plant sources such as carrots, dark green leafy vegetables, red and orange fruits and red palm oil. The Recommended Dietary Allowances (RDAs) for vitamin A vary with age. For healthy breast-fed infants up to six months of age the average RDA is 400 µg/d, and for infants seven to 12 months of age the RDA is 500 µg/d. For children one to three years and four to eight years old, the RDA is 300 µg/d and 400 µg/d, respectively (DRI 2001).
Routine consumption of large amounts of vitamin A over a period of time can result in toxic symptoms which include liver damage, headaches, vomiting, skin desquamation, bone abnormalities, joint pain and alopecia. Hypervitaminosis A appears to be due to abnormal transport and distribution of vitamin A and retinoids that is caused by overloading of the plasma transport mechanisms (Smith 1976). A very high single dose can also cause transient acute toxic symptoms that may include a bulging fontanelle in infants; headaches in older children and adults; and vomiting, diarrhoea, loss of appetite and irritability in all age groups. Toxicity from ingestion of food sources of preformed vitamin A is rare (Hathcock 1997).
How the intervention might work
During pregnancy, women need additional vitamin A (an additional increment of 100 µg/day above basal requirements during the full gestation period) to sustain the growth of the fetus and to provide a limited reserve in the fetal liver as well as to maintain the woman's own tissue growth. Because therapeutic levels of vitamin A are generally higher than preventive levels, the safe intake level recommended during pregnancy is 800 µg retinol equivalents (RE)/day. Women who are or who might become pregnant should carefully limit their total daily vitamin A intake to a maximum of 3000 µg RE (10,000 IU) to minimize the risk of fetal toxicity (WHO/NUT 1998). Infants have very low levels of vitamin A stored in the liver at birth and are dependent on breast milk as a source of vitamin A in the first few months of life. Thus, maternal vitamin A deficiency during lactation, early weaning or artificial feeding may result in vitamin A deficiency in infants (Underwood 1994). The physiologic vitamin A needs of infants born to vitamin A-adequate mothers and fed breast milk with adequate vitamin A (in excess of 30 µg/dL or 1.05 µmol/L) are met for at least the first six months of life (Underwood 1994). Because of the need for vitamin A to support the growth rate in infancy, which can vary considerably, a requirement estimate of 180 µg RE/d seems appropriate. Average consumption of human milk by such infants is about 750 ml/day during the first six months (WHO/NUT/98.1 1998). Assuming an average concentration of vitamin A in human milk of about 1.75 mmol/l, the mean daily intake would have to be about 375 µg RE, which is therefore the recommended safe level.
Why it is important to do this review
The role of vitamin A supplementation in children greater than six months of age is well established (Beaton 1993; Imdad 2010; Rice 2004). Beaton and colleagues in their meta-analysis showed that vitamin A supplementation in children six months to five years of age significantly reduced mortality by 23% (Beaton 1993). A recent Cochrane Review concluded that two oral doses of 200,000 IU of vitamin A on consecutive days in children less than two years of age with measles were associated with a reduced risk of overall mortality (RR 0.18; 95% CI 0.03 to 0.61); similarly with pneumonia-specific mortality (RR 0.33; 95% CI 0.08 to 0.92) (Huiming 2005). The World Health Organization (WHO) recommends administration of vitamin A during vaccination contacts in order to prevent vitamin A deficiency (WHO 1998). The policy has been to supplement 100,000 IU of vitamin A at the earliest possible opportunity after six months of age. However, it has now been recommended that an additional 50,000 IU of vitamin A be administered with each of the diphtheria-tetanus-pertussis (DTP) and polio vaccinations, which are usually given at six, 10 and 14 weeks of age (Sommer 2002). National and regional programmes of vitamin A supplementation are in place in over 60 countries worldwide and target children greater than six months of age. These programs are not only highly effective in reducing mortality and morbidity but, in countries in which vitamin A deficiency constitutes a public health problem, the programmes appear to be among the most cost-effective public health interventions available. Such programs address child survival in children greater than six months of age; this group accounts for a quarter of under five years of age deaths. In order to address the major proportion of deaths in children under five, children less than six months of age should be targeted. Supplementation with vitamin A between one and five months of age has not been found to have a beneficial effect (Daulaire 1992; Rahman 1995; WHO/CHD 1998). Supplementation of neonates has been suggested as a feasible approach to bolstering body stores of vitamin A in early infancy and, therefore, having an impact on mortality and morbidity (Sommer 1995; Sugana 1978).
To evaluate the role of vitamin A supplementation in term neonates in developing countries with respect to the prevention of mortality and morbidity.
We prespecified the following subgroups to investigate heterogeneity:
- maternal vitamin A supplementation;
- birth weight of neonates;
- HIV status of the mother and infant;
- dose and frequency of vitamin A used;
- high baseline infant mortality;
- timing of vitamin A supplementation (either within the first 48 to 72 hours or later).
Criteria for considering studies for this review
Types of studies
All randomised controlled trials, both individual and cluster randomised and irrespective of publication status and language, evaluating the effects of vitamin A supplementation in term neonates in developing countries were included in the review. Studies using factorial design and quasi-randomised trials were also included.
Types of participants
All term neonates (born between 37 to 42 weeks of gestational age) up to 28 days after birth were included.
Types of interventions
Supplementation with vitamin A within the first 28 days of life was compared against a control (placebo or no supplementation). Any trial with continued supplementation beyond the first 28 days of life was excluded from the review. Co-interventions, if any, should have been identical in the two groups.
Types of outcome measures
- All-cause infant mortality at six and 12 months
- Cause-specific infant mortality associated with acute respiratory infections and diarrhoea at six and 12 months
- Infant morbidity at six months of age, associated with acute respiratory infections and diarrhoea, measured as at least one episode of morbidity
- Biochemical indicator values of vitamin A deficiency (vitamin A deficiency measured as serum retinol < 0.70 µmol/L)
- Blindness and signs of xerophthalmia (Bitot's spots and corneal lesions)
- Mean haemoglobin level or anaemia defined as haemoglobin less than the age-specific cut-off value as stated by the authors
- Adverse events reported in trials due to vitamin A toxicity such as bulging fontanelles, vomiting and diarrhoea
Search methods for identification of studies
See: Cochrane Neonatal Review Group methods used in reviews
We used the standard search strategy of the Cochrane Neonatal Review Group. The Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library, 14 June 2010), EMBASE and MEDLINE (1966 to May 2010) via PubMed were searched using the following search terms: (Newborn OR infan* OR neonat*) AND (vitamin A OR retino*) Limit: publication type clinical trial.
We limited the searches to human studies. We did not apply any language restrictions. We also searched related conference proceedings for relevant abstracts. We contacted organizations and researchers in the field for information on unpublished and ongoing trials. We searched reference lists of all trials identified by the above methods. For further identification of ongoing trials the websites www.clinicaltrials.gov and www.anzctr.com were searched.
Data collection and analysis
Selection of studies
Two review authors, Batool Haider (BAH) and Zulfiqar Bhutta (ZAB), independently assessed all the potential studies we identified as a result of the search strategy for inclusion. We resolved any disagreement through discussion.
Data extraction and management
We designed a form to extract data. For eligible studies, two review authors (BAH and ZAB) extracted the data using the agreed form. We resolved discrepancies through discussion. Data were entered into the Review Manager software (RevMan 2008) and checked for accuracy.
Assessment of risk of bias in included studies
Two review authors (BAH and ZAB) independently assessed the risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2008). We resolved any disagreement by discussion.
(1) Sequence generation (checking for possible selection bias)
We described for each included study the method used to generate the allocation sequence, if it was in sufficient detail to allow an assessment of whether it should produce comparable groups.
We assessed the method as:
- low risk (adequate) (any truly random process, e.g. random number table; computer random number generator);
- high risk (inadequate) (any non-random process, e.g. odd or even date of birth; hospital or clinic record number);
- unclear risk.
(2) Allocation concealment (checking for possible selection bias)
We described for each included study the method used to conceal the allocation sequence if in sufficient detail to determine whether intervention allocation could have been foreseen in advance of or during recruitment, or changed after assignment.
We assessed the methods as:
- low risk (adequate) (e.g. telephone or central randomisation; consecutively numbered sealed opaque envelopes);
- high risk (inadequate) (open random allocation; unsealed or non-opaque envelopes, alternation; date of birth);
- unclear risk.
(3) Blinding (checking for possible performance bias)
We described for each included study the methods used, if any, to blind study participants and personnel from knowledge of which intervention a participant received. Studies were judged at low risk of bias if they were blinded or if we judged that the lack of blinding could not have affected the results.
We assessed the methods as:
- low risk (adequate), high risk (inadequate) or unclear risk for participants;
- low risk (adequate), high risk (inadequate) or unclear risk for personnel;
- low risk (adequate), high risk (inadequate) or unclear risk for outcome assessors.
(4) Incomplete outcome data (checking for possible attrition bias through withdrawals, dropouts, protocol deviations)
We described for each included study, and for each outcome or class of outcomes, the completeness of data including attrition and exclusions from the analysis. We stated whether attrition and exclusions were reported, the numbers included in the analysis at each stage (compared with the total number of randomised participants), reasons for attrition or exclusion where reported, and whether missing data were balanced across groups or were related to outcomes. We assessed methods as:
- low risk (adequate);
- high risk (inadequate);
- unclear risk.
(5) Selective reporting bias
We described for each included study how we investigated the possibility of selective outcome reporting bias and what we found.
We assessed the methods as:
- low risk (adequate) (where it was clear that all of the study’s prespecified outcomes and all expected outcomes of interest to the review have been reported);
- high risk (inadequate) (where not all the study’s prespecified outcomes have been reported; one or more reported primary outcomes were not prespecified; outcomes of interest were reported incompletely and so cannot be used; the study failed to include results of a key outcome that would have been expected to have been reported);
- unclear risk.
(6) Other sources of bias
We described for each included study any important concerns we have about other possible sources of bias. We assessed whether each study was free of other problems that could put it at risk of bias as:
(7) Overall risk of bias
We made explicit judgements about whether studies were at high risk of bias, according to the criteria given in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2008). We explored the impact of the overall risk of bias by undertaking sensitivity analysis for primary outcomes. We considered a study to be of high quality if it was judged to have adequate sequence generation and allocation concealment, with either adequate blinding or methods for dealing with incomplete outcome data.
Measures of treatment effect
For dichotomous data, we presented results as summary risk ratios or rate ratios with 95% confidence intervals (CI).
There were no continuous outcomes in this review.
Unit of analysis issues
There were two cluster randomised trials (Klemm 2008; West 1995) included in this review. Klemm et al reported that the observed design effect was 0.9%. In West 1995, the 95% confidence intervals (CIs) of the effect estimates were inflated by 10% to account for the impact of design on the study findings. We estimated that the 10% increase in the 95% CIs gave an intracluster correlation coefficient (ICC) of 0.04 for the cohort of infants administered vitamin A.
We analysed the data using a generic inverse variance approach to meta-analysis, using Review Manager software (RevMan 2008), and generated risk ratio or rate ratio estimates with 95% CIs for the dichotomous outcomes. For this approach, the data were entered as natural logarithms (as log risk ratios and SE of log risk ratio or log rate ratios and SE of log rate ratio) for each individual study, with data either extracted from the published papers or obtained from the authors if not presented in the papers. Data used for the infant mortality analyses along with their source are presented in 'Additional tables'. We used the fixed-effect method for combining data where trials were examining the same intervention and the trial populations and methods were judged to be sufficiently similar.
The review objective was to evaluate the effect of vitamin A supplementation in term neonates. The studies included in this review had enrolled all births that were identified in their study settings without using a restriction for gestational age, of either < 37 or ≥ 37 weeks, which would have allowed us to use the term data only. Birthweight was used as a criterion in two studies: Benn 2008 enrolled normal birthweight neonates (birthweight ≥ 2500 g) and Benn 2010 recruited low birthweight neonates (birthweight < 2500 g) only. Data for term neonates was presented separately only in the published paper of one study, for the infant mortality outcome at six months (Klemm 2008), whereas mortality data for term neonates only for other studies was obtained by contacting the study authors. Information about the gestational age was not available in West 1995 (Keith West; personnel communication 2008). Considering the small number of studies included in the review and the availability of data for primary outcomes, we analysed data for term neonates, where available, followed by the analysis for all infants. For all secondary outcomes, data in published papers were presented for all infants together and have been analysed as such. As the inclusion criterion of Benn 2008 was birthweight at least 2500 g, we assumed that a greater proportion of neonates would be term babies and have analysed its data as such in our term neonate analysis. We used the term 'all infants' to refer to aggregated term and preterm infants data throughout this review.
There were two studies which had maternal supplementation with vitamin A, either in the postpartum period (Malaba 2005) or during pregnancy (Klemm 2008). Malaba et al randomised mother-infant pairs to the four treatment arms (described in detail in the table 'Characteristics of included studies') whereas Klemm et al randomised neonates within each of three previously randomised treatment arms of a maternal supplementation trial of vitamin A. This resulted in two neonatal treatment arms in Klemm 2008 which were balanced across the maternal supplementation arms. Both studies reported no significant interaction between maternal and neonatal supplementation with vitamin A and we have used data for all neonates included in these studies on the basis of their randomisation to the neonatal vitamin A intervention or control group.
Subgroup analysis and investigation of heterogeneity
We measured heterogeneity among the trials by calculating the I
We planned to investigate publication bias for outcomes with more than 10 included studies. However, we did not investigate the presence of this bias as the number of studies included was small.
Description of studies
Seven studies including 51,446 neonates were included in this review.
The study by Humphrey et al was conducted in a single tertiary care hospital in Indonesia (Humphrey 1996) as a safety trial for vitamin A supplementation at the time of birth. This was a randomised double blind placebo controlled trial of 2067 infants with birthweight > 1500 g and without any critical illness. The infants were randomly assigned to receive a single oral dose of vitamin A (50,000 IU) or placebo within 24 hours of delivery. The two groups were similar at baseline for maternal, infant and household characteristics.
The study conducted by West et al (West 1995) in Nepal was part of a large cluster randomised, double blind, placebo controlled trial of vitamin A supplementation in preschool children. A total of 11,918 infants less than six months of age, of which 1621 were neonates, were enrolled and administered vitamin A (50,000 IU in < one month old infants and 100,000 in one to five month old infants) or placebo. Baseline characteristics of the two groups were similar.
The study conducted in India by Rahmatullah et al (Rahmathullah 2003) was also a randomised, double blind, placebo controlled trial in which all live born infants resulting from pregnancies within the participating villages were eligible for inclusion. A total of 11,619 newborn infants born to consenting mothers who were residing in the study area were enrolled. Infants were given two doses of vitamin A or placebo with the first dose being administered within the first 48 hours of delivery and the second dose within 24 hours of the first dose. Baseline characteristics of the families, mothers and infants were similar between the treatment groups.
The Zimbabwe study (Malaba 2005) was a randomised, double bind, placebo controlled trial using a two by two factorial design. Mother-infant pairs were eligible for inclusion if the mother planned to reside in the study area after delivery. None of the two had any life threatening illness and the infant's birthweight was > 1500 g. Around 14,110 infant-mother pairs were enrolled within 96 hours of delivery and were assigned to either of the following groups: Aa (vitamin A supplementation to both the mother and infant), Ap (vitamin A to the mother and placebo to infant), pa (placebo to mother and vitamin A to infant) and pp (placebo to both the mother and infant). The vitamin A dose for mothers was 400,000 IU and for infants it was 50,000 IU. All the treatment groups were similar at baseline for maternal, household and other related variables.
The study conducted by Klemm et al in Bangladesh (Klemm 2008) was a cluster randomised, double bind, placebo controlled trial which was nested within an ongoing parent trial of vitamin A supplementation in pregnant women. All infants born to consenting mothers of the original trial were included in the current trial. A total of 15,948 infants were administered vitamin A (50,000 IU) or placebo at home as soon as possible after birth. Baseline characteristics of the mothers and infants in this study were comparable at baseline.
Two studies were conducted in Guinea Bissau by Benn et al. Benn 2008 was a randomised, double blind, placebo controlled trial which included 4345 normal birthweight infants (birthweight at least 2500 g). For births occurring at the national hospital or local health centres, mothers were invited to participate in the study at the time of Bacille Calmette-Guérin (BCG) vaccination. For home births, mothers were invited to participate at the time of their visit to the local health centres for BCG vaccination. All infants with birthweight at least 2500 g, without any serious medical condition or malformation, for whom parental consent was available were randomised to either oral drops of vitamin A (50,000 IU) or placebo. The treatment groups were similar at baseline for various baseline characteristics. The other study by these investigators was conducted in parallel with Benn 2008. Benn 2010 was a two by two factorial, randomised, double blind, placebo controlled trial in low birth weight neonates (birthweight < 2500 g). This study included 1736 neonates randomised to either 25,000 IU vitamin A or placebo, as well as to early BCG vaccine or the usual late BCG vaccine.
See the table 'Characteristics of included studies' for further details.
There are four ongoing studies that are being conducted in Pakistan, India, Ghana and Tanzania. All four studies are randomised, double blind, placebo controlled trials, with the one in Pakistan using a cluster randomised design.
The inclusion criteria for Pakistan 2008 were live born infants without congenital malformations or serious birth injury from all pregnancies within participating villages, with a sample size of 7,400 infants. The study was designed as an effectiveness trial with vitamin A delivery through the Lady Health Workers program of the government of Pakistan. Intervention included routine postpartum care and vitamin A supplementation (50,000 IU) to the newborn within 48 to 72 hours of birth whereas the control group received routine postpartum care only. Outcomes to be evaluated were all-cause and cause-specific infant mortality at six months, incidence of serious infections (sepsis, pneumonia and diarrhoea), measurement of serum retinol values and rates of breastfeeding in the two groups. Recruitment and follow up for this study have been completed and data analysis is in progress.
The studies being planned in India, Ghana and Tanzania also aim to evaluate the effect of supplementation of 50,000 IU of vitamin A against control. The Indian study is being conducted in two districts in the state of Haryana with an estimated sample size of 40,200 neonates (India 2010). All births in the study area contacted by the enrolment team within the eligible age window and with the parent's consent to participate will be included. The eligible age window has been defined as up to 60 hours after birth. Outcomes that will be evaluated include infant mortality at six months, mortality in the neonatal period (during the first month of life); incidence of severe morbidity, defined as hospitalizations due to any illness in the first six months of infancy; potential adverse effects of vitamin A; and vitamin A status in a subgroup of newborns at two weeks and three months of age and their caregivers.
The methodologies of the studies in Ghana and Tanzania were found to be similar. The Ghana study will be conducted in seven contiguous districts in the Brong Ahafo region of central rural Ghana (Ghana 2010) with a target sample size of 28,000 neonates. The Tanzanian study will be conducted in Dar-es-Salaam and the Kilombero and Ulanga districts in Ifakara (Tanzania 2010). The estimated sample size for this study is 32,000 neonates. Inclusion criteria in both studies are all births in the study area that are contacted by the study team on the day of birth or in the next two days. Both singleton and multiple births are eligible for inclusion and each infant will be provided a unique identification number. Intervention includes vitamin A 50,000 IU once orally within the first three days of life, keeping a minimum period of two hours between birth and dosing. Similar outcomes will be evaluated in the two studies, which are all-cause infant mortality assessed at six months of age, all-cause neonatal mortality assessed at one month of age, incidence of severe morbidity defined as hospitalisations due to any illness in the first six months of infancy, potential adverse effects of vitamin A, and vitamin A and C reactive protein (CRP) status in a subsample of infants at two weeks and three months of age.
See the table 'Characteristics of ongoing studies' for further details.
Two studies were excluded from the review (Bezzera 2009; Bhaskaram 1998). Bezzera 2009 included vitamin A supplementation of mothers only in the immediate postpartum period; their neonates were not supplemented. Bhaskaram 1998 supplemented mothers only with vitamin A within 24 hours of delivery while all neonates were given oral poliovirus vaccine (OPV) between 48 and 72 hours after birth.
Risk of bias in included studies
Three studies adequately randomised neonates to the treatment groups (Benn 2008; Benn 2010; Malaba 2005) with a clear description of the method used for generating the randomisation sequence. Four studies did not provide sufficient details to allow judgement of the adequacy of their methods (Humphrey 1996; Klemm 2008; Rahmathullah 2003; West 1995). The method of allocation concealment was clearly described in four studies (Benn 2008; Benn 2010; Humphrey 1996; Malaba 2005) whereas it was not described in sufficient detail in Klemm 2008, Rahmathullah 2003 and West 1995. Blinding of participants, study personnel and outcome assessors was clearly described and achieved in all included studies. The post randomisation attrition and exclusion of participants was: 1.6% (Benn 2008), 18.7% (Benn 2010), 11% (Humphrey 1996), 7% (Klemm 2008), 41.8% (Malaba 2005) and 18.9% (Rahmathullah 2003), with reasons for attrition and exclusion of the participants described in the papers. Exclusion and attrition were 1.04% in West 1995 and details of these were not provided. Evaluation of selective outcome reporting by reviewing either trial registration documents, if available, or methodology in published papers showed that all trials have reported their findings for prespecified or expected outcomes except for Humphrey 1996 where it was unclear. We identified three trials with potential high risk of other bias: Benn 2008 and Benn 2010 conducted post hoc analyses after assuming that vitamin A might be more beneficial to boys whereas Klemm 2008 was terminated after randomisation of two-thirds of the planned number of infants due to the significantly higher mortality in the control group. Malaba 2005 was found to be free of other bias and the risk of other bias was uncertain in the remaining three trials due to insufficient information (Humphrey 1996; Rahmathullah 2003; West 1995).
See the table 'Characteristics of included studies' for further details on risk of bias in included studies. A graphical presentation of our individual judgments per item per study is provided in Figure 1 and a summary graph is given in Figure 2.
|Figure 1. Methodological quality summary: review authors' judgements about each methodological quality item for each included study.|
|Figure 2. Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies.|
Effects of interventions
A summary of findings table based on the outcomes in term neonates has been included in this review, which is in accordance with the methodology recommended by GRADE ( Summary of findings for the main comparison).
Neonatal vitamin A supplementation versus placebo
All-cause infant mortality at six months of age
An overview of the type and source of data for this outcome is presented in Table 1.
Six included studies (Benn 2008; Benn 2010; Humphrey 1996; Klemm 2008; Malaba 2005; Rahmathullah 2003) measured infant mortality at six months of age. West 1995 measured mortality at four months of age and this data has been included in the six month mortality analysis.
All-cause infant mortality at six months of age: risk ratios based on cumulative risk (%) (Outcome 1.1)
Data from five studies were measured as risk ratios based on cumulative risk.
The pooled estimate of data for term infants from three studies (Humphrey 1996; Klemm 2008; Malaba 2005) suggests that the risk of death from any cause at six months of age for neonates who were supplemented with vitamin A is 18% lower than control, which is statistically significant (typical RR 0.82; 95% CI 0.68 to 0.99) ( Analysis 1.1.1). The level of statistical heterogeneity in this analysis was 63%. As the number of studies included was small, a subgroup analysis to investigate heterogeneity was not considered reliable. Given substantial statistical heterogeneity and the small number of included studies, these findings should be interpreted with caution.
The pooled estimate of the data for all infants from five studies (Humphrey 1996; Klemm 2008; Malaba 2005; Rahmathullah 2003; West 1995) showed a statistically significant reduction of 14% in the risk of death from any cause for neonates supplemented with vitamin A as compared to control (typical RR 0.86; 95% CI 0.77 to 0.97). The level of statistical heterogeneity for this analysis was less than 50% (I
There was only one high quality study included in these analyses (Malaba 2005), hence a sensitivity analysis on the basis of study quality was not undertaken.
All-cause infant mortality at six months of age: rate ratios (per years of follow up) (Outcome 1.3)
Data from four studies were analysed as rate ratios (per year of follow up).
Pooled estimates for term neonates from Benn 2008 and Rahmathullah 2003 showed no evidence of a significant effect on the rate of death from any cause at six months of age in those that received vitamin A as compared to control (typical rate ratio 0.91; 95% CI 0.73 to 1.13) ( Analysis 1.3.1). Analysis for all infants data from four studies (Benn 2008; Benn 2010; Rahmathullah 2003; West 1995) also did not reach statistical significance (typical rate ratio 0.91; 95% CI 0.77 to 1.06) ( Analysis 1.3.2). The levels of statistical heterogeneity were: I
Analysis of two high quality studies showed similar effects of vitamin A on the rate of death at six months as compared to control (typical rate ratio 0.89; 95% CI 0.75 to 1.05) (Rahmathullah 2003; Benn 2008; Benn 2010) (data not shown).
All-cause infant mortality at 12 months of age (Outcomes 1.3 and 1.4)
An overview of the type and source of data for this outcome is presented in Table 2.
Analysis of term neonate data from two studies as rate ratios (Benn 2008; Humphrey 1996) showed no evidence of a significant effect on infant mortality from any cause at 12 months of age in neonates supplemented with vitamin A as compared to control, with statistical heterogeneity of 82% (typical rate ratio 0.95; 95% CI 0.72 to 1.26) (Analysis1.5.1). Further subgroup analysis was not undertaken due to the small number of studies included. The pooled estimate for all infants data from four studies also showed no evidence of a significant effect of supplementation of neonates with vitamin A on infant mortality at 12 months of age compared to controls (typical risk ratio 1.02; 95% CI 0.87 to 1.20; typical rate ratio 1.03; 95% CI 0.87 to 1.23) ( Analysis 1.4 and 1.5.2). The level of statistical heterogeneity was lower than 50% (I
Cause-specific infant mortality at six months of age: diarrhoea and acute respiratory infections (Outcomes 1.5 and 1.6)
Infant mortality related to diarrhoea and acute respiratory infections at six months of age was measured by two studies (Humphrey 1996; Rahmathullah 2003). Data for all infants for Humphrey 1996 were measured as risk ratios based on cumulative risk, which showed no significant effect of vitamin A supplementation on diarrhoea and respiratory infections as compared to control (diarrhoea-specific infant mortality: risk ratio 0.20; 95% CI 0.02 to 1.68; and acute respiratory infection-specific infant mortality: risk ratio 0.66; 95% CI 0.11 to 3.91). Data for all infants from Rahmathullah 2003 were presented as rate ratios (per years of follow up) and showed a similar non-significant effect of vitamin A on the rate of diarrhoea-specific and acute respiratory infection-specific infant mortality at six months of age as compared to control (diarrhoea-specific infant mortality: rate ratio 0.67; 95% CI 0.32 to 1.39; and acute respiratory infection-specific infant mortality: rate ratio 1.00; 95% CI 0.56 to 1.79).
Cause-specific infant mortality at 12 months of age: diarrhoea and acute respiratory infections (Outcomes 1.8 and 1.9)
Data for all infants for Humphrey 1996 were measured as risk ratios based on cumulative risk and showed no evidence of a significant effect of vitamin A on death due to diarrhoea and acute respiratory infections as compared to control (diarrhoea-specific infant mortality: risk ratio 0.40; 95% CI 0.08 to 2.03 and acute respiratory infection-specific infant mortality: risk ratio 0.66; 95% CI 0.11 to 3.95). Benn 2008 and Malaba 2005 analysed data for all infants as rate ratios. Pooled data suggested no evidence of a significant effect of vitamin A on diarrhoea-specific and acute respiratory infections-specific infant mortality at 12 months of age as compared to control (typical rate ratios: 1.32; 0.80 to 2.16; I
Cause-specific infant morbidity at 6 months of age: diarrhoea and acute respiratory infection (Outcomes 1.10 and 1.11)
Two trials (Malaba 2005; Rahmathullah 2003) measured infant morbidity at six months of age as rate ratios (per year of follow up). Pooled estimates showed no significant effect of vitamin A as compared to control on the rate of diarrhoea and acute respiratory infections in infants at six months of age (typical rate ratio: 1.05; 0.99 to 1.10; I
Vitamin A deficiency (Outcomes 1.12 and 1.13)
Vitamin A deficiency defined as serum retinol value < 0.70 µmol/L was available for all infants from one study only (Benn 2008), which showed no evidence of a significant effect of vitamin A supplementation on vitamin A deficiency as compared to control (at 6 weeks: risk ratio 0.94; 0.75 to 1.19; and at four months: risk ratio: 1.02; 0.64 to 1.62).
Anemia (Outcome 1.14)
The impact on anaemia was measured in only one study (Malaba 2005), for all infants born to both HIV positive and negative women. Vitamin A supplementation in neonates did not lead to a significant impact on anaemia (haemoglobin (Hb) < 105 g/L) at 8 to 14 months of age (risk ratio 0.97; 95% CI 0.87 to 1.07) compared to control.
Adverse events (Outcomes 1.15 and 1.16)
Data for adverse events in all infants during the first 48 to 72 hours could be pooled only from two studies (Benn 2008; Humphrey 1996), and only one study (Benn 2008) presented adverse events at one month of age. Pooled estimates suggested no evidence of a significant increase in adverse events during the first 48 to 72 hours, specifically bulging fontanelle (typical risk ratio 1.41; 95% CI 0.91 to 2.19), diarrhoea (typical risk ratio 0.93; 95% CI 0.63 to 1.38) and vomiting (typical risk ratio 0.88; 95% CI 0.74 to 1.05), in the vitamin A group versus the control group. Benn 2008 showed no evidence of a significant increase in adverse events during the first month post supplementation (risk ratio diarrhoea 1.07; 0.46 to 2.51; and vomiting 1.22; 0.57 to 2.58).
The included studies did not measure the impact of neonatal vitamin A supplementation on blindness and xerophthalmia.
The objective of this review was to evaluate the effect of supplementing term neonates with vitamin A as compared to unsupplemented controls. As the term neonatal outcome data were only available for a small number of studies, and then for infant mortality outcomes only, we analysed and presented estimates for both term neonates (where specified) and all infants for the various prespecified outcomes. Our analysis for all infants provided evidence of a 14% (95% CI 3% to 23%) reduction in the risk of death at six months of age in the vitamin A supplemented group as compared to the controls. This was statistically significant. Analysis of the term neonatal outcome included data from a subset of studies included in the all infant analysis and also showed a significant reduction in the risk of death in the first six months of life (reduction of 18%; 95% CI 1% to 32%). These findings should be interpreted with caution due to the small number of studies contributing data to these analyses, statistical heterogeneity and wide confidence intervals that are close to the null effect. Three studies (Benn 2008; Benn 2010; Rahmathullah 2003) had analysed data as rates (per year of follow up) which precluded their inclusion in this analysis. They have been analysed separately, whereas the study by West et al did not include information on gestational age. Further assessment of the effect in term neonates is needed, with data for all studies included in the review.
Analysis of the effect on infant mortality at 12 months of age of vitamin A supplementation in neonates provided no evidence of a significant effect on this outcome. Of four included studies, only one by Humphrey et al in Indonesia (Humphrey 1996) showed a highly significant effect on mortality at the end of the first year of life. Overall, the review findings suggest a potential effect of this intervention in the first half of infancy only.
Deficiency of vitamin A is a major nutritional concern in many countries of the world. All studies included in this review were conducted in developing countries with varying levels of vitamin A deficiency and infant mortality. The first trial of vitamin A in neonates was conducted as a safety trial in Indonesia (Humphrey 1996). This trial had shown significant reduction in the risk of mortality, with a difference in survival between the groups notable after the first month of life and becoming consistent after four months of age. However, maternal serum retinol levels from a subset of this study population showed mean (± SD) levels of 1.79 (± 0.53) and 1.75 (± 0.56) µmol/L in the vitamin A and control groups respectively, suggesting little vitamin A deficiency. The infant mortality rate in the control group (7.2 per 1000 child years) was well below that in the general population and the authors state that the families included in their study were relatively privileged (Humphrey 1996). It has been suggested that though the serum retinol levels were adequate, this finding does not preclude low hepatic reserves in this study group (Tielsch 2008a). Of the other included trials, two were conducted in India (Rahmathullah 2003) and Bangladesh (Klemm 2008), which are characterized by high infant mortality and vitamin A deficiency. The trial in India showed a reduction of 22% in the risk of infant mortality at six months whereas the trial in Bangladesh showed a 15% reduction in the vitamin A group compared with the control group. In the Indian setting, 5% to 6% of included women reported a history of night blindness, which is a clinical manifestation of vitamin A deficiency, and it was not significantly different between the two groups (P = 0.26). The authors noted that the impact on mortality was evident from two weeks of age and continued until three months, after which no further effect was observed. Similar observations were noted in the Bangladesh trial, which reported a difference in the mortality of infants as early as after the first week of life that persisted till four months of age. There were approximately 9.5% of the pregnant women in each group who reported night blindness in their most recent pregnancy. A subsample of this study population was measured for serum retinol in the first trimester and showed a suboptimal vitamin A status (defined as serum retinol < 1.05 µmol/L) in approximately 41% of women in the vitamin A group and 36% of women in the control group. These findings of an effect of vitamin A on the risk of mortality between the early weeks of life until four months does indicate a common biological mechanism of vitamin A.
Conflicting results were shown by studies conducted in Nepal, Zimbabwe and Guinea Bissau. The Nepal trial (West 1995) was conducted as part of a larger vitamin A supplementation trial in preschool children. The infants supplemented in the neonatal period did not show a significant effect on infant mortality, which was evaluated at four months of age. To note, this study setting was characterized by endemic vitamin A deficiency and high infant mortality. This finding was in contrast to findings from other studies conducted in similar settings. Studies conducted in Zimbabwe (Malaba 2005) and Guinea Bissau (Benn 2008) reported non-significant effects of vitamin A on both six and 12 months infant mortality outcomes. Infant mortality was measured in the other trial by Benn et al in Guinea Bissau (Benn 2010) at 12 months, which also showed no effect on the outcome. The vitamin A status of mothers provided evidence of minimal deficiency in these study settings with mean (± SD) serum retinol values in a subset of Zimbabwean women in the control group of 1.09 (± 0.29) and 1.19 (± 0.42) µmol/L at six weeks postpartum. In Guinea Bissau, less than 1% women were found to have low retinol binding proteins (retinol binding proteins < 1.11 µmol/L). This study also measured serum retinol values in infants at six weeks and four months of age and showed no evidence of an effect of vitamin A supplementation on vitamin A deficiency. An important feature of this trial was that all eligible children were provided free consultations and essential drugs for any illness during the first year of life. The mechanism through which vitamin A supplementation in children older than six months of age improves survival has partly been explained by a reduction in the severity rather than the incidence of infections (Sommer 1996). We believe that provision of free consultations and essential drugs for illnesses in the Guinea Bissau trial masked any beneficial effect of vitamin A supplementation that would have occurred through reducing severe illness episodes. The reasons for possible differences in the results of vitamin A supplementation trials conducted in different geographic regions are uncertain and could be a chance observation. However, these findings could represent genuine differences in population attributable risks of micronutrient deficiencies. The studies included neonates of mothers with varying levels of baseline vitamin A deficiency, both low birth weight and normal birth weight neonates and varying rates of baseline infant mortality. These factors could affect the generalizability of study findings. It should be noted that in contrast to the observed benefits of vitamin A supplementation among mothers in Nepal (West 1999), a large trial of maternal vitamin A supplementation in Ghana (Kirkwood 2010) did not show any benefits.
Reasons for these conflicting findings are unclear and there is no clear indication of the biological mechanisms through which vitamin A could lower the risk of death when given in the neonatal period. Various mechanisms have been proposed. Newborns have marginal reserves of vitamin A in their liver and they depend on breast milk as a source of this vitamin in the first few months of life. Hence, low maternal vitamin A levels translate into vitamin A deficiency in the newborns (Underwood 1994). Deficiency of vitamin A could also begin very early in life with the colostrum being discarded or breastfeeding being inadequate. Colostrum and early breast milk have been found to be very rich sources of vitamin A, which can significantly augment vitamin A stores in the neonates (Wallingford 1986). Along with inadequate breastfeeding, introduction of artificial feeds also hinders with the establishment of good breastfeeding practices, thereby denying infants of this critical source of vitamin A throughout the breastfeeding period (Haskell 1999). Artificial feeds early in life also increase the risk of gastrointestinal infections in these infants. Vitamin A supplementation has also been proposed to have an impact on infant mortality through the development and maintenance of the integrity of the intestinal and respiratory epithelia, and enhanced local and systemic immunity (Sommer 1996; Tielsch 2007). These pathways may provide an explanation of the effect in settings where the practice of discarding colostrum, inadequate breastfeeding or artificial feeds and infections are common. Alternatively, the early initiation of feeding of colostrum and exclusive breastfeeding could explain an absence of a beneficial effect of vitamin A received as a supplement. However, in this review we could not study these proposed mechanisms as only a few included studies presented limited information on breastfeeding practices and the use of artificial feeds.
There has been considerable debate on the issue of supplementing neonates with vitamin A due to conflicting findings from the studies and variability in the results of pooled analyses (Abrams 2008; Bhutta 2008; Gogia 2009; Sachdev 2008; Tielsch 2008a). The current review includes data from several new studies published since these earlier reviews and additional data that were obtained by contacting the study authors. Our findings corroborate those published in the earlier Lancet Undernutrition Series (Bhutta 2008) that included three neonatal supplementation trials published until then and used available data for analysing infant mortality outcomes separately at six months and 12 months. It showed a reduction of 20% (95% CI 4% to 34%) in the vitamin A supplemented group compared to the unsupplemented control group at six months of age. The earlier review also showed a non-significant impact of vitamin A on infant mortality at 12 months of age compared to the unsupplemented control (RR 0.90; 95% CI 0.61 to 1.32). A similar review by Gogia et al (Gogia 2009) analysed infant mortality data from six trials by pooling all deaths between the period of initiation of intervention till the last follow up, either at six or 12 months, and suggested no protective effect on mortality during the first year of life (RR 0.92; 95% CI 0.75 to 1.12). We believe that such an approach would mask any beneficial effects of the intervention in early infancy and could also be influenced by routine vitamin A supplementation practices after six months of age in the given population. Given that vitamin A supplementation to infants after six months of age is relatively well established in developing countries, adjunctive benefits of neonatal vitamin A supplementation could provide complementary benefits in young infants.
Limited data were available for the outcomes of cause-specific mortality and morbidity; and vitamin A deficiency, measured as serum retinol values in infants. Data on adverse events, specifically bulging fontanelle, vomiting or diarrhoea, were also limited and showed no significant increase within the first 48 to 72 hours of supplementation. An evaluation of these outcomes is needed with the inclusion of data from all studies that have measured these outcomes and additional new trials that are being planned in Ghana (Ghana 2010), India (India 2010) and Tanzania (Tanzania 2010). These trials have been designed to evaluate the effect of 50,000 IU of vitamin A on infant mortality at six months of age, their primary outcome. Another smaller effectiveness trial in Pakistan (Pakistan 2008) has completed recruitment and follow up and the data are being analysed. These additional trials will greatly contribute towards the evidence base and consensus on the value or otherwise of neonatal vitamin A supplementation.
Implications for practice
Considering the high burden of deaths of children under the age of five years in developing countries, and mortality in infancy being a major contributory cause, it is critical to obtain sound scientific evidence of the effect of vitamin A supplementation in the neonatal period on infant mortality and morbidity. Evidence provided in this review does indicate a potential beneficial effect of supplementing neonates with vitamin A at birth in reducing mortality in the first half of infancy. Considering the absence of a clear indication of the biological mechanism through which vitamin A could affect mortality in early infancy, substantial conflicting findings from individual studies in settings with potentially varying levels of maternal vitamin A deficiency and infant mortality; and given that data from at least four new trials will be available in the foreseeable future, we propose to delay any policy recommendations for neonatal vitamin A supplementation pending these findings.
Implications for research
Future research and trials should examine the effects of vitamin A supplementation in the neonatal period on infant mortality in the first half of infancy. These trials should also include measures of maternal micronutrient status (vitamin A deficiency), the effect of maternal vitamin A supplementation, dose of vitamin A, maternal HIV status, breastfeeding patterns and breast milk vitamin A concentrations. Efforts should be made to stratify effects by age after birth of vitamin A administration, prematurity and intrauterine growth retardation. Research should also be conducted to identify biologic mechanisms and indicators for vitamin A in reducing the risk of death and to explain the differences observed in vitamin A supplementation trials conducted in settings with varying levels of baseline vitamin A deficiency.
We would like to thank the trial authors who provided additional data for this review; the Cochrane Editorial Unit, particularly Toby Lasserson who provided statistical advice and drafted the summary of findings tables, Karla Soares-Weiser and Harriet MacLehose for comments on the 'Risk of bias' tables; and Furqan Bin Irfan who assisted in protocol writing. In addition, we would like to thank the staff at the editorial office of the Neonatal Cochrane Review Group for their support in the preparation of this review and, in particular, the Coordinating Editor Dr Roger Soll and Managing Editor Diane Haughton. We would also like to thank the peer reviewers who provided helpful feedback on the review.
Data and analyses
- Top of page
- Summary of findings [Explanations]
- Authors' conclusions
- Data and analyses
- Contributions of authors
- Declarations of interest
- Sources of support
- Differences between protocol and review
- Index terms
Protocol first published: Issue 1, 2008
Review first published: Issue 10, 2011
Contributions of authors
The review protocol was written by Batool A Haider (BAH) under the guidance of Zulfiqar A Bhutta (ZAB). Data extraction was done by BAH and ZAB. BAH entered the data, created the comparisons, did data analysis and wrote the text of the review. ZAB provided supervision and contributed to the writing process of the review.
Declarations of interest
Zulfiqar A Bhutta is a principal investigator of an ongoing neonatal vitamin A supplementation trial in Pakistan.
Sources of support
- The Aga Khan University Hospital, Pakistan.
- World Health Organization, Switzerland.A grant was provided by the WHO to fund the completion of this Cochrane Review.
Differences between protocol and review
Medical Subject Headings (MeSH)
*Developing Countries; *Infant Mortality; *Morbidity; Infant, Newborn; Randomized Controlled Trials as Topic; Vitamin A [*administration & dosage]; Vitamin A Deficiency [prevention & control]; Vitamins [*administration & dosage]
MeSH check words
* Indicates the major publication for the study