1. Objectives

To evaluate the effect of vitamin A supplementation for preventing morbidity and mortality in children aged six months to five years of age living in areas endemic for vitamin A deficiency

2. How studies were identified

The following databases were searched in March 2016:

• CENTRAL (The Cochrane Library 2016, Issue 2)
• MEDLINE
• EMBASE
• Science Citation Index
• Conference Proceedings Citation Index - Science (Web of Science)
• Cochrane Database of Systematic Reviews (2016, Issue 2)
• Database of Abstracts of Reviews of Effects
• LILACS
• African Index Medicus
• World Health Organization International Clinical Trials Registry (ICTRP)

Reference lists were also searched and the authors directly contacted researchers and organizations

3. Criteria for including studies in the review

3.1 Study type

Randomized controlled trials, including cluster-randomized trials

3.2 Study participants

Children aged six months to five years living in the community

(Studies that exclusively enrolled children in hospital and children with disease or infection were excluded, and study authors were requested to supply disaggregated data if some children met inclusion criteria while others did not. If disaggregated data were not available, the study was included if >50% of the children met the inclusion criteria)

3.3 Interventions

Oral synthetic vitamin A supplementation, at various doses and frequencies, compared to placebo or usual treatment

(Co-interventions must have been identical in both groups. Studies evaluating the effects of food fortification, the consumption of vitamin A rich foods, and beta-carotene supplementation were excluded)

3.4 Primary outcomes

• All-cause mortality

Secondary outcomes included cause-specific mortality due to diarrhoea, measles, meningitis and lower respiratory tract infection (LRTI); cause-specific morbidity including the incidence and prevalence of diarrhoea, measles, malaria, meningitis, LRTI, Bitot’s spots, night blindness, xerophthalmia and hospitalization; side effects and vitamin A deficiency status as measured by serum retinol. Pneumonia and LRTI outcomes were combined post hoc

4. Main results

4.1 Included studies

Forty-seven randomized trials, enrolling approximately 1,223,856 children, were included in this review

• Five trials did not report data that could be included in meta-analyses, reducing the number of included children to 1,223,607; and individual study sample size ranged from 35 to approximately one million children
• Seven trials compared vitamin A supplementation to usual treatment while 40 were placebo-controlled
• Twenty studies excluded children with clinical signs of vitamin A deficiency, one trial specifically recruited participants with biochemical vitamin A deficiency, four trials accepted individuals with clinical signs of vitamin A deficiency, and 23 trials did not specify vitamin A status as an inclusion or exclusion criterion
• Length of follow-up was around one year for most studies, and the median age of participants was 33 months
• The majority of trials (42/47) administered large doses of vitamin A ranging from 50,000 IU to 200,000 IU, and retinol palmitate was most commonly used

4.2 Study settings

• Australia (2), Bangladesh (2), Belize, Brazil (2), China (4), Congo, Ecuador, Ghana (2), Guinea-Bissau (4), Haiti, India (14), Indonesia (5), Mexico (2), Nepal (3), the Philippines, Sudan, and Thailand
• Eighteen studies were conducted in urban settings and 26 in rural settings

4.3 Study settings

How the data were analysed
Oral synthetic vitamin A supplementation was compared to placebo or treatment as usual for the prevention of morbidity and mortality in children. Data were meta-analysed using fixed effect models. For dichotomous outcomes, risk ratios (RR) and 95% confidence intervals (CI) were calculated, and for continuous outcomes, standardized mean differences (SMD) and corresponding 95% CI were calculated using Hedges g. For data provided without numerators and denominators, the generic inverse variance method was used. In overall meta-analyses, data from the last follow-up were used when multiple time points were available, and separate analyses were also performed grouped according to follow-up period (zero to 12 months; 13 to 59 months, and ≥60 months since randomization). Data from cluster-randomized trials were adjusted for clustering, and only data from the first time period of crossover trials were used in analyses. Investigation of heterogeneity was limited to study-level data to avoid aggregation bias, with the following subgroup analyses planned:

• By dose: standard (≤100,000 IU for children six to 11 months of age, and ≤200,000 IU for children 12 months to five years of age) versus high (above standard dose)
• By frequency of supplementation: low (≥ six months between doses) versus high (< six months between doses)
• By location: continent
• By age group: six to 12 months versus one to five years
• By sex

Sensitivity analyses were conducted for the primary outcome excluding studies at high risk of bias for sequence generation, and studies with imputed intra-class correlation coefficients (ICC). Random effects models were also conducted and funnel plots drawn for all outcomes with ten or more studies

Results
Mortality
All-cause mortality
Nineteen trials including 1,202,382 children reported on the primary outcome of all-cause mortality, in which supplementation with vitamin A was associated with a 12% reduction in the risk of death (RR 0.88, 95% CI [0.83 to 0.93], p<0.0001; 1,202,382 children). Moderate statistical heterogeneity was present (I²=61%, p=0.0003). Results were similar when grouped by follow-up periods of zero to 12 months (RR 0.83, 95% CI [0.75 to 0.92], 13 trials) and 13 to 59 months (RR 0.88, 95% CI [0.81 to 0.97], 6 trials).

Subgroup and sensitivity analyses
The overall effect did not differ significantly by location, age group, or sex (all p≥0.22 for differences between subgroups). Subgroup analysis by dose and frequency were not conducted due to lack of data. Pooled analysis of data from 17 studies set in countries with high child mortality demonstrated a similar effect (RR 0.89, 95% CI [0.84 to 0.94]), while meta-analysis of data from two studies in countries with low child mortality showed no effect (RR 1.00, 95% CI [0.14 to 7.08]). In sensitivity analyses, bias due to sequence generation did not influence the overall effect, and increasing the ICC did not appreciably change the effect size. The effect size was increased in random effects modelling (RR 0.76, 95% CI [0.66 to 0.88]), indicating that heterogeneity may be due to larger effect sizes in smaller studies; however, the confidence interval overlapped that of the fixed effect model.

Cause-specific mortality
Diarrhoea-related mortality was statistically significantly reduced in the vitamin A group compared to the control group (RR 0.88, 95% CI [0.79 to 0.98], p=0.023; 9 trials/1,098,538 participants). Meta-analysis of data from six trials reporting on measles mortality produced a non-significant reduction in risk with vitamin A supplementation (RR 0.88, 95% CI [0.69 to 1.11], p=0.27; 1,088,261 participants). Treatment with vitamin A also did not significantly reduce the risk of mortality from meningitis (RR 0.57, 95% CI [0.17 to 1.88], p=0.36; 3 trials), or from LRTI (RR 0.98, 95% [0.86 to 1.12], p=0.79; 9 trials/1,098,538 participants).

Morbidity
Infection and hospitalization
The incidence of diarrhoea was statistically significantly reduced by 15% with vitamin A supplementation (RR 0.85, 95% CI [0.82 to 0.87], p<0.00001; 15 trials/77,946 participants), although a high degree of heterogeneity was present (I²=94%, p<0.00001). In pooled analysis of three trials reporting on the prevalence of diarrhoea, vitamin A supplementation was associated with a six percent increase in the risk of having diarrhoea (RR 1.06, 95% CI [1.03 to 1.10], p<0.0002), also with a high degree of heterogeneity (I²=90%, p<0.001). In six trials reporting on the incidence of measles, a statistically significant 50% reduction in risk with vitamin A treatment was found (RR 0.50, 95% CI [0.37 to 0.67], p<0.00001; 19,556 children). The incidence of malaria was significantly reduced in the treatment group in one trial (RR 0.73, 95% CI [0.60 to 0.88], p=0.0013; 174,132 children), while in two trials reporting on malaria prevalence, no significant difference was found between groups (RR 0.73, 95% CI [0.41 to 1.28], p=0.27). The incidence of LRTI was not reduced with vitamin A supplementation (RR 0.99, 95% CI [0.92 to 1.06], p=0.74; 11 trials/27,540 children); however, the prevalence of LRTI at longest follow-up was reduced by 40% (RR 0.60, 95% CI [0.45 to 0.81]; p=0.00069; 2 trials). No trials reported data on measles prevalence, or on meningitis incidence or prevalence. Vitamin A supplementation did not significantly reduce the risk of being hospitalized once or more during follow-up (RR 0.64, 95% CI [0.40 to 1.02], 1 trial/1185 children), or the risk of being hospitalized for diarrhoea (RR 0.25, 95% CI [0.01 to 6.11], 1 trial/172 children) or for LRTI (RR 0.11, 95% CI [0.01 to 2.06], 1 trial/172 children).

Vision
The incidence of Bitot’s spots was measured in one trial, in which no significant effect of vitamin A treatment was observed (RR 0.93, 95% CI [0.76 to 1.14]); however, there was a statistically significant reduction in the prevalence of Bitot’s spots in combined analysis of five trials (RR 0.42, 95% CI [0.33 to 0.53], p<0.00001; 1,063,278 children). Night blindness was reduced in both incidence (RR 0.53, 95% CI [0.28 to 0.99], p=0.047; 1 trial) and prevalence (RR 0.32, 95% CI [0.21 to 0.50], p<0.00001; 2 trials/22,972 children) in the treatment group relative to controls. While three trials reported no combined effect on the incidence of xerophthalmia (RR 0.85, 95% CI [0.70 to 1.03], p=0.11), pooled analysis of two trials demonstrated a 69% reduction in its prevalence (RR 0.31, 95% CI [0.22 to 0.45], p<0.00001).

Vitamin A deficiency
In meta-analysis of four trials, a 29% reduction in the risk of vitamin A deficiency was observed (RR 0.71, 95% CI [0.65 to 0.78], p<0.00001; 2262 children), although heterogeneity was substantial (I²=78%, p=0.004). Serum vitamin A levels were significantly increased in the treatment group compared to controls (SMD 0.26, 95% CI [0.22 to 0.30], p<0.00001; 14 trials/11,788 children), again with significant heterogeneity (I²=95%, p<0.00001).

Adverse effects
Vomiting within 48 hours of treatment was increased by 97% in the vitamin A group compared to controls (RR 1.97, 95% CI [1.44 to 2.69], p<0.000022; 4 trials/4427 children). The risk of having a bulging fontanel did not differ significantly by treatment group in pooled analysis of four trials (RR 1.24, 95% CI [0.74 to 2.08], p=0.42; 2318 children).

5. Additional author observations*

Evidence for the primary outcome all-cause mortality was assessed using GRADE criteria as being of high quality. While moderate statistical heterogeneity was present, and the random effects model produced a different effect size to that of the fixed effect model, both analyses demonstrated a meaningful treatment effect of vitamin A on the risk of all-cause mortality in children. Given that many studies (20/47) excluded children with vitamin A deficiency at baseline, the effect may be greater among children in resource-poor settings who are at risk of vitamin A deficiency. Evidence for the outcome mortality due to diarrhoea was also regarded as high quality, while evidence for the outcomes measles incidence, Bitot’s spots incidence, night blindness incidence, vitamin A deficiency, and vomiting were judged as being of moderate quality. Evidence for the outcomes mortality due to measles, mortality due to LRTI, diarrhoea incidence, and LRTI incidence were assessed as being of low quality.

Vitamin A supplementation reduced all-cause mortality and the incidence of diarrhoea and measles in children aged six months to five years. Night blindness, Bitot’s spots and xerophthalmia were also reduced with vitamin A supplementation. Although few studies reported on adverse effects, the risk of vomiting was increased in the short-term following vitamin A supplementation. Overall, these findings support vitamin A supplementation for the prevention of morbidity and mortality in children under five years of age in areas at risk of vitamin A deficiency.

Further placebo-controlled trials in populations with known vitamin A deficiency would be unethical. Future trials could evaluate different doses of vitamin A, investigating the potential for a reduction in vomiting with smaller, yet still beneficial, doses. Other long-term outcomes such as growth could be evaluated in future editions of this review, and further reviews could investigate different delivery methods of vitamin A, such as food fortification and interventions to increase the access to and uptake of vitamin A-rich foods.