1. Objectives

To evaluate vitamin A supplementation in very low birth weight infants on the incidence of mortality, neonatal chronic lung disease, and long-term neurodevelopmental disability, also considering the effects of route, dose, and timing of supplementation

2. How studies were identified

The following databases were searched in May 2016:

• CENTRAL (The Cochrane Library 2016, Issue 4)
• MEDLINE
• EMBASE
• CINAHL
• ClinicalTrials.gov
• WHO International Trials Registry and Platform
• ISRCTN Registry

Relevant journals and conference proceedings were also searched

3. Criteria for including studies in the review

3.1 Study type

Randomized controlled trials, including quasi-randomized trials

3.2 Study participants

Very low birth weight infants (≤1500 g or <32 weeks’ gestation)

3.3 Interventions

Vitamin A supplementation compared to placebo or no supplementation, or compared with different vitamin A dosing regimens

3.4 Primary outcomes

• Neonatal death (up to 28 days post-birth)
• Chronic lung disease (oxygen use at 28 days post-birth or at 36 weeks’ post-menstrual age)
• Death or chronic lung disease (oxygen use at 28 days post-birth or at 36 weeks’ post-menstrual age)

Secondary outcomes included serum retinol concentrations (in studies comparing dosages), nosocomial infection, patent ductus arteriosus, necrotising enterocolitis, intraventricular haemorrhage, periventricular leukomalacia, retinopathy of prematurity, neurodevelopment at 18 to 24 months, death or neurodevelopmental impairment at 18 to 24 months, and adverse effects including hypervitaminosis A (raised intracranial pressure, mucocutaneous lesions)

4. Main results

4.1 Included studies

Eleven randomized controlled trials, enrolling 1580 infants, were included in this review

• Ten trials enrolling 1460 infants compared vitamin A supplementation with a control and one trial enrolling 120 infants compared three different vitamin A dosing regimens
• Birth weights ranged from 401 g to up to 1500 g, and gestational ages ranged from 26 weeks to ≤34 weeks
• Most studies excluded infants with congenital anomalies or infection, and several studies included only those infants requiring at least 72 hours of mechanical ventilation in the first week of life
• Intramuscular injection was the route of vitamin A supplementation in eight of the 10 trials comparing vitamin A versus control (2000 to 10,000 IU vitamin A per injection, three to 14 injections occurring between day four to up to one month post-birth); control injections used normal saline or were mock injections
• Oral vitamin A (5000 IU/day until day 28) and intravenous vitamin A (1300 to 3300 IU/kg/day from 48 to 96 hours) were used in the other two trials comparing vitamin A supplementation to control; controls were an oral placebo and no intravenous vitamin A, respectively
• Most infants in both treatment and control groups received some vitamin A via parenteral nutrition
• In the vitamin A dosage regimen trial, infants were randomized to receive for four weeks via intramuscular injection a control vitamin A dose of 5000 IU 3 days/week (standard regimen), or 10,000 IU 3 days/week, or 15,000 IU 1 day/week

4.2 Study settings

• Greece, South Africa, Thailand, the United Kingdom of Great Britain and Northern Ireland (2 trials) and the United States of America (6 trials)
• Studies were hospital-based

4.3 Study settings

How the data were analysed
Three comparisons were made: i) supplemental vitamin A versus placebo or no supplementation; ii) a standard vitamin A dose of 5000 IU, 3 days/week versus 10,000 IU of vitamin A, 3 days/week; and iii) a standard vitamin A dose of 5000 IU, 3 days/week versus 15,000 IU of vitamin A, 1 day/week. Dichotomous outcomes were summarized using risk ratios (RR) and risk differences (RD) with corresponding 95% confidence intervals (CI). Using the RD, the number needed to treat to benefit was calculated. Continuous outcomes were summarized using mean differences (MD) with corresponding 95% CI. Fixed-effects models were used for all meta-analyses. Were sufficient data available, investigation of heterogeneity was planned using sensitivity analyses by risk of bias. Subgroup analyses by route, dose, and timing of supplementation were also planned to investigate potential sources of heterogeneity.

Results
Supplemental vitamin A versus placebo or no supplementation
Neonatal death (up to day 28)
Overall, vitamin A supplementation did not affect the risk of neonatal mortality (RR 0.86, 95% CI [0.66 to 1.11], p=0.24; RD -0.02, 95% CI [-0.06 to 0.02], 6 trials/1165 infants). In subgroup analysis by route of supplementation, no effect was found with intramuscular injection (RR 0.86, 95% CI [0.62 to 1.18], 5 trials/1011 infants) or with oral supplementation (RR 0.86, 95% CI [0.56 to 1.33], 1 trial/164 infants).

Chronic lung disease (oxygen use at day 28)
In pooled analysis of seven trials involving 1070 infants, vitamin A supplementation had no statistically significant effect on the risk of chronic lung disease (RR 0.93, 95% CI [0.86 to 1.01], p=0.07; RD -0.05, 95% CI [-0.10 to 0.00]). In subgroup analysis by route of supplementation, no effect was found with intramuscular injection (RR 0.93, 95% CI [0.86 to 1.01], 5 trials/884 infants), intravenous injection (RR 0.95, 95% CI [0.54 to 1.70], 1 trial/86 infants), or oral supplementation (RR 0.95, 95% CI [0.80 to 1.11], 1 trial/100 infants).

Neonatal death (up to day 28) or chronic lung disease (oxygen use at day 28)
Meta-analysis of six trials involving 1165 infants demonstrated a statistically significant 7% reduction in the risk of death or chronic lung disease with vitamin A supplementation (RR 0.93, 95% CI [0.88 to 0.99], p=0.03; RD -0.05, 95%CI [-0.10 to -0.01]; number needed to treat to benefit 20, 95% CI [10 to 100]). In subgroup analysis by route of supplementation, intramuscular injection reached borderline statistical significance (RR 0.93, 95% CI [0.86 to 1.00], p=0.04; 5 trials/1011 infants), while oral supplementation had no clear effect (RR 0.96, 95% CI [0.86 to 1.06], 1 trial/164 infants).

Death before 36 weeks’ post-menstrual age
In pooled analysis of four trials involving 1089 infants, no effect of vitamin A supplementation on the risk of death before 36 weeks’ post-menstrual age was demonstrated (RR 1.00, 95% CI [0.77 to 1.29]). Subgroup analysis by route of supplementation also failed to show a treatment effect (intramuscular injection: RR 1.06, 95% CI [0.77 to 1.46], 3 trials/935 infants; oral: RR 0.86, 95% CI [0.56 to 1.33], 1 trial/164 infants).

Chronic lung disease (oxygen use at 36 weeks’ post-menstrual age)
Overall, the risk of chronic lung disease was reduced by a statistically significant 13% among the group receiving vitamin A supplementation (RR 0.87, 95% CI [0.77 to 0.99], p=0.03; RD -0.07, 95% CI [-0.13 to -0.01]; number needed to benefit 14, 95% CI [8 to 100]; 5 trials/986 infants). This remained statistically significant in trials employing intramuscular injections (RR 0.85, 95% CI [0.74 to 0.98], p=0.02; 4 trials/886 infants), but not in the single trial using oral supplementation (RR 1.00, 95% CI [0.81 to 1.24], 100 infants).

Death before 36 weeks’ post-menstrual age or chronic lung disease (oxygen use at 36 weeks’ post-menstrual age)
No clear effect on death or chronic lung disease up to 36 weeks’ post-menstrual age was found in pooled analysis of the four trials reporting on this outcome (RR 0.92, 95% CI [0.84 to 1.01], 1089 infants), or in subgroup analysis by route of supplementation (intramuscular injection: RR 0.90, 95% CI [0.81 to 1.01], 3 trials/935 infants; oral: RR 0.98, 95% CI [0.86 to 1.12], 1 trial/164 infants).

Additional outcomes
No statistically significant effect of vitamin A supplementation was found for the following outcomes: death at 18 to 24 months and neurodevelopmental impairment at 18 to 24 months; failure of ductus arteriosus closure by day 14; necrotising enterocolitis; intraventricular haemorrhage; periventricular leukomalacia; and retinopathy of prematurity requiring laser therapy. A 36% reduction in the risk of retinopathy of prematurity, any grade, was observed among infants receiving intramuscular vitamin A (RR 0.64, 95% CI [0.42 to 0.99], p=0.044; 3 trials/159 infants), but no difference was found overall (RR 0.81, 95% CI [0.65 to 1.01], p=0.062; 4 trials/255 infants) or among those receiving oral supplementation (RR 0.95, 95% CI [0.75 to 1.20], 1 trial/96 infants). There was a non-significant trend towards a reduction in the risk of one or more episodes of sepsis in pooled analysis of three trials (RR 0.89, 95% CI [0.76 to 1.04], p=0.13; 947 infants).

Adverse effects
In five studies reporting on adverse effects including fontanelle pressure/tension, clinical or biochemical toxicity, seizures and vomiting, no statistically significant differences were noted between treatment and control groups. In one trial including 49 infants, vitamin A injections were observed to be more painful than normal saline or mock injections.

Higher dose vitamin A (10,000 IU, 3 days/week) versus a standard dose (5000 IU, 3 days/week), and a once-weekly higher dose of vitamin A (15,000 IU, 1 day/week) versus a standard dose (5000 IU, 3 days/week)
Death before 36 weeks’ post-menstrual age
Higher dose vitamin A versus the standard dose had no effect on the risk of death (10,000 IU: RR 0.73, 95% CI [0.33 to 1.62], 1 trial/80 infants; 15,000 IU: RR 0.73, 95% CI [0.33 to 1.62], 1 trial/80 infants).

Chronic lung disease (oxygen use at 36 weeks’ post-menstrual age)
The risk of chronic lung disease was not statistically significantly different between treatment groups (10,000 IU: RR 2.72, 95% CI [0.81 to 9.08], 1 trial/61 infants; 15,000 IU: RR 1.81, 95% CI [0.50 to 6.59], 1 trial/61 infants).

Death before 36 weeks’ post-menstrual age or chronic lung disease (oxygen use at 36 weeks’ post-menstrual age)
The combined risk of death or chronic lung disease was not different between higher and standard vitamin A dosage regimens (10,000 IU: RR 1.21, 95% CI [0.70 to 2.12], 1 trial/80 infants; 15,000 IU: RR 1.00, 95% CI [0.55 to 1.82], 1 trial/80 infants).

Serum retinol concentrations (μg/L)
There was no significant difference in serum retinol concentrations between higher (10,000 IU vitamin A) and standard dosing regimens at day 28. The once-weekly dose (15,000 IU vitamin A) resulted in a significantly lower serum retinol concentration than the standard regimen, however (MD -0.97 μg/L, 95% CI [-1.56 to -0.38], 50 infants), and a higher proportion of infants had serum retinol concentrations <200 μg/L (RR 2.52, 95% CI [1.24 to 5.09], 50 infants; 65% versus 26%).

Additional outcomes
No statistically significant differences between groups were found for the outcomes necrotising enterocolitis or sepsis. A non-significant trend towards a reduced risk of retinopathy of prematurity was observed with higher dosage regimens. Transient fullness of the anterior fontanelle was noted in one infant in the higher-dose group.

5. Additional author observations*

The overall methodological quality of the included trials was fair, with eight of the 11 studies reporting adequate methods of allocation concealment. For the comparison supplemental vitamin A versus no supplementation, GRADE quality of evidence criteria were used to assess the main outcomes. The evidence for the outcomes neonatal death, chronic lung disease (at both 28 days and before 36 weeks’ post-menstrual age), death or chronic lung disease (oxygen use at 28 days), death before 36 weeks’ post-menstrual age, and death or chronic lung disease (before 36 weeks’ post-menstrual age) was rated as moderate quality, while for neurodevelopmental impairment at 18 to 24 months the evidence was rated as low quality. The evidence was downgraded in all cases on the basis of imprecision.

Vitamin A supplementation of very low birth weight infants may reduce the combined risk of neonatal mortality and chronic lung disease at one month of age, and may reduce the risk of chronic lung disease at 36 weeks’ post-menstrual age. Supplementation with vitamin A may also reduce the incidence of retinopathy of prematurity and sepsis. The effect sizes were small, and long-term follow-up data from the largest included trial demonstrated no evidence of an effect of vitamin A supplementation on neurodevelopmental impairment or death from 18 to 22 months’ corrected age. The decision to use vitamin A supplementation for the prevention of chronic lung disease may therefore depend on the local incidence of this outcome.

Further trials in low- and middle-income countries investigating the effect of vitamin A supplementation for the prevention of retinopathy of prematurity are warranted, as these populations suffer the greatest burden of this disease and have a higher risk of poor perinatal vitamin A status. In addition, the effects of delivering vitamin A in an intravenous lipid emulsion compared with delivery via intramuscular injection should be investigated, including outcomes such as safety and acceptability.