Vitamin A for Vitamin A Deficiency

If you have vitamin A deficiency, replacing vitamin A is the treatment, and the evidence that it works is strong. A 2022 Cochrane review of 47 trials in about 1.2 million children found supplements cut deaths from all causes by 12% (19 trials), and reduced deficiency itself (risk ratio 0.71; 4 trials), night blindness (0.32; 2 trials) and Bitot's spots (0.42; 5 trials) (Cochrane review, 2022). In night-blind pregnant women, 6 weeks of small doses 6 days a week brought dark adaptation back to the level of women who were not night-blind (Nepal dark-adaptation trial, 2005).

We rate it 4 out of 5 rather than 5 because most of the large trials tested prevention in children in low-income countries, not treatment of adults whose deficiency comes from surgery or malabsorption. For those adults the doses come from guidelines and case series, and one trial found that RDA-level doses failed to raise blood retinol in children with sickle cell disease (sickle cell trial, 2012). The question is rarely whether vitamin A works, but how much, in what form, and under whose supervision.

Vitamin A is part of rhodopsin, the light-sensitive protein in the retina, and it keeps the conjunctiva and cornea (the surface of the eye) healthy (NIH Office of Dietary Supplements). When stores run low, rhodopsin falls first, so night blindness is usually the earliest sign; continued deficiency dries and damages the eye surface and can end in permanent blindness (NIH Office of Dietary Supplements).

Supplying retinol directly, or beta-carotene that the body converts to retinol, refills liver stores and lets these tissues recover. That is why dark adaptation can return within weeks once intake is restored, from capsules or from foods such as liver, amaranth leaves and carrots (Nepal dark-adaptation trial, 2005); our guide to vitamin A foods lists the richest sources.

Knowing why vitamin A matters does not say how much a given person needs. Absorption depends on fat digestion, so people with pancreatic insufficiency or a malabsorptive bariatric operation may need more than the usual dose, and the US bariatric guideline advises checking iron and copper because low levels can slow recovery (ASMBS guideline, 2017).

Liver disease complicates things. In chronic liver disease, serum retinol and retinol-binding protein (the carrier that releases vitamin A from the liver) still indicate status, but the relative dose-response test cannot measure liver stores because the liver makes less of the carrier; the authors cautioned that supplementation may cause harm in these patients (liver disease study, 2015).

Doses depend on how severe the deficiency is. For eye signs of deficiency (xerophthalmia) in everyone except women of reproductive age, WHO gives an age-based dose immediately on diagnosis, the same dose the next day, and again at least 2 weeks later: 50,000 IU under 6 months, 100,000 IU at 6-12 months and 200,000 IU over 12 months (WHO treatment guide, 1997). WHO uses the same schedule for children with measles where deficiency may be present.

For adults after bariatric surgery without corneal changes, the US guideline recommends 10,000-25,000 IU a day by mouth until symptoms clearly improve, usually 1-2 weeks; with corneal changes, 50,000-100,000 IU by intramuscular injection for 3 days, then 50,000 IU a day by injection for 2 weeks (ASMBS guideline, 2017). These recommendations are graded D, meaning they rest on expert opinion rather than trials. Lower doses work for prevention: weekly 7,000 mcg retinol equivalents cut new night blindness in pregnancy by more than a third (Nepal pregnancy trial, 1998).

The strongest evidence is from community trials in children aged 6 months to 5 years. The 2022 Cochrane review pooled 47 trials and found moderate-certainty reductions in vitamin A deficiency (RR 0.71, 95% CI 0.65-0.78; 4 trials, 2,262 children), night blindness (RR 0.32, 0.21-0.50; 2 trials, 22,972 children) and Bitot's spots (RR 0.42, 0.33-0.53; 5 trials), plus a 12% drop in all-cause mortality (RR 0.88, 0.83-0.93; 19 trials, high certainty) (Cochrane review, 2022). An early trial in 25,000 Indonesian preschool children found twice-yearly 200,000 IU doses were associated with a dramatic fall in xerophthalmia (Sumatra trial, 1988).

Not every trial was so clear. In DEVTA, about 1 million children in north India given 200,000 IU every 6 months had half as much severe deficiency (6% vs 13%) but no significant drop in deaths (mortality ratio 0.96, 95% CI 0.89-1.03) (DEVTA trial, 2013). The biochemical effect was large; the effect on survival was smaller than earlier trials suggested.

In pregnancy, a 2015 Cochrane review of 19 trials found vitamin A reduced maternal night blindness (RR 0.79, 95% CI 0.64-0.98) and anaemia (RR 0.64), without changing maternal or newborn deaths (Cochrane pregnancy review, 2015). In one trial of 9,932 pregnancies in Nepal, weekly vitamin A cut night blindness in pregnancy from 10.7% to 6.7% (RR 0.62); weekly beta-carotene did less (RR 0.83, not significant) (Nepal pregnancy trial, 1998).

Two treatment trials tested people who were already deficient. In night-blind pregnant women, 6 weeks of vitamin A 6 days a week, from capsules, liver or vegetables, improved dark adaptation significantly (P<0.0001), to a level no different from women who were not night-blind (Nepal dark-adaptation trial, 2005). In 510 Senegalese children with abnormal eye-surface cells, beta-carotene and retinyl palmitate were equivalent: about 50% had normal cytology at 7 weeks in both groups (Senegal equivalence trial, 1993).

For adults whose deficiency comes from surgery or disease, there are no placebo-controlled treatment trials. A 2026 meta-analysis found deficiency in 18.0% of people after gastric bypass and 59.0% after biliopancreatic diversion, and 13 of 14 treatment records with eye findings improved or resolved with repletion (post-bariatric meta-analysis, 2026). A 2026 review of 61 children in high-income countries with eye signs found restrictive eating behind 75% of cases and autism in 49% (high-income xerophthalmia review, 2026).

Under-treatment is a real gap. At one US hospital, 27.7% of tested adults were below the hospital's cut-off (330 of 2,725, about 12%, below the WHO threshold), but only 9.6% of the deficient patients had an eye examination and vitamin A was documented for just over half of those with symptoms (New England hospital review, 2026). Evidence is thin for specific conditions: a Cochrane review found no trials of vitamin A in cystic fibrosis (Cochrane cystic fibrosis review, 2018), and RDA-level doses for 12 months did not raise blood retinol in 62 children with sickle cell disease (sickle cell trial, 2012).

Short supervised courses are usually well tolerated. High single doses cause transient vomiting or, in babies, a bulging fontanelle in 2-3% of recipients, which typically settles within 24-48 hours (WHO treatment guide, 1997); the Cochrane review found vomiting roughly doubled within 48 hours of a dose (RR 1.97) (Cochrane review, 2022).

The bigger risk is taking too much for too long. Preformed vitamin A is stored in the liver, and chronic excess can cause dry skin, joint pain, fatigue and abnormal liver tests; the upper limit is 3,000 mcg (10,000 IU) a day for adults and lower for children, from 600 mcg at ages 1-3 to 2,800 mcg at 14-18 (NIH Office of Dietary Supplements). In early pregnancy, women taking more than 10,000 IU a day from supplements had a 4.8-fold higher prevalence of birth defects of cranial neural crest origin (pregnancy cohort, 1995). Beta-carotene is not known to cause birth defects, but high doses raised lung cancer risk in smokers (NIH Office of Dietary Supplements).

Retinoid medicines such as acitretin and bexarotene add to vitamin A and raise the risk of toxicity (NIH Office of Dietary Supplements). Orlistat lowers absorption of vitamin A and beta-carotene, and its makers advise a multivitamin (NIH Office of Dietary Supplements). If you take these, your prescriber should set any vitamin A dose.