Modelling and false assumptions have led to the an inflated number
Dr Clare Craig
From 13 December 2026, the Bread and Flour (Amendment) Regulations 2024 require folic acid to be added to all non-wholemeal wheat flour milled in the United Kingdom. The entire justification for mandatory fortification rests on a modelled claim that 200 babies will be saved a year. When it was announced in September 2021, the Department said fortification would “help avoid around 200 neural tube defects each year”, which amounts to roughly 20 per cent of the annual UK total.
It is worth paying attention to exactly how that was worded.
- The verb is “help avoid” rather than “prevent”.
- The word “around” is included because the number is based on modelling.
- The unit is neural tube defects (NTDs) not live births with a neural tube defect. Therefore it includes every affected terminated pregnancy and other pregnancy losses. It also covers more than one condition: primarily spina bifida with rarer cases of brain malformations.
In the same announcement then Health and Social Care Secretary, Sajid Javid is quoted as saying:
“We are committed to giving more children a healthy start in life. With the safe and taste-free folic acid baked into the national diet, hundreds more babies will be born healthy each year.“
The evidence does not come anywhere close to that claim as we shall see. First, the 200 figure does not relate to living babies with a future life of disability, and second the modelling is based on flawed assumptions.
How many pregnancies are affected?
According to the most up-to-date NHS data, in 2022 there were 621 pregnancies affected by NTD in England. Extrapolating to the UK that would be approximately 731 pregnancies.
Of the 621, 57 are recorded alongside a genomic condition, a chromosomal or single-gene disorder that folic acid cannot affect. At most 564 could even in principle be folate-related.
How many babies are actually born?
Of the 621 affected pregnancies in England, 460 were terminated, 4 miscarried late and 13 were stillborn, leaving 144 live births. Extrapolated to the UK that is approximately 170 live births a year.
| Outcome (2022) | England (recorded) | UK (extrapolated) |
| Affected pregnancies | 621 | ~731 |
| Terminations | 460 | ~540 |
| Late miscarriage or stillbirth | 17 | ~20 |
| Live births | 144 | ~170 |
Neural tube defect pregnancy outcomes for 2022, from the NCARDRS Congenital Condition Statistics Report for England; UK figures extrapolated from England by relative birth numbers.
Even if every affected pregnancy were prevented, there are only around 170 live births each year with a neural tube defect. The slogan of “200 babies saved” therefore cannot literally describe additional babies born.
There aren’t enough babies
The answer is that the model counts pregnancies, not babies. Most of the modelled benefit concerns pregnancies that never become live births. The government claim their intervention would reduce affected pregnancies by 20 per cent. Applied to the recorded figures that works out at approximately 146 fewer pregnancies and 34 fewer live births. Once scaled to the UK, a 20% reduction would work out at 170 fewer affected fetuses. There is no way a 20 per cent reduction could produce 200 additional live-born babies.
What conditions are included?
A neural tube defect is not one condition. If the model were correct, the pregnancies it prevents would be roughly two fifths anencephaly, one in eight encephalocele, and about half spina bifida.
| Condition | Typical outcome |
| Anencephaly | Not survivable |
| Encephalocele | Rarely survivable |
| Spina bifida | Wide spectrum; the only group with substantial numbers of surviving children |
Anencephaly is not survivable and encephalocele rarely so; only the spina bifida group contains substantial numbers of surviving children.
Even accepting the government’s assumptions, the principal measurable effect would be around a hundred fewer terminations following prenatal diagnosis each year, roughly 0.04 per cent of terminations nationally, not hundreds of additional children growing up healthy.
Where does the model come from?
Rather than count the cases actually recorded, the government’s impact assessment began from an assumed 1,000 affected pregnancies a year. That is far higher than the actual record of ~731 UK pregnancies affected in 2022. The 1,000 baseline was built by taking a birth projection about 14 per cent above the outturn and multiplying it by an affected rate about 18 per cent above the registry’s, leaving the starting figure roughly a third too high before any modelling. A fifth of 1,000 is the 200. The 20 per cent reduction itself is the output of a model built on three assumptions.
Every model depends on its assumptions. If those assumptions are wrong, the answer is wrong.
Why are its assumptions doubtful?
Assumption 1: folate explains most of the variation in risk
The model assumes the observational link between maternal blood folate stores and neural tube defects is entirely causal. No clearly identifiable folate-deficient subgroup has been demonstrated, nor is there a high folate group who are protected from the risk of neural tube defects. On the modellers’ own analysis there is no protective threshold, so risk remains even for women with the highest folate levels. If folate were the principal determinant of population risk, countries with higher folate levels would consistently have lower neural tube defect rates. That relationship has never been demonstrated. For decades now, the United Kingdom and Ireland had the world’s highest rates without being folate-deficient; large American case-control studies of pregnancies conceived after fortification found folate intake, from diet or supplements, was unrelated to the risk, and Chile, fortifying at more than three times the American dose, saw no correspondingly larger fall in risk.
Assumption 2: low-dose folic acid is highly effective
The modelling assumes folic acid is highly effective at preventing NTDs. The strongest evidence comes from animal studies and trials in high-risk women with a previous affected pregnancy given a pharmaceutical dose of 4 mg daily, more than ten times the recommended daily amount. Evidence for preventing first affected pregnancies in average-risk women is much weaker, based on a single randomised controlled trial that also recorded an unexplained excess of foetal deaths in the treated group, discussed below.
Assumption 3: those relationships predict a 20 per cent reduction in Britain
Finally, the model applies these assumptions to British folate levels and neural tube defect rates to arrive at around 20 per cent of cases prevented. The same dataset analysed using four published models predicts reductions ranging from 15 to 83 per cent, so even the prediction is unstable. The headline “200” is not a measured reduction but the output of a model whose predictions vary widely depending on the shaky assumptions used.
| Could the apparent benefit simply reflect selective foetal loss? The sole randomised trial of low dose folic acid in healthy women, conducted in Hungary by Czeizel and Dudás, reported no neural tube defects in the arm given a multivitamin containing folic acid, but six in the placebo group given trace elements. The same trial’s own final report, five years later, showed every category of foetal death higher in the treated arm: seventy excess losses in total. It was statistically significant and worked out at roughly nine foetal deaths for each defect prevented. In 1997 the trial’s principal investigator, Czeizel, with Ernest Hook, asked in the Lancet whether folic acid reduces defects by selectively killing affected embryos. They concluded that this selective loss could account for “the entire apparent protective effect”. If there is no evidence that folic acid interventions reduce the risk then the models are based on a false assumption. |
Does real-world evidence support it?
If the model were correct, a reduction of around 20 per cent should be visible after population fortification. It is not. Across Europe, total neural tube defect prevalence has remained broadly unchanged despite decades of supplementation advice and voluntary fortification. The United States is often presented as the exception, but live-birth rates had already been falling for decades before fortification and are strongly influenced by prenatal diagnosis and termination, which were excluded from the CDC analysis. Cross-country comparisons likewise show little difference between countries that fortify and those that do not. Overall, population evidence does not demonstrate the effect size assumed by the model.
How was the benefit valued?
The impact assessment also put a price on each prevented defect. It valued a full healthy life at just under £3 million. That figure rests on valuing each quality-adjusted life-year at £60,000, double the £30,000 that NICE ordinarily applies, on the argument that no other NHS spending is displaced. At the usual £30,000, reaching £3 million would take a hundred years of perfect health.
What the assessment did not do was weigh that against the alternative. It priced the outcome hoped for, a healthy child, without setting beside it what these pregnancies actually face, which is a disabled life or, far more often, a termination. Anencephaly is almost always terminated: of 246 such pregnancies in England in 2022, 223 ended in termination. So on the appraisal’s own figures, preventing a pregnancy that would have been terminated is scored above preventing a whole life lived with spina bifida.
This is not how terminations are counted anywhere else in healthcare. Valued at £3 million each, the roughly 300,000 terminations carried out in the UK every year would represent a health loss of around £900 billion, more than four times the entire NHS budget. If that were how they were counted, reducing them would be the health service’s overriding priority. It is not. The £3 million is applied only where it flatters the case for fortification.
The appraisal was also never finished. In the published impact assessment, every figure for costs and benefits is left to be confirmed.
| Option: fortify all non-wholemeal flour (£m) | Low | High | Best estimate |
| Total cost | Tbc | Tbc | Tbc |
| Total benefit | Tbc | Tbc | Tbc |
| Net benefit | Tbc | Tbc | Tbc |
Summary of costs and benefits for the preferred option, reproduced from the government’s consultation-stage impact assessment.
When the regulations were finally laid in 2024, the only economic document beside them was a de-minimis assessment, the light-touch form used when the cost to business is judged too small to warrant a full one. A measure changing the flour eaten by the whole country reached the statute book with no completed account of its costs and benefits.
Conclusion
Mandatory folic acid fortification has been justified by a modelled claim that it will save babies. The government’s headline invites the public to picture 200 children who would otherwise have been born disabled. On the available evidence the headline number of “200” corresponds principally to pregnancies that would have ended in termination or were incompatible with life. The number of children who would otherwise have survived, with spina bifida or encephalocele, is smaller by an order of magnitude. Every figure used to reach that conclusion is drawn from the evidence base the policy itself relies upon.
If we took the modelled figure at face value, around two dozen children a year could be spared a disability, while folic acid is added to the flour eaten by all 69 million people. Among them are groups for whom added folic acid carries potential harm: some 3.5 million people living with cancer, in whom high folate intake may promote the growth of established tumours; around 100,000 a year who receive a coronary stent, for whom folic acid has been shown to increase restenosis; the many older adults with undiagnosed B12 deficiency, in whom it can mask the anaemia while neurological damage advances; and men, who at a milligram a day had a 2.6-fold higher rate of prostate cancer in the one randomised trial to look. Other harms have yet to be considered or measured.
The public was told this policy would help hundreds more babies to be born healthy each year. The evidence assembled here does not support that claim.
