Fathers' B12 appears in a birth-defects cohort, and folate's curve flattens early

Maternal folic acid prevents neural tube defects; that much is settled [s1]. What has stayed unclear is nearly everything adjacent to it — whether vitamin B12 matters, whether more folate is better than enough, and whether the father's nutritional status has any bearing at all [s1]. A prospective cohort published on 31 August measured all three [s1].

The design

The study drew on the Shanghai Preconception Cohort in China. Two samples were analysed: 3,032 couples with biomarkers measured within four months before conception, among whom there were 73 birth defect cases; and 17,765 women with biomarkers sampled at four months' gestation or earlier, among whom there were 327 cases [s1].

Birth defects were ascertained among live births, stillbirths and abortions due to fetal abnormalities — a broader ascertainment than live births alone, which matters because the most severe defects are the least likely to reach delivery. Standardised predicted prevalences were estimated using weighted logistic regression [s1].

What the biomarkers tracked

The strongest signal was preconception vitamin B12, and it ran in both parents.

Across maternal preconception B12 categories from the lowest (<148 pmol/L) to the highest (≥590 pmol/L), predicted birth-defect prevalence fell from 49.6 to 16.2 cases per 1,000 pregnancies [s1]. Paternal B12 showed a similar but attenuated pattern, from 55.5 to 24.0 cases per 1,000 [s1].

Preconception parental red blood cell folate, by contrast, showed imprecise patterns of lower predicted prevalence — the direction was there, the precision was not [s1].

After conception the folate picture sharpened, and its shape is the practically interesting part. Predicted prevalence was lower at maternal RBC folate levels of 906 to 1,131 nmol/L than at levels below 453 nmol/L — 16.5 versus 25.7 cases per 1,000 — with little further reduction at higher levels [s1]. That is a threshold, not a gradient: the benefit associated with folate appeared between deficiency and sufficiency and then stopped.

Postconception maternal B12 changed little across the lower categories but was lower at higher levels: 11.8 versus 20.4 cases per 1,000 at ≥590 versus 148 to 294 pmol/L [s1].

The paternal question

The paternal association is the finding that will attract attention, and it is also the one most exposed to confounding. Couples share a diet, a household, an income and a set of health behaviours. A father's B12 concentration is partly a measurement of the kitchen he eats in, which is the same kitchen the mother eats in. The attenuation of the paternal pattern relative to the maternal one is consistent with a real but weaker paternal contribution, and it is equally consistent with the paternal measure acting as a proxy for shared household nutrition.

The authors list residual confounding as the study's limitation, without qualification [s1]. An accompanying editorial in the same journal is titled "Both Paternal and Maternal Nutrition Status Matter: An Exemplar From Folate and Vitamin B12 in Association With Risks for Birth Defects" [s2], which indicates the direction editors read the finding — but a title is not an analysis, and nothing in this cohort randomised anything.

What it does not establish

This is observational. It measures nutrient concentrations and counts outcomes; it does not test supplementation, and it cannot say that raising a biomarker would lower a risk. The distinction is not pedantic: the trial history of nutrient supplementation is full of associations that did not survive randomisation.

Absolute numbers are small. Seventy-three birth defect cases among 3,032 couples is the entire evidentiary basis for the preconception parental analysis [s1]. Predicted prevalences derived from that many events carry substantial uncertainty regardless of how the modelling is done.

The cohort is a single-city Chinese population, funded by the National Key Research and Development Program of China, the National Natural Science Foundation of China and the Chinese Academy of Medical Sciences [s1]. Background diet, fortification policy and baseline B12 status differ across countries, and this study cannot say whether the same categories would separate the same way elsewhere.

What it means for a reader

Nothing here is a recommendation, and this paper does not establish a target level for anyone. Periconception nutrition is managed with a clinician, and folate advice in particular is set by national guidance that this study does not supersede.

The two ideas worth carrying out of it are structural rather than personal. First, the folate association appears to plateau — more was not associated with better below the top of the observed range [s1]. Second, the parental unit under study here is a couple, not a woman, and the design that would test whether that matters — measuring both parents before conception — is now demonstrably feasible at scale.

What to watch is replication in a fortified population with a different dietary baseline. The threshold shape of the folate association is the specific claim most worth testing again.

Sources

• [s1] Periconception Parental Red Blood Cell Folate, Vitamin B12, and Birth Defects in Offspring: A Prospective Cohort Study. Annals of Internal Medicine, 31 August 2026. https://doi.org/10.7326/ANNALS-25-05243
• [s2] Both Paternal and Maternal Nutrition Status Matter: An Exemplar From Folate and Vitamin B12 in Association With Risks for Birth Defects. Annals of Internal Medicine, 31 August 2026. https://doi.org/10.7326/ANNALS-26-03045