Maternal blood sugar during pregnancy is linked to children's behavioral problems — and omega-3 fat status may buffer that risk

Abstract-only disclosure: He et al. (2026) published June 6, 2026, in Nutrients is currently available in early-access format. Only the abstract has been released; full methods, results tables, figures, funding sources, and conflict-of-interest disclosures are not yet publicly accessible. All numbers and details reported below come from that abstract. This article will be updated when the complete paper is available.

A prospective birth cohort study published today in Nutrients reports that higher fasting blood glucose, one-hour post-load glucose, and insulin resistance during pregnancy are each associated with increased odds of behavioral and emotional problems in children at age 5. The paper's secondary finding — that higher maternal n-3 polyunsaturated fatty acid (PUFA) levels and lower n-6 PUFA levels in red blood cells attenuated that association — is the more novel contribution, and also the one that carries the most caveats before it can be translated to dietary advice. 1

The study is observational. The association between maternal glucose and child behavior does not establish that elevated blood sugar during pregnancy causes behavioral problems in children. The PUFA finding is similarly associative: women with higher n-3 PUFA status and lower n-6 PUFA status showed a weaker glucose–behavior link in their children, but the mechanism, the relevant PUFA threshold, and the degree to which diet explains PUFA status all require the full paper's data to evaluate.

Study design and population

Study type: Prospective birth cohort · Journal: Nutrients (MDPI, Q1, impact factor 5.0, open access CC BY 4.0) · Vol. 18, Issue 12, Article 1840 · Published: June 6, 2026 · Peer-review status: Published (early access; full article pending)

Lead institution: Department of Maternal and Child Health, School of Public Health, Sun Yat-sen University (Guangzhou, China). Corresponding author: Li Cai, Sun Yat-sen University. Co-authors are affiliated with the National Center of Technology Innovation for Dairy, the Yili Maternal and Infant Nutrition Institute (YMINI, a research division of Inner Mongolia Yili Industrial Group, a major Chinese dairy company), and the Institute of Nutrition and Health, Qinghai University.

Funding and conflicts of interest: Not yet disclosed. The presence of co-authors from a commercial dairy and infant nutrition enterprise (Yili/YMINI) is relevant context for interpreting the study's framing and conclusions; the conflict-of-interest declarations will be available when the full article is published. 1

Participants: 481 mother-child pairs from China. Maternal glucose metabolism was assessed during pregnancy using a 75-gram oral glucose tolerance test (OGTT), capturing fasting plasma glucose (FPG), one-hour and two-hour post-load glucose, HbA1c, insulin, and HOMA-IR (Homeostatic Model Assessment of Insulin Resistance). Maternal erythrocyte (red blood cell) PUFAs were quantified by gas chromatography — a method that reflects dietary fatty acid intake over roughly the preceding two to three months.

Primary outcome: Children's emotional and behavioral problems assessed at age 5 using the Strengths and Difficulties Questionnaire (SDQ), a parent-reported standardized screening instrument widely validated in pediatric research. Scores above the validated threshold were classified as "abnormal."

A smiling toddler playing outdoors with colorful toys — the SDQ measures emotional and behavioral functioning in children around this age — The SDQ (Strengths and Difficulties Questionnaire) is a 25-item parent-completed screening tool used to assess behavioral and emotional functioning in children aged 4–17. 1

What the abstract reports

Three glucose metabolism measures were each independently associated with higher odds of abnormal SDQ scores in children at age 5: 1

Fasting plasma glucose: OR = 1.63 (95% CI: 1.08–2.47)
OGTT one-hour glucose: OR = 1.84 (95% CI: 1.08–3.12)
HOMA-IR (insulin resistance index): OR = 1.52 (95% CI: 1.01–2.27)

All three confidence intervals are borderline: the lower bounds sit between 1.01 and 1.08, meaning the statistical significance is marginal. A larger sample or different covariate adjustment could shift these intervals across the null. The abstract does not specify the units for the ORs (per standard deviation increase? per mmol/L?), which matters for interpreting effect magnitude — that information will be in the methods section when the full paper is released.

One result was sex-stratified: maternal insulin was associated with abnormal SDQ scores only in girls, with a statistically significant sex interaction (p < 0.05). The abstract does not give the separate OR values for boys and girls or the insulin OR for the full sample.

The PUFA finding is described in the abstract as "effect modification": higher maternal erythrocyte n-3 PUFA levels and lower n-6 PUFA levels attenuated the association between glucose metabolism measures and child behavioral problems. The specific ORs for the high n-3 versus low n-3 subgroups, and the nature of the interaction (additive or multiplicative), are not available in the abstract.

The RCT evidence says something different

This is where the study's interpretation becomes complicated, and where any dietary translation needs to be careful.

The observational finding — that higher n-3 PUFA status in pregnant women is associated with better behavioral outcomes in their children — appears straightforward. But the best available intervention evidence says prenatal DHA supplementation does not improve child behavioral outcomes.

Gould et al. (2021) reviewed 25 randomized controlled trials of prenatal DHA supplementation and reported that out of 401 outcome comparisons across behavioral measures, there were no findings of a positive effect of DHA, and 23 instances where the DHA group had worse scores than the control group. 2 Two large RCTs detected small but statistically significant negative effects on SDQ and BRIEF-P scores (effect sizes 0.29–3.61 points). Gould et al. conclude: "There were no findings of a positive effect of DHA, and 23 instances where the DHA group had worse scores compared with the control group." 2

This "observational benefit vs. RCT null or harm" pattern is familiar in nutritional epidemiology — antioxidant supplementation and cardiovascular outcomes followed a similar arc — and it is worth naming directly. There are several possible explanations that are not mutually exclusive:

Erythrocyte PUFA status reflects a dietary pattern (high fish intake, low seed oil intake, balanced n-6:n-3 ratio) that produces benefits through channels other than DHA per se.
The relevant intervention window or PUFA species in observational studies may differ from what DHA supplement trials tested.
Confounding: women with higher n-3 PUFA status in a Chinese cohort may have systematically different diets, socioeconomic circumstances, or glucose control profiles.
The observational finding is simply confounded in a way the cohort study could not fully adjust for.

A parallel Dutch cohort study (Vrijkotte et al., 2021, Clinical Nutrition) found that higher maternal DHA and EPA status in early pregnancy was associated with fewer internalizing behavioral problems in children at ages 5–6, with β = −0.11 for DHA and β = −0.22 for EPA (both p < 0.05). 3 That study used 1,717 pairs and plasma phospholipids rather than erythrocyte PUFAs, and the association held for mother-rated SDQ scores but not teacher-rated scores — suggesting at minimum that the signal is modest and rater-dependent.

mdpi.comhttps://www.mdpi.com/2072-6643/18/12/1840/htmExternal link

Loading content card…

What is still unknown

Because the full paper is not yet available, several questions that directly bear on dietary interpretation remain open:

What PUFA levels were protective? The abstract describes a direction (higher n-3, lower n-6) but not a threshold. Whether this reflects typical fish consumption or unusually high n-3 status in this Chinese cohort cannot be determined yet.
What covariates were adjusted for? The OR values are described as multivariable-adjusted, but the covariate list (maternal age, pre-pregnancy BMI, income, dietary patterns, GDM diagnosis, parity, birth outcomes, child feeding) is unknown.
What drove the sex-specific insulin effect? The girls-only insulin association is intriguing but the mechanism and the male/female OR contrast are not reported.
What are the authors' stated limitations? Standard limitations for this type of study include self-reported covariates, SDQ as a screening tool rather than a clinical diagnosis, and the challenge of separating glucose dysregulation from GDM diagnosis effects.
Funding and COI: Whether Yili/YMINI funding shaped the study design or interpretation cannot be assessed without the full declarations section.

Dietary takeaway

Fish oil capsules — the observational PUFA finding in this cohort study does not translate directly to a DHA supplement recommendation — Erythrocyte PUFAs are a biomarker of dietary fat intake over roughly 2–3 months; they do not directly measure supplementation. 1

There are two separate pieces of actionable information in this study, and they should not be collapsed into one.

On blood glucose during pregnancy: The association between continuous glucose measures and child behavioral outcomes — if it replicates in larger, more diverse samples — would support paying attention to glucose control throughout pregnancy, not only at the GDM diagnostic threshold. Women and their clinicians already have strong reasons (maternal and infant metabolic outcomes) to monitor glucose carefully during pregnancy. This study adds a preliminary neurobehavioral signal to that picture, while acknowledging the ORs are borderline significant and the full analysis is not yet public.

On PUFAs specifically: The observational finding of a protective PUFA effect warrants caution before translation. Erythrocyte PUFAs are a biomarker of dietary fat intake patterns over approximately two to three months, not a direct measure of supplementation. The strongest intervention evidence (Gould et al. 2021, 25 RCTs) does not support DHA supplementation for child behavioral outcomes. That does not mean diet is irrelevant — the PUFA signal in He et al. and Vrijkotte et al. may reflect dietary pattern effects that DHA supplements alone cannot replicate. But it does mean the appropriate dietary framing is food pattern, not supplement.

The concrete recommendation that this study's results support, stated as conservatively as the data permit: Pregnant women already advised to eat fatty fish (salmon, sardines, mackerel) twice per week for established fetal brain development reasons have an additional, preliminary observational signal suggesting that maintaining adequate n-3 PUFA status throughout pregnancy — with a low n-6:n-3 ratio, achievable partly by limiting high-linoleic seed oils — may be relevant to child neurobehavioral trajectories. This study cannot quantify a protective dose, does not demonstrate causation, and should not be read as a recommendation to take DHA supplements above evidence-based guidelines.

For dietitians advising pregnant clients: the same food-first PUFA pattern supported by prior evidence (two servings of low-mercury fatty fish per week, lower reliance on soybean and corn oil for daily cooking) remains the operationally consistent recommendation. This cohort adds a tentative neurobehavioral rationale to reinforce it — pending the full paper's methods and the RCT evidence that will be needed to confirm the causal direction.

Cover image: photo by RDNE Stock project via Pexels