How Breastfeeding Duration Links to Behavior and Speech at 5 Years

Bulgarian researchers followed 92 full-term infants from birth to age 5 and looked at how the duration of breastfeeding was associated with the development of speech, behavior, motor skills, and intelligence. In a “crude” comparison of groups, they noticed that the longer the breastfeeding, the better the language (p=0.037), and the best behavior was in those who were breastfed for 6–12 months (p=0.001). In multivariate regression, the period of 6–12 months was indeed associated with better behavior at age 5 (estimate -5.88; p=0.026) compared with <6 months. But after strict adjustment (partial correlations), no stable independent association with outcomes was found, indicating that other environmental and family factors also play a role. The work was published on August 8, 2025 in Nutrients.

Background

• Why even look at the "long" trace of breastfeeding? Breastfeeding remains the basic recommendation of the largest organizations (exclusively ~6 months, then complementary feeding and continuation of breastfeeding at the family's discretion), and early feeding is considered one of the key determinants of brain development in the first years of life. These guidelines are unified by the WHO and the American Academy of Pediatrics.
• What large reviews of cognitive outcomes showed. Meta-analyses and long-term cohorts often found a small advantage in infants on intelligence tests — on the order of +3–4 IQ points — and, in some cohorts, higher educational and economic outcomes in adulthood (Brazilian data). This created an expectation that breastfeeding duration might be associated with language/behavior in preschool.
• But causality is not guaranteed: the role of family and environment is great. “Sibling designs” (comparison of siblings with different breastfeeding experiences) significantly weakened many associations, indicating a selection effect: parental education and income, migration/cultural context, access to classes and kindergarten. Hence the need to carefully control for covariates and be prepared for nonlinear relationships.
• Mechanistic clues for the effect of breastfeeding. Long-chain polyunsaturated fatty acids (DHA/ARA), human milk oligosaccharides (HMOs), and other bioactive components of human milk have attracted attention; for HMOs, there are signals in observational studies of association with cognitive/language outcomes in preterm and term infants. This provides biological plausibility but does not obviate the need for rigorous designs.
• Why age 5 and domain assessment are important. By age 5, language and behavioral trajectories become distinguishable, and an integrated “total score” can “blur” domain differences. Using a validated instrument (like the Bulgarian study’s NDT5) allows us to separate speech, behavior, motor skills, and nonverbal intelligence and check whether there is a “window of optimum” for breastfeeding duration (e.g., 6–12 months), rather than a simple linear relationship.
• Nutrition policy context: Global recommendations converge: complementary feeding from 6 months while continuing breastfeeding; in real life, duration of breastfeeding is influenced by maternity leave, family/healthcare system support, urbanization and culture – so it is reasonable to validate the results in specific regions (as in the Eastern European cohort).
• What the current work adds is a prospective regional cohort with a detailed domain assessment at 5 years and an attempt to statistically disentangle the effect of breastfeeding duration itself from family and social factors. The important research question here is not “is breastfeeding always better and longer”, but whether there is a range of durations associated with better behavioral/language outcomes, and whether the association holds after stringent adjustments.

What did they do?

• Design: Prospective cohort in Varna, Bulgaria, 2017–2024; 92 children (full-term, without major perinatal problems) reached assessment at 5 years. Parents completed questionnaires on feeding and environmental factors.
• Groups by breastfeeding duration: ≤6 months; 6–12 months; >12 months. Neurodevelopmental assessment — NDT5 test validated for Bulgaria in five domains (motor skills, speech/language, articulation, non-verbal intelligence, behavior; lower scores — better).
• Statistics: Welch ANOVA for between-group differences; partial correlations controlling for covariates; multivariate regressions (behavioral and language domains in separate models).

What did they find?

• Language and behavior: differences between groups are significant (language p=0.037; behavior p=0.001): “long” breastfed babies have better language and optimal behavior at 6–12 months. The total “overall” development score was not statistically different.
• Regressions: for behavior, breastfeeding duration of 6–12 months is associated with better indicators vs <6 months (−5.88; p=0.026), while >12 months does not give a doppler - possibly a plateau effect or small subgroups. For language, rural area (worse, p=0.004) and mixed ethnicity (worse, p=0.045) turned out to be significant predictors; breastfeeding duration itself was not.
• Correlations with corrections of the independent linear relationship “duration of breastfeeding → outcome” did not confirm - the signal is probably nonlinear and “stitched” with the context of the family and environment.

Context and nuances

• Not just food. The authors remind us that breastfeeding supports immunity and microbiota, and the “first 1,000 days” are a window when nutrition has a particular impact on the brain. But the scale and durability of the effects on cognitive domains over the long term are debatable and depend on the country, family, and social environment.
• Social factors are strong. Region of residence, father's education, and ethnicity "emerge" in the models; and the authors rightly consider the unexpected "plus" to behavioral assessments for paternal vaping/smoking to be spurious/residual confounding rather than causality.
• Fits into the literature, but lacks a consistent pattern. There are large cohorts with an IQ advantage for breastfeeding ≥6 months (ALSPAC, +4–5 points), but there are also intra-family comparisons in the US, where the effect is leveled after taking into account family factors. The new Bulgarian array adds regional data and shows a domain-specific nature of the relationship.

What does this mean for parents and the system?

• Breastfeeding should be supported (exclusively for 6 months, then optionally with complementary foods): in addition to short-term benefits, behavioral benefits are possible by age 5, especially if breastfeeding lasted 6–12 months. But be careful when interpreting this as a “direct cause”: the language environment, access to kindergarten/classes, parental literacy, etc. are important.
• Policy: strengthening support for breastfeeding (mothers’ schools, breastfeeding-friendly maternity hospitals) + targeted measures for rural regions and multicultural families in language development - will probably have a greater integral effect than “focusing only on duration”.

Restrictions

Small cohort (n=92), unequal duration groups, self-reported diet (memory risk), possible residual confounding; some signals diverge between methods (ANOVA vs partial correlations). Large longitudinal samples with rich family and environmental data are needed.

Source: Zhelyazkova D. et al. Beyond Infant Nutrition: Investigating the Long-Term Neurodevelopmental Impact of Breastfeeding. Nutrients, 17(16):2578, published August 8, 2025. https://doi.org/10.3390/nu17162578