Cohort name Country Reference Follow‐up Funding	Original Cohort (N total) Exclusion criteria Study population (n, sex and age at baseline)(a)	Ascertainment of outcome	Exposure groups(a) n/person‐years	Model covariates	Results
INMA Project (2003–2005) Spain Amorós et al. (2018a) Approx. 1.7 yr Cohort Public	N = 787 Population Sampled: Pregnant women aged ≥ 16 yr, 10–13 weeks of gestation, singleton pregnancy, intention of undergoing follow‐up and delivery at the corresponding centre of reference, and no impediment for communication. Excluded: missing exposure or outcome variables. n = 651 mother–child pairs Sex (% F): 47 Ethnicity: NR Mothers' age (yr): 30.1 ± 4.4	Neuropsychological development assessed around 12 mo of age (12.3 ± 0.7 mo) using the BSID‐II, mental scale and psychomotor scale. Testing carried out at the children's reference hospital in the presence of their mothers, by four trained psychologists.	Maternal serum Se at 1st trimester of pregnancy (μg/L) 79.74 ± 7.92	BSID mental score: child's sex, maternal BMI, parity, area of residence (urban, metropolitan, semi‐urban and rural), maternal age and intake of seafood per 100 g/d. BSID psychomotor score: social class, paternal age, attendance to nursery, season of birth.	Maternal Se concentrations and BSID‐II scores; β‐coefficient (95% CI) Mental: −0.13 (−0.29, 0.03); p = 0.122 Psychomotor: −0.08 (−0.24, 0.07); p = 0.283 Non‐linearity (splines analysis): inverted U‐shape for the associations between maternal serum Se and mental score (break point at 86.4 μg/L Se (95% CI 79.3, 93.5)) and psychomotor score (break point at 86.2 μg/L Se (95% CI 69.3, 103.0))
INMA Project (2003–2012) Spain Amorós et al. (2018b) Approx. 5.7 yr Cohort Public	N = 787 Population Sampled: Pregnant women aged ≥16 yr, 10–13 wk of gestation, singleton pregnancy, intention of undergoing follow‐up and delivery at the corresponding centre of reference, and no impediment for communication. Excluded: missing exposure or outcome variables. n = 409 mother–child pairs Sex (% F): 48.6 Ethnicity: NR Mothers' age (yr): 30.1 ± 4.4 yr	Neuropsychological development assessed at 5 yr of age (5.8 ± 0.16 yr) by using a standardised version of the MSCA adapted to the Spanish population	Maternal serum Se at 1st trimester of pregnancy (μg/L) 79.9 ± 8.1	Base model (BM): sex, age at evaluation, psychologist, maternal age, maternal educational level or maternal intelligence Verbal: BM + maternal country of birth, parity, type of zone, maternal working status and BMI before pregnancy. Perceptual‐performance: BM + type of zone, smoking during pregnancy, parity, breastfeeding, maternal smoking at evaluation and maternal working status Quantitative: BM + parental educational level, type of zone, parity General cognitive: BM + parity, type of zone, paternal working status at evaluation, and seafood intake during pregnancy. Working memory: BM + maternal country of birth, paternal educational level, parity, maternal smoking at evaluation. Global memory: BM + parity, type of zone, maternal country of birth, and seafood intake during pregnancy. Fine motor: BM + breastfeeding, parity, type of zone and maternal smoking at evaluation. Global motor: BM + breastfeeding. Executive function: BM + maternal country of birth, maternal age, parental educational level, parity, type of zone, maternal smoking at evaluation	Maternal Se concentrations and MSCA scores; β‐coefficients (95% CI) Verbal: −0.052 (−0.164, 0.061); p = 0.366 Perceptual‐performance: −0.046 (−0.128, 0.036); p = 0.268 Quantitative: 0.015 (−0.050, 0.080); p = 0.653 General cognitive: −0.085 (−0.283, 0.112); p = 0.395 Working memory: −0.006 (−0.054, 0.042); p = 0.793 Global memory: −0.041 (−0.110, 0.027); p = 0.234 Fine motor: −0.022 (−0.064, 0.021); p = 0.317 Global motor: −0.048 (−0.115, 0.019); p = 0.160 Executive function: −0.057 (−0.175, 0.060); p = 0.338 Non‐linearity (splines analysis): inverted U‐shape for the associations maternal serum Se and the verbal score (break point at 83.7 μg/L Se (95% CI 77.2, 90.1)) and global memory score (break point at 85.6 μg/L Se (95% CI 77.4, 93.9))
China Li et al. (2020a) Approx. 2 yr Prospective Cohort Public	N = 545 mother–child pairs Population Sampled: Pregnant women recruited from the Wuhan Women and Children Medical Care Center Excluded: Participants with missing values. n = 544 mother–child pairs Sex (% F): 44 Ethnicity: Asian Mothers' age (yr): 28.97 ± 3.43	Neurocognitive development assessed by BSID at 2 yr of age; translated to Chinese and locally standardised.	Maternal urinary Se (Geometric mean ± SD, (P5th, P95th), μg/L) 11.47 ± 2.34 (2.59, 43.03) Based on one spot urine sample collected before delivery (36–42 wk)	Single‐metal model (SMM): child's sex, maternal education, paternal education, weight gain during pregnancy, annual household income and parity. Multi‐metal model (MMM): as above + maternal urinary zinc, copper, manganese, strontium, nickel, cobalt, rubidium, chromium and vanadium.	Maternal Se concentrations and BSID scores; β‐coefficients* (95% CI), by IQR change in lnSe Mental: SMM 2.02 (−0.18, 4.22); MMM 2.28 (−1.23, 5.80) Psychomotor: SMM −0.10 (−1.81, 1.62); MMM −1.09 (−3.83, 1.65) Non‐linearity (Bayesian kernel machine regression): inverted U‐shaped between log(maternal urinary Se) and mental and psychomotor scores in girls, but not boys.
RHEA study Greece Kippler et al. (2016) 4 yr Public	N = 628 mother–child pairs Population Sampled: early stage pregnant women (before 15 wk), residency within the study area, aged ≥16 yr and good understanding of the Greek language. Excluded: Participants with missing data. n = 575 mother–child pairs Sex (% F): 50 Ethnicity: NR Mothers' age (yr): 30 ± 5.1	Neuropsychological development assessed by two trained psychologists, with MSCA at 4 yr (4.2 ± 0.23 yr). Executive and cognitive functions of posterior cortex derived from reorganisation of the MSCA subtests in accordance with their association with specific neurocognitive function areas.	Maternal urinary Se (Mean ± SD, median (5‐95th percentile), μg/L) 23 ± 8.6; 22 (12–39) Samples collected at median 13 wk of gestation (IQR: 4 wk)	Single‐element model (SEM): adjusted for examiner, child sex, age at testing, and maternal age, parity, marital status, education and tobacco smoking. Multi‐element model (MEM): as above + urinary iodine, cadmium, lead.	Maternal urinary Se and MSCA scores; β‐coefficients (95% CI), per every doubling of maternal urinary Se General cognitive: SEM 1.9 (−0.57, 4.3); MEM 2.2 (−0.38, 4.8) Verbal: SEM 2.2 (−0.31, 4.7); MEM 2.4 (−0.14, 5.0) Quantitative: SEM −0.072 (−2.7, 2.6); MEM −0.27 (−3.0, 2.5) Memory: SEM 1.9 (−0.66, 4.4); MEM 1.9 (0.72, 4.5) Perceptive‐performance: SEM 1.5 (−0.78, 3.9); MEM 2.0 (−0.66, 4.7) Motor: SEM 0.54 (−1.6, 2.7); MEM 0.87 (−1.7, 3.5) Executive functions: SEM 1.5 (−0.98, 4.0); MEM 1.4 (−1.3, 4.0) Cognitive functions: SEM 1.9 (0.50, 4.2); MEM 2.5 (−0.055, 5.1) No indication of non‐linearity (scatterplots).
The Tohoku Study of Child Development Japan Tatsuta et al. (2017) 18 mo	N = 879 Population Sampled: mother–child pairs (term infants) Excluded: mothers: in vitro fertilisation, preeclampsia or gestational diabetes mellitus (DM), use of antidiabetic agents, thyroid dysfunction, mental or psychological disease, hepatitis, immune deficiency, malignant tumours, diseases affecting foetus growth; infants: congenital anomalies or severe disease, non‐singleton, preterm birth (< 36 wk), BW < 2,500 g. n = 566 Sex (% F): 49 Ethnicity: Asian Mothers' age (yr): NR	Neurocognitive development assessed by BSID‐II and KSPD at 18 mo of age (range, 17–20 mo). A Japanese version of the BSID‐II was prepared by the researchers.	Cord‐plasma Se (ng/g) Boys: 66.3 ± 10.2 Girls: 67.0 ± 9.6	Pearson correlations were used to determine the association of maternal blood and cord‐blood Se with KPSD/BSID‐II scores.	Cord‐plasma Se and KPSD/BSID‐II scores, r‐coefficients KPSD Developmental quotient: 0.015, p = 0.729 Cognitive‐adaptive: 0.004, p = 0.918 Language‐social: 0.018, p = 0.675 Postural‐motor: 0.020, p = 0.643 BSID‐II Mental: 0.56, p = 0.181 Psychomotor: 0.007, p = 0.876
Norway Varsi et al. (2017) From 18 wk pregnancy to 6 mo postpartum; infants from birth to 6 mo old	N = 140 Population Sampled: mother (pregnant at 18 wk)‐child pairs Excluded: Women with pregnancy related or chronic disease, except those with hypothyroidism under control. Missing outcome. n = 112 Sex (% F): 47 Ethnicity: Caucasian Mothers' age (yr): 31.5 ± 4.3	Neurodevelopment assessed by Ages and Stages Questionnaire (ASQ), completed by parents at 6 mo.	Maternal serum Se at 18 wks pregnancy (μmol/L, median (P2.5, P97.5)) 0.96 (0.71, 1.35) Maternal serum Se at 28 and 36 wk pregnancy; infant serum Se at 6 mo	Birthweight, BW at 6 mo, gender, mo of exclusive breastfeeding, maternal age, education, and parity	Maternal serum Se at 18 wks and ASQ scores, β‐coefficients (95% CI NR) Total: 11, p = 0.007 Problem solving: 3, p = 0.006 Fine motor scores: 3, p = 0.04 Communication: 1, p = 0.22 Personal‐social functioning: 2, p = 0.11 Gross motor score: 1, p = 0.56 No associations observed between ASQ scores and maternal serum Se at other timepoints or infant serum Se Non‐linearity not explored.
PHIME Croatia Močenić et al. (2019) Approx. 18 mo Public	N = 205 mother–child pairs Population Sampled: pregnant women that were permanent residents in the study area for at least 2 years Excluded: Participants with missing values. n = 154 mother–child pairs Sex (% F): 46 Ethnicity: NR Mothers' age (yr): 30.1 (20–43)	Neurodevelopment assessed with the BSID‐III at the age of 18 mo.	Maternal blood Se at delivery (ng/g, mean ± SD (min‐max)) 92.6 ± 22.4 (40.9–182.4) Cord‐blood Se at delivery (ng/g, mean ± SD, (min‐max)) 98.7 ± 21.8 (59.6–162.7)	Pearson correlations were used to determine the association of maternal blood and cord‐blood Se with BSID‐III scores.	Maternal blood Se correlation with BSID‐III scores; r‐coefficients Cognitive: 0.176 (p = 0.029) Language: 0.138 (p = 0.089) Motor: 0.128 (p = 0.113) Cord‐blood Se correlation with BSID‐III scores; r‐coefficients Cognitive: 0.001 (p = 0.993) Language: 0.100 (p = 0.218) Motor: 0.032 (p = 0.692) Non‐linearity not explored.
REPRO_PL Poland Polanska et al. (2017) Approx. 3 yr Public	N = 539 mother–child pairs Population Sampled: women up to 12 wk of single pregnancy, no assisted conception, no pregnancy complications and no chronic diseases. Excluded: Participants with missing values. n for analysis 1‐y old children = 239 n for analysis 2‐y old children = 168 Sex (% F): 53 Ethnicity: NR Mothers' age (yr): 28.9 ± 4.4	Neuropsychological development assessed with BSID‐III in children at 1 year (12.6 ± 1.4 mo) and 2 years of age (24.8 ± 2.5 mo).	Maternal plasma Se in the 1st trimester (μg/L, mean ± SD (min‐max)) 48.3 ± 10.6 (16.1–91.4) Cord‐blood Se at delivery (μg/L, mean ± SD (min‐max)) 31.1 ± 8.2 (13.8–56.3)	Examiner, maternal age, maternal education, child gender and maternal smoking status during pregnancy based on the cotinine level.	Maternal plasma Se (1st trimester) and BSID‐III scores, β‐coefficients (95% CI) 1‐y‐old children Cognitive: 0.11 (−0.03, 0.25) p = 0.13 Language: 0.18 (0.02, 0.34) p = 0.03 Motor: 0.25 (0.08, 0.42) p = 0.005 2‐y‐old children Cognitive: 0.22 (−0.02, 0.47) p = 0.07 Language: 0.07 (−0.15, 0.29) p = 0.52 Motor: 0.15 (−0.07, 0.38) p = 0.18 Cord‐blood Se and BSID‐III scores, β‐coefficients (95% CI) 1‐y‐old children Cognitive: 0.005 (−0.16, 0.17) p = 0.95 Language: 0.10 (−0.08, 0.29) p = 0.28 Motor: 0.17 (−0.03, 0.37) p = 0.10 2‐y‐old children Cognitive: −0.06 (−0.32, 0.19) p = 0.62 Language: 0.09 (−0.14, 0.32) p = 0.44 Motor: 0.07 (−0.16, 0.31) p = 0.55 Non‐linearity not explored.
MINIMat Bangladesh Skröder et al. (2015) Approx. 2 yr Public	N = 4,436 pregnant women; 3,267 had singleton live births Population Sampled: women in early pregnancy, at gestational wk 8 on average; gestational age ≤ 14 wk by ultrasound examination; no severe illness. n = 750 mother–child pairs Sex (% F): 47.5 Ethnicity: NR Mothers' age (yr): 27 (14–44)	Neuropsychological development assessed at 18 mo of age 17–19.2) with BSID‐II and tests for children's comprehensive and expressive language development using a Bangladeshi version of MacArthur's Communicative Development Inventory	Maternal erythrocyte Se (Ery‐Se) at gestational wk 30 (μg/g Hb): Low Ery‐Se: 0.39 ± 0.056 High Ery‐Se: 0.53 ± 0.068	Model 1 (M1): age at testing and gender. Model 2 (M2): M1 + gestational age, maternal age, maternal BMI, socio economic status (SES), home observation for measurement of the environment (HOME), weight for height (WHZ) and birth weight. Model 3 (M3): M2 + Ery‐Zn, Ery‐Mn and urinary iodine. Model 4 (M4): M2 + urinary arsenic, urinary cadmium and Ery‐Pb.	Maternal Ery‐Se and BSID‐II, comprehensive and expressive language scores, β‐coefficients (95% CI), by 0.5 μg/g Hb increase in Ery‐Se Mental: M1 3.2 (−1.7, 8.0); M2 3.3 (−1.7, 8.3); M3 3.8 (−2.2, 9.9); M4 4.0 (−1.4, 9.4) Psychomotor: M1 9.2 (3.9, 15); M2 8.8 (3.3, 14); M3 10 (3.7, 17); M4 10 (4.6, 16) Comprehension: M1 7.2 (3.3, 11); M2 3.7 (0.40, 7.1); M3 3.1 (−0.65, 6.9); M4 3.5 (−0.079, 7.0) Expression: M1 2.4 (−0.066, 4.9); M2 0.75 (−1.6, 3.1); M3 0.087 (−2.7, 2.9); M4 1.3 (−1.3, 3.9) No indication of non‐linearity for mental, psychomotor and expression scores (scatterplots). For comprehension, estimates given between the knots at 0.32 and 0.60 μg/g Hb.
MINIMat Bangladesh Skröder et al. (2017) Approx. 10 yr Public	N = 1,530 children Population Sampled: mother–child pairs, women in early pregnancy, at gestational wk 8 on average gestational age ≤ 14 wk by ultrasound examination; no severe illness. Excluded: Participants with missing values. Analysis maternal Ery‐Se, n = 1,260 Analysis children U‐Se: <34 μg/L, n = 1,214; ≥34 μg/L, n = 20 Sex (% F): 47.6 Ethnicity: NR Mothers' age (yr): NR	Children's cognition assessed at 5 yr (5.4 ± 0.13) and 10 yr (9.5 ± 0.095) using the WPPSI and the WISC	Maternal Se erythrocyte (Ery‐Se) at gestational wk 14 (μg/g Hb): 0.45 ± 0.11 Children urinary Se (U‐Se) at 5 yr (μg/L): 14 ± 6.6	Model 1 (M1): gender, parity and family socio economic status (SES) at enrolment, birthweight, Hb at gestational week 14, age at testing, height‐for‐age z‐score (HAZ), modified Home Observation for Measurement of the Environment score (HOME), testers, school type, mothers' cognitive function, and paternal education (all assessed at 5‐yr follow‐up) Model 2 (M2): M1 + erythrocyte zinc and manganese at gestational wk 14 (prenatal analyses). No M2 for U‐Se as Zn and Mn not measured Model 3 (M3): M1 + erythrocyte cadmium, lead, and arsenic at gestational wk 14 (prenatal analyses) or urinary arsenic, cadmium, and lead at 5 yr	Maternal Ery‐Se and WPPSI/WISC scores; β‐coefficients (95% CI), by 0.1 μg/g Hb increase in Ery‐Se At 5 yr Full developmental: M1 0.94 (0.027, 1.9); M2 0.97 (−0.16, 2.1); M3 0.99 (0.0051, 2.0) Verbal: M1 0.47 (−0.0072, 0.95); M2 0.41 (−0.18, 1.0); M3 0.46 (−0.054, 0.98) Performance: M1 0.27 (−0.078, 0.62); M2 0.32 (−0.11, 0.76); M3 0.25 (−0.13, 0.63) At 10 yr Full developmental: M1 2.2 (1.0, 3.3); M2 2.6 (1.2, 3.9); M3 2.2 (1.0, 3.4) Verbal: M1 0.50 (0.12, 0.89); M2 0.50 (0.051, 0.95); M3 0.51 (0.10, 0.92) Perceptual reasoning: M1 0.53 (0.079, 0.98); M2 0.75 (0.22, 1.3); M3 0.56 (0.081, 1.0) Working memory: M1 0.38 (0.14, 0.62); M2 0.39 (0.11, 0.67); M3 0.37 (0.12, 0.62) Processing speed: M1 0.75 (0.29, 1.2); M2 0.94 (0.40, 1.5); M3 0.78 (0.30, 1.3) Children urinary Se at 5 yr and WISC scores at 10 yr, β‐coefficients (95% CI), by 10 μg/L increase in U‐Se U‐Se < 34 μg/L Full developmental: M1 2.5 (−0.033, 5.0); M3 2.5 (−0.12, 5.1) Verbal: M1 0.49 (−0.35, 1.3); M3 0.52 (−0.34, 1.4) Perceptual reasoning: M1 1.2 (0.18, 2.2); M3 1.2 (0.18, 2.2) Working memory: M1 0.34 (−0.18, 0.86); M3 0.25 (−0.28, 0.78) Processing speed: M1 0.52 (−0.50, 1.5); M3 0.53 (−0.50, 1.6) U‐Se ≥ 34 μg/L Full developmental: M1 1.1 (−12, 14); M3 1.0 (−12, 14) Verbal: M1–1.5 (−5.8, 2.8); M3–1.4 (−5.8, 3.0) Perceptual reasoning: M1 3.1 (−20, 8.1); M3 3.2 (−2.0, 8.3) Working memory: M1–0.13 (−2.8, 2.5); M3–0.28 (−3.0, 2.4) Processing speed: M1–0.37 (−5.5, 4.8); M3–0.44 (−5.7, 4.8)
HEALS project Croatia, Slovenia, Poland Calamandrei et al. (2020) 18 mo for PHIME and 24 mo for REPRO_PL Public	N = 984 Population Sampled: Mothers and their infants from the PHIME and REPRO_PL cohorts. Excluded PHIME: preterm births, babies with congenital malformations or severe perinatal problems or severe health problems that presented in the following mo and potentially compromised their neurological development. REPRO_PL: Poor‐quality BSID tests or missing data. PHIME Croatia, n = 141 PHIME Slovenia, n = 212 REPRO_PL, n = 311 Sex (% F): 51.6 Ethnicity: NR Mothers' age (yr): 29.6 ± 4.5	Neuropsychological development assessed by BSID‐III at 18 mo in PHIME cohort and at 24 mo in REPRO_PL cohort	Cord‐blood Se at delivery (μg/L, mean ± SD (min‐max)) PHIME Croatia 42.5 ± 9.0 (24.0–71.0) n = 141 PHIME Slovenia 40.3 ± 8.0 (15.0–61.0) REPRO_PL 31.1 ± 8.2 (13.8–56.3)	Sex, age at examination (mo), maternal pre‐pregnancy BMI, maternal education level (primary vs secondary vs university degree), and mode of delivery (caesarean vs natural)	Cord‐blood Se and BSID‐III scores, β‐coefficients (95% CI) PHIME CRO Cognitive: 0.045 (−0.201, 0.292) p = 0.717 Language: 0.178 (−0.144, 0.499) p= 0.278 Motor: 0.005 (−0.265, 0.276) p = 0.969 PHIME SLO Cognitive: 0.118 (−0.155, 0.391) p = 0.394 Language: 0.172 (−0.174, 0.518) p = 0.329 Motor: 0.050 (−0.224, 0.324) p= 0.719 REPRO_PL Cognitive: −0.024 (−0.320, 0.273) p= 0.876 Language: 0.071 (−0.186, 0.327) p = 0.587 Motor: −0.029 (−0.325, 0.266) p= 0.846 Non‐linearity not explored.
NHBCS USA Doherty et al. (2020) 3 yr Public	N = 2000 mother–child pairs Population Sampled: mothers aged 18–45 yr, carrying a singleton pregnancy, literate in English, having a private water system as the primary source of water at their residence n children = 371 for SRS‐2 and 318 for BASC‐2 Sex (% F): 51 Ethnicity: Majority white, non‐Hispanic Mothers' age (yr, median (IQR)): 31 (29,34)	Neurobehavior at 3 years of age assessed by mothers using the Social Responsiveness Scale, 2nd edition (SRS‐2) and the Behaviour Assessment System for Children, 2nd edition (BASC‐2)	Maternal prenatal toenail Se (at 25–30 gestational wk) (μg/g, median (IQR)) 0.97 (0.87, 1.07) Infant toenail Se at 2–8 wk after birth (μg/g, median (IQR)) 1.21 (0.90, 1.81)	Maternal age (quadratic), maternal BMI (quadratic), highest level of parental education (high school or less, any college, any graduate), sex, parity (0, ≥1), smoking status (no second‐ or first‐hand, ever second‐hand only, ever first‐hand), age at last breastfeeding (< 365 d, ≥ 365 d), maternal marital status (married, other), birth year, Healthy Eating Index (linear), Parenting Relationship Questionnaire (first three principal components), and age at assessment (linear). Toenail As, Cu, Mn, Pb, and Zn fixed at their medians	Maternal toenail Se and SRS‐2/BASC‐2 scores; effect estimate* (95% CI) SRS‐2 Total: −0.04 (−0.15, 0.07) BSC‐2 Behavioural Symptoms: 0.03 (−0.08, 0.14) Externalising Problems: 0.07 (−0.04, 0.17) Internalising Problems: −0.02 (−0.14, 0.10) Adaptive Skills: 0.02 (−0.09, 0.14) Infant toenail Se and SRS‐2/BASC‐2 scores; effect estimate* (95% CI) SRS‐2 Total: 0.00 (−0.05, 0.05) BSC‐2 Behavioural Symptoms: 0.03 (−0.02, 0.08) Externalising Problems: 0.04 (−0.01, 0.09) Internalising Problems: 0.06 (0.01, 0.12) Adaptive Skills: 0.01 (−0.04, 0.07) *Difference in the mean predicted outcome (standardised) between Se fixed at 75% versus 25% Associations appeared linear (after log2 transformation)
PHIME Italy Castriotta et al. (2020) 40 mo Public	N = 900 mother–child pairs Population sampled: mother–child pairs enrolled in the Northern Adriatic Cohort II in Italy Excluded: preterm births, babies with congenital malformations or severe perinatal problems or with severe health problems that presented in the following mo and potentially compromised their neurological development; missing outcome assessment n = 456 Sex (% F): 47.8 Ethnicity: NR Mothers' age (yr): 33.4 ± 4.3	Neuropsychological development assessed by BSID‐III at 40 mo old.	Cord‐blood Se (ng/g) 117.4 ± 27.1 Cord‐blood Se (ng/g), tertiles T1: Se ≤ 105 T2: 105 < Se ≤ 125 T3: Se > 125	Home size, children fish intake up to 18 mo	Dichotomised BSID‐III cognitive composite score*, OR (90% CI) T1 vs T2: 2.13 (1.18, 3.85) T3 vs T2: 1.41 (0.76, 2.62) *Children who scored under the 20th percentile of the cognitive composite score were considered as having suboptimal cognitive development and were compared with children who scored above that cut‐off

BSID: Bayley Scales of Infant Development; BW: body weight; d: day; F: females; HEALS: Health and Environment‐wide Associations based on Large population Surveys; INMA: Spanish Childhood and Environment Project; KSPD: Kyoto Scale of Psychological Development; M: males; MINIMat: Maternal and Infant Nutrition Interventions in Matlab; mo: month; MSCA: McCarthy Scales of Children's Abilities; NA: not applicable; NHBCS: New Hampshire Birth Cohort Study; NR: not reported; PHIME: Public Health Impact of Long‐term Low level Mixed Element Exposure in Susceptible Population Strata; REPRO_PL: Polish Mother and Child Cohort; SD: standard deviation; USA: United States of America; WISC: Wechsler Intelligence Scale for Children; wk: week; WPPSI: Wechsler Preschool and Primary Scale of Intelligence; yr: year.

(a) Mean ± SD, unless specified otherwise.