da Silva, Santos, De Marco, Rezende, Martins, Silva, Romano-Silva, Miranda 2025 — Heavy metals and child neurodevelopment systematic review (68 studies, 215,195 individuals, 23 countries)
This Federal University of Minas Gerais (UFMG) systematic review, published open-access in International Journal of Environmental Research and Public Health (IJERPH 22:1308, 21 August 2025) and registered at PROSPERO under CRD420250653229, synthesizes 68 primary studies (48 longitudinal + 20 cross-sectional) examining associations between early-life heavy-metal exposure and child neurodevelopmental outcomes. The pooled sample is 215,195 individuals from 23 countries with publication dates 2006-2025. Searches were conducted on 15 February 2025 across PubMed, Scopus, Web of Science, EMBASE, Lilacs, and PsycInfo using “children AND heavy metals AND neurodevelopment” descriptors. Study quality was assessed with the Joanna Briggs Institute (JBI) Critical Appraisal Tool; 43 of 68 studies (63%) were rated Good quality. Heterogeneity in exposure assessment methodology, exposure thresholds, and outcomes precluded meta-analysis; the review is therefore a narrative synthesis with PRISMA-aligned reporting. The five most consistently investigated metals across the corpus were Pb (51 studies; 75%), Hg (33 studies; 48.53%), Cd (20 studies; 29.41%), As (16 studies; 23.53%), and Mn (9 studies; 13.24%). Headline narrative findings: Pb shows the most consistent adverse associations across cognitive, motor, and behavioral domains; Hg and Cd are consistently linked to cognitive deficits with some behavioral and motor effects; As is most consistently associated with behavioral and motor outcomes; prenatal exposure (particularly the first trimester) yields more robust associations than postnatal exposure. Ten studies report beneficial associations for Hg, attributed in the discussion to co-occurring fish consumption (omega-3 and other nutritional factors offsetting MeHg neurotoxicity).
Key numbers
PRISMA flow (Figure 1, p. 5):
- Records identified across six databases (n=686): Embase 68, Lilacs 9, PsycInfo 6, PubMed 411, Scopus 95, Web of Science 97. Registers 0.
- Duplicates removed before screening: 187.
- Records screened by title/abstract: 499. Excluded at title/abstract: 379.
- Reports sought for retrieval: 120. Reports not retrieved: 10.
- Reports assessed for eligibility: 110. Excluded at full-text: 42 (anti-intoxication treatment 3; co-exposure with non-extractable heavy-metal data 9; indirect association 11; not a validated heavy-metal measurement instrument 6; out of scope 13).
- Included studies: 68.
Study-design split (Section 3.1, p. 5): 48 longitudinal + 20 cross-sectional. Publication years 2006-2025.
Sample by country (Table 1, p. 5-6; selected rows; full total n=215,195):
| Country | N studies | N sample |
|---|---|---|
| Japan | 5 | 181,145 |
| China | 12 | 8,485 |
| Spain | 5 | 4,479 |
| Norway | 2 | 4,375 |
| South Korea | 6 | 2,863 |
| Saudi Arabia | 3 | 1,970 |
| United Kingdom | 1 | 1,558 |
| Mexico | 4 | 1,481 |
| Brazil | 4 | 1,167 |
| USA | 4 | 1,057 |
| India | 2 | 992 |
| Taiwan | 5 | 762 |
| Italy | 2 | 760 |
| Canada | 2 | 727 |
| Seychelles | 1 | 643 |
| Greece | 1 | 575 |
| Bangladesh | 1 | 525 |
| Tanzania | 1 | 439 |
| Poland | 1 | 402 |
| Nepal | 3 | 300 |
| Bolivia | 1 | 246 |
| Malaysia | 1 | 155 |
| Democratic Republic of Congo | 1 | 89 |
| Total | 68 | 215,195 |
Gender distribution across the 68 studies (Section 3.1, p. 6): 47 studies (69.12%) male-majority, 17 (25%) female-majority, 4 (5.88%) not reported.
Age distribution across the 68 studies (Section 3.1, p. 6): 24 studies early childhood (0-2 years), 17 preschool (2-6 years), 10 middle childhood (6-9 years), 7 pre-adolescent (9-12 years), 5 spanned early childhood and preschool, 1 spanned preschool and middle childhood, 4 did not specify age.
Metal-investigation frequency across included studies (Section 3.2, p. 6-7):
| Metal | N studies | % of 68 |
|---|---|---|
| Lead (Pb) | 51 | 75.00% |
| Mercury (Hg) | 33 | 48.53% |
| Cadmium (Cd) | 20 | 29.41% |
| Arsenic (As) | 16 | 23.53% |
| Manganese (Mn) | 9 | 13.24% |
| Cesium (Cs) | 5 | 7.35% |
| Cobalt (Co) | 4 | 5.88% |
| Barium (Ba) | 4 | 5.88% |
| Thallium (Tl) | 4 | 5.88% |
| Nickel (Ni) | 3 | 4.41% |
| Uranium (U) | 3 | 4.41% |
| Strontium (Sr) | 3 | 4.41% |
| Rubidium (Rb) | 3 | 4.41% |
| Tin (Sn) | 2 | 2.94% |
| Titanium (Ti) | 2 | 2.94% |
| Vanadium (V) | 2 | 2.94% |
| Aluminum (Al), Silver (Ag), Tungsten (W), Cerium (Ce), Lanthanum (La), Rhenium (Re), Beryllium (Be), Platinum (Pt), Bismuth (Bi) | 1 each | 1.47% each |
Biological-sample types (Section 3.2, p. 7):
- Maternal samples (across 68 studies): blood 32 studies (47.06%), urine 10 (14.71%), hair 7 (10.29%), breast milk 4 (5.88%).
- Child samples (across 68 studies): blood 37 studies (54.41%), hair 10 (14.71%), urine 7 (10.29%), nails 2 (2.94%).
- Aggregate by sample type: blood 43 studies (63.23%), urine 9 studies exclusive (13.23%), hair or hair+nails 8 studies (11.76%).
Analytical methodology (Section 3.2, p. 7): ICP variations 34 studies (50%), atomic absorption variations 24 (34.29%), multi-method 8 (11.76%), Gold Amalgam Mercury Analyzer 1 (1.47%), portable LeadCare II 1 (1.47%).
Quality (Section 3.3, p. 7):
- Cross-sectional (n=20): Good 13 (65%), Fair 7 (35%), Low 0.
- Longitudinal (n=48): Good 30 (62.50%), Fair 17 (35.40%), Low 1 (2.1%).
Effect-size distribution (Section 3.4, p. 7-8): 29 of 68 studies (42.6%) reported sufficient data to estimate effect size. Among those 29, distribution was Small 79.3%, Moderate 13.8%, Large 17.2%. Of those 29 effect-sized studies: 55.2% Good, 41.4% Fair, 3.4% Low quality.
Adverse associations by metal × neurodevelopmental domain (Figure 2, p. 8; cells are N studies reporting that combination):
| Domain | Pb | Hg | Cd | As | Mn |
|---|---|---|---|---|---|
| Cognitive | 17 | 10 | 3 | 2 | 2 |
| Motor | 5 | 4 | 1 | 3 | 0 |
| Neurobehavioral | 3 | 1 | 1 | 3 | 1 |
| General | 1 | 3 | 1 | 2 | 0 |
| Sensory | 1 | 0 | 0 | 0 | 0 |
Beneficial associations by metal × neurodevelopmental domain (Figure 3, p. 8; cells are N studies reporting that combination):
| Domain | Pb | Hg | Cd | As | Mn |
|---|---|---|---|---|---|
| Cognitive | 2 | 2 | 0 | 0 | 0 |
| Motor | 1 | 0 | 1 | 0 | 0 |
| Neurobehavioral | 1 | 2 | 0 | 0 | 0 |
| General | 0 | 1 | 0 | 0 | 0 |
| Sensory | 0 | 0 | 0 | 0 | 0 |
Null associations by metal × neurodevelopmental domain (Figure 4, p. 8; cells are N studies reporting that combination):
| Domain | Pb | Hg | Cd | As | Mn |
|---|---|---|---|---|---|
| Cognitive | 8 | 6 | 4 | 1 | 1 |
| Motor | 2 | 2 | 2 | 0 | 0 |
| Neurobehavioral | 0 | 0 | 0 | 0 | 0 |
| General | 4 | 4 | 1 | 0 | 0 |
| Sensory | 0 | 0 | 0 | 0 | 0 |
Representative effect-size findings cited in Section 3.4.1-3.4.3 (p. 9; small/medium/large categorization is the authors’):
- Small Pb effects (Section 3.4.1): twelve studies report negative associations across lower cognitive performance, gross motor function, ADHD risk, sensory processing dysfunction, lower psychomotor development, lower global neurodevelopment.
- Small Hg/MeHg effects: two fair-quality studies report associations with improved global and motor scores (attributed to co-exposure fish-derived nutrients); six studies report adverse Hg/MeHg associations with social skills, behavioral cognition, mental development, and global delay.
- Small As, Cd, Mn effects: As linked to cognition, social skills, motor development, ASD risk, global neurodevelopment. Cd linked to ADHD risk, ASD risk, motor function. Mn linked to ASD risk and cognitive performance.
- Medium effects (Section 3.4.2): Hg-mental development (Kim 2018), Hg-visuomotor integration (Al-Saleh 2020), Pb-cognition (one low-quality study).
- Large effects (Section 3.4.3): Pb-cognitive/behavioral (Nie 2011, good quality), Hg-psychomotor (Kim 2018 reported elsewhere as well), Cd-cognitive (Kao 2023, good quality), Pb-cognitive (Halabicky 2023, fair quality), Cd-global neurodevelopment (Nyanza 2021, fair quality).
Per-study findings are tabulated in Table 2 (pp. 10-18) across the 68 included primary studies; bio-sample type and the major reported association are reported per study, with quality stars (* fair to *** good).
Methods (brief)
Systematic review designed per PRISMA 2020 (reference [33], Page et al. 2021) and registered at the International Prospective Register of Systematic Reviews (PROSPERO) under CRD420250653229. Search performed 15 February 2025 across PubMed, Scopus, Web of Science, EMBASE, Lilacs, and PsycInfo using descriptors “children AND ‘heavy metals’ AND neurodevelopment” (full strategy in Supplementary Material Table S1). PECO framework: Population = children; Exposure = presence of heavy metals in biological samples from mothers during pregnancy or lactation, or in children after birth, including placenta and umbilical cord blood; Comparator = absent in observational studies, or unexposed populations or different exposure levels; Outcome = general developmental milestones or neurodevelopmental disorders.
Inclusion criteria: studies of children with or without diagnosed neurodevelopmental disorders, English-language, no publication-date restriction. Exclusion criteria: neurodegenerative diseases (e.g., leukodystrophies, lysosomal storage diseases); studies without diagnoses based on validated instruments (e.g., Bayley Scales of Infant and Toddler Development, Wechsler Intelligence Scales, DSM-5/ICD-10 criteria); studies without validated individual biomonitoring of heavy-metal exposure; co-exposure studies that did not allow extraction of metal-specific data; studies focused on trace-element supplementation or iron-deficiency anemia; markers of iron metabolism (transferrin, ferritin); reviews of all types, case reports, case series, gray literature, letters, editorials, and meeting documents.
Screening was performed in pairs with conflicts resolved by a third reviewer. Data extraction used a standard form covering first author, date, country, study design, sample characterization (mother and/or child), sample type and analytical method, neurodevelopmental assessment instruments, and main results.
Quality assessment used the Joanna Briggs Institute (JBI) Critical Appraisal Tool, scored in pairs with conflicts resolved by a third reviewer; cross-sectional scale max 8 points, cohort scale max 11 points. The JBI does not publish a Good/Fair/Low cutoff; the review team weighted items addressing selection bias, measurement bias, and confounding (cross-sectional items 1, 3, 4, 5, 6, 7) and follow-up data and items 1, 3-9 for cohort studies. A “Good” classification required 100% Yes responses on these key items.
Speciation as reported across the underlying primary studies: most Hg measurements are total Hg, with a subset of studies separately reporting MeHg (notably Marques 2012/2015 fish-MeHg in the Brazilian Amazon, Al-Saleh 2016/2020 hair MeHg in Saudi infants, Orenstein 2014 MeHg memory study). Most As measurements are total As. Pb, Cd, and Mn are reported as elemental concentrations without speciation distinctions.
Implications
- Certification (HMTc): Exposure-effect evidence, not food-occurrence evidence. The review contributes corroborating biomarker-to-neurodevelopment association data for Pb, Hg, Cd, As, and Mn across 215,195 individuals in 23 countries, with particular weight on prenatal-exposure first-trimester critical windows. This is the population-health context that threshold-setting workflows draw on; the paper does not propose threshold values and neither does this page.
- Courses: Useful as a teachable reference for the narrative-synthesis approach when meta-analysis is precluded by methodological heterogeneity. The metal × domain heat-map presentation (Figures 2-4) is a useful pedagogical artifact for showing how scattered primary-study findings cluster across cognitive, motor, behavioral, sensory, and general developmental domains. The Hg-beneficial-association subset attributed to co-exposure fish consumption is a useful case study in confounding by nutritional co-exposure.
- App: Does not contribute to ingredient
contamination_profilevalues. This is biomarker/exposure-outcome data, not food-occurrence data. - Microbiome: Section 4 discussion (p. 19) references gut-microbiota modulation of metal bioavailability and excretion, citing animal-model and ASD-microbiota-transfer evidence; this is literature-citation framing within the discussion rather than primary measurement.
Verification notes
- Cite-key derived from first author + year + topic per the system-prompt naming rule.
- Metals frontmatter uses
Pb,tHg,MeHg,Cd,tAs,Mnper Part 14 speciation rule. The review covers 22 named metals across the corpus; only the five consistently investigated (Pb 75%, Hg 48.53%, Cd 29.41%, As 23.53%, Mn 13.24%) plus the Hg and As speciation forms most commonly reported in primary studies are listed. The rarer metals (Cs, Co, Ba, Tl, Ni, U, Sr, Rb, Sn, Ti, V, Al, Ag, W, Ce, La, Re, Be, Pt, Bi) are reported in the Key numbers table but not in the metals routing list because their evidence base in this review is single-study or near-single-study and not a primary focus of the synthesis. - Matrices field uses
blood,urine,hair,breastmilk,umbilical-cord-blood,neurodevelopment— biological-sample types and the outcome modality, consistent with the matrix vocabulary used in other exposure-outcome systematic reviews ingested in this folder (cf. ding2023-heavy-metals-asd-meta-analysis). jurisdictions: [international]follows the global-review convention; included primary studies span 23 countries across five continents.- Products and ingredients arrays are both empty because no food or personal-care product is measured. The review mentions in the Introduction (Section 1, p. 2) that “personal hygiene products and cosmetics may contain Pb, As, and Hg [24,25]” but this is one paragraph citing two background references; the included 68 primary studies measure heavy metals in biological samples from mothers and children, not in consumer products. This source contributes to the five metal pages only and does not feed any product-page Literature Evidence Summary.
- Effect-size percentage discrepancy: Section 3.4 (p. 7) reports the distribution of effect-size magnitudes among the 29 effect-sized studies as Small 79.3%, Moderate 13.8%, Large 17.2%. These sum to 110.9% rather than 100%. The text explicitly notes “some studies assessed more than one outcome,” which accounts for the overlap; some studies reported a mix of small, moderate, and large effects across multiple outcome domains. Recorded the values as printed.
- Hg beneficial associations (Section 4, p. 19): the discussion attributes the ten studies reporting positive Hg-neurodevelopment associations [41,43,66,73,103] to higher fish consumption in the studied populations, citing Hibbeln et al. [117] and Lyall et al. [118] reviews on seafood/omega-3 attenuation of MeHg neurotoxicity. Reference [66] is Nie et al. 2011 (USA cross-sectional, n=11), which Table 2 lists as showing positive associations with Verbal Comprehension, Working Memory, and IQ — but the same study also shows positive associations with externalizing problems and behavioral symptoms (adverse). Recorded the discussion’s attribution as printed; the Nie 2011 study is a special case where the small sample size (n=11) supports both adverse and beneficial association reads.
- Author/year/journal/DOI verified against the PDF title page (p. 1): da Silva et al., IJERPH 2025, 22, 1308, https://doi.org/10.3390/ijerph22081308, published 21 August 2025, MDPI, CC BY 4.0.
- The folder placement (
babycare_05_Other_Categories/05_Other_Categories) reflects that this paper does not align with a single product-category page — it is a cross-cutting exposure-outcome systematic review. Routing destinations are the five metal pages plus the umbrellainfant-and-child-foods-masterregulatory-context page is not declared here because no specific food product is measured; the review’s relevance to infant-and-child food is via the biomarker-to-outcome link, which is mediated by the metal pages.
Wiki pages this source may touch
Page history
The five most recent substantive edits to this page. The full version history lives in git; when DOI minting comes online (see schema docs), each entry below will also link to a version-pinned DataCite DOI.