ECHA 2023 — Investigation report on CMR 1A or 1B substances in childcare articles (REACH Article 68(2))

This European Chemicals Agency (ECHA) investigation report, prepared at the request of the European Commission under REACH Article 68(2) and dated 31 October 2023 (Version 2.0 Final), assembles the evidence base for a future EU restriction on carcinogenic, mutagenic, and reproductive-toxicant category 1A or 1B (CMR 1A/1B) substances in childcare articles. The report is not a primary-measurement study; it aggregates measurements and regulatory limits from 48 sources to support seven deliverables specified in the Commission’s November 2022 request: (1) a list of CMR 1A/1B substances that may be present in childcare articles, (2) tonnage/concentration/frequency-of-finding evidence, (3) release/exposure-potential evidence, (4) an inventory of existing Generic and Specific Concentration Limits (GCL/SCL), (5) identification of where GCL/SCL deviations are needed, (6) a summary of available analytical methods, and (7) an overview of stakeholder consultations. Of the 1,559 measurement entries collected across all CMR groups, 39 % (≈ 608 entries) cover metals, with cobalt compounds the single most-reported metal class at 373 entries and lead compounds second at 164 entries (p. 21, confirmed by the p. 17 narrative that “the CMR 1A or 1B substances most frequently detected in childcare articles are metals, being cobalt and lead the most reported followed by phthalates”); phthalates account for 15 % of measurement entries, organophosphorus compounds for 21 %, and all other chemical groups for ≤ 10 % each. The report’s principal HMI-actionable outputs are (a) the per-metal proposed concentration limits, both content and extractable, deviating from the default 10 mg/kg homogeneous-material limit ECHA proposes for the broader CMR-substance class (Table 3, pp. 34); and (b) the reported concentration-range exceedances of existing CLP/REACH limits for Pb in textiles and synthetic polymers (6 of 106 measurements > 300 mg/kg, non-EU data only) and Co in coatings (3 of 194 measurements > 1,000 mg/kg, non-EU data only). The report explicitly excludes risk-assessment-based threshold derivation in favour of LOQ-anchored content limits, on the rationale that the GCL/SCL framework “may not be sufficiently protective for children” because CMR 1A/1B substances cover genotoxic carcinogens for which “no safe threshold can usually be established” (p. 28).

Key numbers

Aggregate substance and measurement scope (Sections 2.1, 2.2, 3.2)

• CLP ATP-18 entries with harmonised CMR 1A or 1B classification: 1,116 entries; 1,188 individual substances after expansion of defined-set group entries; 108 additional members of open group entries also considered (p. 15).
• Additional substances considered: 27 with CMR 1A or 1B classification agreed by RAC but not yet in CLP Annex VI, and 12 under RAC discussion as of 27 January 2023 (p. 16).
• Substances identified as actually present in childcare articles: 180–181 (the report quotes both numbers; 180 in p. 28 — “For 149 of the 180 substances identified…” — and 181 in p. 30 — “ECHA collected and analysed existing concentration limits and related analytical methods for all 181 substances identified…”); split into 65 measured and 116 suspected (p. 16).
• Measurement entries collected: 1,559 reporting measurements of CMR 1A/1B substances in childcare articles (p. 21).
• Distribution by chemical group (Figure 1, p. 16; shares of the 2,340 “information-collected” entries on measured/suspected CMR-substance presence — per the p. 16 surrounding statement that “the information collected in this report for CMR 1A or 1B substances (2 340 entries) … is categorised in two ways: substances measured and/or substances suspected in childcare articles”): Metals 29 %; Phthalates 23 %; Organophosphorus compounds 15 %; Aldehydes 5 %; Bisphenols 5 %; Aromatic amines 4 %; PAHs 4 %; PFAS 3 %; Organotins 3 %; with the remaining 26 of 35 total chemical groups (per p. 17 statement that “the substances can be divided in 35 different chemical groups”) comprising the residual share. The metal share is the largest single category.
• Distribution by chemical group (measurement-entry shares of the 1,559 measurement entries; p. 21): Metals 39 % (≈ 608 entries; per p. 21 verbatim, “cobalt and lead compounds the most reported (373 and 164 entries, respectively)” — Co compounds 373 entries the single most-reported metal/substance class, Pb compounds 164 entries the second), Organophosphorus compounds 21 % (mainly tris(2-chloroethyl)phosphate, TCEP), Phthalates 15 % (mainly DEHP, DnBP, BBP), all other groups < 10 % each.
• Distribution by childcare-article category (Figure 2, p. 18; measurement-entry shares): Car seats 17 %, Bibs 16 %, Toilet-related 15 %, Bed/mattress 12 %, Diaper/nappy 10 %, Feeding/drinking 8 %, Pram/pushchair/buggy 6 %, Sucking-related 4 %, Playing/walking 4 %, Seating-for-eating 3 %, Other-seating 2 %, with 10 further sub-categories at 1 % or below.
• Substances measured outside EU only (selected, p. 21): acetaldehyde, benzene, tetrabromobisphenol A, vinyl chloride, arsenic compounds, mercury, and cobalt. The report explicitly carries non-EU measurements into the analysis because “the global/internet market” is a relevant source of childcare articles placed on the EU market.

Reported concentration-range exceedances of CLP GCL/SCL (Section 3.2, p. 21–22)

Reported concentrations exceeding the CLP Generic or Specific Concentration Limit (1,000 mg/kg for carcinogenic/mutagenic substances or 3,000 mg/kg for substances toxic to reproduction) — verbatim from p. 21–22:

The report also notes (p. 22) that “very high concentrations of eight different lead compounds were reported for a baby dish sterilizer in the SCIP database. However, it is not reported in which part of the sterilizer the lead substances are included and if the substances were inaccessible or in contact with the dishes to be later used by children.” No numerical Pb concentration is reported for this SCIP entry in the investigation report.

Proposed content and extractable concentration limits for metals (Table 3, pp. 34–35; Section 5.2 per-metal narrative, pp. 40–42)

Table 3 (pp. 34–35) is captioned “Substances or groups of substances with a considered concentration limit in homogeneous material different from 10 mg/kg (content) and/or with considered extractable-related concentration limits.” Table 3 therefore lists only the metals with a substance-specific deviation from the 10 mg/kg default and/or an extractable limit (As, Cd, Cr-VI, Co, Pb, Hg, organotins DOT/DBT). Lithium, vanadium, and ammonium bromide (Br⁻) take the default 10 mg/kg content limit without an extractable and are therefore not in Table 3; their rows below are sourced from the Section 5.2 per-metal narrative on pp. 42 (Li, V) and p. 38 (Br, in the brominated-flame-retardants sub-section).

Concentration limits in the content column apply to the total elemental concentration of the metal atom or ion in homogeneous material of the childcare article, not to the specific CMR-classified compound. Quoting the report (p. 28, last paragraph): “For metals such as arsenic (As), cadmium (Cd), cobalt (Co), lead (Pb), mercury (Hg), chromium (VI) (CrVI), for dioctyl tin (DOT) or dibutyl tin (DBT) compounds, for boron (B) and bromine (Br⁻), the concentration limits (content, extractable content or migration) reflect the concentration of the species (atom or ion) analysed, and not the concentration of the substances that may be present in childcare articles. Therefore, for the concentration limits the species concerned is indicated, e.g. 10 mg As/kg in case of arsenic.”

ECHA’s general criteria flow-chart for content-limit selection (Figure 4, p. 33): if a REACH restriction limit exists and is lower than 10 mg/kg, use the REACH limit; if higher, use the REACH limit unless technical reasons justify deviation to a lower industry-RSL-derived limit; if no REACH restriction exists, default to 10 mg/kg unless an industry RSL or other concentration limit below 10 mg/kg with a suitable LOQ is available, in which case use the lower limit.

For comparison, ECHA’s default for substances without a substance-specific consideration is 10 mg/kg in homogeneous materials = 0.001 % w/w, justified on the grounds that “state-of-art analytical methods/techniques for the quantification of almost any type of substance guarantees LOQs below or at least of 10 mg/kg” (p. 31). The default 10 mg/kg is considered for 78 of 168 substances (46 %) for which a concentration limit is identified (p. 31).

Inventory of existing limits and analytical-method coverage (Section 5.1, p. 27; Section 6, p. 46)

• Substances with an existing concentration limit identified (content, extractable, or migration): 149 of 180 substances ≈ 83 % (p. 28).
• Substances with legally binding concentration limits from REACH restrictions, the Toys Safety Directive, or other EU legislation: 91 ≈ 50 % (p. 28).
• Substances for which the 10 mg/kg default is considered to apply: 78 of 168 with an identified limit ≈ 46 % (p. 31).
• Analytical-method availability conclusion (p. 46): standard analytical methods exist for the majority of identified substances; methods are largely from CEN, DIN, ISO, US EPA, and US CPSC; portable ED-XRF allows on-site screening for As, Cd, Co, Cr, Pb, V (p. 48); portable FT-IR allows screening for total phthalate content in synthetic polymers (mainly PVC).

Childcare-article categorisation under scope (Table 1, pp. 10–11)

The report defines childcare articles per a merged REACH Annex XVII entry 51–52 + CEN/TC 252 definition: “any product intended to facilitate seating, sleep, relaxation, hygiene such as bathing, changing and general body care, feeding, sucking, sleeping, transportation and protection of children” (p. 8). Six main categories with 17 sub-categories:

ECHA’s age-of-children scope aligns with the Toys Safety Directive (2009/48/EC) at under 14 years of age, rather than the more restrictive under-36-months or under-3-years conventions used by ecolabels and the EU Ecolabel for textiles (p. 9).

Other ECHA-considered concentration limits relevant to the broader CMR-substance scope (Table 3, p. 34–35)

(Recorded here for completeness; non-metal substances are outside the HMI metal panel but appear in the same childcare-articles regulatory context.)

Methods (brief)

Source aggregation and substance inventory (Sections 2.1–2.2). Starting set is the 1,116 CLP ATP-18 entries with harmonised CMR 1A or 1B classification, expanded by member-substance enumeration for defined-set group entries (e.g., the group entry covering three phthalates, index 607-426-00-1) and by inclusion of 108 identified members of open group entries (e.g., 2-ethylhexanoic acid and its salts, index 607-230-00-6). Substances with RAC-agreed CMR 1A/1B classification but not yet in CLP Annex VI (27 substances) and substances under RAC discussion (12 substances, CLH proposal submitted) as of 27 January 2023 are also considered. Identification of substances actually present in childcare articles draws on 48 sources: literature search, REACH restrictions, REACH registration dossiers, the SCIP (Substances of Concern In articles, as such or in complex objects) database, Substance in Articles (SiA) notifications, RAPEX rapid-alert recalls, the 2019 COM study on childcare articles, industry Restricted Substances Lists (AFIRM, AAFA, ALDI, Amazon, Bed Bath & Beyond, Bluesign, Versace, Cradle to Cradle), ecolabels (OEKO-TEX Standard 100, EU Ecolabel for textiles and shoes for children, GOTS, Blue Angel textiles, Bluesign, Nordic Swan), Member State authority reports (BfR, BAUA, ANSES, RIVM, KEMI, Danish EPA), and stakeholder consultations. The full substance inventory and per-substance evidence summary live in the report’s Appendix_A_Summary Excel workbook (columns Q tonnage, N concentration ranges, R exposure potential, V existing concentration limits, W extractable content limits, X considered limits).

Concentration-range and frequency aggregation (Section 3.2). ECHA collected 1,559 entries reporting measurements of CMR 1A/1B substances in childcare articles. 39 % cover metals (per p. 21 verbatim, Co compounds the most reported at 373 entries; Pb compounds second at 164 entries); 21 % organophosphorus (mainly tris(2-chloroethyl)phosphate, TCEP); 15 % phthalates (mainly DEHP, DnBP, BBP); other chemical groups < 10 % each. ECHA notes that the dataset cannot support frequency-of-finding conclusions because negative measurements are rarely reported, only “positive” results; this is identified Uncertainty 2 in Table 5 (p. 64). The dataset cannot discriminate EU-produced vs non-EU-produced articles in most cases; specific substances were measured only outside the EU (acetaldehyde, benzene, tetrabromobisphenol A, vinyl chloride, arsenic compounds, mercury, cobalt), and the report carries non-EU data into the analysis because globally/internet-sourced products reach the EU market.

Exposure-potential estimation (Section 3.3). For the 107 substances under REACH registration with physico-chemical data, oral/dermal/inhalation exposure potential was estimated using log Kow, aqueous solubility, and vapour pressure thresholds per the 2019 COM study’s Table 5.1. Concentration detected only at trace level (≤ 0.01 % = 100 mg/kg) was treated as potential low exposure. ECHA did not perform substance-specific risk assessments; REACH Article 68(2) does not require them. Of the 107 substances assessed, exposure potential was estimated as low for 39, medium for 8, and high for 59 substances.

Concentration-limit derivation (Section 5.2, Figure 4 flow-chart, Table 3). The default content concentration limit is 10 mg/kg in homogeneous material (0.001 % w/w), justified because state-of-art analytical methods (CEN, ISO, US EPA, OEKO-TEX, EN ISO standards) guarantee LOQs at or below 10 mg/kg for almost any substance type. Deviations from 10 mg/kg are considered when (i) a REACH-restriction concentration limit is already in place, (ii) technical reasons make 10 mg/kg infeasible, (iii) the LOQ for a standard analytical method is higher than 10 mg/kg, or (iv) industry RSLs and ecolabels consistently apply a stricter limit. For metals, both content limits and extractable-content limits are considered; the extractable limit for As, Cd, Co, Pb, Hg in textiles is anchored to the EU Ecolabel for textiles and shoes for children, OEKO-TEX, and GOTS ecolabels. For chromium VI, no method exists to measure total CrVI content via acid digestion without loss of the +VI oxidation state; the existing REACH entry 47/72 extractable limits are retained.

Testing-strategy guidance (Section 6.2). Five-step testing strategy for enforcement: (1) identify whether the article meets the childcare-article definition; (2) identify the materials accessible to children (skin, mouth, or inhalation contact); (3) determine which substances need testing for those materials, using Table 4 testing-priority matrix; (4) perform screening tests (e.g., portable ED-XRF for As, Cd, Co, Cr, Pb, V in any material; portable FT-IR for total phthalate content in synthetic polymers, mainly PVC); (5) perform targeted analytical tests for confirmation. CEN/TC 252 standards (EN 14372 child cutlery and feeding utensils; EN 14350 child drinking equipment; EN 1400 + A2 soothers; EN 12586 + A1 soother holders; EN 14988 + A1 children’s highchairs) provide chemical-safety requirements and test methods for several childcare-article categories (p. 14).

Speciation discipline (HMI Part 14). The report’s per-metal concentration limits are anchored to total elemental content/extractable measurements (the atom or ion analysed, not the specific CMR-classified compound), per the explicit statement at p. 28. The HMI frontmatter accordingly uses:

• tAs for the diarsenic-trioxide substance entry because the analytical methods measure total elemental As content without speciation between inorganic and total arsenic;
• tHg for mercury because the report does not separate methylmercury from total mercury;
• Cr-VI for chromium VI compounds because the report explicitly speciates Cr-VI in its substance entry, regulatory framework (REACH entries 47, 72), and concentration-limit derivation;
• Sn for the organotin entries (dioctyl tin DOT and dibutyl tin DBT compounds) because the metals abbreviation vocabulary uses elemental Sn; routing to [[metals/organotins]] carries the speciated context;
• Pb, Cd, Co, Li, V follow standard elemental tags.

Stakeholder process (Chapter 7). 233 stakeholders informed (203 EEA + 30 non-EEA across UK, Canada, US, Switzerland) via two calls for evidence (15 February – 31 March 2023 and 10 May – 7 June 2023), ten individual meetings (Environment Protection Agency of Austria, a worldwide childcare-article manufacturer, BEUC, Danish Consumer Council THINK, Finnish Customs Laboratory TULLI, Swedish Chemicals Agency KEMI, Norwegian Environment Agency, CEN, TÜV SÜD Product Service GMBH, SGS), the ECHA Forum for Exchange of Information on Enforcement, the RIME+ platform (informal Risk Management and Evaluation), CARACAL (Competent Authorities for REACH and CLP), and the draft-report consultation (28 August – 29 September 2023). 41 of the 233 stakeholders provided relevant substantive input. Stakeholder distribution by type (Figure 6, p. 58): Authority/Agency 30 %, Producer 21 %, Testing laboratories 18 %, Industry Association 16 %, Environmental organisation 6 %, Consumer Association 5 %, ECO label 1 %, Consultant 1 %, Children-protection Association 1 %, Chemicals-safety Association 1 %. EEA stakeholder counts by country (Figure 7, p. 59): Germany 38 (largest), Belgium 26, Italy 17, France 16, Sweden 10, Denmark 9, Austria 7, Ireland 7, Portugal 6, Poland 6, Netherlands 5, Finland 4, with smaller counts for the remaining 19 EEA countries.

Uncertainty analysis (Chapter 8, Table 5, p. 64). Seven key uncertainties identified per EFSA’s 2018 uncertainty-communication guidance: (1) completeness of the CMR-substance list as potentially-present in childcare articles; (2) representativeness of findings and share of childcare articles potentially containing CMR 1A/1B; (3) exposure potential and consequent risk; (4) suitability of some proposed concentration limits in specific materials; (5) applicability of the 10 mg/kg default to substances for which no concentration-limit or method data were collected; (6) applicability of identified analytical methods to some substances/materials; (7) availability of analytical methods for substances assigned the 10 mg/kg default without method-data collection. None of the uncertainties is judged to alter the report’s main conclusion that CMR 1A/1B substances should be banned in childcare articles.

Implications

• Direction of finding — broad-context regulatory evidence for childcare-articles category. This is an aggregating regulatory investigation report, not a primary-measurement study. Its HMI role is broad regulatory context for childcare-articles product categories where heavy-metals data was the largest single-class signal across the assembled 1,559 measurement entries (39 % of entries; Co compounds the single most-reported metal class at 373 entries and Pb compounds second at 164 entries, per p. 21). For HMTc product categorisation, the report supplies regulatory baseline and surveillance context across at least 16 HMI-mapped childcare-article product slugs (car seats, bibs, high chairs and booster seats, cribs and bassinets, strollers, swings and bouncers, walkers, play yards and gates, diapers and components, baby wipes, infant bottles, infant bottle nipples, sippy cups and toddler drinkware, pacifiers and sucking/teething aids, toy pacifiers, blankets and sleep sacks). Where HMTc product taxonomy has direct measurement coverage (feeding-related items in particular: bibs, infant bottles, infant bottle nipples, sippy cups, pacifiers), this report contributes regulatory-baseline context but does not substitute for direct measurement data, which the report itself notes is sparse (Uncertainty 2, p. 64).

• Metals coverage and proposed regulatory anchoring. The proposed extractable limits for As (0.2 mg/kg), Cd (0.1 mg/kg), Co (1.0 mg/kg), Pb (0.2 mg/kg), and Hg (0.02 mg/kg) — applicable to babies and children under 3 years of age and anchored to EU Ecolabel for textiles and shoes for children, OEKO-TEX, and GOTS limits — are the strictest harmonised regulatory limits ECHA could derive without exceeding the LOQ of available standard methods. The proposed content limits (10 mg/kg for As, Cd, Co, Pb; 0.5 mg Hg/kg) deviate from existing CLP GCL (1,000 mg/kg) and from the existing REACH entry 23 plastic-material Cd limit (100 mg/kg) and REACH entry 63 consumer-articles Pb limit (500 mg/kg) on the rationale that the GCLs/SCLs are not substance-specific risk-based limits and “may not be sufficiently protective for children” because CMR 1A/1B substances cover genotoxic carcinogens for which no safe threshold can usually be established (p. 28). For HMTc certification work, the report provides defensible baseline anchoring for Pb, Cd, As, Hg, and Co limits in the childcare-articles domain (LOQ-based 10 mg/kg content + 0.1–0.2 mg/kg extractable), with the caveat that the report’s scope is the EU regulatory framework rather than the food-and-personal-care HMTc scope.

• Chromium-VI speciation discipline. ECHA explicitly notes (p. 41) that “no methods are available to analyse the total content of the chromium VI ion (CrVI) in articles through acid digestion of the material without losing the +VI oxidation state,” and therefore retains the existing REACH extractable limits (1 mg CrVI/kg textiles per entry 72; 3 mg CrVI/kg leather per entry 47) without proposing a content limit. This is the regulatory analogue of HMI’s speciation-discipline rule (Part 14): the Cr-VI species cannot be derived from total Cr by acid digestion, and any HMI source that claims Cr-VI quantification without species-preserving methodology (alkaline extraction, EN ISO 17075 or similar) should be flagged.

• Phthalate, organophosphorus, and TBBPA non-EU exceedances are the leading concentration-range exceedance signals. Of the 1,559 entries, the highest-frequency exceedances of CLP GCLs/SCLs are for DEHP (85 of 202 measurements > 3,000 mg/kg), TCEP (9 of 310 > 3,000 mg/kg), and TBBPA (4 of 5 > 1,000 mg/kg, non-EU only). Pb exceedances in textiles/synthetic polymers (6 of 106 > 300 mg/kg, non-EU only) and Co in coatings (3 of 194 > 1,000 mg/kg, non-EU only) are the leading metal exceedance signals. The report notes that the high-DEHP exceedances are “mainly reported in articles placed on the market before 2020 and/or in old articles (e.g., second hand),” consistent with REACH entry 51’s 2007–2020 progressive restriction of DEHP, DBP, BBP, and DIBP in toys and childcare articles.

• App and Courses. Courses: the report’s five-step testing strategy (identify article → identify accessible materials → identify substances by material × substance matrix → screen with portable ED-XRF/FT-IR → confirm with targeted methods) and the explicit role of portable ED-XRF for As, Cd, Co, Cr, Pb, V in any material is a directly transferable Cochrane-quality enforcement playbook for the childcare-articles domain (p. 47–48). App: the proposed per-metal content and extractable limits anchored to OEKO-TEX/GOTS/EU Ecolabel limits provide the regulatory floor against which the app’s downstream HMTc-aligned thresholds for feeding-related childcare products (bibs, infant bottles and nipples, sippy cups, pacifiers/teethers) should be benchmarked when those rows reach the Step 0 Lock workflow. Microbiome: not addressed by this source.

Wiki pages this source may touch

• lead
• cadmium
• chromium-hexavalent
• cobalt
• arsenic-total
• mercury-total
• tin
• organotins
• lithium
• vanadium
• car-seats
• bibs
• high-chairs-and-booster-seats
• cribs-and-bassinets
• strollers
• swings-and-bouncers
• walkers
• play-yards-and-gates
• diapers-and-components
• baby-wipes
• infant-bottles
• infant-bottle-nipples
• sippy-cups-toddler-drinkware
• pacifiers-and-sucking-teething-aids
• toy-pacifiers
• blankets-and-sleep-sacks

Verification notes

• 2026-06-01 fresh ingest (Claude Opus 4.7, autonomous v2.0 manual-fetch skill, NEW path). Three identity checks against wiki/sources/ returned no match: full-title search (“Investigation report … CMRs in Childcare Articles … REACH Article 68(2)”) returned no existing pages; raw_handle MFK_04-unknown returned only the sibling page kjolholt2015-car-safety-seats-childrens-textiles which uses MFK_02-unknown (different file in the same folder); cite-key search for echa2023* returned no existing pages. The pre-existing sccs2023-notes-of-guidance-cosmetic-ingredients-v12 is the SCCS Notes of Guidance (a different EU agency report from the same year). Source page written from scratch.
• Source identification. European Chemicals Agency (ECHA), “Investigation report to support the Commission on the preparation of a restriction proposal for the use and presence of CMR 1A or 1B substances in childcare articles based on REACH Article 68(2),” Version 2.0 (Final), 31 October 2023. ECHA, Helsinki. Version history per the report’s front matter: 1.0 draft for third-party consultation (23 August 2023); 2.0 Final (31 October 2023). The report is a deliverable to the European Commission under REACH Article 68(2), in response to the Commission’s request of 29 November 2022. Corporate author “European Chemicals Agency” (no named individual authors; the report is an institutional ECHA output).
• DOI. ECHA investigation reports under REACH Article 68(2) are not minted with DOIs; doi: null and no_doi_assigned: true are correct. The report is available via the ECHA website (echa.europa.eu); the access_url: "https://echa.europa.eu" records the agency root (specific PDF URL not transcribed in the source).
• Source-tier rationale. evidence_tier: A. Per CLAUDE.md Part 13, A-tier covers “government/agency reports” and “authoritative scientific opinions.” This is an agency report from ECHA — the EU’s primary chemicals regulator — prepared at the European Commission’s formal request, version-controlled with a public draft consultation, endorsed by the Forum for Exchange of Information on Enforcement, and consulted with 233 stakeholders across the EEA + UK/CA/US. It is the regulatory-side analogue of EFSA scientific opinions for the chemicals-restriction domain. The report aggregates measurement data from 48 sources rather than performing primary measurement; this is documented in the Methods section and is the standard pattern for agency restriction-preparation reports, not a tier downgrade.
• Source-type rationale. source_type: regulatory-agency-report. Established ECHA agency output with corporate author, version numbering, public consultation, and Commission-deliverable status. Distinct from regulatory-text (statutes/directives such as REACH itself) and from peer-reviewed (academic journal articles).
• License rationale. license: "ECHA public report; standard ECHA report distribution applies." ECHA agency reports are publicly accessible via echa.europa.eu; standard government-publication open redistribution with attribution applies.
• Frontmatter products: field. Sixteen slugs selected from the 2026-05-18 taxonomy snapshot that match the paper’s measurement-coverage-weighted childcare-article categories per Figure 2 (p. 18) and Table 1 (p. 10–11):
  • car-seats — Table 1 sub-category 2-2; 17 % of measurement entries (the single largest category); Figure 5 (p. 47) uses a car-seat illustration as the report’s lead example of accessible-material identification (textile, plastic, metal accessible to children);
  • bibs — Table 1 sub-category 5-1; 16 % of measurement entries (second largest);
  • high-chairs-and-booster-seats — Table 1 sub-category 1-3 (“seating specific for eating”: highchairs, highchair back pads, feeding booster seats, table-mounted chairs, hook-on chairs); CEN/TC 252 standard EN 14988 + A1 specifically named (p. 14);
  • cribs-and-bassinets — Table 1 sub-category 2-1; covers crib mattresses, baby cot mattresses, bassinets, baby cribs and a substantial fraction of the 12 % bed-and-mattress measurement entries;
  • strollers — Table 1 sub-category 3-1 (“buggies, prams, pushchairs, strollers, carriages”); 6 % of measurement entries; only CN customs code (871500) that ECHA identified for childcare-article identification per p. 49;
  • swings-and-bouncers — Table 1 sub-category 1-4 (“baby bouncing cradles/rocker seats — non-powered and powered — baby swings, infant bouncer seats, infant and toddler rockers”);
  • walkers — Table 1 sub-category 4-2 (“baby playpens, baby play dens, baby walkers, crawling blankets, baby play rugs, stationary activity centres”);
  • play-yards-and-gates — Table 1 sub-category 4-3 (“baby safety protection, safety gates, baby monitors, cabinet safety locks and latches, expandable gates and enclosures, play yards, portable bed rails”);
  • diapers-and-components — Table 1 sub-category 1-2 (“baby single-use diapers/nappies, baby reusable diapers/nappies, baby diaper accessories, baby wipes”); 10 % of measurement entries; ANSES 2019 single-use diapers opinion explicitly cited (p. 21);
  • baby-wipes — Table 1 sub-category 1-2 (“baby wipes”); EDANA stakeholder discussion extensively cited;
  • infant-bottles, infant-bottle-nipples, sippy-cups-toddler-drinkware — Table 1 sub-category 5-2 (“baby bottles, baby feeding products, baby and children’s tableware (cutlery, plate, cup), sippy cups, feeding teats, food feeders, baby dishes sterilizer, children water bottles”); CEN/TC 252 standard EN 14350 named (p. 14); 8 % of measurement entries; the SCIP-database baby dish sterilizer with high Pb concentrations (p. 22) maps here;
  • pacifiers-and-sucking-teething-aids — Table 1 sub-category 5-3 (“pacifiers/soothers/dummies/comforters, soother with thermometer, soother holders/chains, teethers, teething rings, baby rattles and teethers, soft and hard biters”); CEN/TC 252 standards EN 1400 + A2 (soothers) and EN 12586 + A1 (soother holders) named (p. 14); 4 % of measurement entries;
  • toy-pacifiers — sub-category 5-3 (“soother holders/chains with playing function”); explicit Toys Safety Directive crossover noted (footnote 8, p. 11);
  • blankets-and-sleep-sacks — Table 1 sub-category 6-1 (“baby blankets, baby and children sleeping bags”).
  • Slugs deliberately not included: food-contact-* slugs (the report’s scope is childcare articles broadly, including feeding-related; the food-contact regulatory regime is treated as a derogated parallel regulatory channel per p. 13–14, not as an HMI matrix here); infant-clothing (the report’s scope excludes general clothing — “articles produced for adults but also used by children” — and treats textiles primarily as the material of construction within childcare articles, not as the article itself); food-packaging-* (out of childcare-article scope); toy-* slugs other than toy-pacifiers (the report acknowledges Toys Safety Directive crossover but explicitly excludes toys per se from its scope, except where a soother-with-playing-function would fall under TSD instead).
• Frontmatter ingredients: [] is correct — childcare-articles regulatory report; no food ingredients in scope.
• Frontmatter matrices: [] is correct — no childcare-article matrix slug exists in the current matrices vocabulary (which is food/water/soil/cosmetic-personal-care biased). Consistent with sibling regulatory-context sources kjolholt2015-car-safety-seats-childrens-textiles, depa2015-cmr-toys, sccs2023-notes-of-guidance-cosmetic-ingredients-v12, and bfr2009-lead-cadmium-toys (and the rest of the toys-and-childcare-articles corpus) which also leave matrices empty. The products field carries the routing signal.
• Frontmatter metals: [Pb, Cd, "Cr-VI", Co, tAs, tHg, Sn, Li, V] per CLAUDE.md Part 14 speciation discipline:
  • Pb — Lead and its compounds (EC 200-755-8); 164 measurement entries, the single most reported substance class; proposed limits 10 mg Pb/kg content + 0.2 mg Pb/kg extractable; existing REACH entries 63 (500 mg/kg content) and 72 (1 mg Pb/kg textile extractable); 6 of 106 measurements > 300 mg/kg in textiles/synthetic polymers (non-EU); SCIP-database baby dish sterilizer with high Pb-compound concentrations.
  • Cd — Cadmium and its compounds (EC 231-152-8); proposed 10 mg Cd/kg content + 0.1 mg Cd/kg extractable; existing REACH entry 23 (100 mg/kg plastic material) and entry 72 (1 mg Cd/kg textile extractable).
  • Cr-VI — Chromium VI compounds explicitly speciated in Table 3 and Section 5.2; existing REACH entries 47 (3 mg CrVI/kg leather extractable) and 72 (1 mg CrVI/kg textile extractable); no content limit proposed because no acid-digestion method preserves the +VI oxidation state. The speciation is explicit in the source, so Cr-VI (not Cr) is the correct frontmatter tag.
  • Co — Cobalt and its compounds (EC 231-158-0); no existing REACH content limit; proposed 10 mg Co/kg content + 1.0 mg Co/kg extractable; 3 of 194 measurements > 1,000 mg/kg in coatings (non-EU).
  • tAs — Diarsenic trioxide (EC 215-481-4) is inorganic As(III) by classification, but the report’s analytical methods and concentration limits are anchored to total elemental As content (the atom analysed, not the species) per the explicit statement on p. 28. Per CLAUDE.md Part 14 (“If the paper does not separate inorganic vs total arsenic, the page should say tAs not iAs”), tAs is the correct frontmatter tag because the measurement basis is total As; the underlying substance entry is diarsenic trioxide but the regulatory limit (10 mg As/kg content; 0.2 mg As/kg extractable) is on total As. Verification notes record this disambiguation explicitly for downstream review.
  • tHg — Mercury (Hg, EC 231-106-7) without speciation between MeHg and total Hg; tHg per Part 14.
  • Sn — Organotin compounds (DOT, dioctyl tin; DBT, dibutyl tin); proposed 1 mg DOT or DBT/kg extractable, deviating from existing REACH entry 20’s 1,000 mg/kg mixtures/articles limit. The metals abbreviation vocabulary uses elemental Sn; the wiki-pages-touched list adds [[metals/organotins]] to carry the speciated DOT/DBT context.
  • Li — Lithium hydroxide (EC 215-183-4); default 10 mg Li/kg content.
  • V — Divanadium pentaoxide (EC 215-239-8); default 10 mg V/kg content.
  • Skipped from frontmatter: B (boron; HMI metals taxonomy does not include boron), Br (bromine; not a metal in the HMI panel), Sb (antimony; the report does not have a per-Sb section, only group-level RAPEX surveillance counts), Ni (nickel; not separately addressed in the per-metal sections of this report, distinct from depa2015-cmr-toys which does carry Ni for the same childcare-articles regulatory domain).
• Frontmatter jurisdictions: [EU] — ECHA’s regulatory scope is the European Union (CLP Regulation EU/1272/2008; REACH Regulation EC/1907/2006; Toys Safety Directive 2009/48/EC; RoHS Directive 2011/65/EU; Cosmetic Products Regulation EC/1223/2009; Food Contact Material Regulation EC/1935/2004; Biocidal Products Regulation EU/528/2012; Medical Devices Regulation EU/2017/745). Stakeholder consultation included UK, Canada, US, and Switzerland (Figure 8, p. 61), but these are consultation participants, not regulatory-scope jurisdictions; the report’s binding regulatory frame is EU.
• Brand firewall (Part 12, strict reading locked 2026-05-17). The report does not name specific childcare-article brands or manufacturers in its concentration-range or per-substance findings; substances are reported at the chemical-group × material × product-category level. One stakeholder reference names “Versace RSL 10 mg/kg” as an industry restricted-substances-list source for the Cd content limit consideration (Table 3 narrative, p. 40); this is a regulatory-source reference (RSL = restricted substances list) at the regulatory-source-attribution level rather than a product-contamination ranking, so it falls outside the strict-reading Part 12 violation per the regulatory-event-subject exception. The methods section does not name specific instrument vendors (the report aggregates analytical-method data from CEN, ISO, EPA, OEKO-TEX, RSLs at the method-standard level, not at the vendor level). The SCIP-database baby dish sterilizer reference (p. 22) does not name the manufacturer. ECHA’s ten stakeholder-meeting list (p. 60) includes “a worldwide childcare articles manufacturer” unnamed, demonstrating ECHA’s own brand-firewall posture. Trade associations (EUROPUR, EDANA, AAFA, AFIRM, ENPC, BEUC, Toys Industry Europe), Member State authority names (BfR, BAUA, ANSES, RIVM, KEMI, Danish EPA, Norwegian Environment Agency, Austrian EPA), and CMR-substance research consortia are institutional references for methodology and stakeholder attribution, not brand-firewall violations. Clean brand-firewall posture overall.
• Wiki/HMTc firewall (Part 2). No HMTc threshold proposals; no consumer-audience risk advisories; no cross-source synthesis statements added on the wiki side. The Implications section reports ECHA’s proposed per-metal concentration limits as the regulatory baseline (LOQ-anchored ecolabel-derived 0.1–0.2 mg/kg extractable for Pb, Cd, Co, As, Hg) without proposing HMTc certification thresholds; that decision belongs to the Standards Workbench (per Part 19), not the wiki. The report’s own conclusion — that CMR 1A/1B substances should be banned in childcare articles, with deviation from CLP GCL/SCL because the GCL/SCL framework is not based on substance-specific risk assessment and may not be sufficiently protective for children — is reported descriptively as ECHA’s own conclusion and ECHA’s own regulatory rationale, not as HMI’s recommendation for HMTc thresholds.
• Speciation discipline applied to mercury, arsenic, and chromium. ECHA explicitly anchors its metal concentration limits to total elemental content (the atom or ion analysed, not the specific CMR-classified compound), per the statement on p. 28. The wiki page therefore records: tAs (not iAs) for the diarsenic-trioxide entry because the analytical methods measure total As without speciation; tHg (not MeHg) for the mercury entry because the report does not speciate; Cr-VI (not Cr total) for the chromium VI entry because the regulatory framework and substance entry explicitly speciate the +VI species, with the report itself flagging that no acid-digestion method preserves the +VI oxidation state. This is consistent with the speciation discipline in depa2015-cmr-toys (which records tAs and Cr total for RAPEX-derived data that does not speciate), with the difference that ECHA’s per-metal regulatory framework here explicitly speciates Cr-VI from total Cr at the regulatory-entry level.
• Folder context. The PDF lives under raw/Manual Fetch Kimi /May 21 Kimi_Agent_Download Corruption Issue/_extracted_infantdurable_03_Carriers_HighChairs_CarSeats/03_Carriers_HighChairs_CarSeats/ in the Kimi corruption-issue raw tree (handle MFK_04-unknown). The folder name implies the file is one of a set of papers on baby carriers, high chairs, and car seats; this report’s scope is broader (all childcare articles per the REACH Article 68(2) definition), but car seats are the report’s largest single category by measurement-entry count (17 %) and Figure 5 (p. 47) uses a car-seat schematic as the report’s lead example of accessible-material identification. The sibling page kjolholt2015-car-safety-seats-childrens-textiles (raw_handle MFK_02-unknown, the 02_Unknown.pdf in the same folder) is a Danish-EPA primary-measurement study of car safety seats and other textile childcare products that focuses specifically on the car-seats/baby-slings/baby-mattresses product set; this ECHA 2023 report is the regulatory umbrella for the same childcare-articles category. The two pages are complementary (primary-measurement vs regulatory-context) rather than near-duplicates.
• SCIP-database baby-dish-sterilizer Pb entry (Section 3.2, p. 22) — data-integrity caveat. The report states that “very high concentrations of eight different lead compounds were reported for a baby dish sterilizer in the SCIP database. However, it is not reported in which part of the sterilizer the lead substances are included and if the substances were inaccessible or in contact with the dishes to be later used by children.” No numerical Pb concentration is reported; ECHA flags this entry as data-integrity-limited (uncertainty 2 in Table 5: representativeness of findings). The HMI page does not pull through the Pb-concentration value for this SCIP entry because the source does not provide one; the qualitative narrative is preserved.
• DEHP and TCEP non-metal exceedance data. Recorded in Key numbers Section 3.2 table for completeness because they appear in the same source table as the Pb (textiles/synthetic polymers) and Co (coatings) exceedances; these are non-metal substances outside the HMI metal panel but contribute regulatory context (REACH entry 51 progressive phthalate restrictions; the TCEP screening report ECHA 2018b). No downstream HMI metal-page edits triggered by these non-metal entries.
• Raw integrity. raw_sha256 = ea3157029e63e60bb834d07e5531bf949be17b0404f555e2e7c98b8679403b08 verified by shasum -a 256 against the file at raw_path on 2026-06-01.
• Near-duplicates. None identified in the corpus. Sibling regulatory-context sources include kjolholt2015-car-safety-seats-childrens-textiles (Danish-EPA primary measurement of car-safety-seat textiles, same MFK folder), depa2015-cmr-toys (Danish-EPA market surveillance of CMR substances in toys, sibling MFK folder), sccs2023-notes-of-guidance-cosmetic-ingredients-v12 (SCCS notes of guidance for cosmetic-ingredient testing — different scope), bfr2009-lead-cadmium-toys (BfR opinion on Pb/Cd in toys), and fowles2021-lead-toys (academic review of Pb in toys). None duplicates the ECHA 2023 report’s substance-inventory aggregation (1,188 + 108 + 27 + 12 substances) or its per-metal proposed concentration-limit table; this is the unique ECHA-side regulatory-framework anchor for childcare articles.
• 2026-06-01 audit-application notes (Claude Opus 4.7 acting on fresh-context audit subagent verdict REVISE). Audit subagent reported two ❌ definite-error findings on Check 1 numerical fidelity plus four ⚠️ citation-precision concerns; Checks 2, 3, 4, 5 all clean. Independent verification against the source PDF confirmed all six findings.
  • ❌ #1 (Co/Pb most-reported metal class inversion). VERIFIED: p. 21 verbatim states “39 % of the entries cover metals, being cobalt and lead compounds the most reported (373 and 164 entries, respectively),” and p. 17 verbatim states “the CMR 1A or 1B substances most frequently detected in childcare articles are metals, being cobalt and lead the most reported.” Initial draft had Pb described as most-reported (164 entries) and Co as second; this inverted the source. Correction applied in five locations (tier_rationale; body opening narrative; Key numbers Distribution by chemical group bullet; Methods Concentration-range section; Implications first bullet) — Co is now consistently named as the single most-reported metal class at 373 entries and Pb as second at 164 entries, with p. 21 cited as the source.
  • ❌ #2 (Figure 1 denominator conflation). VERIFIED: p. 16 verbatim text immediately under Figure 1 states “The information collected in this report for CMR 1A or 1B substances (2 340 entries) that may be present in childcare articles is categorised in two ways: substances measured and/or substances suspected in childcare articles.” Figure 1 therefore represents the distribution of 2,340 information-collected entries across chemical groups, not a distribution of the 180–181 substances actually identified as present. Initial draft attributed Figure 1’s 29 % metals share to “substance shares of the 180–181 identified” — this conflated two distinct denominators. Correction applied in Key numbers Distribution by chemical group (Figure 1) bullet — the bullet now anchors the Figure 1 distribution to the 2,340 information-collected entries with the p. 16 verbatim quotation, and clarifies that the remaining chemical groups make up the residual share of 35 total groups (per the p. 17 statement that “the substances can be divided in 35 different chemical groups”).
  • ⚠️ #3 (180/181 page-number swap). VERIFIED: p. 28 contains “For 149 of the 180 substances identified…” (180 here); p. 30 contains “ECHA collected and analysed existing concentration limits and related analytical methods for all 181 substances identified…” (181 here). Initial draft transposed these as “181 in p. 29, 180 in p. 30.” Correction applied in Key numbers substances-identified bullet — now correctly cites 180 on p. 28 and 181 on p. 30 with the verbatim source quotations.
  • ⚠️ #4 (Table 3 attribution for Li, V, Br/ammonium-bromide rows). VERIFIED: Table 3 (pp. 34–35) is captioned “Substances or groups of substances with a considered concentration limit in homogeneous material different from 10 mg/kg (content) and/or with considered extractable-related concentration limits.” Li (10 mg Li/kg default content, no extractable), V (10 mg V/kg default content, no extractable), and ammonium bromide / Br⁻ (10 mg Br⁻/kg default content) all take the default and are therefore not in Table 3; their rows are sourced from the Section 5.2 per-metal narrative on pp. 38 (Br) and 42 (Li, V). Initial draft attributed the entire metals-limits table to “Table 3, p. 34; Section 5.2 narrative, pp. 40–42” without distinguishing which rows come from Table 3 vs the narrative. Correction applied in Key numbers metals-limits-table heading — heading now reads “Table 3, pp. 34–35; Section 5.2 per-metal narrative, pp. 40–42” and an introductory paragraph explicitly flags that Li, V, and ammonium bromide rows are not in Table 3 but come from the per-metal narrative pages 38 and 42.
  • ⚠️ #5 (612 → ~608 entries arithmetic). VERIFIED: 39 % of 1,559 = 607.5, which rounds to 608, not 612. Correction applied in tier_rationale and in Key numbers Distribution by chemical group (measurement entries) bullet — ”≈ 612 entries” replaced by ”≈ 608 entries” in both locations; per p. 21 the explicit breakdown is Co 373 + Pb 164 = 537 entries for the two most-reported metal classes, with the remaining ≈ 71 entries split across all other CMR-classified metals (Cr-VI compounds, organotins, mercury, etc.).
  • ⚠️ #6 (Table 3 page-citation precision). Note that Table 3 spans pp. 34–35 rather than just p. 34; the Lead, Mercury, Nitrosamines, Organic dyes, Organotins, PAHs, Pesticides, PFAS, Phthalates, and Quinolines rows are on p. 35. Correction applied in the metals-limits-table heading — heading now cites “Table 3, pp. 34–35” rather than “Table 3, p. 34.”
  • No false-positive subagent findings. All six findings applied. Audit-application commit follows; routing-audit re-run remained clean (16 product-source routing rows unchanged; no new routing_unresolved or routing_malformed entries).

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.