Ricketts et al. 2020 — Mercury in Jamaican skin-lightening products

This study quantifies total mercury in 60 skin-lightening products purchased from 12 vendors across Jamaica (February–May 2017) and surveys 384 users about application patterns. The analytical strategy is two-stage: portable handheld energy-dispersive X-ray fluorescence (XRF) on all 60 products for primary screening, with cold-vapor atomic absorption spectroscopy (CVAAS) on a 25-sample subset whose XRF readings fell below the XRF detection limit. Six of 60 products exceeded the US FDA 1 ppm limit for mercury in cosmetics; three products carried “alarmingly high” concentrations above 400 ppm. The XRF range across the full panel spanned 0.05 ppm to 17,547 ppm — five orders of magnitude. The CVAAS sub-panel (excluded by design from the higher-concentration products) ranged 0.05 ppm to 3.68 ppm. Cream formulations carried the highest mean mercury content in the CVAAS sub-panel; soap, lotion, and “other” formulations were lower. The survey component documents that 51% of female users and 49% of male users apply skin-lightening products more than once per day, and that combining multiple products in homemade mixtures is common.

Key numbers

Full panel of 60 skin-lightening products purchased in Jamaica, Feb–May 2017 — XRF screening (Niton XL3T GOLDD+ handheld; detection limit ≈10 ppm Hg in the cream/soap matrix as calibrated):

CVAAS sub-panel of 25 products (Buck Scientific 400A; detection limit 0.03 µg/g = 30 ppb Hg). This sub-panel was drawn from the 57 products whose XRF screen was below the XRF detection limit; the three highest-XRF products are not represented here.

CVAAS quartile breakdown (source Table 3):

CVAAS sub-panel split by product formulation (source Table 4):

Country of manufacture of skin-lightening products encountered on the Jamaican market (source Table 2; reported n = 109, but the eight tabulated counts sum to 100 — see Verification notes):

The authors note that some products originating from European nations and the Ivory Coast had mercury concentrations greater than 1 ppm.

User survey (n = 384 self-identified users of skin-lightening products in Jamaica):

• Sex: 196 female (51%); 188 male (49%)
• Age: 280 <30 yr; 92 31–50 yr; 12 >50 yr
• Frequency of application: 197 more than once per day; 130 once per day; 57 once per week
• Socioeconomic: 192 employed; 84 unemployed; 108 student
• Mixing of products: 221 of 384 (58%) reported combining different skin-lightening products in homemade mixtures
• Reported health effects: itchiness 55; irritability 52; “other” (headaches, scars, depression) 31 — total 138 reports of side effects, of which 79 were among individuals using a product with Hg above 1 ppm. Of 30 individuals reporting headaches, scars, or depression, 93% used a product containing more than 1 ppm mercury.

Cross-country context cited by the authors (source Table 5; not measured in this study — reproduced as the authors’ literature comparator):

Regulatory anchors as cited within the source:

• US FDA: <1 ppm mercury in skin-lightening cosmetics (the binding action level the authors compare against)
• European Union and many African nations: outright bans on mercury in cosmetics
• Minamata Convention Articles 4, 16, 17, 18 and Annex A of Article 3: cited by the authors as the international framework this study supports

Contextual case report cited by the authors (Al-Saleh 2016, ref. 9 of the source; not measured in this study): A Belgian woman who used a 1%-mercury skin-lightening soap during pregnancy and lactation recorded blood Hg 91 µg/L and urine Hg 784 µg/g; her three-month-old infant recorded blood Hg 19 µg/L and urine Hg 274 µg/g. The authors cite this case as a motivating exposure pathway for the present study; the present study does not measure biomarkers in human subjects.

Methods

Sampling. Sixty skin-lightening products were purchased from twelve vendor locations across Jamaica between February and May 2017. The vendor survey approach paralleled the Zero Mercury Working Group brand-survey method (ref. 10 of the source). Sample selection prioritized the most popular products on the Jamaican retail market as identified by the user survey conducted in parallel. Product name, physical description, and label ingredient list were recorded for each. Product-form categories used downstream: creams, gels, lotions, and bar soaps.

XRF screening. All 60 products were screened on a Niton XL3T GOLDD+ portable handheld energy-dispersive X-ray fluorescence analyzer, using the instrument’s built-in “plastics mode” calibration. Approximately 10 mL of cream, gel, or lotion was placed in a sample cup and covered with Mylar film; bar soaps were cut to 5 g aliquots. Each sample was measured for one minute in triplicate. The analyzer’s stock manufactured calibrations do not include mercury in cream or soap matrices, so the team manually constructed a calibration curve by spiking a representative lotion matrix with mercury chloride standards at 20, 40, and 100 ppm Hg and measuring recovery (R² = 0.99 across the three spike points). Mean mercury recovery in the spiked-lotion matrix was approximately 65%. Using the manual calibration curve, the analyzer’s mercury detection limit in the cream/soap matrix was estimated at approximately 10 ppm. Accuracy was further verified against a multi-element reference material (Niton PN 180-554, batch SN PE-079-N, ThermoFisher Scientific): certified value 1002 ppm; measured value 969 ± 4.5 ppm.

CVAAS confirmatory analysis. Twenty-five of the 60 products whose XRF screens fell below the 10 ppm XRF detection limit were re-analyzed by cold-vapor atomic absorption spectroscopy on a Buck Scientific 400A mercury analyzer. Sample digestion used approximately 0.25 g of product treated with sulfuric acid and 1:1 perchloric and nitric acids; the digestate was diluted to 100 mL. A 50 mL aliquot was reduced with 5 mL of stannous chloride in a biological-oxygen-demand bottle attached to the aeration apparatus; maximum mercury absorbance was recorded at 2 minutes. Each batch included a blank, a standard, and spike samples. Working standards were prepared from a 1000 mg/L ascorbic-acid stock solution at two ranges: low (0.05–0.5 µg Hg/mL) and high (1.0–5.0 µg Hg/mL). CVAAS detection limit: 0.03 µg/g (≈30 ppb Hg).

Speciation. Both XRF and CVAAS measure total mercury. The study does not distinguish elemental, inorganic (Hg²⁺ salts), or organic (methylmercury) species. The authors’ introduction notes that inorganic mercury is the typical added ingredient in skin-lightening products (it inhibits melanin formation) but that methylmercury has also been reported. The mercury concentrations reported throughout this study are total Hg only.

User survey. A convenience sample of 384 self-identified users of skin-lightening products was recruited from the same Jamaican parishes covered by the vendor survey. Participants completed a self-administered questionnaire on product usage patterns, demographics, and self-reported health effects, with informed verbal consent. The instrument is reproduced in the source’s Supplemental Material 1.

Statistics. Data analysis used Microsoft Excel 2013 and SPSS version 17.

Implications

Certification. The paper contributes occurrence data for the skin-lightening cosmetic product class on the Jamaican market in 2017. Within the 60-product XRF panel, 6 of 60 products (10%) exceeded the US FDA 1 ppm cosmetic limit, and 3 of 60 products (5%) exceeded 400 ppm, with a single-sample maximum of 17,547 ppm (body text) — 17,345 ppm (Table 5 attribution) measured by XRF. Within the 25-product CVAAS sub-panel (which excludes the high-XRF tail by construction), 3 of 25 products (12%) exceeded the FDA 1 ppm cosmetic limit; the cream formulation carried the highest mean Hg concentration (0.67 ± 1.04 ppm, n=14) within the CVAAS sub-panel, with soap (0.18 ± 0.11, n=7), lotion (0.11 ± 0.01, n=2), and “other” (0.05 ± 0.01, n=2) lower. The full-panel XRF range spans five orders of magnitude (0.05–17,547 ppm), so a panel-level central-tendency statistic understates the upper-tail public-health signal that the three above-400 ppm products represent. (Routing of this source to the skin-lightening-cream slug and from there to the skin-lightening-products public advisory rather than to a Cat 2 certification row reflects the HMTc Cat 2 Step 0 lock decision recorded in wiki/products/skin-lightening-cream.md and is documented in the Verification notes below; that routing is governance state, not a claim of this paper.)

Courses. The two-stage analytical design (handheld XRF screening followed by CVAAS confirmation on the sub-detection-limit fraction) is a useful field-surveillance template for situations where the contamination distribution spans many orders of magnitude. The paper’s recovery experiment with HgCl₂ spikes of a lotion matrix at three concentrations (20, 40, 100 ppm) — yielding R² = 0.99 across the three points but only ~65% mean recovery — illustrates a common XRF cosmetic-matrix limitation that downstream courses can use as a worked example of calibration-versus-recovery distinctions.

App. Per CLAUDE.md Part 12, the consumer app does not surface brand-level rankings. Skin-lightening products as a category should be flagged for mercury concern; the heterogeneity documented here (95% of products below 10 ppm but a 5% tail above 400 ppm, with the maximum at 17,547 ppm) means a category-level flag is the only defensible app behavior in the absence of per-product testing data.

Microbiome. Not addressed by this paper.

Wiki pages this source may touch

• mercury-total
• mercury
• skin-lightening-cream (retired; routes to skin-lightening-products)
• children-personal-care

Verification notes

• 2026-05-18 merge-enhance ingest (Claude Opus 4.7, autonomous v2.0 manual-fetch skill). Pre-existing page at this cite-key dated updated: 2026-05-14. Three identity checks against wiki/sources/ confirmed a single in-place match on raw_handle kimi-children-personal-care-ricketts2020 and cite-key ricketts2020-mercury-skin-lightening-jamaica; DOI grep returned no matches (source has no assigned DOI per no_doi_assigned: true). Merge-enhance path applied per v2.0 SKILL: cite_key, raw_handle, raw_path, license, access_url, near_duplicates, and no_doi_assigned flag preserved from the prior revision; body content reconstructed against the PDF to clear schema and firewall defects in the prior revision (see below).
• Defects in the prior 2026-05-14 revision, with corrections applied. (a) Frontmatter metals: [tHg, MeHg] corrected to metals: [tHg] — both XRF and CVAAS measure total mercury only; the source’s introduction notes that some skin-lightening products contain methylmercury, but no MeHg measurement was performed in this study, so MeHg should not be in the source’s metal-slug list. (b) “Most-used products” listing of branded products by name (Idole Lotion, Bio Claire Cream, etc.) deleted under the Part 12 strict reading locked 2026-05-17 — the survey-popularity ranking is brand attribution. Replaced with category-level summary of product-form mix. (c) Spike-recovery experiment matrix-identifier brand name (“La Bamakoise Tamarin Lait Extra Tonique (lotion)”) replaced with the product-form descriptor “a representative lotion matrix” in the Methods section; the brand name does not contribute method-reproducibility information beyond “any commercial cream-base lotion.” (d) Speculative slugs in the prior revision’s “Wiki pages updated on ingest” section ([[products/skin-lightening-lotion]], [[products/skin-lightening-soap]], [[regulations/fda-cosmetic-mercury-1ppm]]) removed — none of those pages exist; per CLAUDE.md Part 10, source pages do not propose new product or regulation slugs in the body, and per the Cat 2 Step 0 lock the skin-lightening product class is consolidated under the single retired [[products/skin-lightening-cream]] routing destination regardless of base form (cream/lotion/soap/oil/pill). (e) Implications section rewritten to remove threshold-proposing language (“HMTc Cat 2 skin-lightening thresholds at the FDA 1 ppm level (or tighter) catch 10% of…”, “A precautionary threshold below 0.5 ppm catches…”), which violated Part 2 wiki/HMTc firewall by proposing specific certification values from a single source. The corrected Implications section frames the paper’s contribution to the public advisory rather than to Cat 2 certification row math. (f) Legacy heading ## Wiki pages updated on ingest replaced with ## Wiki pages this source may touch per current v2.0 schema. (g) Source Tables 3 (CVAAS quartiles) and 4 (CVAAS by formulation) added — the prior revision summarized only the XRF range and did not transcribe these tables, but they are the source’s only quantitative within-panel breakdown of the 25-sample CVAAS sub-panel and are the relevant inputs for any downstream synthesis. (h) Source Table 2 (country of manufacture) and Table 5 (cross-country range comparator) added for the same reason. (i) Health-effects survey numbers added.
• Brand-firewall compliance (Part 12 strict, 2026-05-17 lock). This source names brands at three layers: (i) the user-survey ranking of “most-used products” (Figure 3 of the source); (ii) the CVAAS quartile assignment of specific named products (Table 3); (iii) the spike-recovery matrix identifier in the Methods section. All three are stripped from this page under the strict reading — survey popularity, contamination-value rankings, and method-matrix identification by brand are all brand attribution to either consumption or contamination, none of which Part 12 permits in the public wiki. The product-form categories (cream, lotion, soap, gel, oil, pill) and the formulation-type breakdown of Hg means (source Table 4) preserve the analytical information without brand identification. The Niton XRF analyzer, Buck Scientific CVAAS analyzer, Niton PN 180-554 reference material, ThermoFisher Scientific source for the reference material, Mylar film, and SPSS / Excel statistical software are method/instrument/material identifications retained per Exception 2 of the audit-prompt brand-firewall rule (scientific reproducibility depends on knowing the instrument). The Zero Mercury Working Group brand-survey method citation (source ref. 10) is also retained as a method reference.
• Wiki/HMTc firewall (Part 2). The Implications section avoids proposing specific HMTc threshold values; the source’s documented Hg distribution feeds the public advisory at skin-lightening-products rather than a Cat 2 certification row, since skin-lightening is explicitly out of HMTc scope.
• Mercury speciation. The source explicitly notes (p. 1, p. 2, p. 6) that elemental, inorganic, and organic mercury are all possible additives in skin-lightening products, but the present study’s analytical methods (XRF and CVAAS) measure total mercury only. Frontmatter metals: [tHg] accordingly. The page does not infer the inorganic-versus-methylmercury split from any of the reported total-Hg values.
• Table 2 (country of manufacture) sample-size inconsistency. The Table 2 caption in the source reads “(n=109)” but the eight tabulated country counts sum to 100 (Ivory Coast 19 + European Union 19 + Lebanon 15 + United States 15 + England 10 + Jamaica 8 + India 8 + Other 6 = 100). The text on p. 5 states “about 53% of skin lightening products purchased in Jamaica were imported from the Ivory Coast, European Union and Lebanon” — (19+19+15)/100 = 53% confirms n=100 in the body of the table. The “(n=109)” caption appears to be a typographical error in the source. Counts transcribed as printed; the n=100 denominator is the working figure for any downstream calculation.
• Table 5 (cross-country range) US row maximum. The source’s Table 5 prints the US range as “1729.00–45622.00 ppm” with reference 3 (Hamann et al. 2014); the source’s own discussion text on p. 6 reads the maximum as “45,622 ± 322 ppm” for products from Thailand and reports the US maximum elsewhere. The Table 5 row attribution to the United States (1,729–45,622 ppm) is transcribed as printed in Table 5; cross-checking against the original Hamann reference is left for the downstream synthesis pass (if any), as this is the authors’ literature comparator, not the present study’s own measurement.
• Source’s internal inconsistency on the Jamaica panel maximum. Three different values appear in the source for the highest Hg concentration measured in the present study’s 60-product Jamaican panel: (i) abstract (p. 1) and analytical-data narrative (p. 5) read “0.05 ppm to 17,547 ppm”; (ii) Discussion (p. 6) repeats “the highest observed value of 17,547 ppm”; (iii) Table 5 cross-country comparator (p. 7) prints the Jamaica row as “0.05-17345.00”; (iv) Conclusions (p. 8) state “popular products having mercury concentrations up to 17,000 ppm.” The body and Table 5 differ by 202 ppm; the rounded Conclusion figure is the most-conservative restatement. This wiki page transcribes Table 5 as printed (17,345.00 in the Table 5 reproduction) and the body-text figure (17,547 ppm) in the narrative summary and overview table; both are reproduced rather than reconciled because cross-checking the underlying XRF reading is outside the scope of source-page ingest.
• CVAAS sub-panel selection bias (analytical method note). By design, the 25 CVAAS samples were drawn from the 57 products whose XRF screens fell below the XRF detection limit of approximately 10 ppm. The three highest-XRF products (>400 ppm) are not in the CVAAS sub-panel. Any downstream pooling of “CVAAS mean by formulation type” therefore underestimates the central-tendency Hg of the full Jamaican-market panel and should not be used as a panel-mean estimator. The full-panel descriptive statistic is the XRF range (0.05–17,547 ppm) with the three above-400 ppm products as the dominant upper-tail signal.
• XRF recovery caveat. The handheld XRF analyzer’s manual calibration in this study used HgCl₂ spikes of a single lotion matrix; mean recovery was ~65%, with measured concentrations consistently lower than reference standard values, and recovery higher at higher concentrations. The reported XRF values are accordingly conservative (likely under-estimating true Hg by a factor of ~1.5 at low concentrations); the 17,547 ppm maximum should be read as a lower-bound estimate of the true Hg loading in that product.
• Authors verified against PDF byline. Phylicia Ricketts¹, Christopher Knight², Andre Gordon¹, Ana Boischio³, Mitko Voutchkov¹. Affiliations: ¹Department of Physics, Radioecological Lab, University of the West Indies, Mona Jamaica; ²Analytical Services Department, Mines and Geology Division, Hope Gardens, Kingston, Jamaica; ³Pan American Health Organization, Washington DC. Corresponding author: Phylicia Ricketts (phylicia.ricketts02@uwimona.edu.jm).
• License. Page 9 of the PDF states: “This is an Open Access article distributed in accordance with Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0/).” License field updated to CC BY 3.0 (prior revision had CC BY-NC which is inconsistent with CC BY/3.0).
• DOI. The PDF does not print a DOI on its face; the journal-issue identifier on p. 1 reads “J Health Pollution 26: (200601) 2020 © Pure Earth.” The no_doi_assigned: true flag is retained from the prior revision. Reference 11 of the source (Murphy et al. 2015, also J Health Pollut) carries DOI 10.5696/2156-9614-5-9.33, suggesting an analogous DOI pattern may exist for this article (e.g., 10.5696/2156-9614-10-26.200601); cross-vendor DOI verification is left for a future pass.
• Audit subagent (2026-05-18, general-purpose fresh-context, v2.0 skill Phase 2) — verdict REVISE. Two findings: (1) Check 1 ❌ on the Table 5 Jamaica-row maximum — wiki transcribed “17,545.00 ppm” but the PDF Table 5 prints “17,345.00”; verified against PDF p. 7, finding is correct, corrected in this revision (17,545.00 → 17,345.00) and the source’s internal 17,547 / 17,345 / “up to 17,000” inconsistency across body, table, and Conclusions is documented in the immediately-preceding verification note. (2) Check 5 ⚠️ on the Implications → Certification paragraph asserting the HMTc Cat 2 Step 0 lock as in-body governance state — verified against the audit-prompt CHECK 5 rule that source pages may “describe what the paper’s findings contribute to threshold work” but may not assert HMTc-program scope decisions in body text. Finding accepted; the Implications → Certification paragraph rewritten to literature-occurrence framing (panel exceedance proportions, formulation-type means, full-panel range and skew), with the Cat 2 Step 0 lock routing context relocated to a parenthetical reference back to this Verification notes section and to the products/skin-lightening-cream page itself. The audit also flagged ⚠️ in Check 2 on the [[advisory/skin-lightening-products]] slug as not appearing in the taxonomy snapshot; verified by listing wiki/advisory/ — the page exists at wiki/advisory/skin-lightening-products.md (the taxonomy snapshot covers the four indexed namespaces ingredients/products/metals/regulations, not the advisory namespace). Finding rejected as a false positive — the slug resolves; the taxonomy snapshot is the wrong reference for advisory pages. Checks 3 and 4 ✅ with no findings.

Ingest log

• 2019-09-03 (received) / 2019-12-16 (accepted) / 2020-06 (published): Journal of Health & Pollution 10(26): 200601. © Pure Earth. CC BY 3.0 (open access).
• 2026-05-14 prior ingest (claude_opus_47_session): manual-fetch Phase 1 ingest; raw_handle kimi-children-personal-care-ricketts2020 set; products ["[[products/skin-lightening-cream]]"] declared; metals incorrectly set to [tHg, MeHg]; body section labels included the legacy ## Wiki pages updated on ingest heading with speculative non-existent slugs; Most-used-products and CVAAS-quartile sections named brands.
• 2026-05-18 merge-enhance ingest (Claude Opus 4.7, autonomous v2.0 manual-fetch skill): EXISTING path per identity checks. Full PDF re-read (pp. 1–10, including Tables 1–5 and Figures 1–4). Defects in prior revision addressed: metals corrected to [tHg]; brand names stripped from Methods spike-matrix identifier, Most-used products section, and CVAAS-quartile listings; Tables 3, 4, 5, and 2 added with category-level transcriptions; speculative [[products/skin-lightening-lotion]], [[products/skin-lightening-soap]], and [[regulations/fda-cosmetic-mercury-1ppm]] slugs removed; legacy ## Wiki pages updated on ingest heading replaced with current ## Wiki pages this source may touch; Implications section rewritten to remove HMTc threshold-proposing language; license corrected from CC BY-NC to CC BY 3.0 per the PDF’s own copyright page. Routing audit refreshed; ingest committed at b155701; queued for audit.
• 2026-05-18 Phase 2 audit (fresh-context Agent subagent, general-purpose): Verdict REVISE. Two findings applied: (i) Table 5 Jamaica-row maximum corrected from 17,545.00 → 17,345.00 to match PDF Table 5 (Check 1 ❌; the prior 17,545 transcription matched neither Table 5 nor body text); (ii) Implications → Certification paragraph rewritten to remove in-body assertion of the HMTc Cat 2 Step 0 lock scope decision (Check 5 ⚠️), with the governance-routing context relocated to Verification notes and a parenthetical pointer to the products/skin-lightening-cream retirement page. One finding rejected as false positive: Check 2 ⚠️ on the advisory/skin-lightening-products slug (the page exists at wiki/advisory/skin-lightening-products.md; the taxonomy snapshot covers the four indexed namespaces, not the advisory namespace). Source’s internal 17,547/17,345/17,000 inconsistency across body/Table 5/Conclusions documented in Verification notes.

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.