Angelon-Gaetz et al. 2018 — Lead in spices, herbal remedies, and ceremonial powders, North Carolina

This MMWR surveillance report from the North Carolina Childhood Lead Poisoning Prevention Program (NCCLPPP) describes seven years of environmental lead investigations in the homes of 61 children with confirmed elevated blood lead levels (BLLs), in which spices, herbal remedies, and ceremonial powders were sampled. Of 386 samples collected from 59 properties (6.0% of the 983 home lead investigations conducted in NC during January 2011–January 2018), 111 samples were contaminated with ≥1 mg/kg Pb — 28.8% of the full 386-sample set and 62.7% of the 177 samples that fell within the 50 product categories whose mean Pb was ≥1 mg/kg. Mean lead levels of ≥1 mg/kg were identified in 50 product categories — 40 food categories and 10 nonfood categories. Among edible items, turmeric had the highest average lead concentration (66.4 mg/kg, range 0.1–740 mg/kg, n=34); among nonfood items, traditional cosmetic and ceremonial powders (kumkum, sindoor, surma) reached percent-level lead concentrations, with sindoor averaging 41,401 mg/kg (range 0.1–130,000 mg/kg, n=8). The affected children were predominantly of Asian descent (67.2%), and 71% of the investigated properties were built after 1978, ruling out lead paint as the primary exposure source in most homes. The authors directly called for a national maximum allowable limit for lead in spices and herbal remedies, which did not exist at the time of publication.

Key numbers

Study scope:

• 983 home lead investigations conducted in North Carolina, January 2011–January 2018
• 59 properties (6.0% of 983) in 8 primarily urban counties had spice/herbal/ceremonial-powder sampling
• 61 children with confirmed elevated BLLs lived at these 59 properties
• 42 of 59 properties (71%) were built after 1978; in 32 of those 42 (76%) no lead was found in paint, dust, mini-blinds, faucets, bathtub glaze, or furniture finish; in 7 of those 32, spices/herbal remedies/ceremonial powders were the only identified risk
• 386 samples included in analysis (six excluded for different sampling/analysis methods): 344 food (89%), 42 nonfood (11%)
• 111/386 samples (28.8% of the full set; 62.7% of the 177 samples falling within the 50 mean-≥1 mg/kg categories) contained ≥1 mg/kg Pb; among nonfood, 35/42 (83.3%); among food, 76/344 (22.0%)

Blood lead levels in the 61 affected children:

• Average screening (initial) BLL: 17.0 µg/dL (SD 9.6)
• Average diagnostic (confirmatory) BLL: 15.2 µg/dL (SD 7.0)
• Average child age at investigation: 2.3 years (range 0.9–6.6 years)
• Diagnostic BLLs drawn February 28, 2011 – December 5, 2017

Lab method: Atomic absorption mass spectrometer screen at the NC State Laboratory of Public Health; samples <15 mg/kg Pb subsequently re-analyzed by ICP-MS (starting 2011). Below-LOD environmental sample values replaced with LOD/√2 (Croghan & Egeghy, 2003); below-LOD blood lead values standardized to 1 µg/dL.

Nonfood items — categories with average Pb ≥1 mg/kg (N=10 categories, 37 samples; Table):

Food items — categories with average Pb ≥1 mg/kg (N=40 categories, 140 samples; Table):

Spices and condiments (N=26 categories):

Medications, oils, and supplements:

Prepared foods and other food products:

Provenance metadata: Country of purchase was recorded for only 187/386 (48%) samples; of those with known origin, 142/187 (76%) were purchased in the United States. The remaining 199 samples have unknown provenance.

Demographics of the 61 affected children (race/ethnicity available for 58/61, 95%):

• Asian (including Indian and Pakistani descent): 41 children (67.2%)
• Black or African American (including two siblings born in West Africa): 9 children (13%)
• White (including one from Afghanistan): 8 children (≈14%)
• Hispanic ethnicity known for 51 children (84%); 7 of those (11.5%) Hispanic

Regulatory context cited by the report (at time of publication, November 2018):

• No US permissible limit for Pb in spices
• WHO Codex Standard 193–1995 permissible limit: 0.02 mg/kg Pb for infant formula; 2 mg/kg Pb for salt
• FDA limit for Pb in natural-source food color additives (e.g., paprika, saffron, turmeric): 10 mg/kg
• FDA action levels for products intended for consumption by children: 0.1 mg/kg for candy; 0.5 mg/kg for other foods; spices are not classified as food intended for children
• US spice imports increased ~50% since 1998 (>95% of US-consumed spices are imported)

Consumption-frequency context cited by the report:

• EPA What We Eat in America survey: 0.09 g/day cumin, 0.03 g/day turmeric (general US population)
• South Indian cooking pattern (Siruguri & Bhat 2015): 1.22 ± 1.14 g cumin per portion, 0.60 ± 0.46 g turmeric per portion, in dishes prepared daily

Methods (brief)

Retrospective cross-sectional case series. Lead investigators from state and local health departments conducted free home lead investigations for children with confirmed elevated BLLs (defined since July 1, 2017 as two consecutive results ≥5 µg/dL within 12 months; previously two consecutive ≥10 µg/dL within 6 months). Investigators collected information on home build year and documented evidence of lead in paint, water, soil, consumer products, and foods. Samples of lead paint, water, soil, consumer products, and foods (including spices, herbal remedies, and ceremonial powders) were submitted to the NC State Laboratory of Public Health for chemical analysis.

All spice, herbal-remedy, and ceremonial-powder samples were screened for lead by atomic absorption mass spectrometer at the NC State Laboratory of Public Health. Starting in 2011, samples with <15 mg/kg Pb on the AAS screen were subsequently re-analyzed by ICP-MS. Environmental-sample results below the limit of detection (LOD) were replaced with LOD/√2 (Croghan & Egeghy, 2003); blood-lead results below LOD were standardized to 1 µg/dL.

Investigation reports and environmental sample data were entered into the NC childhood lead surveillance system and linked to children’s blood lead test results. Because consumption-frequency information was not collected for most children, descriptive statistics were computed for environmental-sample results separately from blood-lead data; direct toxicologic modeling was not attempted.

This is a surveillance and case-series design, not a market-basket study: samples were collected from homes where children had already been identified with elevated BLLs, so reported concentrations represent a clinically-ascertained sample rather than a population-representative spice market.

Implications

Certification: US-government surveillance documentation of a quantified spice-lead pathway in real-world child health investigations: 28.8% of 386 spice / herbal / ceremonial-powder samples from homes of children with confirmed elevated BLLs were ≥1 mg/kg Pb, with turmeric reaching 740 mg/kg in a single sample and an average of 66.4 mg/kg across 34 samples. Within the food categories captured in this clinically-ascertained sample, chili powder/red pepper (n=23), coriander (n=9), cumin (n=17), and turmeric (n=34) are the categories with the largest within-source N and therefore the most weight-bearing food-category Pb distributions in this paper. The kumkum, sindoor, and surma percent-level Pb concentrations are nonfood ceremonial cosmetics, not directly applicable to food-row certification work, but they are the key documented pathway by which lead enters households as a result of accidental ingestion by children and contamination of household surfaces. The source itself argues that “setting a national maximum allowable limit for lead in spices and herbal remedies might further reduce the risk for lead exposure from these substances”; no numerical HMT&C threshold is proposed here.

Courses: Strong anchor case for the regulatory vacuum around spice lead contamination in 2018 (no US national limit; FDA action levels apply to candy and “other foods” but spices are not classified as food intended for children). Documents the cultural and demographic concentration of risk: 67.2% of affected children were of Asian descent, and many lived in post-1978 housing where lead paint was ruled out by investigation. Documents the failure mode of standard childhood lead risk assessment when the exposure pathway is dietary rather than environmental. The NCCLPPP survey tool developed for systematic spice/herbal-remedy exposure documentation (available in English and Spanish, piloted in one NC county, cultural-sensitivity tested in focus groups with Hispanic and South Asian community members) is a documented example of how to capture exposure data that standard environmental investigations miss.

App: Clinically-ascertained spice Pb distributions for turmeric (n=34, mean 66.4 mg/kg, range 0.1–740), chili powder/red pepper (n=23, mean 12.6 mg/kg, range 0.1–170), cumin (n=17, mean 1.1 mg/kg, range 0.1–6.4), coriander (n=9, mean 4.8 mg/kg, range 0.1–39), and cinnamon (n=2, mean 2.6 mg/kg, range 2.5–2.7). These are upper-tail distributions from homes of lead-exposed children, not population-representative market data; route into ingredient pages as exposure-tail context rather than as central tendency. Per-product-category contamination profiles for app surfacing should use this paper as evidence that high-Pb tails exist in turmeric and chili-powder/red-pepper categories at concentrations far above the FDA 10 mg/kg color-additive limit.

Microbiome: not addressed.

Limitations stated by the authors

The report explicitly notes five limitations: (1) spices are frequently purchased wholesale and decanted, so origin and lot-number metadata are often discarded; (2) many lead investigators did not collect spice/herbal-remedy samples — 11 reports were excluded from the numerator but not the denominator, so the prevalence of spice-related cases may be underestimated; (3) prior to recent policy changes, persons with BLLs 5–9 µg/dL were typically offered education only (no home investigation) unless local ordinance lowered the trigger, which may further underestimate cases; (4) even individually low Pb concentrations may produce elevated BLLs under chronic, combined exposure, but direct toxicologic modeling was not feasible given missing consumption data; (5) the small sample and broad age range (infants through age 6) make modeling exposure effects difficult because Pb metabolism and developmental sensitivity differ markedly between a 1-year-old and a 5-year-old.

Wiki pages this source may touch

• herbs-and-spices
• turmeric
• cinnamon
• spices
• lead

Verification notes

• 2026-05-29 (Claude Opus 4.7, ingest-next-manual-fetch-pdf v2.0 — Phase 1): EXISTING-path merge-enhance. Pre-existing page (last updated: 2026-05-14) carried the legacy raw_handle: manual-fetch-kimi, a truncated raw_path (“Lead in Spices, Herbal Remedies, and Ceremonial Powders Sampled f.pdf”), license: "unknown", no access_url, no raw_sha256, no tier_rationale, empty sampling_locations, null sampling_year_range, and used the legacy ## Wiki pages updated on ingest heading. Numerical content of the previous Key numbers section was spot-checked against the source PDF and confirmed accurate; values are preserved verbatim and the surrounding structure was reorganized into the per-Table layout that matches the source’s Table on pages 1292–1293. Cite-key angelon-gaetz2018-lead-spices-north-carolina preserved.
• Three identity checks: (1) DOI grep on 10.15585/mmwr.mm6746a2 matched only this page; (2) raw_handle grep on MFK_lead-in-spices-herbal-remedies-and-ceremonial-powd was empty (the prior page used the generic manual-fetch-kimi handle); (3) cite-key grep on angelon-gaetz matched only this page. No true-duplicate conflict.
• Frontmatter additions: raw_handle updated to the MFK_ form matching the suggested handle; raw_path corrected to the full filename; raw_sha256: 870822a0727ae5ecc8375b32a947130719b0bb3df406ee12d855599367d781d4 computed via shasum -a 256; access_url set to the canonical CDC MMWR URL https://www.cdc.gov/mmwr/volumes/67/wr/mm6746a2.htm; license: us-government-work (CDC MMWR is a US Government work, public domain; the prior "unknown" was a placeholder); tier_rationale added; sampling_locations set to “North Carolina, US — 8 primarily urban counties; home lead investigations January 2011–January 2018”; sampling_year_range: "2011-01 to 2018-01"; sample_population expanded with the race/ethnicity, age, and food/nonfood breakdown the source reports.
• Body additions: full per-category Pb tables for both nonfood (Table page 1292) and food (Table page 1293) reproduced rather than the prior bulleted subset (which captured only ~11 of the 50 product categories with average Pb ≥1 mg/kg); the source itself provides standard-deviation and range columns for every multi-sample category, which are now preserved. Provenance metadata (187/386 known country of purchase) and the consumption-frequency context from the EPA WWEIA survey and Siruguri & Bhat (2015) added because they are load-bearing for any later exposure-modeling work that touches this source. Stated-limitations section added verbatim from the source’s Discussion paragraph.
• HMT&C firewall (CLAUDE.md Part 2): the prior ## Implications “Certification” subsection contained a threshold proposal (“HMT&C standards for spice-adjacent products … should treat lead as the primary concern and benchmark against the 10 mg/kg FDA color additive limit as a minimum floor, not a safety level”) and two interpretive synthesis claims (“consistent with lead chromate added for color”; “suggesting household-level contamination spread from spice handling”). All three were removed under the strict reading of Part 2 (source pages report what the paper found; no HMT&C threshold proposal; no synthesis across papers). The replacement language reports what the paper itself argued for (a national maximum allowable limit for lead in spices and herbal remedies) and which categories carry the largest N, without proposing a numerical HMT&C threshold.
• Brand firewall (CLAUDE.md Part 12): the source does not name specific brands in the contamination data; it names product categories (Balguti Kesaria as an Ayurvedic-medicine category, kumkum/sindoor/surma as cosmetic-powder categories) and a single unbranded “saffron supplement.” Scientific-method vendor names — atomic absorption mass spectrometer and ICP-MS at the NC State Laboratory of Public Health — are preserved per the 2026-05-17 Exception 2 (scientific reproducibility).
• Speciation: paper measures total Pb only; no speciation issue. Recorded as Pb in metals:.
• Routing: source has products: ["[[products/spices]]"], which is the umbrella product category (provisional scaffold; not yet a locked HMTc row). No routing rows are emitted for this source in data/evidence/product_source_routing_audit.csv because the audit currently emits rows only for locked HMTc product rows; this is the expected state for the spices product page until Step 0 Lock and is not a routing-audit failure.
• 2026-05-29 audit subagent (fresh-context, claude-subagent-general-purpose) verdict: REVISE with 3 ⚠️ presentation findings, 0 ❌ definite errors. All five checks (numerical fidelity, slug vocabulary, speciation/methods, Part 12 brand firewall, Part 2 wiki/HMTc firewall) had no ❌ findings; spot-checked numbers across nonfood and food Tables verified verbatim against the PDF.
  • ⚠️ Finding 1 (numerical clarity, applied): the abstract sentence and the Key numbers bullet describing 111 contaminated samples now disambiguate the two denominators the source reports together on PDF p. 1291 — 111/386 = 28.8% of the full sample set, 111/177 = 62.7% of the 50-category subset. Both percentages are now stated explicitly.
  • ⚠️ Finding 2 (slug vocabulary, applied as documentation): matrices: [..., ceremonial-powders, herbal-remedies] — the bare strings ceremonial-powders and herbal-remedies are not in the controlled matrix vocabulary listed in docs/gpt-collaboration/taxonomy-snapshot.md; they describe the source’s actual scope and are used in two other source pages in the corpus. They pass current routing-audit discipline (source is not in routing_unresolved.csv or routing_malformed.csv). Retained as-is and flagged here per the subagent’s recommendation rather than removed, because removal would silently drop scope information the source actually carries.
  • ⚠️ Finding 3 (sampling-date precision, applied as documentation): sampling_year_range: "2011-01 to 2018-01" matches the broader home-investigation-window phrasing on PDF p. 1290 (“home lead investigations were conducted at 983 properties in North Carolina during January 2011–January 2018”). The narrower investigation-conducted span for this subset of 59 properties / 61 children was March 17, 2011 – January 26, 2018 (PDF p. 1290). The frontmatter range covers the broader window because the sampling frame is the full 2011–2018 investigation program; the narrower subset span is preserved here for reference.

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.