Masri et al. 2025 — Dietary consumption of toxicant-laden foods and beverages in Southern California

This is a population dietary-intake study, not a contamination occurrence study. The authors deployed a one-week online food frequency questionnaire to 186 Southern California residents (60.8% adults, 39.2% adolescents) covering staple foods and the food categories most commonly targeted under California’s Proposition 65 for lead and acrylamide exposure: root vegetables, rice, leafy greens, pasta/noodles, tea, juice, seafood, twenty individual herbs/spices, stuffed grape leaves, tamarind/chili candy, chocolate products, nutritional powders, and traditional beverage powders such as matcha. The study reports daily intake (g/day), portion size, eating-occasion frequency, and home-food behaviors (homegrown food, store-bought soil use, tap-water reliance, pre-1978 housing) by age and race/ethnicity, then uses one Prop 65 consent-judgment threshold to demonstrate the magnitude of underestimated single-day lead exposure from stuffed grape leaves. The contribution to the Heavy Metal Index corpus is exposure-side: it provides population intake denominators that downstream synthesis can multiply against published contamination concentrations to estimate dietary exposure, and it documents underestimation patterns in Prop 65’s product-by-product exposure assessment methodology for herbs and spices (which cumulate to ~8.3 g/day mean across the food category but are conventionally assessed at a 0.5 g RACC per single herb). The study itself measures no metal concentrations and proposes no contamination thresholds.

Key numbers

All values are intake in g/day or g/person/day; consumption frequency in eating occasions (EO) per day or per week. Where stated, “per capita” means averaged across all N = 186 respondents (including non-consumers); “consumers only” means averaged across the subset who reported any consumption of that item over the past 7 days. Daily intake = portion size × consumption frequency × USDA volume-to-mass conversion. No analytical metal measurements were made in this study; lead exposure values quoted below are arithmetic products of reported intake and externally-cited consent-judgment contamination ceilings.

Sample characteristics (N = 186)

Per capita daily intake of selected food categories (g/day, all N=186)

Selection limited to categories most relevant to the heavy-metals corpus; full Table 2 contains additional entries. ”% (consumers)” = percentage of N=186 reporting any consumption of the item in the past 7 days.

Beverage portion-size assumptions (paper-specified)

• Wine: 5 oz = ~150 g per glass.
• Other liquids (water, juice, tea): 8 oz = ~240 g per cup.
• Nutritional powder and chocolate drink powder portion-size from USDA FoodData Central (mean portion 44.3 g and 23.3 g respectively); traditional drink powder 2.1 g per serving.

Herb and spice intake — top categories (Table 4)

Per-capita (“All N=186”) and consumer-only daily mass intake of the 20 herbs/spices surveyed. Eating-occasion (EO) frequency averaged across all participants. Unit conversion (g/tsp) from USDA where available; “1.8 ᵇ” is the paper’s fallback value when not separately reported.

Cumulative herb/spice intake (all 20 categories aggregated to a single food category)

Age and race/ethnicity differentials (Table 3, regression analyses, β estimates in g/day relative to reference)

Reference group for race/ethnicity: White/Caucasian. Reference for age: adolescents (13–17). Bold β = p < 0.01; underlined β = p < 0.05. Values reproduced as printed; only categories with at least one statistically significant differential are listed here.

Adolescent vs adult differential (Section 3.3 summary)

Stuffed grape leaves — single-serving lead exposure calculation (paper’s own arithmetic, Discussion §4.4)

Reference threshold: California Prop 65 consent-judgment lead ceiling for stuffed grape leaves, 0.018 ppm (Superior Court of California 2021, paper Reference 73).

Other Prop 65 consent-judgment ceilings cited in Discussion (paper’s report, not measurements)

The paper references the following lead ceilings as context for population exposure interpretation (not as findings of this study):

Behavioral patterns affecting exposure (Sections 3.5, 4.3)

Race/ethnicity discrepancies — largest gaps reported in Section 3.4 and 3.5

Methods (brief)

Study design. Cross-sectional online dietary survey administered via Qualtrics (Provo, UT) between 1 March and 15 June 2023 to Southern California residents (adults ≥18 and adolescents 13–17). Recruitment by snowball sampling through community contacts (university groups, non-profit organizations) circulating the survey link in English and Spanish via email listservs and social media (Facebook, Instagram). Targeted at popular staple foods consumed in high quantities and/or foods most commonly pursued under California’s Proposition 65 enforcement for lead and acrylamide contamination, as reported by the California Attorney General. Concentration measurements were not performed.

Instrument. Adapted from the NHANES Food Frequency Questionnaire with the recall window reduced from one year to one week. Participants were asked whether they consumed each item in the past 7 days, then (multiple-choice) how often and how much per typical occasion. Beverages were pre-assigned a serving size (5 oz wine, 8 oz other liquids); food/beverage powder serving size estimated from USDA FoodData Central. For 20 individual herbs/spices, only frequency was asked separately; quantity per occasion was asked once for the herb/spice food category as a whole (per FDA’s RACC guidance). Additional questions on tea-bag reuse, cumulative water intake, tap vs bottled water source, home gardening, store-bought soil use, and pre-1978 housing year. Full survey instrument in Supplementary Table S1.

Data reduction. Range-response options converted to range midpoint (e.g., “1–2 cups” → 1.5); “less than” responses converted to half-unit; “greater than” responses converted to plus-one-unit; consumption frequencies converted similarly (“2–4 times per week” → 3). Volume responses (cups, teaspoons) converted to grams via USDA FoodData Central density values. Daily intake (g/day) = portion size × consumption frequency. Per-capita statistics retained all participants including non-consumers; “consumers only” statistics excluded participants reporting zero consumption.

Herb/spice intake calculation. Consumption frequency per herb/spice multiplied by 0.83 tsp/EO (the average consumption quantity per EO across all herb/spice categories) and by the USDA density (g/tsp) of that specific herb/spice to yield mass intake per category. Total Herb/Spice daily intake = sum across all 20 categories per participant.

Statistics. SAS 9.4 (SAS Institute, Cary, NC). Means, medians, and group means by age and race/ethnicity. Multivariate regression with two covariates (age category, race/ethnicity) only — reference groups White/Caucasian and adolescent (13–17). Significance assessed at p < 0.10 and p < 0.05. No correction for multiple comparisons reported. The authors note that small-cell counts in some race/ethnicity strata may have lessened ability to detect intake differences.

Demographic distinctions. “Indian” and “Middle Eastern” were defined as distinct from “Asian” (East/Southeast Asia only) given the differing cuisine and ingredient profiles described in Section 2.2. The “Other” category was 1.1% (n = 2).

Survey-completion quality control. Of 207 surveys completed, 21 were excluded for being below the age threshold, residing outside California, or completion time < 3.38 min (the 1st-percentile threshold). N = 186 retained. Median completion time 8.15 min (IQR 5.9–11.6).

Ethics. IRB-exempt for adult survey (UC Irvine Office of Research, Kuali Research Protocols). Adolescent administration approved by IRB Committee C, approval code 2733, approval date 9 May 2023. Informed consent waived per exempt status.

Funding. NIH/NIEHS R01ES030353.

Limitations

• Convenience snowball sample drawn from Orange County, California, with no probability framework. The retained sample over-represents Asian/Pacific Islander (40.9%) and Middle Eastern (11.8%) residents relative to background county demographics; White/Caucasian (11.8%) and Hispanic (28.0%) are under-represented vs the regional reference statistics the authors cite. Generalizability beyond Orange County is correspondingly limited.
• N per race/ethnicity stratum is small (Indian n = 12, Middle Eastern n = 22, Other n = 2). Several regression coefficients in Table 3 are estimated on cells of n < 25; the authors flag that small-cell counts may have lessened ability to detect differences.
• Two-covariate multivariate model (age and race/ethnicity only). Income, education, gender, household-size, and neighborhood confounders were not adjusted for despite being collected; the authors note residual confounding cannot be ruled out.
• One-week recall is shorter than the NHANES 1-year recall but longer than the 2-day NHANES dietary-recall instrument; results are not directly comparable to either. The authors report cumulative-frequency values ~2× higher than NHANES FFQ and ~30% lower than 2-day NHANES dietary recall, attributed to the difference in window length and demographic composition.
• Self-reported intake is subject to recall bias and social-desirability bias. The authors expect modest bias because the survey was not marketed as a “health” survey.
• Volume-to-mass conversion uses standard USDA densities and may not match individual respondent product choice (e.g., dried vs fresh; powder density variation across brands).
• Pre-added herbs/spices in processed foods are not captured; the authors note herb/spice intake may be “drastically underestimated” because participants do not know the herb content of prepared meals they consume.
• No metal concentration measurements were performed. Exposure estimates require an external contamination ceiling (e.g., Prop 65 consent-judgment value, FDA action level, regulatory cap) to be multiplied against the intake values; the paper presents one such calculation (stuffed grape leaves) and references several other ceilings for context.
• Stuffed grape leaves intake reflects only n = 11 consumers (5.8% of N = 186); the consumer-mean portion size of 242 g and intake of 44.5 g/day are estimated on a very small subgroup and should not be generalized to the broader Middle Eastern population without replication.
• The Discussion’s lead-exposure calculation for stuffed grape leaves uses the consent-judgment ceiling (0.018 ppm) as the contamination value, which represents the legally allowable maximum after a court settlement rather than measured product concentrations; actual product Pb may be lower or higher.
• “Less than” and “greater than” range conversions (e.g., “<1 cup” → 0.5 cups, “>3 cups” → 4 cups) introduce a directional bias toward central tendency; tail consumers are clipped.
• Reuse-of-tea-bags behavior is reported as a frequency question, not as a quantification of additional lead leaching; the inference that fresh-leaching is greater than label-assumption is qualitative.

Implications

For HMI evidence corpus. This source contributes California-specific population intake denominators for several food categories in the corpus: rice (52.6 g/day per capita; 57.8 g/day among consumers), leafy greens (33.6 / 39.5), root vegetables (83.9 / 87.2), tea (70.2 / 151.7), juice (72.4 / 120.5), pasta/noodles (30.9 / 38.0), seafood-other (21.8 / 55.8), fish-not-fried (11.0 / 24.1), chocolate foods (9.6 / 12.2), nutritional powder (6.8 / 24.3), traditional beverage powder including matcha (0.3 / 0.8), stuffed grape leaves (2.6 / 44.5), tamarind/chili candy (1.5 / 6.4), and twenty individual herb/spice categories with a cumulative mean intake of 8.3 g/day across the food category. These intake values may be used by downstream synthesis when multiplied against a per-ingredient or per-product contamination central tendency.

For exposure-assessment methodology coverage. The cumulative herb/spice intake of 8.3 g/day (vs 0.5 g RACC per single herb on FDA labeling guidance) is one of the cleaner exposure-side documentations in the recent literature of why product-by-product Prop 65 exposure assessment systematically underestimates aggregate herb/spice contaminant intake. The herb/spice food-category subdivisions documented in court enforcement (basil, oregano, etc. assessed separately) versus the USDA’s “seasonings, spices and herbs” single-category designation is a methodology gap relevant to regulatory crosswalks.

For Prop 65 / California regulatory crosswalk. The paper documents the following specific consent-judgment lead ceilings as context: stuffed grape leaves 0.018 ppm (2021); rice 0.056 ppm (2018); nutritional powders 3 µg/day; nutritional powders containing chocolate 4 µg/day; chocolate products 0.225 ppm Pb and 0.960 ppm Cd; “naturally occurring” lead in tamarind/chili candy 0.02 ppm under OEHHA Title 27 §28500 (2021). These are settlement-derived legal maxima, not measured concentrations or population exposure benchmarks; they belong in the regulatory crosswalk rather than the contamination occurrence layer.

For age-stratified exposure. Adolescent (13–17) intake exceeded adult intake by an average of +17% in portion size and +24% in daily mass, with statistically significant differentials (p < 0.05 multivariate) for rice, tostadas/taco shells, juice (p < 0.1), tea (p < 0.1), crackers, and popcorn. Combined with lower body weight, the per-kg-bw adolescent exposure differential is larger than the per-capita differential; downstream age-stratified exposure modeling should account for both factors.

For race/ethnicity-stratified exposure. Hispanic/Latino participants reported tamarind/chili candy intake 3.71 g/day higher than the White/Caucasian reference (p < 0.01); Middle Eastern participants reported stuffed grape leaves intake 18.44 g/day higher (p < 0.01); Indian participants reported root-vegetable intake 65.5 g/day higher (p < 0.01) and total herb/spice intake 13.1 g/day higher (p < 0.01). The authors frame these differentials as a Proposition-65 environmental-justice concern when product-specific “naturally occurring” pollution exemptions overlap with foods consumed disproportionately by specific racial/ethnic groups.

For tea exposure pathway. Roughly two-thirds of tea consumers reported reusing tea bags across multiple cups per day, which the authors note creates greater opportunity for lead leaching than the 3–5 min single-infusion model used in conventional Prop 65 exposure assessment. Tea consumer-only intake of 151.7 g/day (~⅔ of an 8 oz cup) is the working denominator for tea-derived dietary lead in this population.

For homegrown-food and tap-water pathway exposure. 23.1% of respondents reported homes built before 1978 (lead-paint and lead-plumbing era), 35% reported home food gardening, 40% of home-growers reported growing food in non-store-bought soil, and 26% reported >70% of daily water from unfiltered tap water. These behaviors are exposure modifiers for lead-from-paint-and-soil and lead-from-plumbing pathways; the paper does not measure but documents their prevalence.

Wiki pages this source may touch

• lead
• cadmium
• arsenic-total
• mercury-methyl
• rice
• tea
• leafy-greens
• root-vegetables
• fruit-juice
• seafood
• fish
• herbs-and-spices
• spices
• matcha-powder
• chocolate
• water
• turmeric
• parsley
• cinnamon
• pasta
• tea
• tea-infusions
• matcha
• fruit-juice-not-canned
• rice-bulk-grain
• leafy-vegetables-other
• root-tuber-vegetables
• pasta-wheat-based
• seafood
• spices
• chocolate
• oehha-lead-prop65
• oehha-cadmium-prop65
• oehha-arsenic-prop65
• fda-juice-haccp-lead-50ppb

Verification notes

• 2026-06-08 — Fresh ingest from raw/Manual Fetch Discovery/. The discovery-time filename masri2025-matcha-heavy-metal.pdf is misleading: the paper is not specifically about matcha. Matcha appears only as one example (“traditional beverage powder”) at 0.3 g/day per-capita / 0.8 g/day among consumers (Table 2). The paper is a cross-sectional dietary intake survey covering staple and Prop-65-targeted foods broadly. Per the el-daouk2020 precedent, the misleading discovery handle is preserved as raw_handle: MFD_masri2025-matcha-heavy-metal to keep the filesystem trail intact; the cite-key reflects the paper’s actual content (masri2025-toxicant-foods-california).
• 2026-06-08 — This is an exposure-side / dietary-intake study, not a contamination occurrence study. No metals were measured. All “lead exposure” calculations in the paper (e.g., stuffed grape leaves 4.3 µg/day) are arithmetic products of intake values from this study × externally-cited consent-judgment contamination ceilings. The frontmatter metals: field lists the four heavy metals the paper discusses in its background, results, and exposure interpretation (Pb, Cd, tAs, MeHg). The paper does not separately speciate arsenic or mercury — tAs and MeHg are used where the paper explicitly references inorganic As (for juice/rice context citing FDA action levels) and methylmercury (for seafood bioaccumulation per Section 4.1, Reference 35).
• 2026-06-08 — Brand firewall (Part 12): the paper does not report any per-brand contamination measurements. Brand names appear only in two contexts, both permissible: (a) Discussion §4.4 references “Whitney R. Leeman v. Starbucks Corporation” (Reference 28) and “Consumer Advocacy Group, Inc. v. Gulf Pacific Rice Co.” (Reference 70) as Prop 65 enforcement case citations — these are public-record regulatory-event references, permitted under Part 12 Exception 1; (b) Methods §2.3 names Qualtrics (Provo, UT) and SAS (SAS Institute, Cary, NC) as the survey platform and statistical software — permitted under Part 12 Exception 2 (scientific-method vendor/material names). No per-brand intake or per-brand contamination ranking is reproduced.
• 2026-06-08 — Slug coverage notes:
  • matrices: includes mixed-diet, herbs-spices, tea-infusions, beverages, candies-confectionery, prepared-mixed-dishes, dietary-survey. These are study-derived bare-string slugs reflecting the paper’s food-category coverage; the routing audit treats matrices as bare strings and should not flag these. dietary-survey is added to signal that this is an exposure-side intake study rather than a matrix-specific occurrence study, useful for the route_kind exposure_only downstream.
  • ingredients/leafy-greens is used as the umbrella; the snapshot also has leafy-vegetables but the paper uses “leafy greens” terminology throughout. Both slugs exist in the wiki.
  • ingredients/herbs-and-spices and ingredients/spices are both included because the paper distinguishes the food-category aggregate (8.3 g/day cumulative) from the per-spice intakes (Table 4). The umbrella and the granular slugs serve different downstream synthesis needs.
  • products/matcha is included because the paper explicitly identifies matcha as the worked example of “traditional beverage powder” in Table 2.
• 2026-06-08 — Per-capita vs consumers-only distinction: the paper consistently reports both. The Key numbers tables preserve both columns where source-reported. Downstream synthesis using this paper for population-exposure modeling should choose the appropriate denominator (per-capita for population-mean exposure; consumers-only for at-risk-subpopulation exposure).
• 2026-06-08 — Stuffed grape leaves sample size: the consumer-mean portion (242 g, ~6 leaves) and intake (44.5 g/day) are estimated on n = 11 consumers (5.8% of N = 186). The high values are reproduced as-printed but a small-cell caveat is noted in Limitations.
• 2026-06-08 — Tea bag reuse: the paper text reports both “roughly one-third” (Section 3.2, of tea users always/sometimes reuse bags) and “two-thirds” (Discussion §4.1, of consumers using “fresh teabags multiple times per day”). These describe overlapping but distinguishable behaviors. Both are reproduced in the Behavioral patterns table without resolution.
• 2026-06-08 — Tamarind/chili candy regression coefficient: the +3.71 g/day Hispanic differential is bolded in Table 3 (p < 0.01), but only 24.1% of the full sample reported any consumption (n = 33 of 186), and 59.5% prevalence among Hispanic/Latino vs 0% among Caucasian and Middle Eastern (Section 3.4 text). The coefficient is reliable as a between-group difference but the absolute intake magnitude (mean 1.5 g/day per capita; 6.4 g/day among consumers) is the right denominator for exposure modeling, not the regression coefficient.
• 2026-06-08 — Evidence tier B: peer-reviewed, transparent methods, NIEHS-funded, sample size N = 186 with documented exclusion criteria. Downgraded from A because of convenience sampling, restricted geography (Orange County), and small per-stratum n. Upgraded from C because the methods are reproducible and the dataset is structured for downstream re-analysis.
• 2026-06-08 — Audit subagent (general-purpose) flagged Table 3 significance labels: (a) Tea/Adult 33.76 labeled (p<0.01) and (b) Rice/Middle Eastern 23.79 labeled (p<0.01). Verified against PDF page 9 Table 3 (bold = p<0.01, underlined = p<0.05 per the table note) and the §3.3 narrative (“rice (p<0.05), tostadas/taco shells (p<0.05), juice (p<0.05), tea (p<0.1), crackers (p<0.05) and popcorn (p<0.05) showed statistically significant differences in daily intake when comparing adults and adolescents”). Findings confirmed correct. Corrected Tea/Adult (p<0.01) → (p<0.1) (matches §3.3 explicit p-value and the cell’s unformatted appearance in Table 3). Corrected Rice/Middle Eastern (p<0.01) → (ns) (Table 3 cell is plain text — neither bold nor underlined; §3.4 narrative makes no claim of significance for this cell). On the same pass, two further Adult-column labels in the same table were also corrected against the §3.3 text: Juice/Adult (p<0.01) → (p<0.05) and Rice/Adult (p<0.01) → (p<0.05); both cells are underlined in Table 3 per the source.
• 2026-06-08 — Audit subagent flagged [[products/chocolate]] as not in the taxonomy snapshot. Verified against docs/gpt-collaboration/taxonomy-snapshot.md Products list: chocolate IS present in the Products section (between childrens-nail-polish and clasps-fasteners-infant-products per the alphabetical list). Finding was a false positive — the slug is valid. No change applied; flagged here per the audit-handling protocol (false positives are documented but not acted on).
• 2026-06-08 — Audit subagent flagged an in-paper race/ethnicity figure discrepancy: §3.1 narrative (PDF page 5) reports Asian/Pacific Islander 38.3%, Latino/Hispanic 22.9%, White/Caucasian 11.2%, while Table 1 (page 6) reports 40.9% (n=76), 28.0% (n=52), 11.8% (n=22) for the same groups. The wiki page reproduces Table 1 (the formal demographic table) rather than the §3.1 narrative because Table 1 carries the cell counts and is the consistent source for the gender (62.9% / 36.0% / 1.1%) and education breakdowns the wiki also uses. The §3.1 narrative figures appear to be an editorial error in the manuscript; the Table 1 figures are internally consistent (sum to 100% across the six race/ethnicity rows excluding rounding). Flagged here so downstream synthesis using this source is aware of the source-internal inconsistency.

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.