Weldegebriel et al. 2025 — Toxic metals (Cd, Cr, Pb, Ni) in 80 packaged fruit juices marketed to children in Gondar, Ethiopia

This peer-reviewed Scientific Reports study (Sci Rep 15:36868, 2025) measured cadmium, chromium, lead, and nickel in 80 packaged fruit juice samples consumed by children in Gondar city, Northwest Ethiopia. Samples covered eight Ethiopian-manufactured brands across four juice types (mango, pineapple, strawberry, cocktail) and five packaging formats (tetra pack, glass, carton, plastic, canned). Health risk to children aged 3-14 years was assessed via target hazard quotient (THQ), hazard index (HI), and cancer risk (CR), parameterized from a separately-conducted consumption survey of n=422 Gondar children. Median pooled concentrations across all 80 samples were Cd 0.08 mg/L, Cr 0.004 mg/L, Pb 0.035 mg/L, Ni 0.078 mg/L (Table 5). Several samples exceeded international safety limits for Cd, Ni, and Pb, but all individual-metal THQ values and HI fell below 1, suggesting non-cancer effects are unlikely at the modeled intake (0.034 L/day, 20 kg body weight, 6-year exposure duration). Cancer risk for Cd and Ni exceeded the USEPA 1×10⁻⁶ acceptable threshold. Cr and Ni concentrations differed significantly by juice type (Kruskal-Wallis p=0.018 for Cr, p=0.001 for Ni; Cd p=0.055 is at the conventional threshold but does not meet p<0.05); the authors also report that juices in cans and glass containers carried higher metal loads than tetra pack or plastic containers. Method: 210VGP flame atomic absorption spectroscopy with HNO₃ wet acid digestion, triplicate measurements, certified reference material from an India-accredited laboratory, calibration r² 0.9983-0.9998 (Table 2; the PDF narrative text on p. 3 prints the range as 0.9983-0.9992, a paper-internal inconsistency).

Key numbers

Pooled medians and ranges across all 80 samples (Table 5, unspiked column; mg/L)

Ni and Cd medians lie close to the reported maxima for the 80-sample set; this is unusual for a median statistic but reflects what the source publishes in Table 5.

Median by juice type (Table 6; mg/L)

*p < 0.05. The source’s Table 6 prints “Pub” in the row that holds Pb values; this is a typographical error in the source and the row is lead.

Health risk assessment for children (Table 7)

Hazard Index (HI = ΣTHQ) = 0.171. All single-metal THQ values are below 1 and the HI is below 1, indicating non-cancer adverse effects are considered unlikely at the modeled intake. Cancer risk ranking is Ni > Cd > Cr > Pb. Cd (1.55×10⁻⁵) and Ni (6.76×10⁻⁵) exceed the USEPA 1×10⁻⁶ acceptable cancer-risk threshold; Cr (1.02×10⁻⁶) is at the threshold; Pb (1.56×10⁻⁷) is below.

Exposure parameters (source p. 5): ingestion rate IRi = 0.034 L/day (converted from 0.238 L/week as stated in the methods text; the Table 4 survey mean of 0.28 L/week ± 0.095 was not used directly in the HRA model); body weight BWi = 20 kg; exposure duration ED = 6 years; exposure frequency EF = 365 d/yr; averaging time AT = ED × EF = 2190 days; age-dependent adjustment factor ADAF = 3 for cancer risk. Reference doses (RfD, mg/kg-day): Cd 0.001, Cr 0.003, Pb 0.0035, Ni 0.008. Cancer slope factors (kg-day/mg): Cd 0.38, Cr 0.5, Pb 0.0085, Ni 1.7.

Analytical performance (Table 3, with caveat below)

Calibration r² values: Pb 0.9998, Cd 0.9983, Cr 0.9997, Ni 0.9995 (Table 2). Table 3 columns are labeled “LOQ(mg/L)” then “LOD(mg/L)” in the source; for Cd, Pb, and Ni this produces the unusual condition LOD > LOQ. The source text (p. 4) states “Table 3 shows the instrumental detection limit ranges from 0.001 to 0.04 mg/L, whereas the method detection limit was 0.0024-0.08 mg/L,” which suggests the column labels may have been transposed or that the two columns mix instrumental and method limits. The values are reproduced here exactly as the source’s Table 3 prints them and the discrepancy is flagged in Limitations.

Methods (brief)

Sample collection: cross-sectional sampling from five purposively-selected Gondar markets (Bilko, Piassa, Chechela, College, Azezo) in 2022 (source prints the date range as “February 30 to July 30 2022”; February 30 is a non-existent date — this is a paper-internal contradiction and the actual start date is presumably an end-of-February or early-March 2022). Eight Ethiopian-manufactured brands were selected by simple random technique; 10 samples per brand were purchased proportional to lot-size, for a total of 80 juice samples. Brand identities are stated in the source as B1-B8 with associated juice and packaging descriptors (mango in tetra pack, glass, and canned formats from Neqemte; strawberry canned and pineapple carton from Addis Abeba; cocktail glass; one mango brand “B8” in plastic). Samples were transported refrigerated to the laboratory; per-brand samples were pooled and homogenized into representative composites for analysis.

Sample preparation: 4 mL aliquot of juice + 7 mL concentrated HNO₃ (69-72%) in 50 mL conical flask, heated on hotplate at 180 °C until clear and colourless; cooled digestate then diluted with 10 mL deionized water and filtered through Whatman No. 42 paper; filtrate transferred to 50 mL volumetric flask and diluted to the mark with distilled water; stored at 4 °C until analysis. Each digest analyzed in triplicate. Reagent-only blanks digested in parallel. Glassware washed with detergent and tap water, rinsed with distilled water, soaked in dilute chromic acid (H₂CrO₄ prepared from concentrated H₂SO₄), and rinsed again with distilled water.

Instrumentation: 210VGP-model flame atomic absorption spectrometer with air-acetylene flame and analyte-specific hollow-cathode lamps. Operating conditions (Table 1): Cd 228.9 nm / slit 0.7 nm / 3.244 eV; Pb 217 nm / slit 0.7 nm / 2.566 eV; Cr 359 nm / slit 0.7 nm / 3.833 eV; Ni 324.7 nm / slit 0.2 nm / 3.792 eV. Acetylene was used as the fuel gas.

Calibration and QA: seven-point calibration curves prepared from serial dilution of Blulux 1000 mg/L stock standards (range per analyte, Table 2: Pb 0.06-8 mg/L, Cd 0.005-10 mg/L, Cr 0.02-7 mg/L, Ni 0.04-6 mg/L). Certified reference material from an India-accredited laboratory was used for standard preparation. Recovery was assessed by spiking juice samples with 100 ppm Cd/Cr/Ni or 10 ppm Pb intermediate standards, redigesting, and computing %R = (C_spiked – C_sample) / C_added × 100; observed recovery 81-115% across all metals (within the 80-120% acceptance window). RSD 1.2-13.3% across metals.

Children’s juice consumption survey: a separately-conducted structured questionnaire of n=422 parents in Gondar (72.9% children age 3-6 years, 27.1% age 7-14 years; 60.6% female, 39.4% male children; mean child age 6.0 ± 2.54 years). Sample size derived from single-population proportion formula (Z=1.96, P=0.50, q=0.50, d=0.05) yielding minimum n=384; inflated to 422 for 10% expected non-response. Survey mean weekly intake 0.28 L/week ± 0.095 (range 0.25-2 L/week, Table 4). The methods text reports 0.238 L/week (= 0.034 L/day) as the value carried into the CDI calculation; this differs from the Table 4 survey mean and the page reproduces both values without resolving the discrepancy.

Statistics: data tabulated and managed in Microsoft Excel; transferred to SPSS for analysis. Non-parametric tests used because of non-normality. Kruskal-Wallis with multiple pairwise comparisons for between-group (juice-type) differences. Significance threshold p ≤ 0.05.

Limitations

• The 80-sample pool mixes tetra pack, glass, carton, plastic, and canned packaging within a single statistic. Strawberry (n=10) is sampled only in canned format; pineapple (n=10) only in carton; cocktail (n=10) only in glass; mango (n=50) spans tetra pack, glass, canned, and plastic. The pooled medians in Table 5 therefore cannot be reproduced as a single packaging-homogeneous estimator. The Figures 2-4 caption/data integrity issues (see next bullet) limit how reliably the by-packaging cross-tabs can be reconstructed from the published paper alone.
• Median values for Cd (0.08 mg/L, max 0.1) and Ni (0.078 mg/L, max 0.08) lie within the IQR of the reported maxima for the 80-sample set; the IQR width relative to the median-to-max distance suggests the distribution is heavily left-skewed or has ties at the upper end. The source does not publish per-sample values or the per-brand composite values, so the distribution shape cannot be independently reconstructed from the published tables.
• Table 3 column labels for LOD and LOQ produce the unusual relationship LOD > LOQ for Cd, Pb, and Ni. The source’s text describes “instrumental detection limit 0.001-0.04 mg/L” and “method detection limit 0.0024-0.08 mg/L” but does not reconcile these ranges with the Table 3 column headings. Sub-LOQ imputation or non-detect handling cannot be inferred reliably from the published table alone.
• Source figure labels are partially corrupted: Figures 2, 3, and 4 are each captioned “Distribution of cadmium concentration” or “Distribution of lead concentration” but the plotted y-axis ranges and values appear to correspond to different analytes in the surrounding text (Fig. 2’s body-text refers to chromium; Fig. 3’s body-text refers to lead; Fig. 4’s body-text refers to nickel; Figs. 3 and 4 are visually identical to each other). Because of these caption/data mismatches, the figure-level packaging × juice-type cross-tabs are not reproduced on this page; Tables 5 and 6 are used as the reliable cross-tabs.
• Sampling date range is stated as “February 30 to July 30 2022,” a paper-internal contradiction (February has no 30th day).
• The HRA intake parameter (0.238 L/week → 0.034 L/day) used in the CDI calculation differs from the survey-derived mean intake (0.28 L/week, Table 4). The source does not explain the discrepancy.
• Only Ethiopian-manufactured brands were included; the conclusion does not generalize to imported juices on the Ethiopian retail market.
• The CSF (cancer slope factor) values used (Cd 0.38, Cr 0.5, Pb 0.0085, Ni 1.7 (kg-day/mg)) are stated by the authors without IRIS/USEPA citation; the Cr CSF in particular (0.5) corresponds approximately to the oral CSF for Cr(VI), not total Cr — the source measures total Cr by FAAS and does not speciate, so the CR estimate for Cr should be treated as an upper bound assuming all Cr is Cr(VI).

Wiki pages this source may touch

• fruit-juice-not-canned
• fruit-juices-non-apple
• fruit-juice
• mango-juice
• cadmium
• chromium
• lead
• nickel

Verification notes

• 2026-05-20 merge-enhanced under the v2.0 manual-fetch ingest skill from the MFK PDF (raw/Manual Fetch Kimi /01_Fruits_and_Fruit_Products/01_Fruits_and_Fruit_Products/Toxic metal contamination and health risk assessment of packaged fruit juices consumed by children in Northwest Ethiopia.pdf; sha256 47f001a31a8fe4c323ac21194da39407dfdf97a4f1425f04343bb04c41dacd59). The MFK PDF and the existing FM_12546707 marker ingest are the same paper (matching DOI 10.1038/s41598-025-20789-x, matching title up to a “Gondar city, Ethiopia” vs “Northwest Ethiopia” subtitle variation between the Nature page and the print PDF). The previous wiki page (dated 2026-04-29) was a Marker-corpus ingest of the FM_12546707 markdown; this revision is the Cochrane-quality re-ingest from the manual-fetch PDF. Preserved cite_key, raw_path, raw_sha256, near_duplicates per skill rule.

• Frontmatter raw_handle previously held the file path raw/markdown/FM_12546707/FM_12546707.md. Corrected to the proper handle form FM_12546707; raw_path retained as the markdown-file path.

• Frontmatter products previously listed [fruit-juice-not-canned, packaged-fruit-juice, "[[products/fruit-juices-non-apple]]", "[[products/fresh-fruit]]"] — two invalid entries. packaged-fruit-juice is not in the taxonomy snapshot; removed. fresh-fruit is the wrong product scope (this study covers packaged liquid juice, not fresh fruit); removed. Retained fruit-juice-not-canned and fruit-juices-non-apple in consistent quoted-wikilink form. The mixing of canned packaging within this study is captured in Limitations rather than by adding a canned-fruit product slug, because canned-fruit (which exists in the taxonomy) refers to canned solid fruit, not canned juice; there is no canned-fruit-juice product slug in the current taxonomy.

• Frontmatter ingredients was empty in the previous version; added [[ingredients/fruit-juice]] and [[ingredients/mango-juice]] (both present in taxonomy and on disk). Pineapple, strawberry, and cocktail juice do not have dedicated ingredient pages and route to the parent fruit-juice.

• Frontmatter matrices retained as [packaged-fruit-juice, mango-juice, pineapple-juice, strawberry-juice, cocktail-juice] — descriptor matrices matching the source’s juice-type cross-tabs, not product slugs.

• Body content substantially enhanced: previous Key Numbers had 3 pooled rows (Ni/Cd/Pb medians and ranges) plus a 4-row Ni-by-fruit-type breakout, totaling 7 rows. Cr was missing from the pooled and the by-fruit-type tables; the entire Health Risk Assessment (Table 7: CDI, THQ, CR per metal, HI total) was absent; the analytical performance table (Table 3 recovery/RSD/LOD/LOQ) was absent; the calibration r² values were absent; the exposure-model parameters (RfD, CSF, BWi, ED, EF, ADAF, IRi) were absent. Methods (brief) was a one-paragraph summary that did not mention the n=422 survey, the per-element instrument operating conditions, the certified reference material source, the digestion solvent volumes, the seven-point calibration design, or the per-brand pooling step. All added. Limitations were limited to a single sentence about canned-vs-not-canned mixing; expanded to cover the seven distinct methodological caveats actually present in the paper (canned mixing; median-near-max distribution; LOD/LOQ inversion; figure caption/data mismatch; February 30 sampling date; intake-value discrepancy; geographic-scope limit; Cr speciation assumption in CR estimate).

• Legacy section headings updated to current Part 6 schema: ## Summary → page-summary paragraph integrated under the title; ## Key Numbers → ## Key numbers; ## Methods Brief → ## Methods (brief). The legacy ## Implications section was removed: per Part 2 firewall, the prior page mixed wiki and HMTc registers (“Useful candidate evidence for fruit juice nickel once canned products are excluded” — operational HMTc guidance; “Supports fruit-type and packaging-type risk features after row-level extraction” — App-product operational guidance). These statements are removed and the underlying methodological observation (packaging mixing requires filtering before percentile math) is preserved in Limitations as a neutral observation about this source rather than as operational guidance for downstream HMTc/App work. The legacy ## Wiki Pages Updated On Ingest section was removed (routing fan-out is system-maintained per Part 5b) and replaced with ## Wiki pages this source may touch per current schema.

• Brand firewall (Part 12) check: the source names eight brand identifiers explicitly in Methods (“pine apple, Wow mango, 3D, 7star mango, strawberry, cocktail, dada mango, an d 7star”) and references them as B1-B8 in Results. No brand names are reproduced on this wiki page; all by-juice-type values are reported at the juice-type and packaging-format descriptor level only. Brand-by-brand attribution does not appear and would not be reproduced even if requested by future sessions.

• Wiki/HMTc firewall (Part 2) check: no synthesis claims, no HMTc threshold proposals, no consumer audience translations, no statements of the form “this confirms the literature consensus that…” Implications section deleted as noted above.

• Paper-internal contradictions surfaced rather than silently corrected: (1) sampling start date “February 30”; (2) Table 3 LOD/LOQ column inversion; (3) Figure 2/3/4 caption/data mismatch; (4) HRA intake (0.238 L/week in methods text) vs survey mean (0.28 L/week in Table 4); (5) Cr cancer-risk uses a CSF that corresponds to Cr(VI) while the analytical method (FAAS) measures total Cr without speciation; (6) calibration r² range printed as 0.9983-0.9992 in the narrative on p. 3 but Table 2 carries values up to 0.9998. These are flagged in Limitations or in the Key numbers narrative; the page does not attempt to correct the source.

• Audit subagent (2026-05-20) flagged Check 1 ⚠️ that the opening paragraph wrote “Cd and Ni concentrations differed significantly by juice type” while the cited p-values were for Cr and Ni; Cd’s p=0.055 does not meet p<0.05. Verified against PDF Table 6 (p. 8): Cr p=0.018*, Ni p=0.001*, Cd p=0.055, Pb p=0.68. Finding correct; opening paragraph corrected to “Cr and Ni… differed significantly” with Cd p=0.055 noted as not meeting the threshold.

• Audit subagent (2026-05-20) flagged Check 3 ⚠️ that the methods description “diluted to 50 mL with deionized water and filtered” compressed the source’s two-stage dilution into one and conflated deionized vs distilled water. Verified against PDF p. 3: source describes cooled digestate diluted with 10 mL deionized water, filtered through Whatman No. 42, then transferred to a 50 mL volumetric flask and made up to the mark with distilled water. Finding correct; methods text rewritten to preserve the two-stage dilution and the deionized/distilled distinction.

• Audit subagent (2026-05-20) flagged Check 3 ⚠️ “Bluelux” misspelling of stock-standard vendor. Verified against PDF p. 2: source spells the vendor “Blulux” (no second “e”). Finding correct; corrected on this page.

• Audit subagent (2026-05-20) flagged Check 5 ⚠️ that the Limitations bullet 1 explicitly named “the locked HMTc Category 1 row is fruit-juice-not-canned” and stated “packaging-format filtering is needed before any percentile math for the not-canned row” — a Part 2 boundary slip in the form of HMTc-row operational guidance on a wiki source page. Verified against Part 2 firewall (CLAUDE.md): wiki source pages report what the literature found at the source level; routing/aggregation discipline for HMTc workflow belongs in the workbench layer (data/workbench/standards/), not on the source page. Finding correct; Limitations bullet 1 reframed as a neutral observation about the source’s packaging-format mixing and what it does to the published pooled estimator, without naming HMTc or referencing percentile-math intent.

• Audit subagent (2026-05-20) flagged Check 1 ⚠️ (optional) that the calibration-r² narrative range in PDF p. 3 (0.9983-0.9992) disagrees with Table 2 (which prints values up to 0.9998). Verified; the wiki page Key numbers uses the Table 2 values (0.9998 for Pb). Added to the Limitations contradiction list as paper-internal inconsistency (6) and added a parenthetical note in the opening paragraph that flags both ranges. The page values themselves remain unchanged because Table 2 is the more granular and primary record.

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.