Quality and safety investigation of commonly used topical cosmetic preparations

Almukainzi M; Alotaibi L; Abdulwahab A; Albukhary N; El Mahdy AM

doi:10.1038/s41598-022-21771-7

Almukainzi et al. 2022 — Heavy metals in 21 topical cosmetic products on the Riyadh, Saudi Arabia market by ICP-MS

This open-access Scientific Reports study tested 21 commercial topical cosmetic and personal-care products purchased from Riyadh, Saudi Arabia retailers for eleven elemental impurities (Al, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Cd, Pb) by inductively coupled plasma mass spectrometry following USP General Chapter <232> / NF 35 acid-digestion procedures, alongside aerobic-plate-count microbial testing and pH/organoleptic checks. Product categories were selected by a 447-respondent online survey of mostly-female adults in Saudi Arabia identifying the most frequently used topical cosmetics; one commercial sample was tested per category (two for toothpaste). The authors compared every analyte against the US FDA cosmetic guidance limits (Pb 10 ppm in general cosmetics, 20 ppm as additive; Cr 50 ppm; As 3 ppm) and USP <232> elemental-impurity acceptance criteria (Cd 0.5 ppm, Co 5 ppm, Ni 20 ppm, Cu 300 ppm), and identified two toothpastes as the most contaminated samples in the panel — both exceeding the FDA Pb limit (75.86 and 78.31 ppm Pb) and the FDA As limit (221.96 and 209.33 ppm As) by one to two orders of magnitude. Cadmium exceeded the USP 0.5 ppm acceptance criterion in five samples, with the deodorant sample reporting an outlying mean Cd value (306.92 ppm) substantially higher than the rest of the panel. Mercury was listed in the abbreviations but was not measured. Evidence tier is B: peer-reviewed open-access journal, transparent methods, n=1 per product category (no within-category replication), and several reported values carry very large coefficients of variation that warrant cautious interpretation (see Verification notes).

Key numbers

Eleven metals quantified in all 21 products by ICP-MS, reported as mean ± SD in ppm (mg/kg) in the as-sold finished product. Calibration standards were prepared at 5, 15, 30, 70, and 150 ppb; coefficients of determination for all elements ranged 0.9995-1. LOD/LOQ were not reported numerically.

Per-product results, Table 4 of the source (ppm, mean ± SD)

Values transcribed by pdftotext -layout of page 8 of the source PDF and verified cell-by-cell. The standard deviation in nearly every cell sits at approximately √2 × the mean (consistent with two-replicate measurement with one near-zero reading); see Verification notes.

Sample	Product type	Al	Cr	Mn	Fe	Co	Ni	Cu	Zn	tAs	Cd	Pb
1	Sunblock cream	93.70 ± 131.51	50.96 ± 71.82	10.44 ± 14.73	597.63 ± 844.63	0.50 ± 0.71	62.45 ± 88.27	102.02 ± 144.17	202.52 ± 286.33	1.06 ± 1.50	4.40 ± 6.23	12.00 ± 16.97
2	BB cream	12042.16 ± 17028.409	67.49 ± 95.44	216.47 ± 306.02	95438.52 ± 134969.89	2.77 ± 3.92	39.42 ± 55.75	123.32 ± 174.40	216.00 ± 305.43	0.78 ± 1.10	0.11 ± 0.14	14.57 ± 20.61
3	Lip balm	60454.08 ± 85494.57	70.69 ± 99.79	53.20 ± 75.22	12737.88 ± 18013.41	1.99 ± 2.82	41.06 ± 57.98	170.97 ± 241.52	262.44 ± 371.06	2.26 ± 3.20	0.18 ± 0.25	22.47 ± 31.78
4	Hand cream	1309.24 ± 1850.39	67.78 ± 95.55	11.61 ± 16.26	908.19 ± 1283.91	0.61 ± 0.86	48.35 ± 68.38	143.56 ± 202.73	462.52 ± 654.06	2.05 ± 2.90	0.00 ± 0.00	15.89 ± 22.46
5	Hair cream	79.50 ± 112.36	71.38 ± 100.84	10.18 ± 14.38	583.46 ± 824.84	1.43 ± 2.02	46.76 ± 66.08	156.82 ± 221.68	289.15 ± 408.48	1.84 ± 2.60	0.54 ± 0.76	16.23 ± 22.96
6	Shampoo	793.26 ± 1121.23	52.35 ± 74.01	7.29 ± 10.31	542.57 ± 767.24	1.16 ± 1.64	38.94 ± 55.07	116.13 ± 164.22	190.42 ± 269.26	1.47 ± 2.08	0.20 ± 0.28	13.27 ± 18.76
7	Facial cleanser	48.63 ± 68.17	56.58 ± 79.99	7.52 ± 10.63	454.55 ± 642.76	1.03 ± 1.46	36.60 ± 51.75	121.87 ± 172.35	201.40 ± 284.78	1.33 ± 1.87	0.00 ± 0.00	13.37 ± 18.91
8	Baby oil	26.81 ± 37.89	21.17 ± 29.88	2.62 ± 3.70	193.16 ± 272.97	0.08 ± 0.12	13.16 ± 18.59	41.72 ± 58.91	71.68 ± 101.33	0.15 ± 0.22	0.23 ± 0.33	5.52 ± 7.81
9	Baby powder	3774.14 ± 5283.13	39.08 ± 54.96	373.94 ± 528.79	3873.85 ± 5302.39	1.83 ± 2.53	24.67 ± 34.61	79.52 ± 112.09	139.13 ± 196.62	1.19 ± 1.67	0.38 ± 0.54	23.18 ± 31.82
10	Bar soap	60.94 ± 86.12	66.10 ± 93.37	7.42 ± 10.48	520.53 ± 735.81	0.49 ± 0.69	42.36 ± 59.87	143.62 ± 202.96	376.24 ± 532.05	1.52 ± 2.15	0.45 ± 0.63	16.11 ± 22.78
11	Hair dye	367.41 ± 519.57	105.00 ± 148.49	23.05 ± 32.59	1197.86 ± 1694.04	1.69 ± 2.39	67.83 ± 95.92	224.09 ± 316.90	672.19 ± 950.61	1.48 ± 2.10	0.87 ± 1.23	28.83 ± 40.77
12	Makeup powder	32775.71 ± 46350.53	56.42 ± 79.68	433.145 ± 610.27	111077.89 ± 157087.28	22.71 ± 32.04	63.27 ± 89.34	44.28 ± 62.50	70.90 ± 100.15	9.24 ± 13.05	0.08 ± 0.11	14.56 ± 20.55
13	Makeup mousse cream	3993.12 ± 5646.54	25.14 ± 35.38	45.31 ± 64.03	47706.29 ± 67465.70	3.36 ± 4.75	19.37 ± 27.32	48.62 ± 68.74	186.20 ± 263.25	3.81 ± 5.39	0.33 ± 0.47	6.89 ± 9.73
14	Mascara	42.77 ± 60.39	20.92 ± 29.41	1437.91 ± 2032.44	273733.78 ± 387116.45	5.81 ± 8.22	16.42 ± 23.13	40.17 ± 56.79	119.77 ± 168.91	3.74 ± 5.29	0.05 ± 0.07	5.27 ± 7.42
15	Deodorant	38170.15 ± 53979.29	119.90 ± 169.41	17.94 ± 25.30	1310.63 ± 1851.15	2.74 ± 3.87	127.16 ± 179.73	181.89 ± 257.19	275.12 ± 389.01	81.21 ± 114.83	306.92 ± 430.10	21.31 ± 30.11
16	Oil (silicone hair serum per Table 1 constituents)	1.83	0.05	0.00	0.19	0.00	0.02	0.06	0.01	0.00	0.00	0.00
17	Shaving cream	47.06 ± 66.25	15.28 ± 21.53	4.74 ± 6.60	180.90 ± 255.72	0.10 ± 0.14	13.46 ± 19.02	35.50 ± 50.20	88.13 ± 124.63	0.51 ± 0.72	0.09 ± 0.12	4.73 ± 6.70
18	Lipstick	15281.48 ± 21596.69	22.12 ± 31.23	92.42 ± 130.70	68523.48 ± 96906.62	1.25 ± 1.77	16.54 ± 23.37	31.79 ± 44.89	83.07 ± 117.47	4.52 ± 6.39	0.24 ± 0.34	4.96 ± 7.01
19	Toothpaste A (NaF / hydrated silica)	574060.74 ± 811820.98	1610.64 ± 2277.26	257.51 ± 364.11	16826.03 ± 23791.75	17.65 ± 24.95	970.83 ± 1372.89	264.00 ± 373.30	556.67 ± 787.23	221.96 ± 313.90	9.19 ± 13.00	75.86 ± 107.27
20	Toothpaste B (CaCO₃ / monofluorophosphate)	479717.95 ± 678423.33	878.37 ± 1242.06	334.78 ± 473.44	15603.46 ± 22066.21	18.29 ± 25.86	600.37 ± 849.01	269.94 ± 381.66	1114.37 ± 1575.89	209.33 ± 296.04	8.80 ± 12.44	78.31 ± 110.75
21	Body lotion	112.90 ± 159.50	65.99 ± 93.32	14.54 ± 20.56	736.09 ± 1040.89	0.98 ± 1.39	62.20 ± 87.95	155.36 ± 219.70	207.94 ± 294.05	1.41 ± 2.00	0.61 ± 0.87	19.61 ± 27.73

Exceedances per the source’s text discussion (p. 8-9)

The source’s Table 4 caption indicates gray-highlighted cells “exceeded the US FDA limits,” but the cell-level gray styling in the PDF table is difficult to verify cell-by-cell at the published image resolution and the discussion text appears to mix FDA and USP <232> criteria. The authoritative per-analyte exceedance list is therefore drawn from the source’s discussion text (p. 8-9), which states the specific samples that exceeded each limit:

Analyte	Limit cited	Source-text-stated exceedances (sample IDs)	Maximum value in panel (ppm)
Pb	20 ppm (US FDA cosmetic additive)	3, 9, 11, 15, 19, 20 (“6 of the tested cosmetic products exceeded this limit”)	78.31 (sample 20, toothpaste B)
tAs	3 ppm (US FDA)	12, 13, 14, 15, 18, 19, 20	221.96 (sample 19, toothpaste A)
Cr	50 ppm (US FDA)	“most of our tested cosmetic products exceeded this finding” — per Table 4 values: samples 1, 2, 3, 4, 5, 6, 7, 10, 11, 12, 15, 19, 20, 21 (14 of 21)	1610.64 (sample 19, toothpaste A)
Cd	0.5 ppm (USP <232>)	1, 15, 19, 20 (text-stated; per Table 4 values, samples 5 [0.54], 11 [0.87], and 21 [0.61] also exceed this cutoff but are not listed in the source’s discussion)	306.92 (sample 15, deodorant)
Co	5 ppm (USP <232>)	12, 14, 19, 20	22.71 (sample 12, makeup powder)
Ni	20 ppm (USP <232>)	“most of the tested cosmetic products in this study exceeded this limit” — per Table 4 values: samples 1, 2, 3, 4, 5, 6, 7, 9, 10, 11, 12, 15, 19, 20, 21 (15 of 21)	970.83 (sample 19, toothpaste A)
Cu	300 ppm (USP <232>)	“The only metal level that did not exceed the USP limit in any of the tested samples was Cu.”	269.94 (sample 20, toothpaste B; below limit)
Al, Fe, Zn, Mn	No FDA or USP limit cited by the source	n/a	Al 574,060 / Fe 273,733 / Zn 1114 / Mn 1437 — all in toothpaste, makeup, or mascara

Headline panel-level findings (authors’ synthesis, p. 8-9)

Lead: The source states 6 of 21 products exceeded the US FDA 20 ppm Pb-in-cosmetic-additives limit; samples 3, 9, 11, 15, 19, 20 carry Pb > 20 ppm. The maximum value was 78.31 ppm in the calcium-carbonate toothpaste (sample 20). The author-cited prior literature reports 23.57 ppm Pb in toothpaste by atomic absorption spectrometry (Orisakwe et al. 2016, ref 29) — the source contextualizes its toothpaste-Pb findings against that prior measurement.
Arsenic (tAs): US FDA limit cited as 3 ppm. Exceeded in seven samples (makeup powder 9.24, makeup mousse cream 3.81, mascara 3.74, deodorant 81.21, lipstick 4.52, both toothpastes 221.96 and 209.33). Maximum was 221.96 ppm in the sodium-fluoride/hydrated-silica toothpaste (sample 19).
Chromium: US FDA limit cited as 50 ppm. The source narrative states “most of our tested cosmetic products exceeded this finding”; per the Table 4 values 14 of 21 products carry Cr > 50 ppm. Maximum was 1610.64 ppm in sample 19 (toothpaste). The authors note an external comparison value of 7000 ppm Cr in eyeshadow from a separate Saudi-market study (ref 34).
Cobalt: USP <232> limit cited as 5 ppm. Source-text-stated exceedances: samples 12 (22.71), 14 (5.81), 19 (17.65), and 20 (18.29).
Nickel: USP <232> limit cited as 20 ppm. The source narrative states “most of the tested cosmetic products in this study exceeded this limit”; per the Table 4 values 15 of 21 products carry Ni > 20 ppm. Maximum 970.83 ppm in sample 19 (toothpaste).
Cadmium: USP <232> limit cited as 0.5 ppm. Source-text-stated exceedances: samples 1 (sunblock cream 4.40), 15 (deodorant 306.92), 19 (toothpaste 9.19), and 20 (toothpaste 8.80). The discussion does not list samples 5 (0.54), 11 (0.87), or 21 (0.61) despite Table 4 values that nominally exceed the 0.5 ppm cutoff; this is a paper-internal inconsistency between the discussion narrative and the table values.
Copper: USP <232> limit cited as 300 ppm. The paper states copper was below the limit in every tested product; the maximum reported value in Table 4 (sample 20 toothpaste, 269.94 ppm) is below that ceiling.
Al, Fe, Zn, Mn: No FDA or USP limit cited by the authors. The highest values for all four were concentrated in the toothpaste samples (Al up to 574,060 ppm in sample 19, equivalent to ~57% Al by mass), in mascara (Fe up to 273,733 ppm in sample 14, equivalent to ~27% Fe by mass), and in makeup powder (Fe up to 111,077 ppm in sample 12, ~11% Fe by mass). The Al and Fe magnitudes are consistent with the abrasive/pigment ingredients in the corresponding formulations (hydrated silica, calcium carbonate, iron-oxide colorants in mascara and powder makeup), but the authors do not separate inert filler from soluble impurity in their reporting and do not address the analytical implication that ICP-MS quantitation of major matrix elements is operating well outside the 5-150 ppb calibration range. See Verification notes.

Microbial findings (briefly; not the primary HMTc-relevant data)

Fourteen of 21 products had aerobic-plate-count microbial loads above the FDA 500 cfu/g eye-area limit for children under 3 years (SCCS limit 100 cfu/g for eye area or children under 3); baby oil reached 1471.5 cfu/ml and hair serum 1128 cfu/ml. Staphylococcus aureus was the dominant isolate (12 of 14 contaminated products), with Bacillus species the second most common (10 of 14).

Organoleptic findings

All samples passed visual inspection (no lumps, grit, discoloration, characteristic odor). pH ranged 6.90-8.10 across the panel.

Methods

Product selection. A closed- and open-ended consent-bearing online questionnaire was distributed via social media in Saudi Arabia to identify frequently used cosmetic and personal-care products. 447 responses were collected (88.8% female, 68.9% aged 19-40 years, 70.2% bachelor’s degree). 21 product categories were selected from the survey results, and one representative commercial sample per category was purchased from local markets and drug stores in Riyadh (two samples were purchased for toothpaste, representing two different abrasive formulations). The institutional review board at Princess Nourah bint Abdulrahman University granted exempt status (Registration 21-0330).

Sample preparation and digestion (heavy metals). Sample preparation followed the elemental-impurity procedure in USP General Chapter <232> in USP 40 / NF 35 (ref 15). Each sample was first digested in a concentrated-acid mixture using closed-vessel digestion. The acid mixture ratios (HNO₃ 69%, H₂O₂ 30%, HF 37%) and sample masses varied by product matrix and are reported individually in Table 2 of the source. The patterns: samples 1-7, 10, 11, 15, and 21 (14 products) used 10 mL HNO₃ alone at sample masses of 250-500 mg; samples 8 and 9 (baby oil, baby powder) used HNO₃ + H₂O₂ (5+3 mL and 7+2 mL respectively) without HF; samples 12-14 and 18 (makeup powder, makeup mousse cream, mascara, lipstick) used 6 mL HNO₃ + 0.50 mL HF; samples 16 and 17 (silicone hair serum labeled “Oil”, shaving cream) used 10 mL HNO₃ + 2 mL H₂O₂; samples 19 and 20 (toothpastes) used 2 mL HNO₃ + 2 mL HF at 200 mg sample mass — the lowest sample mass and highest fluoride load in the panel.

Instrumentation (heavy metals). Inductively coupled plasma mass spectrometry (ICP-MS). The make/model of the ICP-MS instrument is not reported. Calibration was performed against eleven calibration standards prepared at blank, 5, 15, 30, 70, and 150 ppb from a multi-element standard. Coefficients of determination for all eleven elements ranged from 0.9995 to 1. The reported elements were Al, Fe, Zn, Cr, Mn, Cu, Ni, As, Pb, Cd, and Co. Mercury and beryllium are listed in the source’s abbreviations but were not measured.

Microbial analysis. Aerobic plate count via serial dilution on nutrient agar (24 h incubation, 24 °C). Differential identification of growth used MacConkey, potato dextrose, cetrimide, Simmons’ citrate, and mannitol salt agars; S. aureus / Bacillus / fungal status was assessed by colony morphology and growth on selective media.

Organoleptic and pH testing. Visual inspection for lumps, color, odor; texture by skin rub; pH measured with a digital pH meter.

Method-quality reporting gaps. The paper does not report: instrument make/model, certified reference material recoveries, method blanks, procedural-blank limits of detection/quantification, within-run replicate structure for individual product values (the reported ± SD almost certainly represents within-replicate variation on the single product sample rather than between-product variation, but the source does not specify the replicate count), spike recovery, or duplicate sample preparation. The standard-deviation magnitudes reported (often equal to or larger than the corresponding mean) are unusually large for ICP-MS at the concentrations reported and are flagged in Verification notes as a methodological caveat to factor into evidence weighting.

Implications

Certification. The paper contributes to the Cat 2 Children Personal Care occurrence pool for several finished-product categories present in the panel (notably baby oil, baby talcum powder, sunblock cream, lipstick, lotion). Outside Cat 2, the toothpaste results (samples 19 and 20) feed the Cat 5 / oral care category pool with Pb, As, Cr, Ni, Co, and Cd values that substantially exceed both US FDA cosmetic limits and USP <232> oral-route acceptance criteria. The n=1 per product category limits the standalone weight of any single value; the paper is best treated as a contributing data point that pools with other Saudi-market and global cosmetics surveys for the per-row marginal distributions, with B-tier weighting reflecting the n=1 sample structure and the analytical-QC reporting gaps noted in Methods. The deodorant Cd reading (306.92 ppm) is a substantial outlier that warrants flagging for explicit treatment in any synthesis pass; ICP-MS Cd at that concentration in a finished cosmetic is unusual and could be either a true high-contamination finding (e.g., colored or pigmented additive) or a reporting/calibration artifact.

Courses. Useful as a worked example of how a survey-driven product-panel design generates representative-but-low-replication occurrence data for cosmetic regulatory courses. The author-cited cross-comparison values (Orisakwe 2016 Pb 23.57 ppm in toothpaste, ref 34 Cr 7000 ppm in eyeshadow, ref 13 Co in Saudi cosmetics) provide a teachable example of where this study sits in the broader literature.

App. Relevant for the consumer-facing app’s adult-cosmetics scoring layer (sun protection, lip care, hair care, oral care, deodorant). Baby oil and baby talcum powder results extend the pediatric layer. Per CLAUDE.md Part 12 the app must not surface brand-level rankings; the per-product values reported here are linked to product categories (sample types), not to the manufacturer brand the authors purchased — and consistent with that, this source page reports per-category values, not per-brand.

Microbiome. Not the primary focus, but the high S. aureus and Bacillus prevalence across the panel is a relevant background signal for any future federation work on skin- and oral-microbiome interaction with cosmetic preservative systems.

Wiki pages this source may touch

Verification notes

2026-05-18 fresh ingest (Claude Opus 4.7, autonomous v2.0 manual-fetch skill). New page; no prior wiki revision. DOI grep, raw_handle grep, and almukainzi2022 cite-key grep against wiki/sources/ all returned zero matches before ingest. (The cite “Almukainzi et al. 2022” appears in the body of chengappa2025-toothpaste-india-aas.md as a cross-reference to this paper’s Saudi toothpaste values; that is a reference within another source’s narrative, not a wiki source page for this paper.)
Brand-firewall compliance (Part 12 strict, 2026-05-17 lock). The PDF does not name the brand of any of the 21 tested products — the source describes them only by category (sunblock cream, BB cream, etc.) and by the ingredient list printed on each package (Table 1 of the source). No brand attribution had to be stripped from this ingest. Product-form descriptors used in the Key numbers table (“Toothpaste A (NaF / hydrated silica)” and “Toothpaste B (CaCO₃ / monofluorophosphate)”) distinguish the two toothpaste samples by their declared abrasive system, which is a formulation descriptor disclosed in the source’s Table 1 ingredient lists, not a brand identifier.
Wiki/HMTc firewall (Part 2). No synthesis-across-papers claims, no HMTc threshold proposals, no consumer risk advisories appear in this page. The author-cited prior-literature comparison values (Orisakwe 2016 toothpaste Pb, ref 34 eyeshadow Cr) are reported as part of the source’s own narrative inside Key numbers per Part 13 evidence-reporting practice, not as a cross-source synthesis by this wiki.
Lip balm, bar soap, mineral/paraffin oil — product-page gaps. The 21-product panel includes a lip balm (sample 3, adult formulation, not a children’s lip balm), a bar soap (sample 10), and a paraffin/mineral oil (sample 16) that do not map cleanly to existing product pages: the current taxonomy has childrens-lip-balm-plain and childrens-lip-balm-mineral-bearing but no adult-lip-balm slug; hand-soap is liquid-specific; and there is no general mineral-oil slug. These three product values are reported in the Key numbers table by sample but are not declared in frontmatter products:. The routing layer will surface the lip balm to [[products/lipstick]] via matrices fan-out only if the routing rules permit that cross-category fan-out; if not, lip balm coverage from this paper depends on either a new adult-lip-balm slug (Step 0 Lock required) or extension of the children’s-lip-balm slug to cover adult. Flagged here for Karen’s routing/taxonomy attention but not blocking the ingest.
Hg and Be listed but not measured. The source’s abbreviations table lists “Hg = Mercury” and “Ba = Beryllium” but Table 4 reports only eleven elements (Al, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Cd, Pb). Mercury and beryllium are referenced in the introduction as analytes of concern in cosmetics generally but were not part of the ICP-MS panel actually quantified. metals: frontmatter reflects what was measured (no tHg, no MeHg, no Be).
Arsenic speciation. ICP-MS without prior chromatographic separation measures total arsenic. Reported as tAs per the speciation convention (Part 14); the source itself uses “As” (interpreted as total arsenic).
Chromium speciation. ICP-MS without prior chromatographic separation measures total Cr. Reported as Cr (total chromium), not Cr-VI. The source’s discussion compares against the US FDA “Cr 50 ppm” cosmetics limit (which is also a total-Cr limit), consistent with this interpretation.
Large standard-deviation magnitudes are paper-internal, not a transcription error. Across nearly all Table 4 entries the reported SD is approximately 1.414 × the corresponding mean (√2 ratio, suggestive of n=2 replicates with one near-zero reading), and most entries report SD substantially larger than the mean (CV ≈ 141%). Examples: BB cream Al mean 12042 ± SD 17028 (CV 141.4%), mascara Fe mean 273,733 ± SD 387,116 (CV 141.4%), makeup powder Fe mean 111,077.89 ± SD 157,087.28 (CV 141.4%), deodorant Cd mean 306.92 ± SD 430.10 (CV 140.1%). The SD/mean ≈ √2 pattern is the source’s reporting convention for the heavy-metals values, consistent with two-replicate measurements where the SD of a {0, 2·mean} pair equals exactly √2 · mean. The source does not state replicate count for the heavy-metals measurements; values are transcribed as printed, and the SD convention is documented here so future synthesis sessions can weight these point estimates accordingly.
Toothpaste Al values are at percent-of-mass scale. Sample 19 (toothpaste, hydrated silica formulation) reports Al = 574,060.74 ppm = ~57.4% Al by mass; sample 20 (toothpaste, calcium carbonate formulation) reports Al = 479,717.95 ppm = ~47.9% Al by mass. The Table 1 ingredient lists for these toothpastes do not include aluminum hydroxide or any other aluminum-bearing ingredient (only sodium fluoride 0.32%, hydrated silica, and excipients for sample 19; calcium carbonate, hydrated silica, and excipients for sample 20). Aluminum at that concentration would be implausible in those formulations on first-principles grounds; the most likely explanations are (a) calibration extrapolation orders of magnitude beyond the 5-150 ppb calibration range causing systematic over-reporting of major matrix elements after acid digestion, (b) silica matrix interference on Al ICP-MS m/z = 27 from ²⁸Si or polyatomic ⁷Li¹⁶O⁺ / ¹²C¹⁵N⁺, or (c) a reporting/unit error in the source. The source does not flag or discuss this; the values are reported as-printed and the methodological caveat is documented here.
Sample 16 product-label discrepancy. Table 1 of the source labels sample 16 as “Oil” with constituents “Cyclopentasiloxane, dimethiconol, fragrance ingredient, ethylhexyl methoxycinnamate, BHT” — those constituents are silicone-fluid + UV-absorber chemistry typical of a leave-on hair serum, not paraffin/mineral oil (sample 8 “Baby oil” has the actual paraffin in its constituent list: “Paraffin oil light, perfume”). The discussion text on p. 7 calls sample 16 “hair oil serum” with microbial load 1128 cfu/ml. This source page labels sample 16 as “Oil (silicone hair serum per Table 1 constituents)” to capture both the source’s Table 1 label and the discussion’s descriptor.
Sample 16 reports zeros without ± SD. Table 4 shows seven of eleven analytes at “0.00” with no ± SD term, alongside Al 1.83, Cr 0.05, Cu 0.06, and Fe 0.19 with similarly suppressed SD. This is consistent with the paper having reported the silicone-hair-serum values as point measurements (likely below LOQ for most analytes) rather than mean-of-replicates; transcribed as displayed.
License. “Open Access” Creative Commons Attribution 4.0 International License declared explicitly on the article’s last page. license: CC BY 4.0.
DOI verified. DOI 10.1038/s41598-022-21771-7 printed on every page of the article and on the last-page copyright block.
Authors verified against PDF byline. May Almukainzi, Lubna Alotaibi, Anfal Abdulwahab, Nada Albukhary, Areej M. El Mahdy. Corresponding author M. Almukainzi (mkalmukainzi@pnu.edu.sa). Institutional affiliation: Department of Pharmaceutical Sciences and College of Pharmacy, Princess Nourah bint Abdulrahman University (Riyadh, Saudi Arabia); Department of Microbiology and Immunology, Faculty of Pharmacy, Mansoura University (Mansoura, Egypt).
Audit subagent (2026-05-18, general-purpose fresh-context, v2.0 skill Phase 2) — five corrections applied. After the first ingest pass, the audit was re-validated cell-by-cell against the pdftotext -layout extraction of page 8 of the source PDF (the layout-mode extraction preserves cell alignment unambiguously). The audit subagent flagged Table 4 numerical fidelity at ⚠️/❌ on several cells; independent verification against the pdftotext output confirmed and corrected the following:
1. Sample 12 Fe mean corrected from “11077.89” to “111077.89” (a missing leading “1” digit). The SD 157087.28 IS correct as-printed; SD/mean ratio = 1.4142 (√2), consistent with the table-wide reporting convention. This corrects the audit subagent’s hypothesis that the SD should be ~15708; the discrepancy was in the mean, not the SD.
2. Sample 12 Mn mean corrected from “433.04” to “433.145” (paper-internal three-decimal-place reporting in this one cell; the table convention is two decimal places elsewhere, transcribed as printed).
3. Sample 18 (lipstick) Co SD corrected from “2.79” to “1.77” (Co value 1.25 × √2 = 1.768, consistent with the SD convention).
4. Sample 1 (sunblock) Ni SD corrected from “88.25” to “88.27” (pdftotext-verified).
5. Sample 3 (lip balm) Ni SD corrected from “58.07” to “57.98” (pdftotext-verified).
Audit subagent — Methods digestion description corrected. The first ingest pass said the H₂O₂ digestion group was “samples 16-17 (oil, shaving cream) and sample 21 (body lotion)“. Per Table 2 of the source, sample 21 (body lotion) uses HNO₃ alone (10 mL, no H₂O₂), and samples 8 (baby oil, 5 mL HNO₃ + 3 mL H₂O₂) and 9 (baby powder, 7 mL HNO₃ + 2 mL H₂O₂) DO use H₂O₂. The Methods paragraph has been rewritten to reflect Table 2 exactly: samples 8, 9, 16, 17 are the H₂O₂ group.
Audit subagent — Table 4 bolding strategy reformed. The first ingest pass attempted to reproduce the source’s Table 4 gray-cell highlighting by bolding cells in the wiki table. The audit flagged that the source’s discussion text and the cell-level highlighting do not always agree (e.g., the discussion’s Cd-exceedance list does NOT include samples 5 [0.54], 11 [0.87], or 21 [0.61] despite those values exceeding the 0.5 ppm USP cutoff); the audit also pushed back on bolding sample 9 Cr (39.08 < 50 FDA cutoff). On verification this is correct — the source’s text and table convey different exceedance maps. The wiki has been revised to (a) present Table 4 values plain (no bold), and (b) add an explicit “Exceedances per the source’s text discussion” subsection that lists the source-text-stated exceedances per analyte and notes where the table values disagree with the discussion. This is the honest representation: the source is internally inconsistent on the Cd exceedance count, and the wiki documents that rather than papering over it.
Audit subagent — Check 2 finding rejected (false positive on taxonomy slugs). The audit flagged ten of the fourteen products: slugs in frontmatter (sun-suntan-products, makeup-foundation-powders-blush, body-hand-leave-on-skin-care, leave-on-hair-preparations, shampoo-adult, facial-cleansers, permanent-hair-dyes-tints, eye-makeup, deodorants, shaving-cream-gel-foam) as not appearing in docs/gpt-collaboration/taxonomy-snapshot.md. Independent verification: all ten slugs exist as actual wiki product pages in wiki/products/ (confirmed by directory listing; the directory holds 297 product pages). The taxonomy snapshot doc is a curated subset focused on Cat 1 + a few personal-care categories, not the full taxonomy. The routing audit (npm run evidence:source-routes) generated 13 routing rows against the declared product slugs with zero unresolved entries, confirming all slugs are recognized by the routing layer. The audit subagent’s Check 2 ❌ was based on an outdated/partial reference doc, not on the actual taxonomy.
Audit subagent — borderline Check 5 phrasing softened. The first ingest pass said “the toothpaste-as-lead-source pattern is not unique to this study,” which the audit flagged as borderline cross-source synthesis in the wiki voice. Reworded to “the source contextualizes its toothpaste-Pb findings against that prior measurement,” which attributes the cross-comparison to the source itself rather than asserting it as wiki synthesis.
Audit subagent — Microbial 1467.5 vs 1471.5 abstract/Table-3 inconsistency noted. The audit observed that the source’s abstract reports the microbial-load range as “between 1467.5 and 299.5 cfu/ml” while Table 3 shows baby oil at 1471.5 as the actual maximum (1467.5 is the second-highest, hair cream sample 5). The wiki body text quotes Table 3 (1471.5 baby oil), not the abstract, so no transcription correction is needed; flagged here as a paper-internal abstract/table inconsistency for any future synthesis pass that may want to reconcile it.
Audit subagent — Sample 16 paraffin-vs-silicone-serum label discrepancy applied. Audit flagged that Table 1 sample 16 constituents are silicone-based (cyclopentasiloxane/dimethiconol/ethylhexyl methoxycinnamate/BHT), not paraffin, despite the Table 1 product-type label “Oil” and despite the discussion text on p. 7 calling it “hair oil serum.” First-pass wiki label “Oil (paraffin)” corrected to “Oil (silicone hair serum per Table 1 constituents)” and added to a dedicated Verification note above. Sample 8 (“Baby oil”) IS the paraffin (constituents “Paraffin oil light, perfume”).

Ingest log

2022-10-31 (publication): Scientific Reports 12:18299. Received 29 July 2022, accepted 30 September 2022, published 31 October 2022. CC BY 4.0.
2026-05-18 fresh ingest (Claude Opus 4.7, autonomous v2.0 manual-fetch skill): NEW path. Three identity checks against wiki/sources/ (DOI grep, raw_handle grep, cite-key grep) returned zero matches. PDF read in 11-page chunks via the pages parameter; all of abstract, methods, Tables 1-4, Figure 3 (microbial), Figure 4 (graphical summary), discussion, and references read in full. Source page written, routing audit refreshed (13 routing rows, 0 unresolved, advisory-only malformed for missing optional ingredients:), ingest committed, audit queued.
2026-05-18 Phase 2 audit (fresh-context Agent subagent, general-purpose): Verdict REVISE. Five Check-1 numerical-fidelity findings flagged; on independent verification against pdftotext extraction of page 8, all five Table 4 transcription errors confirmed and corrected (sample 12 Fe mean digit-drop, sample 12 Mn precision, sample 18 Co SD, sample 1 and 3 Ni SD rounding). Check-3 Methods H₂O₂ grouping error confirmed against Table 2 and corrected. Check-2 ❌ on product slugs rejected as false positive (taxonomy-snapshot.md is a curated subset; all slugs exist in wiki/products/ and resolve in the routing layer). Check-5 borderline phrasing softened. Bolding strategy reformed to remove unverifiable cell-level highlighting reproduction and replaced with text-derived exceedance summary. All paper-internal inconsistencies (Cd-discussion vs Cd-table, abstract microbial range vs Table 3, sample 16 paraffin vs silicone label, Mn three-decimal-place outlier) flagged in Verification notes for downstream synthesis.

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.

Commit	Date	Description
ce3e07c	2026-05-28	activation \| Vercel DATACITE env slots set, curators.md filled with founder entry + six scoped reviewer invitations, peer-review onboarding playbook drafted
51400b9	2026-05-28	audit-queue: gasparik2017-wild-boar-slovakia-metals audited-revised

Heavy Metal Index

Index