Heavy Metal Index
Ask the IndexAboutIndex Pro
Spinach, kale, collard greens, romaine and iceberg lettuce illustrating kale ingredient context.

Heavy metals in Kale

Lead, cadmium, total arsenic, inorganic arsenic and total mercury in kale: monitoring results, studies, references by form and basis, limits, testing and controls.

Cite this pageVersion history
Quick read

Kale in the HMI library has 117 reported results for lead, cadmium, total arsenic, inorganic arsenic, total mercury, nickel, chromium, aluminium, tin, antimony, uranium, Co, Cu, Fe, Mn, Zn, V, Se, arsenic (As), Ba, mercury (Hg) and Mo from 11 sources.1World Health Organization · 2026GEMS/Food Contaminants database heavy-metal exportsGEMS/Food Contamination Monitoring and Assessment ProgrammeRead source →2U.S. Food and Drug Administration · 2022FY2018-FY2020 TDS Elements Analytical ResultsFDA Total Diet StudyRead source →3Banyen S · 2013Anodic Stripping Voltammetry for Simultaneous Determination of Heavy Metals in VegetablesKMITL Science Journal (Ladkrabang) 22(1), January-June 2556 B.E., pp. 84-99Read source →4Kim BF, Lupolt SN, Santo RE, Bachman G, Zhu X, Yang T, Fukagawa NK, Richardson ML, Green C, Phillips KM, Nachman KE · 2024Nutrients and non-essential metals in darkibor kale grown at urban and rural farms: a pilot studyPLOS ONERead source →5Norton GJ, Deacon CM, Mestrot A, Feldmann J, Jenkins P, Baskaran C, Meharg AA · 2015Cadmium and lead in vegetable and fruit produce selected from specific regional areas of the UKThe Science of the Total Environment 533, 520-527 (2015); accepted author manuscriptRead source →6Saleem M, Wang Y, Pierce D, Sens DA, Somji S, Garrett SH · 2025Concentration and Potential Non-Carcinogenic and Carcinogenic Health Risk Assessment of Metals in Locally Grown VegetablesFoodsRead source →

Kale overview

Kale in the HMI library has 117 reported results for lead, cadmium, total arsenic, inorganic arsenic, total mercury, nickel, chromium, aluminium, tin, antimony, uranium, Co, Cu, Fe, Mn, Zn, V, Se, As, Ba, Hg and Mo from 11 sources.1World Health Organization · 2026GEMS/Food Contaminants database heavy-metal exportsGEMS/Food Contamination Monitoring and Assessment ProgrammeRead source →2U.S. Food and Drug Administration · 2022FY2018-FY2020 TDS Elements Analytical ResultsFDA Total Diet StudyRead source →3Banyen S · 2013Anodic Stripping Voltammetry for Simultaneous Determination of Heavy Metals in VegetablesKMITL Science Journal (Ladkrabang) 22(1), January-June 2556 B.E., pp. 84-99Read source →4Kim BF, Lupolt SN, Santo RE, Bachman G, Zhu X, Yang T, Fukagawa NK, Richardson ML, Green C, Phillips KM, Nachman KE · 2024Nutrients and non-essential metals in darkibor kale grown at urban and rural farms: a pilot studyPLOS ONERead source →5Norton GJ, Deacon CM, Mestrot A, Feldmann J, Jenkins P, Baskaran C, Meharg AA · 2015Cadmium and lead in vegetable and fruit produce selected from specific regional areas of the UKThe Science of the Total Environment 533, 520-527 (2015); accepted author manuscriptRead source →6Saleem M, Wang Y, Pierce D, Sens DA, Somji S, Garrett SH · 2025Concentration and Potential Non-Carcinogenic and Carcinogenic Health Risk Assessment of Metals in Locally Grown VegetablesFoodsRead source → WHO GEMS/Food submissions reported lead in kale, as sold from 484 results in 4 reporting regions, with subgroup medians from 2 to 2,890 µg/kg.1World Health Organization · 2026GEMS/Food Contaminants database heavy-metal exportsGEMS/Food Contamination Monitoring and Assessment ProgrammeRead source → In the US Total Diet Study, kale, cooked had a median lead result below 4 µg/kg across 6 composites from 2019–2020.2U.S. Food and Drug Administration · 2022FY2018-FY2020 TDS Elements Analytical ResultsFDA Total Diet StudyRead source →

Across 4 other studies of ordinary farms and markets, reported lead means and medians on an as-sold or fresh-weight basis ranged from 3.2 µg/kg (Spungen et al. 2024) to 130 µg/kg (Kim et al. 2024).4Kim BF, Lupolt SN, Santo RE, Bachman G, Zhu X, Yang T, Fukagawa NK, Richardson ML, Green C, Phillips KM, Nachman KE · 2024Nutrients and non-essential metals in darkibor kale grown at urban and rural farms: a pilot studyPLOS ONERead source →7Spungen JH, et al. · 2024Infants’ and young children’s dietary exposures to lead and cadmium: FDA total diet study 2018-2020Food Additives & Contaminants: Part ARead source → For cadmium, 4 studies on an as-sold or fresh-weight basis reported means and medians from 6.5 µg/kg (Norton et al. 2015) to 30 µg/kg (Kim et al. 2024).4Kim BF, Lupolt SN, Santo RE, Bachman G, Zhu X, Yang T, Fukagawa NK, Richardson ML, Green C, Phillips KM, Nachman KE · 2024Nutrients and non-essential metals in darkibor kale grown at urban and rural farms: a pilot studyPLOS ONERead source →5Norton GJ, Deacon CM, Mestrot A, Feldmann J, Jenkins P, Baskaran C, Meharg AA · 2015Cadmium and lead in vegetable and fruit produce selected from specific regional areas of the UKThe Science of the Total Environment 533, 520-527 (2015); accepted author manuscriptRead source →

For a buyer the practical points are the form bought, the basis a result is reported on, where the kale came from and the laboratory’s reporting limit. Testing finds contamination but does not remove it.

What counts as kale

This page covers kale in 2 forms kept separate in every comparison: kale, as sold (108 results) and kale, cooked (9 results).1World Health Organization · 2026GEMS/Food Contaminants database heavy-metal exportsGEMS/Food Contamination Monitoring and Assessment ProgrammeRead source →2U.S. Food and Drug Administration · 2022FY2018-FY2020 TDS Elements Analytical ResultsFDA Total Diet StudyRead source →3Banyen S · 2013Anodic Stripping Voltammetry for Simultaneous Determination of Heavy Metals in VegetablesKMITL Science Journal (Ladkrabang) 22(1), January-June 2556 B.E., pp. 84-99Read source →4Kim BF, Lupolt SN, Santo RE, Bachman G, Zhu X, Yang T, Fukagawa NK, Richardson ML, Green C, Phillips KM, Nachman KE · 2024Nutrients and non-essential metals in darkibor kale grown at urban and rural farms: a pilot studyPLOS ONERead source →5Norton GJ, Deacon CM, Mestrot A, Feldmann J, Jenkins P, Baskaran C, Meharg AA · 2015Cadmium and lead in vegetable and fruit produce selected from specific regional areas of the UKThe Science of the Total Environment 533, 520-527 (2015); accepted author manuscriptRead source →6Saleem M, Wang Y, Pierce D, Sens DA, Somji S, Garrett SH · 2025Concentration and Potential Non-Carcinogenic and Carcinogenic Health Risk Assessment of Metals in Locally Grown VegetablesFoodsRead source →

Results stay on the basis each source reports: as sold or fresh weight, dry weight, or unstated. Dry-weight, as-sold and unstated results are never pooled, and no moisture conversion is applied.

Heavy metal contamination profile of kale

The table holds every extracted kale result: one row for each statistic a source reports for a form, group and metal, 117 rows in all, grouped by metal and form; percentiles, ranges and maxima for the same samples as a mean or median are shown on that row.1World Health Organization · 2026GEMS/Food Contaminants database heavy-metal exportsGEMS/Food Contamination Monitoring and Assessment ProgrammeRead source →2U.S. Food and Drug Administration · 2022FY2018-FY2020 TDS Elements Analytical ResultsFDA Total Diet StudyRead source →3Banyen S · 2013Anodic Stripping Voltammetry for Simultaneous Determination of Heavy Metals in VegetablesKMITL Science Journal (Ladkrabang) 22(1), January-June 2556 B.E., pp. 84-99Read source →4Kim BF, Lupolt SN, Santo RE, Bachman G, Zhu X, Yang T, Fukagawa NK, Richardson ML, Green C, Phillips KM, Nachman KE · 2024Nutrients and non-essential metals in darkibor kale grown at urban and rural farms: a pilot studyPLOS ONERead source →5Norton GJ, Deacon CM, Mestrot A, Feldmann J, Jenkins P, Baskaran C, Meharg AA · 2015Cadmium and lead in vegetable and fruit produce selected from specific regional areas of the UKThe Science of the Total Environment 533, 520-527 (2015); accepted author manuscriptRead source →6Saleem M, Wang Y, Pierce D, Sens DA, Somji S, Garrett SH · 2025Concentration and Potential Non-Carcinogenic and Carcinogenic Health Risk Assessment of Metals in Locally Grown VegetablesFoodsRead source → Monitoring medians are HMI calculations from individual submissions; non-detects remain intervals between zero and the reporting limit and are never read as zero.1World Health Organization · 2026GEMS/Food Contaminants database heavy-metal exportsGEMS/Food Contamination Monitoring and Assessment ProgrammeRead source →

36 rows can be rated against an HMI reference for the same form and basis, 0 rows have a direct same-study comparison, and the rest stay unrated with the reason recorded.

Comparison methods and scope

Each row is one statistic a source reports for one form of kale, group and metal; percentiles, ranges and maxima reported for the same samples as a mean or median are shown on that row. Different forms and preparation states are kept apart, and dry-weight results never share a reference with as-sold or fresh-weight results. Ratings place a result against an HMI reference built from monitoring results and ordinary-setting medians for the same form and basis. Same-study contrasts compare a contaminated or treated site with its control, targeted with random sampling, a prepared form with the unprocessed one, or one group of a study with the median of its other groups. Held rows failed a check against their own source and are shown but not compared.

About this table
Food formCollection scopeFindingComparisonEvidence
Kale findings from monitoring programmes, total diet studies and every extracted field, market and contaminated-site study, in µg/kg with each source’s form, basis, statistic and censoring retained.
Food formCollection scopeFinding (µg/kg)ComparisonEvidence
19 findings · 7 studies · 3 spread figures folded in
Kale, as soldWHO GEMS/FoodEURO2003–2019median.
Kale, as soldWHO GEMS/FoodEURO2009–2017median.
Kale, as soldWHO GEMS/FoodEURO2005–2018median.
Kale, as soldWHO GEMS/FoodPAHO2014–2016median.
Kale, as soldWHO GEMS/FoodSEARO2012–2013median.
Kale, as soldWHO GEMS/FoodWPRO2010–2011median.
Kale, as soldWHO GEMS/FoodWPRO2001median.
Kale, as soldBanyen 2013Hua Takhe fresh market, Bangkok2011mean.
Kale, as soldKim et al. 2024rural farms2019mean.
Kale, as soldKim et al. 2024urban farms (Baltimore City)2019mean.
Kale, as soldNorton et al. 2015NE Britain (Aberdeenshire) basket surveyCollection date unreportedmedian.
Kale, as soldNorton et al. 2015NE Britain (Aberdeenshire) basket surveyCollection date unreportedmean.Same samples: 18.7 µg/kg (minimum); 252 µg/kg (maximum)
Kale, as soldNorton et al. 2015SW Britain (Devon and Cornwall) basket surveyCollection date unreportedmedian.
Kale, as soldNorton et al. 2015SW Britain (Devon and Cornwall) basket surveyCollection date unreportedmean.Same samples: 29.3 µg/kg (maximum)
Kale, as soldNorton et al. 2015SW Britain (Devon and Cornwall) basket surveyCollection date unreportedmin.
Kale, as soldSaleem et al. 2025all samplesCollection date unreportedDry weightmean.
Kale, as soldWHO GEMS/FoodSEARO2012Basis not statedmedian.
Kale, as soldCentre for Food Safety (Hong Kong) 2013all samples2010–2011Basis not statedmean.
Kale, as soldNgodhe et al. 2018all plots, sludge-irrigated2018Basis not statedmean.
2 findings · 2 studies
Kale, cookedFDA Total Diet Study6 collection regions2019–2020median.
Kale, cookedSpungen et al. 2024all samples2018–2020mean.
19 findings · 7 studies · 3 spread figures folded in
Kale, as soldWHO GEMS/FoodEURO2003–2019median.
Kale, as soldWHO GEMS/FoodEURO2009–2017median.
Kale, as soldWHO GEMS/FoodEURO2003–2006median.
Kale, as soldWHO GEMS/FoodEURO2003–2018median.
Kale, as soldWHO GEMS/FoodPAHO2015–2016median.
Kale, as soldWHO GEMS/FoodWPRO2010–2011median.
Kale, as soldWHO GEMS/FoodWPRO2001median.
Kale, as soldBanyen 2013Hua Takhe fresh market, Bangkok2011mean.
Kale, as soldKim et al. 2024rural farms2019mean.
Kale, as soldKim et al. 2024urban farms (Baltimore City)2019mean.
Kale, as soldNorton et al. 2015NE Britain (Aberdeenshire) basket surveyCollection date unreportedmedian.
Kale, as soldNorton et al. 2015NE Britain (Aberdeenshire) basket surveyCollection date unreportedmean.Same samples: 8.9 µg/kg (minimum); 25.6 µg/kg (maximum)
Kale, as soldNorton et al. 2015SW Britain (Devon and Cornwall) basket surveyCollection date unreportedmedian.
Kale, as soldNorton et al. 2015SW Britain (Devon and Cornwall) basket surveyCollection date unreportedmean.Same samples: 39.2 µg/kg (maximum)
Kale, as soldNorton et al. 2015SW Britain (Devon and Cornwall) basket surveyCollection date unreportedmin.
Kale, as soldSaleem et al. 2025all samplesCollection date unreportedDry weightmean.
Kale, as soldCentre for Food Safety (Hong Kong) 2013all samples2010–2011Basis not statedmean.
Kale, as soldJoint FAO/WHO Expert Committee on Food Additives 2022GEMS/Food contaminants database2011Basis not statedmean.
Kale, as soldJoint FAO/WHO Expert Committee on Food Additives 2022GEMS/Food contaminants database2011Basis not statedmean.
2 findings · 2 studies
Kale, cookedFDA Total Diet Study6 collection regions2019–2020median.
Kale, cookedSpungen et al. 2024all samples2018–2020mean.
5 findings · 1 study
Kale, as soldWHO GEMS/FoodEURO2022median.
Kale, as soldWHO GEMS/FoodEURO2004–2023median.
Kale, as soldWHO GEMS/FoodEURO2008–2015median.
Kale, as soldWHO GEMS/FoodEURO2004–2006median.
Kale, as soldWHO GEMS/FoodWPRO2001median.
1 finding · 1 study
Kale, cookedFDA Total Diet Study6 collection regions2019–2020median.
6 findings · 2 studies
Kale, as soldWHO GEMS/FoodEURO2016median.
Kale, as soldWHO GEMS/FoodEURO2010median.
Kale, as soldWHO GEMS/FoodWPRO2010–2011median.
Kale, as soldWHO GEMS/FoodWPRO2000median.
Kale, as soldWHO GEMS/FoodEURO2010Dry weightmedian.
Kale, as soldCentre for Food Safety FEHD HKSAR 2012all samples (4 sampling occasions, composites)2010–2011Basis not statedmean.
3 findings · 1 study
Kale, as soldWHO GEMS/FoodEURO2004–2016median.
Kale, as soldWHO GEMS/FoodEURO2001median.
Kale, as soldWHO GEMS/FoodWPRO2001median.
1 finding · 1 study
Kale, cookedFDA Total Diet Study6 collection regions2019–2020median.
4 findings · 3 studies
Kale, as soldKim et al. 2024rural farms2019mean.
Kale, as soldKim et al. 2024urban farms (Baltimore City)2019mean.
Kale, as soldSaleem et al. 2025all samplesCollection date unreportedDry weightmean.
Kale, as soldCentre for Food Safety (Hong Kong) 2013all samples2010–2011Basis not statedmean.
1 finding · 1 study
Kale, cookedFDA Total Diet Study6 collection regions2019–2020median.
3 findings · 2 studies
Kale, as soldKim et al. 2024rural farms2019mean.
Kale, as soldKim et al. 2024urban farms (Baltimore City)2019mean.
Kale, as soldSaleem et al. 2025all samplesCollection date unreportedDry weightmean.
1 finding · 1 study
Kale, cookedFDA Total Diet Study6 collection regions2019–2020median.
1 finding · 1 study
Kale, as soldCentre for Food Safety (Hong Kong) 2013all samples2010–2011Basis not statedmean.
3 findings · 2 studies
Kale, as soldWHO GEMS/FoodWPRO2010–2011median.
Kale, as soldWHO GEMS/FoodWPRO2001median.
Kale, as soldCentre for Food Safety (Hong Kong) 2013all samples2010–2011Basis not statedmean.
1 finding · 1 study
Kale, as soldCentre for Food Safety (Hong Kong) 2013all samples2010–2011Basis not statednot detected.
2 findings · 1 study
Kale, as soldKim et al. 2024rural farms2019mean.
Kale, as soldKim et al. 2024urban farms (Baltimore City)2019mean.
1 finding · 1 study
Kale, cookedFDA Total Diet Study6 collection regions2019–2020median.
1 finding · 1 study
Kale, as soldSaleem et al. 2025all samplesCollection date unreportedDry weightmean.
10 findings · 4 studies
Kale, as soldBanyen 2013Hua Takhe fresh market, Bangkok2011mean.
Kale, as soldKim et al. 2024rural farms2019mean.
Kale, as soldKim et al. 2024urban farms (Baltimore City)2019mean.
Kale, as soldSaleem et al. 2025all samplesCollection date unreportedDry weightmean.
Kale, as soldNgodhe et al. 2018all plots, sludge-irrigated2018Basis not statedmean.
Kale, as soldNgodhe et al. 2018plot 12018Basis not statedmean.
Kale, as soldNgodhe et al. 2018plot 22018Basis not statedmean.
Kale, as soldNgodhe et al. 2018plot 32018Basis not statedmean.
Kale, as soldNgodhe et al. 2018plot 42018Basis not statedmean.
Kale, as soldNgodhe et al. 2018plot 52018Basis not statedmean.
3 findings · 2 studies
Kale, as soldKim et al. 2024rural farms2019mean.
Kale, as soldKim et al. 2024urban farms (Baltimore City)2019mean.
Kale, as soldSaleem et al. 2025all samplesCollection date unreportedDry weightmean.
3 findings · 2 studies
Kale, as soldKim et al. 2024rural farms2019mean.
Kale, as soldKim et al. 2024urban farms (Baltimore City)2019mean.
Kale, as soldSaleem et al. 2025all samplesCollection date unreportedDry weightmean.
7 findings · 4 studies
Kale, as soldBanyen 2013Hua Takhe fresh market, Bangkok2011mean.
Kale, as soldKim et al. 2024rural farms2019mean.
Kale, as soldKim et al. 2024urban farms (Baltimore City)2019mean.
Kale, as soldSaleem et al. 2025all samplesCollection date unreportedDry weightmean.
Kale, as soldNgodhe et al. 2018all plots, sludge-irrigated2018Basis not statedmean.
Kale, as soldNgodhe et al. 2018plot 2 (lowest)2018Basis not statedmean.
Kale, as soldNgodhe et al. 2018plot 5 (highest)2018Basis not statedmean.
3 findings · 2 studies
Kale, as soldKim et al. 2024rural farms2019mean.
Kale, as soldKim et al. 2024urban farms (Baltimore City)2019mean.
Kale, as soldCentre for Food Safety (Hong Kong) 2013all samples2010–2011Basis not statedmean.
1 finding · 1 study
Kale, as soldSaleem et al. 2025all samplesCollection date unreportedDry weightmean.
3 findings · 2 studies
Kale, as soldKim et al. 2024rural farms2019mean.
Kale, as soldKim et al. 2024urban farms (Baltimore City)2019mean.
Kale, as soldSaleem et al. 2025all samplesCollection date unreportedDry weightmean.
2 findings · 1 study
Kale, as soldKim et al. 2024rural farms2019mean.
Kale, as soldKim et al. 2024urban farms (Baltimore City)2019mean.
1 finding · 1 study
Kale, as soldSaleem et al. 2025all samplesCollection date unreportedDry weightmean.
2 findings · 1 study
Kale, as soldKim et al. 2024rural farms2019mean.
Kale, as soldKim et al. 2024urban farms (Baltimore City)2019mean.
Shared dataset methods and interpretation

Source summaries retain their original statistic, material, basis and censoring. Each source drawer identifies methods and overlapping samples. No matched reference or pooled distribution is inferred.

HMI kale Pb reference · Kale, as sold

Reference and scope

HMI kale Pb reference · Kale, as sold. Kale, as sold; As sold, as eaten or fresh weight, µg/kg; no moisture conversion. Distribution of subgroup median concentration values.

Reference (µg/kg)Q1MedianQ3
Primary HMI reference12.815–20 (bounded)102
Alternative weighting12.7–12.8 (bounded)13.9–16.8 (bounded)102

35 reference units; 2 sampling families. Counts describe evidence units, not independent studies or certainty.

Contributing familyReference unitsWeight
WHO GEMS/Food3350%
Norton et al. 2015250%

Methods

HMI pools subgroup medians for kale, as sold on a as sold, as eaten or fresh weight basis: monitoring results summarized as one median per reporting region, sampling type and collection year, total diet composites and printed sample values as single observations, and printed subgroup medians from ordinary farms or markets. Each sampling family has equal total weight. Non-detects stay as zero-to-limit intervals, so quartiles carry lower and upper bounds. Means, ranges and maxima never enter as observations. Other forms, other bases, targeted sampling and results from mining, industrial, wastewater-affected, experimental, fertiliser or manufacturer settings are kept out of this reference.

Inverse weighted empirical CDF; midpoint at exact cumulative-weight boundaries. At least four reference units and a definitely positive Q1–Q3 spread are required. This is a display eligibility rule, not statistical validation.

Limitations

2 sampling families with different designs; none is a probability sample of world supply. Reporting regions and purchase markets are not growing areas. Non-detects with differing limits widen the bounds. A position is descriptive concentration, not a safety or compliance judgment.

Admitted source families

Reference version and interpretation

Frozen evidence selection: 2026-10-22. Method: hmi-pooled-reference-v2.

hmi-pooled-reference-v2:kale-1aad1360db42:40ef46a379b0

The reference is computed by HMI from admitted evidence, not copied from a study baseline. It describes the evidence currently in this scope; it is not a worldwide or current-lot benchmark. Source, year and region filters do not change it. Different HMI scopes are not one common ranking.

HMI kale Pb reference · Kale, as sold, basis not stated

Reference and scope

HMI kale Pb reference · Kale, as sold, basis not stated. Kale, as sold; Basis not stated by the source, µg/kg; no moisture conversion. Distribution of subgroup median concentration values.

Reference (µg/kg)Q1MedianQ3
Primary HMI reference135135135
Alternative weighting135135135

1 reference units; 1 sampling family. Counts describe evidence units, not independent studies or certainty.

Contributing familyReference unitsWeight
WHO GEMS/Food1100%

Methods

HMI pools subgroup medians for kale, as sold on a basis not stated by the source basis: monitoring results summarized as one median per reporting region, sampling type and collection year, total diet composites and printed sample values as single observations, and printed subgroup medians from ordinary farms or markets. Each sampling family has equal total weight. Non-detects stay as zero-to-limit intervals, so quartiles carry lower and upper bounds. Means, ranges and maxima never enter as observations. Other forms, other bases, targeted sampling and results from mining, industrial, wastewater-affected, experimental, fertiliser or manufacturer settings are kept out of this reference.

Inverse weighted empirical CDF; midpoint at exact cumulative-weight boundaries. At least four reference units and a definitely positive Q1–Q3 spread are required. This is a display eligibility rule, not statistical validation.

Limitations

WHO GEMS/Food is the only compatible sampling family for kale, as sold on this basis in the completed kale ledger, so this single-family reference cannot represent other markets. Reporting regions and purchase markets are not growing areas. Non-detects with differing limits widen the bounds. A position is descriptive concentration, not a safety or compliance judgment.

Admitted source families

Reference version and interpretation

Frozen evidence selection: 2026-10-22. Method: hmi-pooled-reference-v2.

hmi-pooled-reference-v2:kale-221432e9e290:200f6761fa7f

The reference is computed by HMI from admitted evidence, not copied from a study baseline. It describes the evidence currently in this scope; it is not a worldwide or current-lot benchmark. Source, year and region filters do not change it. Different HMI scopes are not one common ranking.

HMI kale Pb reference · Kale, cooked

Reference and scope

HMI kale Pb reference · Kale, cooked. Kale, cooked; As sold, as eaten or fresh weight, µg/kg; no moisture conversion. Distribution of subgroup median concentration values.

Reference (µg/kg)Q1MedianQ3
Primary HMI reference≤4≤44.8
Alternative weighting≤4≤44.8

6 reference units; 1 sampling family. Counts describe evidence units, not independent studies or certainty.

Contributing familyReference unitsWeight
FDA Total Diet Study6100%

Methods

HMI pools subgroup medians for kale, cooked on a as sold, as eaten or fresh weight basis: monitoring results summarized as one median per reporting region, sampling type and collection year, total diet composites and printed sample values as single observations, and printed subgroup medians from ordinary farms or markets. Each sampling family has equal total weight. Non-detects stay as zero-to-limit intervals, so quartiles carry lower and upper bounds. Means, ranges and maxima never enter as observations. Other forms, other bases, targeted sampling and results from mining, industrial, wastewater-affected, experimental, fertiliser or manufacturer settings are kept out of this reference.

Inverse weighted empirical CDF; midpoint at exact cumulative-weight boundaries. At least four reference units and a definitely positive Q1–Q3 spread are required. This is a display eligibility rule, not statistical validation.

Limitations

FDA Total Diet Study is the only compatible sampling family for kale, cooked on this basis in the completed kale ledger, so this single-family reference cannot represent other markets. Reporting regions and purchase markets are not growing areas. Non-detects with differing limits widen the bounds. A position is descriptive concentration, not a safety or compliance judgment.

Admitted source families

Reference version and interpretation

Frozen evidence selection: 2026-10-22. Method: hmi-pooled-reference-v2.

hmi-pooled-reference-v2:kale-c7231900e0c0:f75047cf07ef

The reference is computed by HMI from admitted evidence, not copied from a study baseline. It describes the evidence currently in this scope; it is not a worldwide or current-lot benchmark. Source, year and region filters do not change it. Different HMI scopes are not one common ranking.

HMI kale Cd reference · Kale, as sold

Reference and scope

HMI kale Cd reference · Kale, as sold. Kale, as sold; As sold, as eaten or fresh weight, µg/kg; no moisture conversion. Distribution of subgroup median concentration values.

Reference (µg/kg)Q1MedianQ3
Primary HMI reference6.511.6–12.2 (bounded)12.6–14 (bounded)
Alternative weighting6.512.213–14 (bounded)

34 reference units; 2 sampling families. Counts describe evidence units, not independent studies or certainty.

Contributing familyReference unitsWeight
WHO GEMS/Food3250%
Norton et al. 2015250%

Methods

HMI pools subgroup medians for kale, as sold on a as sold, as eaten or fresh weight basis: monitoring results summarized as one median per reporting region, sampling type and collection year, total diet composites and printed sample values as single observations, and printed subgroup medians from ordinary farms or markets. Each sampling family has equal total weight. Non-detects stay as zero-to-limit intervals, so quartiles carry lower and upper bounds. Means, ranges and maxima never enter as observations. Other forms, other bases, targeted sampling and results from mining, industrial, wastewater-affected, experimental, fertiliser or manufacturer settings are kept out of this reference.

Inverse weighted empirical CDF; midpoint at exact cumulative-weight boundaries. At least four reference units and a definitely positive Q1–Q3 spread are required. This is a display eligibility rule, not statistical validation.

Limitations

2 sampling families with different designs; none is a probability sample of world supply. Reporting regions and purchase markets are not growing areas. Non-detects with differing limits widen the bounds. A position is descriptive concentration, not a safety or compliance judgment.

Admitted source families

Reference version and interpretation

Frozen evidence selection: 2026-10-22. Method: hmi-pooled-reference-v2.

hmi-pooled-reference-v2:kale-946ad5b9b6ec:8973b53ac46b

The reference is computed by HMI from admitted evidence, not copied from a study baseline. It describes the evidence currently in this scope; it is not a worldwide or current-lot benchmark. Source, year and region filters do not change it. Different HMI scopes are not one common ranking.

HMI kale Cd reference · Kale, cooked

Reference and scope

HMI kale Cd reference · Kale, cooked. Kale, cooked; As sold, as eaten or fresh weight, µg/kg; no moisture conversion. Distribution of subgroup median concentration values.

Reference (µg/kg)Q1MedianQ3
Primary HMI reference1315.550
Alternative weighting1315.550

6 reference units; 1 sampling family. Counts describe evidence units, not independent studies or certainty.

Contributing familyReference unitsWeight
FDA Total Diet Study6100%

Methods

HMI pools subgroup medians for kale, cooked on a as sold, as eaten or fresh weight basis: monitoring results summarized as one median per reporting region, sampling type and collection year, total diet composites and printed sample values as single observations, and printed subgroup medians from ordinary farms or markets. Each sampling family has equal total weight. Non-detects stay as zero-to-limit intervals, so quartiles carry lower and upper bounds. Means, ranges and maxima never enter as observations. Other forms, other bases, targeted sampling and results from mining, industrial, wastewater-affected, experimental, fertiliser or manufacturer settings are kept out of this reference.

Inverse weighted empirical CDF; midpoint at exact cumulative-weight boundaries. At least four reference units and a definitely positive Q1–Q3 spread are required. This is a display eligibility rule, not statistical validation.

Limitations

FDA Total Diet Study is the only compatible sampling family for kale, cooked on this basis in the completed kale ledger, so this single-family reference cannot represent other markets. Reporting regions and purchase markets are not growing areas. Non-detects with differing limits widen the bounds. A position is descriptive concentration, not a safety or compliance judgment.

Admitted source families

Reference version and interpretation

Frozen evidence selection: 2026-10-22. Method: hmi-pooled-reference-v2.

hmi-pooled-reference-v2:kale-4da1ab9befe8:b36824e5097b

The reference is computed by HMI from admitted evidence, not copied from a study baseline. It describes the evidence currently in this scope; it is not a worldwide or current-lot benchmark. Source, year and region filters do not change it. Different HMI scopes are not one common ranking.

HMI kale tAs reference · Kale, as sold

Reference and scope

HMI kale tAs reference · Kale, as sold. Kale, as sold; As sold, as eaten or fresh weight, µg/kg; no moisture conversion. Distribution of subgroup median concentration values.

Reference (µg/kg)Q1MedianQ3
Primary HMI reference≤2.691.03–5.73 (bounded)3.69–10.7 (bounded)
Alternative weighting≤1.661.05–5.73 (bounded)3.69–10.7 (bounded)

23 reference units; 1 sampling family. Counts describe evidence units, not independent studies or certainty.

Contributing familyReference unitsWeight
WHO GEMS/Food23100%

Methods

HMI pools subgroup medians for kale, as sold on a as sold, as eaten or fresh weight basis: monitoring results summarized as one median per reporting region, sampling type and collection year, total diet composites and printed sample values as single observations, and printed subgroup medians from ordinary farms or markets. Each sampling family has equal total weight. Non-detects stay as zero-to-limit intervals, so quartiles carry lower and upper bounds. Means, ranges and maxima never enter as observations. Other forms, other bases, targeted sampling and results from mining, industrial, wastewater-affected, experimental, fertiliser or manufacturer settings are kept out of this reference.

Inverse weighted empirical CDF; midpoint at exact cumulative-weight boundaries. At least four reference units and a definitely positive Q1–Q3 spread are required. This is a display eligibility rule, not statistical validation.

Limitations

WHO GEMS/Food is the only compatible sampling family for kale, as sold on this basis in the completed kale ledger, so this single-family reference cannot represent other markets. Reporting regions and purchase markets are not growing areas. Non-detects with differing limits widen the bounds. A position is descriptive concentration, not a safety or compliance judgment.

Admitted source families

Reference version and interpretation

Frozen evidence selection: 2026-10-22. Method: hmi-pooled-reference-v2.

hmi-pooled-reference-v2:kale-a13836054e4d:7552926f173c

The reference is computed by HMI from admitted evidence, not copied from a study baseline. It describes the evidence currently in this scope; it is not a worldwide or current-lot benchmark. Source, year and region filters do not change it. Different HMI scopes are not one common ranking.

HMI kale tAs reference · Kale, cooked

Reference and scope

HMI kale tAs reference · Kale, cooked. Kale, cooked; As sold, as eaten or fresh weight, µg/kg; no moisture conversion. Distribution of subgroup median concentration values.

Reference (µg/kg)Q1MedianQ3
Primary HMI reference5.25.99.7
Alternative weighting5.25.99.7

6 reference units; 1 sampling family. Counts describe evidence units, not independent studies or certainty.

Contributing familyReference unitsWeight
FDA Total Diet Study6100%

Methods

HMI pools subgroup medians for kale, cooked on a as sold, as eaten or fresh weight basis: monitoring results summarized as one median per reporting region, sampling type and collection year, total diet composites and printed sample values as single observations, and printed subgroup medians from ordinary farms or markets. Each sampling family has equal total weight. Non-detects stay as zero-to-limit intervals, so quartiles carry lower and upper bounds. Means, ranges and maxima never enter as observations. Other forms, other bases, targeted sampling and results from mining, industrial, wastewater-affected, experimental, fertiliser or manufacturer settings are kept out of this reference.

Inverse weighted empirical CDF; midpoint at exact cumulative-weight boundaries. At least four reference units and a definitely positive Q1–Q3 spread are required. This is a display eligibility rule, not statistical validation.

Limitations

FDA Total Diet Study is the only compatible sampling family for kale, cooked on this basis in the completed kale ledger, so this single-family reference cannot represent other markets. Reporting regions and purchase markets are not growing areas. Non-detects with differing limits widen the bounds. A position is descriptive concentration, not a safety or compliance judgment.

Admitted source families

Reference version and interpretation

Frozen evidence selection: 2026-10-22. Method: hmi-pooled-reference-v2.

hmi-pooled-reference-v2:kale-64c9d6a7fa1b:39ebbe6c7994

The reference is computed by HMI from admitted evidence, not copied from a study baseline. It describes the evidence currently in this scope; it is not a worldwide or current-lot benchmark. Source, year and region filters do not change it. Different HMI scopes are not one common ranking.

HMI kale iAs reference · Kale, as sold

Reference and scope

HMI kale iAs reference · Kale, as sold. Kale, as sold; As sold, as eaten or fresh weight, µg/kg; no moisture conversion. Distribution of subgroup median concentration values.

Reference (µg/kg)Q1MedianQ3
Primary HMI reference≤31.5–6.5 (bounded)53.1–56.6 (bounded)
Alternative weighting≤31.5–6.5 (bounded)53.1–56.6 (bounded)

4 reference units; 1 sampling family. Counts describe evidence units, not independent studies or certainty.

Contributing familyReference unitsWeight
WHO GEMS/Food4100%

Methods

HMI pools subgroup medians for kale, as sold on a as sold, as eaten or fresh weight basis: monitoring results summarized as one median per reporting region, sampling type and collection year, total diet composites and printed sample values as single observations, and printed subgroup medians from ordinary farms or markets. Each sampling family has equal total weight. Non-detects stay as zero-to-limit intervals, so quartiles carry lower and upper bounds. Means, ranges and maxima never enter as observations. Other forms, other bases, targeted sampling and results from mining, industrial, wastewater-affected, experimental, fertiliser or manufacturer settings are kept out of this reference.

Inverse weighted empirical CDF; midpoint at exact cumulative-weight boundaries. At least four reference units and a definitely positive Q1–Q3 spread are required. This is a display eligibility rule, not statistical validation.

Limitations

WHO GEMS/Food is the only compatible sampling family for kale, as sold on this basis in the completed kale ledger, so this single-family reference cannot represent other markets. Reporting regions and purchase markets are not growing areas. Non-detects with differing limits widen the bounds. A position is descriptive concentration, not a safety or compliance judgment.

Admitted source families

Reference version and interpretation

Frozen evidence selection: 2026-10-22. Method: hmi-pooled-reference-v2.

hmi-pooled-reference-v2:kale-b1347739c34a:f1724df4e3e9

The reference is computed by HMI from admitted evidence, not copied from a study baseline. It describes the evidence currently in this scope; it is not a worldwide or current-lot benchmark. Source, year and region filters do not change it. Different HMI scopes are not one common ranking.

HMI kale iAs reference · Kale, as sold, dry weight

Reference and scope

HMI kale iAs reference · Kale, as sold, dry weight. Kale, as sold; Dry weight, µg/kg; no moisture conversion. Distribution of subgroup median concentration values.

Reference (µg/kg)Q1MedianQ3
Primary HMI reference103103103
Alternative weighting103103103

1 reference units; 1 sampling family. Counts describe evidence units, not independent studies or certainty.

Contributing familyReference unitsWeight
WHO GEMS/Food1100%

Methods

HMI pools subgroup medians for kale, as sold on a dry weight basis: monitoring results summarized as one median per reporting region, sampling type and collection year, total diet composites and printed sample values as single observations, and printed subgroup medians from ordinary farms or markets. Each sampling family has equal total weight. Non-detects stay as zero-to-limit intervals, so quartiles carry lower and upper bounds. Means, ranges and maxima never enter as observations. Other forms, other bases, targeted sampling and results from mining, industrial, wastewater-affected, experimental, fertiliser or manufacturer settings are kept out of this reference.

Inverse weighted empirical CDF; midpoint at exact cumulative-weight boundaries. At least four reference units and a definitely positive Q1–Q3 spread are required. This is a display eligibility rule, not statistical validation.

Limitations

WHO GEMS/Food is the only compatible sampling family for kale, as sold on this basis in the completed kale ledger, so this single-family reference cannot represent other markets. Reporting regions and purchase markets are not growing areas. Non-detects with differing limits widen the bounds. A position is descriptive concentration, not a safety or compliance judgment.

Admitted source families

Reference version and interpretation

Frozen evidence selection: 2026-10-22. Method: hmi-pooled-reference-v2.

hmi-pooled-reference-v2:kale-93ee90f1e5c5:ae7c6e8b111b

The reference is computed by HMI from admitted evidence, not copied from a study baseline. It describes the evidence currently in this scope; it is not a worldwide or current-lot benchmark. Source, year and region filters do not change it. Different HMI scopes are not one common ranking.

HMI kale tHg reference · Kale, as sold

Reference and scope

HMI kale tHg reference · Kale, as sold. Kale, as sold; As sold, as eaten or fresh weight, µg/kg; no moisture conversion. Distribution of subgroup median concentration values.

Reference (µg/kg)Q1MedianQ3
Primary HMI reference≤0.30.05–2 (bounded)0.87–4 (bounded)
Alternative weighting≤0.30.05–2 (bounded)0.87–4 (bounded)

13 reference units; 1 sampling family. Counts describe evidence units, not independent studies or certainty.

Contributing familyReference unitsWeight
WHO GEMS/Food13100%

Methods

HMI pools subgroup medians for kale, as sold on a as sold, as eaten or fresh weight basis: monitoring results summarized as one median per reporting region, sampling type and collection year, total diet composites and printed sample values as single observations, and printed subgroup medians from ordinary farms or markets. Each sampling family has equal total weight. Non-detects stay as zero-to-limit intervals, so quartiles carry lower and upper bounds. Means, ranges and maxima never enter as observations. Other forms, other bases, targeted sampling and results from mining, industrial, wastewater-affected, experimental, fertiliser or manufacturer settings are kept out of this reference.

Inverse weighted empirical CDF; midpoint at exact cumulative-weight boundaries. At least four reference units and a definitely positive Q1–Q3 spread are required. This is a display eligibility rule, not statistical validation.

Limitations

WHO GEMS/Food is the only compatible sampling family for kale, as sold on this basis in the completed kale ledger, so this single-family reference cannot represent other markets. Reporting regions and purchase markets are not growing areas. Non-detects with differing limits widen the bounds. A position is descriptive concentration, not a safety or compliance judgment.

Admitted source families

Reference version and interpretation

Frozen evidence selection: 2026-10-22. Method: hmi-pooled-reference-v2.

hmi-pooled-reference-v2:kale-3b5c1cc1cb14:4f71bc53833f

The reference is computed by HMI from admitted evidence, not copied from a study baseline. It describes the evidence currently in this scope; it is not a worldwide or current-lot benchmark. Source, year and region filters do not change it. Different HMI scopes are not one common ranking.

HMI kale tHg reference · Kale, cooked

Reference and scope

HMI kale tHg reference · Kale, cooked. Kale, cooked; As sold, as eaten or fresh weight, µg/kg; no moisture conversion. Distribution of subgroup median concentration values.

Reference (µg/kg)Q1MedianQ3
Primary HMI reference≤11.051.1
Alternative weighting≤11.051.1

6 reference units; 1 sampling family. Counts describe evidence units, not independent studies or certainty.

Contributing familyReference unitsWeight
FDA Total Diet Study6100%

Methods

HMI pools subgroup medians for kale, cooked on a as sold, as eaten or fresh weight basis: monitoring results summarized as one median per reporting region, sampling type and collection year, total diet composites and printed sample values as single observations, and printed subgroup medians from ordinary farms or markets. Each sampling family has equal total weight. Non-detects stay as zero-to-limit intervals, so quartiles carry lower and upper bounds. Means, ranges and maxima never enter as observations. Other forms, other bases, targeted sampling and results from mining, industrial, wastewater-affected, experimental, fertiliser or manufacturer settings are kept out of this reference.

Inverse weighted empirical CDF; midpoint at exact cumulative-weight boundaries. At least four reference units and a definitely positive Q1–Q3 spread are required. This is a display eligibility rule, not statistical validation.

Limitations

FDA Total Diet Study is the only compatible sampling family for kale, cooked on this basis in the completed kale ledger, so this single-family reference cannot represent other markets. Reporting regions and purchase markets are not growing areas. Non-detects with differing limits widen the bounds. A position is descriptive concentration, not a safety or compliance judgment.

Admitted source families

Reference version and interpretation

Frozen evidence selection: 2026-10-22. Method: hmi-pooled-reference-v2.

hmi-pooled-reference-v2:kale-3987bb4e0624:81a50b47d021

The reference is computed by HMI from admitted evidence, not copied from a study baseline. It describes the evidence currently in this scope; it is not a worldwide or current-lot benchmark. Source, year and region filters do not change it. Different HMI scopes are not one common ranking.

HMI kale Ni reference · Kale, cooked

Reference and scope

HMI kale Ni reference · Kale, cooked. Kale, cooked; As sold, as eaten or fresh weight, µg/kg; no moisture conversion. Distribution of subgroup median concentration values.

Reference (µg/kg)Q1MedianQ3
Primary HMI reference≤4060.5120
Alternative weighting≤4060.5120

6 reference units; 1 sampling family. Counts describe evidence units, not independent studies or certainty.

Contributing familyReference unitsWeight
FDA Total Diet Study6100%

Methods

HMI pools subgroup medians for kale, cooked on a as sold, as eaten or fresh weight basis: monitoring results summarized as one median per reporting region, sampling type and collection year, total diet composites and printed sample values as single observations, and printed subgroup medians from ordinary farms or markets. Each sampling family has equal total weight. Non-detects stay as zero-to-limit intervals, so quartiles carry lower and upper bounds. Means, ranges and maxima never enter as observations. Other forms, other bases, targeted sampling and results from mining, industrial, wastewater-affected, experimental, fertiliser or manufacturer settings are kept out of this reference.

Inverse weighted empirical CDF; midpoint at exact cumulative-weight boundaries. At least four reference units and a definitely positive Q1–Q3 spread are required. This is a display eligibility rule, not statistical validation.

Limitations

FDA Total Diet Study is the only compatible sampling family for kale, cooked on this basis in the completed kale ledger, so this single-family reference cannot represent other markets. Reporting regions and purchase markets are not growing areas. Non-detects with differing limits widen the bounds. A position is descriptive concentration, not a safety or compliance judgment.

Admitted source families

Reference version and interpretation

Frozen evidence selection: 2026-10-22. Method: hmi-pooled-reference-v2.

hmi-pooled-reference-v2:kale-64a5c1f6334c:9bfa24316e47

The reference is computed by HMI from admitted evidence, not copied from a study baseline. It describes the evidence currently in this scope; it is not a worldwide or current-lot benchmark. Source, year and region filters do not change it. Different HMI scopes are not one common ranking.

HMI kale Cr reference · Kale, cooked

Reference and scope

HMI kale Cr reference · Kale, cooked. Kale, cooked; As sold, as eaten or fresh weight, µg/kg; no moisture conversion. Distribution of subgroup median concentration values.

Reference (µg/kg)Q1MedianQ3
Primary HMI reference≤50≤5070
Alternative weighting≤50≤5070

6 reference units; 1 sampling family. Counts describe evidence units, not independent studies or certainty.

Contributing familyReference unitsWeight
FDA Total Diet Study6100%

Methods

HMI pools subgroup medians for kale, cooked on a as sold, as eaten or fresh weight basis: monitoring results summarized as one median per reporting region, sampling type and collection year, total diet composites and printed sample values as single observations, and printed subgroup medians from ordinary farms or markets. Each sampling family has equal total weight. Non-detects stay as zero-to-limit intervals, so quartiles carry lower and upper bounds. Means, ranges and maxima never enter as observations. Other forms, other bases, targeted sampling and results from mining, industrial, wastewater-affected, experimental, fertiliser or manufacturer settings are kept out of this reference.

Inverse weighted empirical CDF; midpoint at exact cumulative-weight boundaries. At least four reference units and a definitely positive Q1–Q3 spread are required. This is a display eligibility rule, not statistical validation.

Limitations

FDA Total Diet Study is the only compatible sampling family for kale, cooked on this basis in the completed kale ledger, so this single-family reference cannot represent other markets. Reporting regions and purchase markets are not growing areas. Non-detects with differing limits widen the bounds. A position is descriptive concentration, not a safety or compliance judgment.

Admitted source families

Reference version and interpretation

Frozen evidence selection: 2026-10-22. Method: hmi-pooled-reference-v2.

hmi-pooled-reference-v2:kale-21cc5a11bd0a:89134b12fe11

The reference is computed by HMI from admitted evidence, not copied from a study baseline. It describes the evidence currently in this scope; it is not a worldwide or current-lot benchmark. Source, year and region filters do not change it. Different HMI scopes are not one common ranking.

HMI kale Sn reference · Kale, as sold

Reference and scope

HMI kale Sn reference · Kale, as sold. Kale, as sold; As sold, as eaten or fresh weight, µg/kg; no moisture conversion. Distribution of subgroup median concentration values.

Reference (µg/kg)Q1MedianQ3
Primary HMI reference≤105.5–10.5 (bounded)11
Alternative weighting≤105.5–10.5 (bounded)11

2 reference units; 1 sampling family. Counts describe evidence units, not independent studies or certainty.

Contributing familyReference unitsWeight
WHO GEMS/Food2100%

Methods

HMI pools subgroup medians for kale, as sold on a as sold, as eaten or fresh weight basis: monitoring results summarized as one median per reporting region, sampling type and collection year, total diet composites and printed sample values as single observations, and printed subgroup medians from ordinary farms or markets. Each sampling family has equal total weight. Non-detects stay as zero-to-limit intervals, so quartiles carry lower and upper bounds. Means, ranges and maxima never enter as observations. Other forms, other bases, targeted sampling and results from mining, industrial, wastewater-affected, experimental, fertiliser or manufacturer settings are kept out of this reference.

Inverse weighted empirical CDF; midpoint at exact cumulative-weight boundaries. At least four reference units and a definitely positive Q1–Q3 spread are required. This is a display eligibility rule, not statistical validation.

Limitations

WHO GEMS/Food is the only compatible sampling family for kale, as sold on this basis in the completed kale ledger, so this single-family reference cannot represent other markets. Reporting regions and purchase markets are not growing areas. Non-detects with differing limits widen the bounds. A position is descriptive concentration, not a safety or compliance judgment.

Admitted source families

Reference version and interpretation

Frozen evidence selection: 2026-10-22. Method: hmi-pooled-reference-v2.

hmi-pooled-reference-v2:kale-a9dac0340408:1873dfdbb85c

The reference is computed by HMI from admitted evidence, not copied from a study baseline. It describes the evidence currently in this scope; it is not a worldwide or current-lot benchmark. Source, year and region filters do not change it. Different HMI scopes are not one common ranking.

HMI kale U reference · Kale, cooked

Reference and scope

HMI kale U reference · Kale, cooked. Kale, cooked; As sold, as eaten or fresh weight, µg/kg; no moisture conversion. Distribution of subgroup median concentration values.

Reference (µg/kg)Q1MedianQ3
Primary HMI reference≤1≤11.5
Alternative weighting≤1≤11.5

6 reference units; 1 sampling family. Counts describe evidence units, not independent studies or certainty.

Contributing familyReference unitsWeight
FDA Total Diet Study6100%

Methods

HMI pools subgroup medians for kale, cooked on a as sold, as eaten or fresh weight basis: monitoring results summarized as one median per reporting region, sampling type and collection year, total diet composites and printed sample values as single observations, and printed subgroup medians from ordinary farms or markets. Each sampling family has equal total weight. Non-detects stay as zero-to-limit intervals, so quartiles carry lower and upper bounds. Means, ranges and maxima never enter as observations. Other forms, other bases, targeted sampling and results from mining, industrial, wastewater-affected, experimental, fertiliser or manufacturer settings are kept out of this reference.

Inverse weighted empirical CDF; midpoint at exact cumulative-weight boundaries. At least four reference units and a definitely positive Q1–Q3 spread are required. This is a display eligibility rule, not statistical validation.

Limitations

FDA Total Diet Study is the only compatible sampling family for kale, cooked on this basis in the completed kale ledger, so this single-family reference cannot represent other markets. Reporting regions and purchase markets are not growing areas. Non-detects with differing limits widen the bounds. A position is descriptive concentration, not a safety or compliance judgment.

Admitted source families

Reference version and interpretation

Frozen evidence selection: 2026-10-22. Method: hmi-pooled-reference-v2.

hmi-pooled-reference-v2:kale-c47045f2562a:732ce8646344

The reference is computed by HMI from admitted evidence, not copied from a study baseline. It describes the evidence currently in this scope; it is not a worldwide or current-lot benchmark. Source, year and region filters do not change it. Different HMI scopes are not one common ranking.

1 dataset

How metals reach kale

No study in the library compared kale from contaminated and ordinary settings, so the routes by which metals reach it are not measured here.

Which forms and regions measured lower

The HMI reference for lead in kale, as sold pools 35 subgroup units from 2 sampling families, with a median of between 15 and 20 µg/kg and a middle half spanning 12.8 (lower quartile) and 102 µg/kg (upper quartile).1World Health Organization · 2026GEMS/Food Contaminants database heavy-metal exportsGEMS/Food Contamination Monitoring and Assessment ProgrammeRead source →5Norton GJ, Deacon CM, Mestrot A, Feldmann J, Jenkins P, Baskaran C, Meharg AA · 2015Cadmium and lead in vegetable and fruit produce selected from specific regional areas of the UKThe Science of the Total Environment 533, 520-527 (2015); accepted author manuscriptRead source → For lead in kale, cooked, the reference median was below 4 µg/kg, with a middle half from below 4 to 4.8 µg/kg across 6 units.2U.S. Food and Drug Administration · 2022FY2018-FY2020 TDS Elements Analytical ResultsFDA Total Diet StudyRead source → For cadmium in kale, as sold, the reference median was between 11.6 and 12.2 µg/kg, with a middle half from 6.5 to between 12.6 and 14 µg/kg across 34 units.1World Health Organization · 2026GEMS/Food Contaminants database heavy-metal exportsGEMS/Food Contamination Monitoring and Assessment ProgrammeRead source →5Norton GJ, Deacon CM, Mestrot A, Feldmann J, Jenkins P, Baskaran C, Meharg AA · 2015Cadmium and lead in vegetable and fruit produce selected from specific regional areas of the UKThe Science of the Total Environment 533, 520-527 (2015); accepted author manuscriptRead source →

Among GEMS/Food reporting regions, kale, as sold had the lowest lead subgroup median in SEARO (2 µg/kg) and the highest in SEARO (135 µg/kg); reporting regions are markets, not growing areas.1World Health Organization · 2026GEMS/Food Contaminants database heavy-metal exportsGEMS/Food Contamination Monitoring and Assessment ProgrammeRead source → For cadmium, the lowest regional median was in EURO (4.45 µg/kg) and the highest in PAHO (40.2 µg/kg).1World Health Organization · 2026GEMS/Food Contaminants database heavy-metal exportsGEMS/Food Contamination Monitoring and Assessment ProgrammeRead source →

Targeted kale, as sold samples in WPRO had lead higher at 2,890 µg/kg, against 12.5 µg/kg for WPRO in the same study.1World Health Organization · 2026GEMS/Food Contaminants database heavy-metal exportsGEMS/Food Contamination Monitoring and Assessment ProgrammeRead source → In another study, targeted kale, as sold samples in WPRO had cadmium higher at 2,430 µg/kg, against 18.5 µg/kg for WPRO in the same study.1World Health Organization · 2026GEMS/Food Contaminants database heavy-metal exportsGEMS/Food Contamination Monitoring and Assessment ProgrammeRead source →

Processing, cooking and preparation

No study in the library measured the same kale before and after washing, peeling, drying or cooking for metals, so no processing effect can be stated. Heat does not destroy metals; cooking or drying changes a concentration by moving water or metal, not by removing the element.

Mitigation for kale

Controls with evidence for kale act at the stages below. Grow leaves away from coal-fired power plants (evidence from leafy greens): leaf mercury fell with distance from two coal-fired power plants, from 86.7 µg/kg at 1 km to 23.9 µg/kg at 10 km in water spinach; grocery-store leaves from more than 55 km away were below 1 µg/kg.8Li R, Wu H, Ding J, Fu W, Gan L, Li Y · 2017Mercury pollution in vegetables, grains and soils from areas surrounding coal-fired power plantsScientific ReportsRead source → Rinse leaves before eating (evidence from leafy greens): rinsing lowered leaf mercury by 19–63%.8Li R, Wu H, Ding J, Fu W, Gan L, Li Y · 2017Mercury pollution in vegetables, grains and soils from areas surrounding coal-fired power plantsScientific ReportsRead source → Vinegar soak after washing (evidence from leafy greens): a 10% vinegar soak lowered market-lettuce lead from 155 to 95 µg/kg and cadmium from 46.9 to 36.2 µg/kg, but left farm lettuce lead unchanged.9Bora FD, Bunea A, Pop SR, Banita SI, Dusa DS, Chira A, Bunea CI · 2022Quantification and Reduction in Heavy Metal Residues in Some Fruits and Vegetables: A Case Study Galați County, RomaniaHorticulturaeRead source →

Test lots against leaf-specific limits (evidence from leafy greens): lead 0.30 mg/kg and cadmium 0.10 mg/kg (0.20 mg/kg for spinach, similar leaves and fresh herbs), wet weight, after washing and separating the edible part.10European Commission · 2023Commission Regulation (EU) 2023/915 of 25 April 2023 on maximum levels for certain contaminants in food and repealing Regulation (EC) No 1881/2006Official Journal of the European UnionRead source → Buy from controlled supply rather than roadside stalls (evidence from leafy greens): washed roadside leaves still carried more cadmium (6.84 mg/kg) than leaves from large stores (2.10–2.50 mg/kg dry weight).11Adhikari S, Struwig M · 2024Concentrations and health risks of selected elements in leafy vegetables: a comparison between roadside open-air markets and large stores in Johannesburg, South AfricaNot reportedRead source → Irrigate with arsenic-free water (evidence from leafy greens): leaf arsenic rose roughly in proportion to irrigation-water arsenic; at 500 µg/L spinach reached 6.57 µg/g dry matter against 1.05 µg/g with arsenic-free water.12Clair-Caliot G, Marks SJ, Hug SJ, Bretzler A, N’guessan ND, Tihe SFK, Lalanne F · 2021Uptake of Arsenic by Irrigated Vegetables and Cooked Food Products in Burkina FasoFrontiers in WaterRead source →

About this table
ControlMaterialEffectsEvidence
Controls by production stage, retaining all paired analyte outcomes. Seasons and reused guidance do not add independent studies. No effectiveness ranking is implied.
ControlMaterialEffectsEvidence
Grow leaves away from coal-fired power plantscultivation · As reportedLeafy greens
tHg · Decreased
tHg · Decreased
1

Li et al. 2017

HMI source record · Original source

One dataset family. Linked analytes, repeated seasons and reports are not independent studies.

Effects and comparators

Total mercuryDecreasedLower mercury farther from the plant (observational)Water spinach mean total mercury1 km versus 10 kmFresh weightHigh · HMI calculates the relative change from the retained source means: (23.88 − 86.69) / 86.69 × 100 = -72.5%. This is a descriptive contrast of means; no confidence interval or significance test is inferred.

Total mercuryDecreasedLower mercury farther from the plant (observational)Lettuce mean total mercury1 km versus 10 kmFresh weightHigh · HMI calculates the relative change from the retained source means: (7.23 − 39.04) / 39.04 × 100 = -81.5%. This is a descriptive contrast of means; no confidence interval or significance test is inferred.

Farms at 1, 3, 5 and 10 km from the plants and grocery-store produce.

How magnitude is classified

HMI editorial magnitude bands: Low is a change greater than 0% and below 25%; Moderate is 25% to below 50%; High is 50% or more, relative to that control’s own comparator. Direction is shown separately. These are descriptive display bands, not validated efficacy cutoffs, occurrence-baseline positions, statistical significance, evidence certainty, safety or net benefit. A range spanning bands retains both labels; a qualitative label has no implied percentage.

Control in context

Leaf mercury fell with distance from two coal-fired power plants, from 86.7 µg/kg at 1 km to 23.9 µg/kg at 10 km in water spinach; grocery-store leaves from more than 55 km away were below 1 µg/kg.

ContextSelected control
MaterialLettuce, amaranth and water spinach, Guizhou, China
Stagecultivation
AudienceIngredient suppliers · Brands
PeriodAs reported
ConditionsAreas with point-source air emissions.
Tested scaleField survey
Sample unitsSource sampling units; no independent intervention units unless stated.
Trade-offsEffects on other metals, yield and cost are given where the source reports them.
VerificationConfirm with lot results by leaf on a fresh-weight basis with an adequate quantification limit.

Design, applicability and operating conditions

Observational field or market comparison. Areas with point-source air emissions.

One study family.

As reported by the source

Reduces metal reaching or remaining in the edible leaf, or detects it before release.

Source measurements and linked outcomes

AnalytePeriodEffectConverted valuesOriginal means (SD)BasisLocator
Total mercury (tHg)As reportedLower mercury farther from the plant (observational)86.69 → 23.88 µg/kgAbsolute values unextractedFresh weightTable 2
Total mercury (tHg)As reportedLower mercury farther from the plant (observational)39.04 → 7.23 µg/kgAbsolute values unextractedFresh weightTable 2

Observational; distance is confounded with soil mercury.

No percentage calculated unless shown in outcomes.

Basis as stated in each outcome; no conversion.

Trade-offs and implementation

As reported in outcomes.

Not assessed unless stated.

Not assessed.

Not quantified.

Not assessed.

Applicability and verification

Areas with point-source air emissions.

Lot testing.

See source.

Extraction limits and next actions

Unresolved fieldNext action
commercial-scale effectMost evidence is from single sites or pot trials; keep effects source-scoped.

Provenance and assessment

As disclosed by the source.

Source facts and HMI calculations are distinguished above. These selected source findings do not constitute a systematic review; outcome-level appraisal remains incomplete.

Irrigate with arsenic-free watercultivation · As reportedLeafy greens
tAs · Increased
tAs · Increased
1

Clair-Caliot et al. 2021

HMI source record · Original source

One dataset family. Linked analytes, repeated seasons and reports are not independent studies.

Effects and comparators

Total arsenicIncreasedHigher arsenic with contaminated irrigation waterSpinach leaf total arsenic0 versus 500 µg/L irrigation arsenicDry weightHigh · HMI calculates the relative change from the retained source means: (6570 − 1050) / 1050 × 100 = 525.7%. This is a descriptive contrast of means; no confidence interval or significance test is inferred.

Total arsenicIncreasedHigher arsenic with contaminated irrigation waterAmaranth leaf total arsenic0 versus 500 µg/L irrigation arsenicDry weightHigh · HMI calculates the relative change from the retained source means: (2230 − 120) / 120 × 100 = 1758.3%. This is a descriptive contrast of means; no confidence interval or significance test is inferred.

Arsenic-free irrigation versus water spiked to 100, 500 and 1,000 µg/L arsenic(V), same soil.

How magnitude is classified

HMI editorial magnitude bands: Low is a change greater than 0% and below 25%; Moderate is 25% to below 50%; High is 50% or more, relative to that control’s own comparator. Direction is shown separately. These are descriptive display bands, not validated efficacy cutoffs, occurrence-baseline positions, statistical significance, evidence certainty, safety or net benefit. A range spanning bands retains both labels; a qualitative label has no implied percentage.

Control in context

Leaf arsenic rose roughly in proportion to irrigation-water arsenic; at 500 µg/L spinach reached 6.57 µg/g dry matter against 1.05 µg/g with arsenic-free water.

ContextSelected control
MaterialSpinach, lettuce and amaranth, greenhouse trial, Burkina Faso
Stagecultivation
AudienceIngredient suppliers · Brands
PeriodAs reported
ConditionsAreas with arsenic-bearing groundwater.
Tested scaleGreenhouse trial
Sample unitsSource sampling units; no independent intervention units unless stated.
Trade-offsEffects on other metals, yield and cost are given where the source reports them.
VerificationConfirm with lot results by leaf on a fresh-weight basis with an adequate quantification limit.

Design, applicability and operating conditions

Measured paired comparison or trial. Areas with arsenic-bearing groundwater.

One study family.

As reported by the source

Reduces metal reaching or remaining in the edible leaf, or detects it before release.

Source measurements and linked outcomes

AnalytePeriodEffectConverted valuesOriginal means (SD)BasisLocator
Total arsenic (tAs)As reportedHigher arsenic with contaminated irrigation water1050 → 6570 µg/kgAbsolute values unextractedDry weightTable 5
Total arsenic (tAs)As reportedHigher arsenic with contaminated irrigation water120 → 2230 µg/kgAbsolute values unextractedDry weightTable 5

Greenhouse experiment; control-plant values differ between the paper’s tables.

No percentage calculated unless shown in outcomes.

Basis as stated in each outcome; no conversion.

Trade-offs and implementation

As reported in outcomes.

Not assessed unless stated.

Not assessed.

Not quantified.

Not assessed.

Applicability and verification

Areas with arsenic-bearing groundwater.

Lot testing.

See source.

Extraction limits and next actions

Unresolved fieldNext action
commercial-scale effectMost evidence is from single sites or pot trials; keep effects source-scoped.

Provenance and assessment

As disclosed by the source.

Source facts and HMI calculations are distinguished above. These selected source findings do not constitute a systematic review; outcome-level appraisal remains incomplete.

Cover contaminated soil with clean soilsoil remediation · As reportedLeafy greens
Cd · Decreased
Cd · Decreased
1

Zhang et al. 2024

HMI source record · Original source

One dataset family. Linked analytes, repeated seasons and reports are not independent studies.

Effects and comparators

CadmiumDecreasedLower cadmium with a clean-soil coverLettuce cadmiumUntreated versus 15 cm coverDry weightHigh · HMI calculates the relative change from the retained source means: (980 − 2570) / 2570 × 100 = -61.9%. This is a descriptive contrast of means; no confidence interval or significance test is inferred.

CadmiumDecreasedLower cadmium with deep ploughingLettuce cadmiumUntreated versus 40 cm ploughingDry weightModerate · HMI calculates the relative change from the retained source means: (1840 − 2570) / 2570 × 100 = -28.4%. This is a descriptive contrast of means; no confidence interval or significance test is inferred.

Covered or ploughed plots versus untreated plots in the same greenhouse.

How magnitude is classified

HMI editorial magnitude bands: Low is a change greater than 0% and below 25%; Moderate is 25% to below 50%; High is 50% or more, relative to that control’s own comparator. Direction is shown separately. These are descriptive display bands, not validated efficacy cutoffs, occurrence-baseline positions, statistical significance, evidence certainty, safety or net benefit. A range spanning bands retains both labels; a qualitative label has no implied percentage.

Control in context

A 15 cm layer of uncontaminated soil cut lettuce cadmium by 62%; deep ploughing to 30–50 cm cut it by 11–28%.

ContextSelected control
MaterialLettuce, greenhouse field trial on cadmium-enriched soil, China
Stagesoil remediation
AudienceIngredient suppliers · Brands
PeriodAs reported
ConditionsProtected cultivation on naturally cadmium-enriched soil.
Tested scaleField trial
Sample unitsSource sampling units; no independent intervention units unless stated.
Trade-offsEffects on other metals, yield and cost are given where the source reports them.
VerificationConfirm with lot results by leaf on a fresh-weight basis with an adequate quantification limit.

Design, applicability and operating conditions

Measured paired comparison or trial. Protected cultivation on naturally cadmium-enriched soil.

One study family.

As reported by the source

Reduces metal reaching or remaining in the edible leaf, or detects it before release.

Source measurements and linked outcomes

AnalytePeriodEffectConverted valuesOriginal means (SD)BasisLocator
Cadmium (Cd)As reportedLower cadmium with a clean-soil cover2570 → 980 µg/kgAbsolute values unextractedDry weightFigure 1
Cadmium (Cd)As reportedLower cadmium with deep ploughing2570 → 1840 µg/kgAbsolute values unextractedDry weightFigure 1

One greenhouse; the abstract and figure give different values (figure used).

No percentage calculated unless shown in outcomes.

Basis as stated in each outcome; no conversion.

Trade-offs and implementation

As reported in outcomes.

Not assessed unless stated.

Not assessed.

Not quantified.

Not assessed.

Applicability and verification

Protected cultivation on naturally cadmium-enriched soil.

Lot testing.

See source.

Extraction limits and next actions

Unresolved fieldNext action
commercial-scale effectMost evidence is from single sites or pot trials; keep effects source-scoped.

Provenance and assessment

As disclosed by the source.

Source facts and HMI calculations are distinguished above. These selected source findings do not constitute a systematic review; outcome-level appraisal remains incomplete.

Irrigate with well water rather than treated wastewatercultivation · As reportedApple
tAs · Increased
tAs · Increased
tAs · Context
1

Nasirpour et al. 2026

HMI source record · Original source

One dataset family. Linked analytes, repeated seasons and reports are not independent studies.

Effects and comparators

Total arsenicIncreasedAbout 2.8 times higher with 100% wastewater (year 2)Fruit total arsenic, dry weight100% wastewater versus well water, year 2Whole fruit, dry weightHigh · HMI calculates the relative change from the retained source means: (390 − 140) / 140 × 100 = 178.6%. This is a descriptive contrast of means; no confidence interval or significance test is inferred.

Total arsenicIncreasedAbout 2.7 times higher with 50% wastewater (year 2)Fruit total arsenic, dry weight50% wastewater versus well water, year 2Whole fruit, dry weightHigh · HMI calculates the relative change from the retained source means: (380 − 140) / 140 × 100 = 171.4%. This is a descriptive contrast of means; no confidence interval or significance test is inferred.

Total arsenicNot clearly different in year 1 (about 780 against 690 µg/kg)Fruit total arsenic, dry weight100% wastewater versus well water, year 1Whole fruit, dry weightContext · No supported numerical treatment contrast is retained for this finding. A direction, market association, source-selection strategy or guidance statement alone does not establish a percentage change.

Well-water control against 50% and 100% treated wastewater, same orchard, cultivar and year; three rootstocks pooled.

How magnitude is classified

HMI editorial magnitude bands: Low is a change greater than 0% and below 25%; Moderate is 25% to below 50%; High is 50% or more, relative to that control’s own comparator. Direction is shown separately. These are descriptive display bands, not validated efficacy cutoffs, occurrence-baseline positions, statistical significance, evidence certainty, safety or net benefit. A range spanning bands retains both labels; a qualitative label has no implied percentage.

Control in context

Fruit from trees irrigated with 100% or 50% treated municipal wastewater carried about 2.5–2.8 times the arsenic of well-water fruit in the second year; first-year results did not separate clearly.

ContextSelected control
MaterialGolden Delicious apples, orchard near Mashhad, Iran, 2021–2022
Stagecultivation
AudienceIngredient suppliers · Brands
PeriodAs reported
ConditionsOne orchard and cultivar over two seasons.
Tested scaleField
Sample unitsSource sampling units; no independent intervention units unless stated.
Trade-offsOther metals, yield, taste and cost not quantified unless stated.
VerificationConfirm with lot-specific lead and cadmium results on the fresh fruit, using an adequate quantification limit.

Design, applicability and operating conditions

Orchard field experiment. One orchard and cultivar over two seasons.

One study family.

As reported by the source

Reduces metal reaching or remaining on apples, or detects it before release.

Source measurements and linked outcomes

AnalytePeriodEffectConverted valuesOriginal means (SD)BasisLocator
Total arsenic (tAs)As reportedAbout 2.8 times higher with 100% wastewater (year 2)140 → 390 µg/kgAbsolute values unextractedWhole fruit, dry weightFig. 1U p.7; Table 3 p.4
Total arsenic (tAs)As reportedAbout 2.7 times higher with 50% wastewater (year 2)140 → 380 µg/kgAbsolute values unextractedWhole fruit, dry weightFig. 1U p.7
Total arsenic (tAs)As reportedNot clearly different in year 1 (about 780 against 690 µg/kg)Numerical baseline not extractedAbsolute values unextractedWhole fruit, dry weightFig. 1U p.7; Duncan letters

No detection limit, reference material or recovery reported; values read from a chart.

Ratios are calculated from the source values shown in outcomes.

Basis as stated in outcomes; no mass balance reported.

Trade-offs and implementation

Pesticide residues are outside this table.

Not assessed.

Not assessed.

Not quantified.

Not assessed.

Applicability and verification

One orchard and cultivar over two seasons.

Lot testing.

See source.

Extraction limits and next actions

Unresolved fieldNext action
controlled commercial-scale apple intervention effectNo controlled commercial apple intervention study in the library; keep effects unquantified unless measured.

Provenance and assessment

As disclosed by the source.

Source facts and HMI calculations are distinguished above. These selected source findings do not constitute a systematic review; outcome-level appraisal remains incomplete.

Test lots against leaf-specific limitslot acceptance and release · As reportedLeafy greens
Cd · Context
Pb · Context
Guidance

Regulation (EU) 2023/915

HMI source record · Original source

Official guidance; not an independent experimental dataset. Reused across actions without adding studies.

Effects and comparators

CadmiumLimit 0.10–0.20 mg/kgCadmium maximum levelLeaf vegetables; spinachFresh weightContext · No supported numerical treatment contrast is retained for this finding. A direction, market association, source-selection strategy or guidance statement alone does not establish a percentage change.

LeadLimit 0.30 mg/kgLead maximum levelLeaf vegetablesFresh weightContext · No supported numerical treatment contrast is retained for this finding. A direction, market association, source-selection strategy or guidance statement alone does not establish a percentage change.

Regulatory maximum levels.

How magnitude is classified

HMI editorial magnitude bands: Low is a change greater than 0% and below 25%; Moderate is 25% to below 50%; High is 50% or more, relative to that control’s own comparator. Direction is shown separately. These are descriptive display bands, not validated efficacy cutoffs, occurrence-baseline positions, statistical significance, evidence certainty, safety or net benefit. A range spanning bands retains both labels; a qualitative label has no implied percentage.

Control in context

Lead 0.30 mg/kg and cadmium 0.10 mg/kg (0.20 mg/kg for spinach, similar leaves and fresh herbs), wet weight, after washing and separating the edible part.

ContextSelected control
MaterialLeaf vegetables and leafy brassica (EU)
Stagelot acceptance and release
AudienceIngredient suppliers · Brands
PeriodAs reported
ConditionsLeaves placed on the EU market.
Tested scaleRegulation
Sample unitsSource sampling units; no independent intervention units unless stated.
Trade-offsEffects on other metals, yield and cost are given where the source reports them.
VerificationConfirm with lot results by leaf on a fresh-weight basis with an adequate quantification limit.

Design, applicability and operating conditions

Regulatory or author guidance. Leaves placed on the EU market.

One study family.

As reported by the source

Reduces metal reaching or remaining in the edible leaf, or detects it before release.

Source measurements and linked outcomes

AnalytePeriodEffectConverted valuesOriginal means (SD)BasisLocator
Cadmium (Cd)As reportedLimit 0.10–0.20 mg/kgNumerical baseline not extractedAbsolute values unextractedFresh weightAnnex I 3.2.6
Lead (Pb)As reportedLimit 0.30 mg/kgNumerical baseline not extractedAbsolute values unextractedFresh weightAnnex I 3.1.6

A limit detects non-compliant lots; it does not lower contamination.

No percentage calculated unless shown in outcomes.

Basis as stated in each outcome; no conversion.

Trade-offs and implementation

As reported in outcomes.

Not assessed unless stated.

Not assessed.

Not quantified.

Not assessed.

Applicability and verification

Leaves placed on the EU market.

Lot testing.

See source.

Extraction limits and next actions

Unresolved fieldNext action
commercial-scale effectMost evidence is from single sites or pot trials; keep effects source-scoped.

Provenance and assessment

As disclosed by the source.

Source facts and HMI calculations are distinguished above. These selected source findings do not constitute a systematic review; outcome-level appraisal remains incomplete.

Buy from controlled supply rather than roadside stallssourcing and procurement · As reportedLeafy greens
Cd · Decreased
1

Adhikari et al. 2024

HMI source record · Original source

One dataset family. Linked analytes, repeated seasons and reports are not independent studies.

Effects and comparators

CadmiumDecreasedLower cadmium in large-store leaves (observational)Group mean cadmiumWashed roadside versus supermarketDry weightHigh · HMI calculates the relative change from the retained source means: (2500 − 6840) / 6840 × 100 = -63.5%. This is a descriptive contrast of means; no confidence interval or significance test is inferred.

Roadside versus supermarket and vegetable-market purchases.

How magnitude is classified

HMI editorial magnitude bands: Low is a change greater than 0% and below 25%; Moderate is 25% to below 50%; High is 50% or more, relative to that control’s own comparator. Direction is shown separately. These are descriptive display bands, not validated efficacy cutoffs, occurrence-baseline positions, statistical significance, evidence certainty, safety or net benefit. A range spanning bands retains both labels; a qualitative label has no implied percentage.

Control in context

Washed roadside leaves still carried more cadmium (6.84 mg/kg) than leaves from large stores (2.10–2.50 mg/kg dry weight).

ContextSelected control
MaterialFive leafy vegetables, Johannesburg
Stagesourcing and procurement
AudienceIngredient suppliers · Brands
PeriodAs reported
ConditionsUrban retail sourcing.
Tested scaleMarket survey
Sample unitsSource sampling units; no independent intervention units unless stated.
Trade-offsEffects on other metals, yield and cost are given where the source reports them.
VerificationConfirm with lot results by leaf on a fresh-weight basis with an adequate quantification limit.

Design, applicability and operating conditions

Observational field or market comparison. Urban retail sourcing.

One study family.

As reported by the source

Reduces metal reaching or remaining in the edible leaf, or detects it before release.

Source measurements and linked outcomes

AnalytePeriodEffectConverted valuesOriginal means (SD)BasisLocator
Cadmium (Cd)As reportedLower cadmium in large-store leaves (observational)6840 → 2500 µg/kgAbsolute values unextractedDry weightTable 2

Observational; growing sites not identified.

No percentage calculated unless shown in outcomes.

Basis as stated in each outcome; no conversion.

Trade-offs and implementation

As reported in outcomes.

Not assessed unless stated.

Not assessed.

Not quantified.

Not assessed.

Applicability and verification

Urban retail sourcing.

Lot testing.

See source.

Extraction limits and next actions

Unresolved fieldNext action
commercial-scale effectMost evidence is from single sites or pot trials; keep effects source-scoped.

Provenance and assessment

As disclosed by the source.

Source facts and HMI calculations are distinguished above. These selected source findings do not constitute a systematic review; outcome-level appraisal remains incomplete.

Rinse leaves before eatinghome preparation · As reportedLeafy greens
tHg · Decreased
1

Li et al. 2017

HMI source record · Original source

One dataset family. Linked analytes, repeated seasons and reports are not independent studies.

Effects and comparators

Total mercuryDecreasedLower mercury after rinsingLettuce total mercuryUnwashed versus washed, 1 km siteFresh weightModerate · HMI calculates the relative change from the retained source means: (21.03 − 35.7) / 35.7 × 100 = -41.1%. This is a descriptive contrast of means; no confidence interval or significance test is inferred.

Unwashed versus rinsed leaves from the same site.

How magnitude is classified

HMI editorial magnitude bands: Low is a change greater than 0% and below 25%; Moderate is 25% to below 50%; High is 50% or more, relative to that control’s own comparator. Direction is shown separately. These are descriptive display bands, not validated efficacy cutoffs, occurrence-baseline positions, statistical significance, evidence certainty, safety or net benefit. A range spanning bands retains both labels; a qualitative label has no implied percentage.

Control in context

Rinsing lowered leaf mercury by 19–63%.

ContextSelected control
MaterialLettuce and amaranth near coal power plants
Stagehome preparation
AudienceConsumers · Ingredient suppliers
PeriodAs reported
ConditionsLeaves with deposited dust or fly ash.
Tested scaleField survey
Sample unitsSource sampling units; no independent intervention units unless stated.
Trade-offsEffects on other metals, yield and cost are given where the source reports them.
VerificationConfirm with lot results by leaf on a fresh-weight basis with an adequate quantification limit.

Design, applicability and operating conditions

Measured paired comparison or trial. Leaves with deposited dust or fly ash.

One study family.

As reported by the source

Reduces metal reaching or remaining in the edible leaf, or detects it before release.

Source measurements and linked outcomes

AnalytePeriodEffectConverted valuesOriginal means (SD)BasisLocator
Total mercury (tHg)As reportedLower mercury after rinsing35.7 → 21.03 µg/kgAbsolute values unextractedFresh weightTable 4

Removes surface mercury only.

No percentage calculated unless shown in outcomes.

Basis as stated in each outcome; no conversion.

Trade-offs and implementation

As reported in outcomes.

Not assessed unless stated.

Not assessed.

Not quantified.

Not assessed.

Applicability and verification

Leaves with deposited dust or fly ash.

Lot testing.

See source.

Extraction limits and next actions

Unresolved fieldNext action
commercial-scale effectMost evidence is from single sites or pot trials; keep effects source-scoped.

Provenance and assessment

As disclosed by the source.

Source facts and HMI calculations are distinguished above. These selected source findings do not constitute a systematic review; outcome-level appraisal remains incomplete.

Vinegar soak after washinghome preparation · As reportedLeafy greens
Pb · Decreased
Pb · Decreased
1

Bora et al. 2022

HMI source record · Original source

One dataset family. Linked analytes, repeated seasons and reports are not independent studies.

Effects and comparators

LeadDecreasedLower lead after vinegar soakMarket lettuce leadTap water versus 10% vinegarFresh weightModerate · HMI calculates the relative change from the retained source means: (94.52 − 155.07) / 155.07 × 100 = -39.0%. This is a descriptive contrast of means; no confidence interval or significance test is inferred.

LeadDecreasedEssentially unchangedFarm lettuce leadTap water versus 10% vinegarFresh weightLow · HMI calculates the relative change from the retained source means: (73.14 − 73.25) / 73.25 × 100 = -0.2%. This is a descriptive contrast of means; no confidence interval or significance test is inferred.

Tap-water washing versus vinegar soak.

How magnitude is classified

HMI editorial magnitude bands: Low is a change greater than 0% and below 25%; Moderate is 25% to below 50%; High is 50% or more, relative to that control’s own comparator. Direction is shown separately. These are descriptive display bands, not validated efficacy cutoffs, occurrence-baseline positions, statistical significance, evidence certainty, safety or net benefit. A range spanning bands retains both labels; a qualitative label has no implied percentage.

Control in context

A 10% vinegar soak lowered market-lettuce lead from 155 to 95 µg/kg and cadmium from 46.9 to 36.2 µg/kg, but left farm lettuce lead unchanged.

ContextSelected control
MaterialLettuce and cabbage, Galați, Romania
Stagehome preparation
AudienceConsumers
PeriodAs reported
ConditionsHousehold or commercial washing.
Tested scaleLaboratory test
Sample unitsSource sampling units; no independent intervention units unless stated.
Trade-offsEffects on other metals, yield and cost are given where the source reports them.
VerificationConfirm with lot results by leaf on a fresh-weight basis with an adequate quantification limit.

Design, applicability and operating conditions

Measured paired comparison or trial. Household or commercial washing.

One study family.

As reported by the source

Reduces metal reaching or remaining in the edible leaf, or detects it before release.

Source measurements and linked outcomes

AnalytePeriodEffectConverted valuesOriginal means (SD)BasisLocator
Lead (Pb)As reportedLower lead after vinegar soak155.07 → 94.52 µg/kgAbsolute values unextractedFresh weightTables 1 and 3
Lead (Pb)As reportedEssentially unchanged73.25 → 73.14 µg/kgAbsolute values unextractedFresh weightTables 1 and 3

Analytical replicates; lot identity across treatments not stated.

No percentage calculated unless shown in outcomes.

Basis as stated in each outcome; no conversion.

Trade-offs and implementation

As reported in outcomes.

Not assessed unless stated.

Not assessed.

Not quantified.

Not assessed.

Applicability and verification

Household or commercial washing.

Lot testing.

See source.

Extraction limits and next actions

Unresolved fieldNext action
commercial-scale effectMost evidence is from single sites or pot trials; keep effects source-scoped.

Provenance and assessment

As disclosed by the source.

Source facts and HMI calculations are distinguished above. These selected source findings do not constitute a systematic review; outcome-level appraisal remains incomplete.

Intervention evidence

Trade-offs and limits of controls

Metals do not move together, so a lot that passes a lead limit says little about cadmium, arsenic or nickel unless those are tested as well.

Testing kale for metals

Results depend on the method and its reporting limit. Reporting limits for lead ranged from 0.5 µg/kg (WHO GEMS/Food) to 150 µg/kg (WHO GEMS/Food), so a non-detect from a less sensitive method says much less.1World Health Organization · 2026GEMS/Food Contaminants database heavy-metal exportsGEMS/Food Contamination Monitoring and Assessment ProgrammeRead source → Reporting limits for cadmium ranged from 0.08 µg/kg (WHO GEMS/Food) to 29 µg/kg (WHO GEMS/Food), so a non-detect from a less sensitive method says much less.1World Health Organization · 2026GEMS/Food Contaminants database heavy-metal exportsGEMS/Food Contamination Monitoring and Assessment ProgrammeRead source → Reporting limits for total arsenic ranged from 1.25 µg/kg (WHO GEMS/Food) to 250 µg/kg (WHO GEMS/Food), so a non-detect from a less sensitive method says much less.1World Health Organization · 2026GEMS/Food Contaminants database heavy-metal exportsGEMS/Food Contamination Monitoring and Assessment ProgrammeRead source →

A useful lot result names the form, the basis and a quantification limit well below the limit that applies.

Buying and verifying kale

No maximum level for metals in kale was verified from primary text for this page; buyers should confirm the limit that applies in the destination market.

Exposure from kale

No source in the library converted kale concentrations into a verified estimate of intake for a defined population. Concentrations describe the food, not individual risk.

Evidence review and remaining gaps

Every source record that names kale was decided on its content: 11 occurrence extracted, 11 records in all. The occurrence table renders every extracted result; held values remain visible with their reason.

References identify the sources used in these answers. The Sources index preserves additional discovery links; inclusion there does not mean a source has been reviewed for this assessment. Corrections, completed searches, new direct evidence and changes in requirements should reopen the relevant answer.

Kale FAQs

What heavy metals are found in kale?

The library reports lead, cadmium, total arsenic, inorganic arsenic, total mercury, nickel, chromium, aluminium, tin, antimony, uranium, Co, Cu, Fe, Mn, Zn, V, Se, As, Ba, Hg and Mo in kale, most often lead, cadmium and total arsenic.1World Health Organization · 2026GEMS/Food Contaminants database heavy-metal exportsGEMS/Food Contamination Monitoring and Assessment ProgrammeRead source →2U.S. Food and Drug Administration · 2022FY2018-FY2020 TDS Elements Analytical ResultsFDA Total Diet StudyRead source →3Banyen S · 2013Anodic Stripping Voltammetry for Simultaneous Determination of Heavy Metals in VegetablesKMITL Science Journal (Ladkrabang) 22(1), January-June 2556 B.E., pp. 84-99Read source →4Kim BF, Lupolt SN, Santo RE, Bachman G, Zhu X, Yang T, Fukagawa NK, Richardson ML, Green C, Phillips KM, Nachman KE · 2024Nutrients and non-essential metals in darkibor kale grown at urban and rural farms: a pilot studyPLOS ONERead source → WHO GEMS/Food submissions reported lead in kale, as sold from 484 results in 4 reporting regions, with subgroup medians from 2 to 2,890 µg/kg.1World Health Organization · 2026GEMS/Food Contaminants database heavy-metal exportsGEMS/Food Contamination Monitoring and Assessment ProgrammeRead source →

Which regions report the lowest levels in kale?

Among GEMS/Food reporting regions, kale, as sold had the lowest lead subgroup median in SEARO (2 µg/kg) and the highest in SEARO (135 µg/kg); reporting regions are markets, not growing areas.1World Health Organization · 2026GEMS/Food Contaminants database heavy-metal exportsGEMS/Food Contamination Monitoring and Assessment ProgrammeRead source → For cadmium, the lowest regional median was in EURO (4.45 µg/kg) and the highest in PAHO (40.2 µg/kg).1World Health Organization · 2026GEMS/Food Contaminants database heavy-metal exportsGEMS/Food Contamination Monitoring and Assessment ProgrammeRead source →

How much of the arsenic in kale is inorganic?

Inorganic arsenic, the more toxic form, was measured in 6 results from 2 sources, far fewer than total arsenic.1World Health Organization · 2026GEMS/Food Contaminants database heavy-metal exportsGEMS/Food Contamination Monitoring and Assessment ProgrammeRead source →13Centre for Food Safety FEHD HKSAR · 2012The First Hong Kong Total Diet Study: Inorganic ArsenicCentre for Food Safety, Food and Environmental Hygiene Department, Government of the Hong Kong Special Administrative RegionRead source → Centre for Food Safety FEHD HKSAR 2012 reported inorganic arsenic at 3 µg/kg (mean) in kale, as sold from all samples (4 sampling occasions, composites).13Centre for Food Safety FEHD HKSAR · 2012The First Hong Kong Total Diet Study: Inorganic ArsenicCentre for Food Safety, Food and Environmental Hygiene Department, Government of the Hong Kong Special Administrative RegionRead source →

References

Works cited on this page, in citation-number order. See Sources for this page’s source inventory. Each title links to a source record with bibliographic details, reported findings, and document provenance.

  1. GEMS/Food Contaminants database heavy-metal exportsWorld Health Organization · GEMS/Food Contamination Monitoring and Assessment Programme · 2026 · extranet.who.intDataset
  2. FY2018-FY2020 TDS Elements Analytical ResultsU.S. Food and Drug Administration · FDA Total Diet Study · 2022 · www.fda.govDataset
  3. Anodic Stripping Voltammetry for Simultaneous Determination of Heavy Metals in VegetablesBanyen S · KMITL Science Journal (Ladkrabang) 22(1), January-June 2556 B.E., pp. 84-99 · 2013 · li01.tci-thaijo.orgPeer-reviewed
  4. Nutrients and non-essential metals in darkibor kale grown at urban and rural farms: a pilot studyKim BF, Lupolt SN, Santo RE, Bachman G, Zhu X, Yang T, et al. · PLOS ONE · 2024 · doi.org/10.1371/journal.pone.0296840Peer-reviewed
  5. Cadmium and lead in vegetable and fruit produce selected from specific regional areas of the UKNorton GJ, Deacon CM, Mestrot A, Feldmann J, Jenkins P, Baskaran C, et al. · The Science of the Total Environment 533, 520-527 (2015); accepted author manuscript · 2015 · doi.org/10.1016/j.scitotenv.2015.06.130Peer-reviewed
  6. Concentration and Potential Non-Carcinogenic and Carcinogenic Health Risk Assessment of Metals in Locally Grown VegetablesSaleem M, Wang Y, Pierce D, Sens DA, Somji S, and Garrett SH · Foods · 2025 · doi.org/10.3390/foods14132264Peer-reviewed
  7. Infants’ and young children’s dietary exposures to lead and cadmium: FDA total diet study 2018-2020Spungen JH and et al. · Food Additives & Contaminants: Part A · 2024 · doi.org/10.1080/19440049.2024.2396910Peer-reviewed
  8. Mercury pollution in vegetables, grains and soils from areas surrounding coal-fired power plantsLi R, Wu H, Ding J, Fu W, Gan L, and Li Y · Scientific Reports · 2017 · doi.org/10.1038/srep46545Peer-reviewed
  9. Quantification and Reduction in Heavy Metal Residues in Some Fruits and Vegetables: A Case Study Galați County, RomaniaBora FD, Bunea A, Pop SR, Banita SI, Dusa DS, Chira A, et al. · Horticulturae · 2022 · doi.org/10.3390/horticulturae8111034Peer-reviewed
  10. Commission Regulation (EU) 2023/915 of 25 April 2023 on maximum levels for certain contaminants in food and repealing Regulation (EC) No 1881/2006European Commission · Official Journal of the European Union · 2023 · eur-lex.europa.euRegulation
  11. Concentrations and health risks of selected elements in leafy vegetables: a comparison between roadside open-air markets and large stores in Johannesburg, South AfricaAdhikari S and Struwig M · 2024 · doi.org/10.1007/s10661-023-12283-6Peer-reviewed
  12. Uptake of Arsenic by Irrigated Vegetables and Cooked Food Products in Burkina FasoClair-Caliot G, Marks SJ, Hug SJ, Bretzler A, N’guessan ND, Tihe SFK, et al. · Frontiers in Water · 2021 · doi.org/10.3389/frwa.2021.667308Peer-reviewed
  13. The First Hong Kong Total Diet Study: Inorganic ArsenicCentre for Food Safety FEHD HKSAR · Centre for Food Safety, Food and Environmental Hygiene Department, Government of the Hong Kong Special Administrative Region · 2012 · www.cfs.gov.hkGovernment

Sources

Browse Kale Sources for sources cited on this page and additional discovery candidates. The searchable index distinguishes verified citation use from discovery relationships whose relevance has not been established.

Update history

22 October 2026: Rebuilt from every kale record in the library: 11 occurrence sources, references by form and basis, same-study contrasts and stage-led controls.

Graph View