Health Canada 2020 — Guideline Technical Document, Cadmium in drinking water

This Health Canada Guideline Technical Document, prepared in collaboration with the Federal-Provincial-Territorial Committee on Drinking Water, establishes a maximum acceptable concentration (MAC) of 0.007 mg/L (7 µg/L) for total cadmium in drinking water based on a sample taken at the consumer’s tap. The guideline replaces earlier Canadian drinking-water cadmium guidance and is derived from kidney effects (renal tubular damage measured by β2-microglobulinuria) as the critical health endpoint, using the JECFA (2011) tolerable monthly intake of 25 µg/kg body weight per month as the toxicological anchor. The document is the operative federal-level reference for cadmium in Canadian drinking-water programs and incorporates an extensive Canadian provincial/territorial occurrence dataset, an analytical-method inventory, treatment-technology review (with emphasis on galvanized-steel plumbing as the dominant source), and an international comparison with U.S. EPA, Australian NHMRC, WHO, and EU drinking-water values.

Key numbers

Health-based value derivation (Section 10.0, p. 29-30):

• Maximum acceptable concentration (MAC): 0.007 mg/L (7 µg/L) total cadmium, sample at tap.
• Tolerable daily intake (TDI) adopted: 0.8 µg/kg body weight per day (= JECFA 2011 tolerable monthly intake of 25 µg/kg bw/month).
• Allocation factor for drinking water: 0.20 (“floor value”; food is the main exposure source).
• Body weight: 70 kg (adult; Health Canada 1994).
• Drinking-water intake: 1.5 L/day (adult).
• HBV calculation: 0.0008 mg/kg/day × 70 kg × 0.20 ÷ 1.5 L/day = 0.007 mg/L (rounded).

Toxicological reference points (Section 9.1.2.1, p. 21-22):

• EFSA (2009a) BMDL₀₅ for urinary cadmium (UCd) of 4.0 µg/g creatinine based on a 300 µg/g creatinine cut-off for urinary β2-microglobulin (B2M); after applying an adjustment factor of 3.9 (per WHO 2005), EFSA’s reference value is 1 µg/g creatinine UCd.
• JECFA (2011) breakpoint UCd of 5.24 µg/g creatinine (5th–95th percentile interval 4.95–5.57) in adults ≥50 y; with toxicodynamic variability factor of 3, the corresponding dietary exposure is 0.8 µg/kg bw/day (5th percentile) — adopted as the TDI.
• JECFA tolerable monthly intake: 25 µg/kg bw/month (Health Canada selected as basis).

International drinking-water values for comparison (Section 2.4 and 10.1, p. 2 and 30):

Canadian occurrence data — drinking and source (raw) water (Table 2, p. 5-6; sampling years 2000–2019):

No samples were provided from Nunavut or the Northwest Territories. Source attributions (footnotes 1–12) are reproduced verbatim from the document and credit provincial/territorial environment, health, and water-security agencies plus Environment and Climate Change Canada (2017) for the national dataset.

Plumbing-system observations (Section 7.1.6.1, p. 16-17, Viraraghavan et al. 2000 dataset cited in document):

• Copper-plumbed dwellings, first round (three sequential sample volumes drawn per dwelling — 125 mL faucet-stagnant, 500 mL plumbing-stagnant, 125 mL distribution-main): Cd concentrations <DL–171 µg/L (first sample), <DL–39 µg/L (second sample), <DL–102 µg/L (third sample). Maximum Cd concentrations of 133 µg/L and 101 µg/L were measured in the second and third rounds (subsequent rounds over time), respectively.
• Plastic-plumbed dwellings: Cd 8–38 µg/L in the first samples taken during the first round; <10 µg/L in all samples during the second and third rounds.
• Brass faucet leaching (Samuels and Meranger 1984 cited in document): chrome-plated brass faucets after 24 h stagnation gave <0.05–10 µg/L (raw, filtered, treated, groundwater, fulvic-acid solutions); highest value 10 µg/L for treated water. After a second 24 h period: <0.05–4 µg/L.
• Cement-mortar lining leaching (Guo et al. 1998 cited in document): up to 1.1 µg/L under static stagnation conditions over five days.

Canadian dietary cadmium exposure (Section 5.2, p. 6; Health Canada 2017a):

• Median dietary exposure: 0.30 µg/kg bw/day in males aged 51–71+ years.
• 0.83 µg/kg bw/day in both sexes aged 4–8 years.
• Multi-sample basis 2009–2015 Canadian food supply.

Canadian biomonitoring data — CHMS (Section 5.6.2, p. 7-8):

• Blood cadmium geometric mean (GM) ages 6–79: cycle 1 (2007–2009) 0.34 µg/L (95% CI 0.31–0.37, n=5,319); cycle 2 (2009–2011) 0.30 µg/L (95% CI 0.27–0.33, n=5,575); cycle 3 (2012–2013) 0.34 µg/L (95% CI 0.31–0.37, n=5,067).
• Blood Cd consistently higher in females (cycle GMs 0.38, 0.33, 0.39 µg/L) than males (0.30, 0.27, 0.31 µg/L).
• Urinary Cd GM ages 6–79: cycle 1 0.34 µg/L (95% CI 0.31–0.38, n=5,491); cycle 2 0.40 µg/L (95% CI 0.36–0.44, n=5,738).
• Creatinine-adjusted UCd: cycle 1 0.42 µg/g creatinine (95% CI 0.40–0.44, n=5,478); cycle 2 0.37 µg/g creatinine (95% CI 0.34–0.41, n=5,719).
• Age trend (Statistics Canada 2015 cited in document, cycle 3 data): blood cadmium GM 0.42 µg/L in adults 20–79 vs 0.12 µg/L in ages 3–19; statistically significant. Urinary cadmium also showed age-dependent increases (60–79 > 40–59 > 20–39) per Garner and Levallois (2016).

Approved analytical methods (Table 3, p. 9):

U.S. EPA practical quantitation limit (PQL): 2 µg/L (U.S. EPA 2009).

Residential and material standards (Section 7.2, p. 18-19):

• NSF/ANSI 53, 58, 62 certified residential treatment device performance: average influent of 0.03 mg/L → maximum effluent 0.005 mg/L.
• NSF/ANSI 61 (components) and 60 (treatment chemicals) single product allowable concentration: 0.0005 mg/L.

Carcinogenicity and other toxicology (Section 9-10):

• IARC (2012) classification: Group 1, “carcinogenic to humans” — based on sufficient evidence of lung, kidney, and prostate cancer in occupationally inhalation-exposed workers; epidemiological evidence linking oral cadmium exposure to cancer is limited (document, p. 24, 29).
• Oral LD₅₀ in rats and mice: 100–300 mg/kg (JECFA 2001 cited in document, p. 25).
• Bone-effect BMDL₀₅ values from epidemiology (range 0.5 to ~2 µg/g creatinine UCd) considered inconsistent and not used as critical effect (Section 9.1.2.2, p. 23-24).

Methods (brief)

Regulatory risk-assessment guideline document, not an analytical study. The document compiles and synthesizes:

1. Health risk assessment: based on Health Canada’s 2018 cadmium-in-foods hazard assessment (Health Canada 2018a) and prior risk assessments by EFSA (2009a, 2011) and JECFA (2011). Critical effect = renal tubular damage measured by urinary β2-microglobulin (B2M). Dose-response derived from EFSA (2009a) meta-analysis of 35 epidemiological studies aggregating 165 matched group-mean pairs of UCd and B2M (>30,000 individuals; predominantly female Asian populations). Toxicokinetic modelling used Amzal et al. (2009) one-compartment model with Monte Carlo simulation; toxicodynamic variability factor of 3 applied by JECFA (2011). Health Canada adopted the JECFA tolerable monthly intake of 25 µg/kg bw/month (= 0.8 µg/kg bw/day).
2. Occurrence dataset compilation: drinking-water Cd data from each Canadian province/territory (footnote-credited submissions 2017–2020) plus national Environment and Climate Change Canada (2017) raw-water dataset of n=18,998. No data from Nunavut or the Northwest Territories.
3. Analytical-methods inventory: five approved methods with method-detection-limit values (Table 3).
4. Treatment-technology review: municipal (coagulation, precipitation, ion exchange, membrane filtration, adsorption) and residential (NSF/ANSI 53, 58, 62) options.

No new analytical measurements are reported. The document operates as a synthesis-and-rule-derivation publication: dose-response selection → TDI adoption → HBV calculation (0.0008 mg/kg/day × 70 kg × 0.20 ÷ 1.5 L/day = 0.007 mg/L rounded) → MAC.

Limitations

• The MAC is derived from a dietary critical effect (kidney) with a 20% allocation factor reflecting that drinking water is a minor exposure source compared to food. The MAC does not protect against any incremental risk associated with simultaneous high-end dietary cadmium exposure beyond the 20% allocation, nor against inhalation exposure from cigarette smoke (which the document acknowledges produces blood Cd levels four to five times higher in smokers; Section 5.4, p. 7).
• The critical EFSA (2009a) meta-analysis used group means rather than individual data points; an adjustment factor of 3.9 was applied to address residual interindividual variability. Health Canada adopted JECFA’s slightly different reanalysis (biexponential breakpoint model) yielding numerically similar reference value.
• Oral carcinogenicity dose-response data are considered insufficient for quantitative risk assessment (Section 10.0, p. 29); the MAC is not based on a cancer endpoint despite IARC Group 1 classification (which derives from inhalation exposure).
• Bone-effect data are inconsistent and were not used as critical effect, although some studies report effects at UCd levels lower than the renal endpoint (Section 9.1.2.2, p. 23-24).
• The Canadian occurrence dataset has no Nunavut or Northwest Territories samples. Sampling protocols differed across jurisdictions (raw, source, treated, distributed, tap, well, distribution system); cross-jurisdiction comparisons should be made cautiously.
• The Alberta well dataset is highly skewed (mean 13.4 µg/L, median 15.0 µg/L, only 0.30% above DL of n=1,686) and likely reflects a small population of high-cadmium private wells rather than the typical distribution.
• The document references but does not reproduce the underlying Viraraghavan et al. (2000), Samuels and Meranger (1984), Schock and Neff (1988), and Subramanian et al. (1991) plumbing-leaching datasets; users seeking primary first-draw data should consult those publications directly.

Implications

• Certification: This guideline establishes the operative Canadian federal cadmium-in-drinking-water MAC (7 µg/L total Cd at tap), the underlying toxicological reference value (TDI 0.8 µg/kg bw/day; tolerable monthly intake 25 µg/kg bw/month), and the residential treatment performance expectation (NSF/ANSI certified devices reducing 0.03 mg/L influent to ≤0.005 mg/L effluent). It provides the Canadian regulatory backdrop for any HMT&C threshold work touching drinking water as an ingredient or process water in food production. The 0.20 drinking-water allocation factor (food is dominant exposure source) is the operative national assumption for partitioning the TDI across exposure media.
• Courses: Useful as a worked example of a national risk assessment that adopts an international body’s reference dose (JECFA tolerable monthly intake) rather than re-deriving from first principles, and as an illustration of how galvanized-steel plumbing legacies (pre-1980 National Plumbing Code Canada) drive present-day metal exposure independent of source-water quality. The plumbing-versus-source-water decoupling is a transferable lesson for any drinking-water lead/cadmium curriculum.
• App: Drinking water (tap, bottled) is a minor but non-zero Cd exposure pathway in the Canadian dataset compiled here. The document’s distributed-water medians across provinces sit in the 0.01–0.10 µg/L band, while the source-water and well distributions reach the tens of µg/L for a small fraction of samples (Alberta wells 0.30% above DL with detected-fraction median 15.0 µg/L; BC Interior raw and treated 97.0% above DL with detected-fraction median 0.02 µg/L). The document attributes the dominant tap-side contribution to galvanized-steel service lines, well components, and pre-1980 plumbing rather than to source-water quality. Central-tendency values across the full sample set do not summarize this within-population heterogeneity.
• Microbiome: Not addressed by the document.

Verification notes

• Cite-key uses the hc (Health Canada) agency abbreviation with the publication year (2020), consistent with other Health Canada source pages in wiki/sources/ (e.g., hc2008-aluminum-food-additives-industry-request, hc2008-review-dietary-exposure-aluminum).
• The sibling file drinking-water-quality-guideline-cadmium.pdf in the same folder is byte-identical (SHA-256 match) to raw_path; listed in near_duplicates.
• matrices includes general drinking-water descriptors (drinking-water, tap-water, source-water, raw-water, treated-water, distribution-water) that span the document’s data tables; no closer-fitting canonical matrix slug for treated/distributed water exists in the system-prompt list.
• No matching [[regulations/canada-cadmium-drinking-water-mac]] page exists yet in the taxonomy snapshot; the document IS the operative Canadian federal MAC for cadmium in drinking water. Flagged here as a backlog regulation-page candidate but NOT created (regulation pages require a hard agency identifier and a Karen-driven Step 0 Lock per CLAUDE.md Part 10).
• Canadian provincial table (Table 2) reproduces the document’s footnote citations 1–12 in the table header explanation; values are quoted exactly as the document reports them, including the visually unusual entries (e.g., Quebec mean 0.20 µg/L vs median 0.01 µg/L — a wide right tail; Alberta well median 15.0 µg/L > mean 13.4 µg/L — an artifact of the 0.30% detection rate with very small detected-fraction n).
• Document references but does not reproduce the underlying primary leaching datasets (Viraraghavan et al. 2000, Samuels and Meranger 1984, Schock and Neff 1988, Subramanian et al. 1991); Key numbers explicitly attributes leaching ranges to the document’s own citation chain rather than direct measurement here.
• No brand names appear in the source’s contamination data; vendor names in the analytical-methods table (U.S. EPA, APHA) are method-publishing standards bodies, not Part 12 violations.
• Page numbers in Key numbers cite the PDF’s internal page numbers (e.g., “p. 29-30” = the document’s Section 10.0 heading on PDF-page 29 of 50).
• Audit subagent (2026-06-04) flagged the original Statistics Canada 2015 age-trend block as “UCd GM 0.42 µg/L in adults 20–79 vs 0.12 µg/L in ages 3–19”; verified against source p. 8 — the source explicitly attributes these values to blood cadmium, not urinary cadmium. Block corrected to BCd and a separate sentence added noting the Garner and Levallois (2016) age-dependent UCd increases.
• Audit subagent (2026-06-04) flagged the Viraraghavan et al. (2000) copper-plumbing block for conflating “samples” (the three sequential within-round sample volumes — 125/500/125 mL — drawn per dwelling) with “rounds” (monthly repeats); verified against source p. 17 — finding correct. Block restructured to separate within-round samples (first/second/third sample 125/500/125 mL, values <DL–171/<DL–39/<DL–102 µg/L) from cross-round maxima (133, 101 µg/L in second and third rounds).
• Audit subagent (2026-06-04) flagged App-audience phrasing (“tail risk for app users with old housing stock”, “typical baseline”) as edging into the Part 2 wiki/HMTc firewall (consumer-risk-advisory framing and app-data threshold proposal); reframed to describe what the document’s dataset contains (median band, well/raw outliers, plumbing attribution) without prescribing a downstream HMI app default value or advisory.
• Audit subagent (2026-06-04) flagged “Oral LD₅₀ in rats” as narrower than source attribution; verified against source p. 25 — source says “in rats and mice”. Corrected.

Wiki pages this source may touch

• cadmium
• water
• bottled-drinking-water
• epa-iris-cadmium-rfd
• jecfa-cadmium-ptmi
• efsa-cadmium-twi

Page history

The five most recent substantive edits to this page. The full version history lives in git; when DOI minting comes online (see schema docs), each entry below will also link to a version-pinned DataCite DOI.