Hexavalent Chromium (Cr-VI) Epidemiology and Lung Cancer

RoseyFields evaluates the human epidemiological evidence for hexavalent chromium as follows.

Study Populations and Data Sources

RoseyFields’ epidemiological review of hexavalent chromium [Cr(VI)] draws on four distinct categories of human population data, and treats the diversity of these sources—not any single study—as the foundation for causal inference.

Occupational cohorts are the backbone of the Cr(VI) evidence base and include chromate production workers in Baltimore, Maryland (Gibb et al., 2000b; Gibb et al., 2020) and Painesville, Ohio (Luippold et al., 2003; Proctor et al., 2016); chrome plating and electroplating workers across multiple countries (Lindberg and Hedenstierna, 1983; Bright et al., 1997); stainless steel and mild-steel welders (Bonde and Ernst, 1992; Kumar et al., 2005); and tannery workers exposed to chromium-based tanning agents (Khan et al., 2013). These cohorts offer well-characterized, often quantitative exposure histories (air monitoring, urinary or blood chromium, job-exposure matrices), long follow-up periods, and—critically—exposure intensities well above general-population background, which gives these studies the statistical power to detect effects that would be undetectable in the general population.

Community and contamination-site cohorts complement the occupational data by capturing non-occupational, often lower-level, chronic environmental exposure, typically via contaminated drinking water. The most extensively studied such population is a set of villages in Jinzhou, Liaoning Province, China, situated near a ferrochromium factory that contaminated local groundwater with Cr(VI) for decades; this population has been analyzed and re-analyzed multiple times using the same underlying mortality data (Zhang and Li, 1987; Beaumont et al., 2008), a history discussed further in Section 5. These community studies are essential precisely because they capture the ingestion pathway and general-population susceptibility (including children and the elderly) that occupational cohorts—by their nature, healthy working-age adults—cannot.

General-population biomonitoring studies, most notably the U.S. National Health and Nutrition Examination Survey (NHANES), provide cross-sectional blood and urinary chromium measurements in a nationally representative sample, along with a wide range of concurrently measured health indicators (cardiovascular, metabolic, renal). These studies are valuable for characterizing background exposure distributions and generating hypotheses, but—as discussed in Section 5—chromium’s short biological half-life and NHANES’s single-timepoint, cross-sectional design limit their ability to support causal inference for chronic disease outcomes on their own.

Finally, a smaller set of clinical and case-series reports—acute poisoning cases, dermatological patch-testing series, and small cross-sectional worker panels assessing biomarkers of genotoxicity or organ-specific injury—rounds out the evidence base, offering mechanistic and hazard-identification value even where they lack the statistical power or design rigor for quantitative risk estimation.

RoseyFields’ assessors treat the diversity of these population types as a genuine strength rather than a mere inventory requirement. Consistency of an association across occupational cohorts on different continents, community populations exposed via an entirely different route (ingestion versus inhalation), and independent research groups using different methods and adjusting for different confounders is one of the classic Bradford Hill considerations supporting causation, and it is considerably more persuasive than a strong finding replicated only within a single research group’s own cohort. Conversely, where an association is seen in only one population type or one research group’s cohort—as is true for several outcomes discussed below—RoseyFields’ assessors treat that isolation as a genuine limitation on causal inference, not merely a data gap to be filled later.

Health Outcomes by Body System and Strength of Evidence

RoseyFields’ review organizes the full range of human health outcomes investigated for Cr(VI) into three evidentiary tiers, consistent with how EPA’s 2024 toxicological review and independent systematic reviews (Hessel et al., 2021; den Braver-Sewradj et al., 2021) have characterized the same literature.

Respiratory system. Ulceration and perforation of the nasal septum is the strongest, most consistent, and most extensively quantified human health effect of inhaled Cr(VI), documented across occupational cohorts in multiple countries with a clear exposure-response relationship (Gibb et al., 2000a; Cohen et al., 1974; Lindberg and Hedenstierna, 1983). Occupational asthma and respiratory sensitization to Cr(VI) is supported by well-documented case series and outbreak investigations (Bright et al., 1997) and is now a recognized, if less quantitatively characterized, occupational entity. Lung cancer from inhaled Cr(VI) is supported by strong, consistent, quantitatively modeled evidence across multiple independent cohorts and is discussed in detail in Section 4. Decrements in spirometric lung function and chronic bronchitis-type symptoms are reported in some but not all occupational cohorts and are best characterized as suggestive.

Gastrointestinal system. Stomach and other gastrointestinal cancers following oral Cr(VI) exposure represent the paradigm case of suggestive-but-inconsistent human evidence, discussed in detail in Section 4 and Section 5. Direct caustic and irritant GI effects (abdominal pain, ulceration, hemorrhage) are well documented but almost exclusively in acute, high-dose ingestion case reports (accidental or intentional poisoning) rather than in chronic low-level exposure cohorts, limiting their relevance to typical environmental or occupational exposure scenarios.

Dermatologic system. Allergic contact dermatitis from Cr(VI) sensitization is among the best-established human health effects of chromium exposure of any valence state, with a large, consistent literature spanning decades of clinical patch-testing series across multiple countries (Bregnbak et al., 2015) and well-documented historical associations with cement work, leather tanning, and electroplating. RoseyFields’ assessors treat this as strong, consistent evidence, notwithstanding that it reflects an immunologic sensitization mechanism rather than the systemic toxicity pathways relevant to most other outcomes discussed here.

Renal system. Sub-clinical proximal tubular injury—elevated low-molecular-weight urinary proteins such as beta-2- microglobulin and retinol-binding protein—is reported with reasonable consistency across occupational cohorts at sufficiently high cumulative exposure (Franchini and Mutti, 1988), representing suggestive-to-moderate evidence of a genuine but subclinical effect; progression to overt clinical nephropathy in occupationally exposed humans is rare and reported mainly in acute high-dose poisoning cases.

Hematological system. In contrast to the well-documented microcytic, hypochromic anemia produced by Cr(VI) in rodent bioassays, human occupational evidence for hematological effects is inconsistent: some cohorts report reduced hemoglobin, hematocrit, and red cell counts (Khan et al., 2013), while other similarly exposed cohorts report no significant hematological differences from controls. RoseyFields’ assessors characterize this as an outcome with suggestive but materially inconsistent human evidence.

Male reproductive system. Evidence is mixed and cohort-dependent: some studies of welders and industrial workers report reduced sperm concentration and motility correlating with blood or urinary chromium (Kumar et al., 2005), while a comparably designed study of TIG and mild-steel welders found no semen-quality deficits with increasing internal chromium exposure (Bonde and Ernst, 1992). This is a clear example of suggestive-but-inconsistent evidence, likely reflecting differences in co-exposure (welding fume composition varies substantially by process and base metal), exposure intensity, and study power.

Female reproductive and developmental outcomes. Human evidence is inadequate to assess female reproductive toxicity; the available literature consists of a single low-confidence epidemiology study and is not corroborated by higher-confidence data. Developmental outcomes (birth weight, growth, congenital anomalies) following maternal Cr(VI) exposure have not been adequately studied in humans; the hazard identification for this outcome rests almost entirely on animal bioassay data.

Hepatic system. Human epidemiological data on Cr(VI)-associated liver injury are sparse; the hazard identification for hepatic effects, as with developmental toxicity, is substantially animal-bioassay-driven, with only limited, largely cross-sectional human biomarker data available.

Immune system (non-sensitization). Beyond the well-established sensitization/allergic pathway discussed above, evidence for broader Cr(VI)-related immunomodulation or immunosuppression in humans is limited to small biomarker studies and is best characterized as inadequate for firm conclusions.

Cardiovascular and metabolic outcomes. Cross-sectional NHANES analyses have reported associations between blood chromium and cardiovascular disease, diabetes, or depression, but—as discussed in Section 5—these associations are difficult to interpret causally given chromium’s essential-nutrient role at trace levels, the cross sectional design, and the absence of speciation between Cr(VI) and the nutritionally essential Cr(III) in most blood chromium assays; RoseyFields’ assessors treat this literature as inadequate to support either a hazard or a protective causal conclusion regarding Cr(VI) specifically.

Detailed Evidence for the Strong/Consistent Outcomes

Lung cancer (inhalation). The single most extensively replicated human health effect of Cr(VI) is lung cancer mortality among chromate production, chrome pigment, and chrome plating workers. The Baltimore, Maryland chromate production cohort (2,354 workers first employed 1950–1974) demonstrated a clear exposure-response relationship between cumulative Cr(VI) exposure and lung cancer mortality (Gibb et al., 2000b), and a subsequent re analysis found that the relative risk of lung cancer mortality varied significantly by age at exposure, with the youngest exposed age group showing the highest hazard ratio (Gibb et al., 2020). The Painesville, Ohio chromate production cohort, followed through 2011 (average follow-up 34.4 years; 24,535 person-years), found a standardized mortality ratio for lung cancer of 186 (95% CI 145–228) among workers hired before 1959, when exposures were highest (Proctor et al., 2016), corroborating an earlier update of the same cohort (Luippold et al., 2003). A comprehensive meta-analysis pooling 47 cohort studies published between 1985 and 2016 found Cr(VI) exposure was associated with elevated respiratory cancer mortality overall (summary SMR = 1.33; 95% CI 1.19–1.48) (Deng et al., 2019). RoseyFields’ assessors regard this body of evidence—independent cohorts on two continents, multiple industrial processes, a demonstrated exposure-response gradient, biological plausibility from the mutagenic mode of action, and a quantitative meta-analytic summary estimate—as satisfying essentially every Bradford Hill consideration for causation, which is why lung cancer is the one Cr(VI) health outcome supported by formal agency cancer-slope-type quantification (discussed in Section 6) rather than qualitative hazard characterization alone.

Nasal septum ulceration and perforation (inhalation). The second strongly established outcome is direct, dose related irritant and corrosive injury to the nasal mucosa and septum, observed with striking consistency in occupationally exposed cohorts studied by independent research groups: American chromate production workers (Gibb et al., 2000a), a large series of electroplate workers (Cohen et al., 1974), and Swedish chrome-plating workers (Lindberg and Hedenstierna, 1983) all documented nasal irritation, ulceration, or septal perforation at broadly overlapping air concentrations, with incidence rising sharply with exposure intensity and with nasal septal perforation reported in roughly two-thirds of workers exposed to peak concentrations of 20 µg/m³ or more in the Swedish cohort. Because this effect is a direct contact/corrosive injury rather than a systemic or stochastic effect, the close agreement across independent cohorts and countries is unsurprising mechanistically, but it is nonetheless the human evidentiary basis EPA uses directly for the noncancer inhalation reference concentration, as RoseyFields’ companion risk assessment methodology page for Cr(VI) describes.

Allergic contact dermatitis (dermal, any route of skin contact). Chromium sensitization and allergic contact dermatitis represent a large, mechanistically distinct, and highly consistent body of clinical evidence. Patch-testing surveys across multiple countries and decades consistently identify chromium as one of the most common metal allergens, with sensitization prevalence in dermatitis-clinic populations reported at roughly 2 to 4% in Scandinavian series and considerably higher in some U.S. dermatitis populations (Bregnbak et al., 2015). The historical epidemiology of this outcome also provides an unusually clean natural experiment: Sweden’s 1989 legislation reducing Cr(VI) content in cement was followed by a marked decline in cement-related occupational contact dermatitis, a before-after pattern that itself constitutes meaningful evidence of a causal, dose-dependent relationship. RoseyFields’ assessors note that while sensitization and dermatitis are not life-threatening outcomes, the strength and consistency of this evidence base makes it useful for corroborating that Cr(VI) reaches and biologically interacts with human tissue at occupationally realistic exposure levels—a point sometimes relevant to broader causation arguments in litigation.

Study Quality Considerations

RoseyFields’ assessors evaluate every Cr(VI) epidemiological finding against a standard set of validity threats, and the chemical’s literature offers unusually instructive examples of several of them.

Confounding and co-exposure are pervasive in the occupational literature. Chromate production, welding, and chrome plating workers are frequently co-exposed to nickel, cadmium, asbestos, silica, or diesel exhaust, any of which independently affects lung cancer risk or other outcomes under study; RoseyFields’ assessors specifically scrutinize whether a given cohort’s exposure was reasonably specific to Cr(VI) or confounded by these co-exposures before relying on its quantitative effect estimate. Smoking is a related and pervasive confounder for lung cancer specifically; well-designed cohort studies (e.g., Gibb et al., 2000b; Proctor et al., 2016) address this through internal comparison and standardized mortality ratios referenced to appropriate populations, but older or smaller studies frequently lack adequate smoking data, and RoseyFields’ assessors weight studies accordingly.

Exposure misclassification is a second major limitation, and the Cr(VI)-specific literature illustrates why it matters: total chromium measurements—common in older industrial hygiene and environmental datasets—do not distinguish the substantially more toxic Cr(VI) from the far less toxic, indeed nutritionally essential, Cr(III), so studies relying on total chromium (including much of the NHANES cardiometabolic literature) cannot support Cr(VI)-specific causal conclusions without additional speciation data or reasonable process-based inference (e.g., that a given industrial bath or process is essentially pure Cr(VI)). Reconstructing historical occupational exposure decades after the fact— necessary in most of the chromate production cohorts discussed above—introduces further, generally non-differential misclassification that tends to bias effect estimates toward the null, meaning the true effect may be understated rather than overstated in many of these cohorts.

The Jinzhou, China drinking-water cohort is RoseyFields’ primary illustration of how the same underlying data can support opposite conclusions depending on analytic choices. The original 1987 Chinese-language report found higher stomach and lung cancer mortality in Cr(VI)-contaminated villages than in “the general population” without formal statistical testing (Zhang and Li, 1987). When Beaumont et al. (2008) later obtained more granular village-level data and applied modern statistical methods, they found an association between drinking-water Cr(VI) and stomach cancer mortality; when other investigators reanalyzed essentially the same underlying data using differently selected comparison villages, they reported no significant increase. RoseyFields’ assessors present this history not to dismiss the Jinzhou findings but to illustrate, concretely, how comparison-group selection, exposure-boundary definitions, and small-area mortality statistics in a historical, non-Western dataset can each independently drive a study toward or away from statistical significance—precisely the kind of methodological sensitivity that warrants caution before treating any single reanalysis as definitive, and that explains why EPA’s own 2024 characterization of oral-route human cancer evidence is “slight” rather than “strong,” notwithstanding a “robust” animal database.

Reverse causation is a more significant concern for the general-population biomonitoring literature (e.g., NHANES cardiometabolic associations) than for the occupational cancer and nasal-injury literature, since a prospective, decades-long occupational cohort with mortality as the endpoint is not vulnerable to the concern that pre-existing disease altered exposure; a cross-sectional blood-chromium-versus-current-health-status comparison is considerably more vulnerable to this concern, along with the related problem that chromium’s short blood and urine half-life means a single biomonitoring measurement may poorly represent an individual’s relevant historical exposure window for a chronic disease with a long latency.

Weight-of-Evidence and Agency Determinations

Independent expert panels and regulatory agencies have reached converging, though not identical, weight-of-evidence conclusions that RoseyFields’ assessors treat as an important cross-check on the primary literature review above, without treating any single agency determination as a substitute for that review. The International Agency for Research on Cancer classifies chromium(VI) compounds as Group 1, carcinogenic to humans, based on sufficient evidence in both humans and experimental animals together with strong supporting mechanistic evidence—a classification made in its 1990 monograph and not altered in subsequent IARC reviews. The U.S. National Toxicology Program’s Report on Carcinogens likewise lists chromium hexavalent compounds as known to be a human carcinogen. OSHA and NIOSH have independently concluded that occupational Cr(VI) exposure causes lung cancer in humans, a conclusion that underpinned OSHA’s 2006 rulemaking substantially lowering the permissible exposure limit. EPA’s most recent (2024) IRIS toxicological review distinguishes route of exposure in its cancer classification—”carcinogenic to humans” via inhalation, reflecting the long-standing, consistent epidemiologic lung cancer evidence just described, versus only “likely to be carcinogenic to humans” via the oral route, reflecting a database EPA itself characterizes as having “robust” animal evidence but only “slight” direct human evidence for GI tract cancer. California’s Office of Environmental Health Hazard Assessment lists Cr(VI) compounds as carcinogens under Proposition 65. Two recent systematic reviews conducted for the Dutch National Institute for Public Health and the Environment independently reached materially consistent conclusions: that Cr(VI) causes lung cancer and sinonasal cancer in humans, is suspected of causing stomach and laryngeal cancer, and causes a defined set of irreversible non-cancer effects (nasal, respiratory, dermal, ocular, dental) via the occupational literature reviewed above (Hessel et al., 2021; den Braver-Sewradj et al., 2021).

Application to Litigation: General and Specific Causation

RoseyFields’ assessors use this epidemiological evidence, alongside the toxicological dose-response methodology described in the companion Cr(VI) risk-assessment page, to support both general and specific causation opinions, while being careful not to conflate the two.

For general causation—whether Cr(VI) exposure is capable of causing a given disease in humans at all—the strong, multi-cohort, quantitatively consistent lung cancer and nasal-injury evidence described in Section 4 provides a well supported foundation largely independent of any single study, reinforced by the converging agency and expert-panel determinations in Section 6 and by the mechanistic and animal-toxicological evidence developed in RoseyFields’ companion methodology page. For outcomes in the suggestive-but-inconsistent tier—GI cancer, hematological effects, male reproductive effects—RoseyFields’ assessors present general causation opinions more cautiously, explicitly identifying the specific sources of inconsistency (co-exposure, exposure misclassification, comparison group selection) discussed in Section 5, since a general causation opinion resting on suggestive-but-contested epidemiology is considerably more vulnerable to Daubert-type challenge than one resting on the lung cancer or nasal injury literature.

For specific causation—whether Cr(VI) exposure caused a given disease in a particular individual—epidemiological relative risk and attributable risk estimates from the strong-evidence-tier studies are combined with individualized exposure and dose reconstruction (as described in RoseyFields’ risk-assessment methodology), latency analysis (time from first substantial exposure to diagnosis, benchmarked against the exposure-response and latency patterns in the underlying cohorts), differential diagnosis addressing alternative causes (smoking history, co-exposures, family history, idiopathic disease base rates), and, where relevant, biomarker or clinical findings (e.g., documented nasal septal pathology, tubular proteinuria) that corroborate a substantial individual exposure. RoseyFields’ assessors are explicit, in every specific-causation work product, about which tier of evidence supports the opinion being offered and disclose the same study-quality limitations described in Section 5 as they apply to the individual case, since an epidemiological foundation that is honestly characterized—strong for some outcomes, genuinely contested for others—is both more scientifically defensible and more useful to the retaining party than one that overstates the consistency of a mixed literature.

Bibliography