Hydrogen Sulfide Epidemiology and Respiratory Effects

RoseyFields evaluates the human epidemiological evidence for H2S as follows, drawing on occupational cohort and controlled-exposure studies, community and contamination-site cohorts, and case-series data published in the peer-reviewed literature.

1. Overview of Study Populations and Data Sources

The human evidence base for H2S is unusual among industrial chemicals in that it is dominated by community, rather than occupational, cohort studies, because several large population centers sit atop naturally emitting geothermal or geologic H2S sources and have been followed for decades using routine health administrative data. RoseyFields draws on four principal categories of study population, and treats their diversity as a strength rather than a limitation, because convergent findings across independent populations, exposure sources, and study designs increase confidence that an observed association reflects a real biological effect rather than an artifact of a single population’s confounding structure or exposure measurement approach.

Community/geothermal contamination-site cohorts. The single richest body of chronic, low-level H2S epidemiology comes from Rotorua, New Zealand, a city built on an active geothermal field and the subject of a sustained research program spanning mortality, cancer incidence, respiratory function, asthma, cognitive function, and ocular outcomes (Bates et al., 1997; Bates et al., 1998; Bates et al., 2002; Bates et al., 2013; Bates et al., 2015; Reed et al., 2014). A parallel, more recent research program has followed residents of the Mt. Amiata geothermal district in Tuscany, Italy, using a formally reconstructed residential exposure cohort of over 33,000 residents linked to regional mortality and hospital discharge registries (Nuvolone et al., 2019; Nuvolone et al., 2020). A third community data stream comes from populations living near industrial (rather than geologic) H2S sources: residents near a Finnish kraft pulp mill in South Karelia (Jaakkola et al., 1990; Haahtela et al., 1992; Marttila et al., 1994; Marttila et al., 1995; Partti Pellinen et al., 1996), and residents of two Nebraska communities near a meatpacking/rendering complex (Campagna et al., 2004). Independent community-based investigations by Legator and colleagues (Legator et al., 2001) and by Kilburn (Kilburn and Warshaw, 1995; Kilburn, 1997, 1999, 2004) examined additional community cohorts in the United States reporting chronic low-level exposure symptoms.

Occupational cohorts. Occupational data include a retrospective cohort of female petrochemical workers in Beijing, China, assessed for reproductive outcomes (Xu et al., 1998); Finnish pulp and paper industry workers assessed for respiratory function (Jäppinen et al., 1990); a cohort of workers with chronic occupational exposure assessed for olfactory function (Hirsch and Zavala, 1999); Norwegian sewage treatment and sewage-net workers assessed for sensory-motor neuropsychological function (Goffeng et al., 2023); and a national database of occupational H2S fatalities in the United States from 1984 to 1994 (Fuller and Suruda, 2000).

Controlled human exposure (experimental) studies. A distinct and methodologically important body of evidence comes from controlled inhalation exposure of healthy volunteers under precisely quantified, laboratory-verified H2S concentrations (0.5-10 ppm) during rest and exercise, conducted primarily by a single Canadian research group over the 1990s (Bhambhani and Singh, 1991; Bhambhani et al., 1994). Because exposure concentration is directly measured and controlled rather than estimated or self reported, these studies provide the most internally valid dose-response data available for H2S in humans, at the cost of examining only short-duration exposures in healthy volunteers rather than chronic, real world, or susceptible-population exposures.

Acute poisoning case series. A separate literature consists of clinical case series and case reports of individuals who survived high-concentration, life-threatening acute H2S exposure, generally in occupational settings, and were followed for persistent sequelae (Wasch et al., 1989; Tvedt et al., 1991a; Tvedt et al., 1991b; Snyder et al., 1995; Bassindale and Hosking, 2011).

General-population biomonitoring. RoseyFields did not identify a validated, population-scale biomonitoring program (comparable to NHANES biomonitoring for metals or many persistent organics) that tracks H2S-specific internal-dose biomarkers in the general population. This gap reflects a genuine toxicokinetic constraint rather than a research oversight: H2S is rapidly metabolized and does not form a stable, long-lived blood or urine biomarker suitable for population surveillance, and even in acute poisoning deaths, sulfide is unstable postmortem and degrades rapidly in stored samples (Bassindale and Hosking, 2011). RoseyFields treats the absence of a biomonitoring-based exposure metric as a material data gap (Section 5) rather than as evidence of an absence of exposure.

Diversity of populations as a strength. RoseyFields places particular weight on the fact that the strongest, most consistent H2S epidemiological findings (respiratory morbidity and mortality, described below) have been independently replicated across geographically and demographically distinct populations — geothermal-exposed New Zealanders, geothermal-exposed Italians, industrially-exposed Finns, and industrially-exposed Americans — using different study designs (cohort, case-crossover, cross-sectional, ecological) and different exposure assessment methods (passive monitoring networks, dispersion modeling, and fixed-site monitors). This cross-population, cross-design convergence is a classical Hill criteria consideration supporting causal inference, and is markedly stronger evidence than any single large study standing alone.

2. Comprehensive Overview of Health Outcomes Investigated in Humans, by Body System

Respiratory system — strong, consistent, quantitatively-modeled evidence. This is the best-supported target organ system in the human H2S literature. Multiple independent populations show concordant findings of respiratory symptoms, respiratory morbidity, and respiratory mortality associated with H2S exposure (see Section 3 for detail).

Ocular system — strong, consistent evidence for irritant symptoms; suggestive evidence for cataract. Eye irritation is one of the most consistently reported acute symptoms across community exposure studies (Haahtela et al., 1992; Partti-Pellinen et al., 1996), and a geothermal community investigation reported an association between chronic H2S exposure and cataract (Bates et al., 2002), though this finding has not been independently replicated in the other major geothermal cohorts.

Central nervous system — acute, high-concentration poisoning: strong, consistent evidence; chronic, low-level exposure: suggestive but inconsistent evidence. Persistent neurological sequelae following acute, high-concentration poisoning (memory, motor, visual, and auditory deficits; electrophysiological abnormalities) are well and consistently documented in case series (Wasch et al., 1989; Tvedt et al., 1991a; Tvedt et al., 1991b; Snyder et al., 1995). By contrast, chronic, low-level exposure and neurobehavioral/cognitive function is an area of genuine scientific disagreement: some community-based investigations report neurobehavioral deficits and mood symptoms in chronically exposed populations (Kilburn and Warshaw, 1995; Kilburn, 1997, 1999; Kilburn, 2004; Legator et al., 2001), while the single largest, most methodologically rigorous cognitive-function study conducted to date — a cross-sectional evaluation of 1,637 Rotorua residents using a comprehensive neuropsychological test battery and individually modeled 30-year exposure histories — found no evidence of cognitive impairment associated with chronic ambient H2S exposure (Reed et al., 2014).

Peripheral nervous system — suggestive evidence. The Mt. Amiata residential cohort study found an association between H2S exposure and hospitalization for disorders of the peripheral nervous system (Nuvolone et al., 2019; Nuvolone et al., 2020), and a Norwegian occupational study identified alterations in sensory-motor neuropsychological function among sewage workers (Goffeng et al., 2023). RoseyFields characterizes this endpoint as suggestive because it currently rests on a smaller number of studies than the respiratory endpoint and requires independent replication.

Cardiovascular system — predominantly null or inconsistent evidence. The Mt. Amiata residential cohort study — arguably the most methodologically advanced cardiovascular investigation available — found no positive association between H2S exposure and cardiovascular disease mortality or hospitalization (Nuvolone et al., 2019). Earlier Rotorua mortality data showed statistically significant elevations in standardized mortality ratios for specific circulatory sub-categories (rheumatic heart disease, hypertensive disease), but the original investigators themselves concluded that the data did not provide convincing evidence of elevated overall mortality in the exposed population, citing substantial concerns about confounding (Bates et al., 1997).

Cancer — inconsistent, and confounded by co-exposure in the one positive study. An ecological analysis of cancer registry and hospital discharge data for Rotorua found significantly elevated incidence of nasal and lung/respiratory tract cancers among geothermally-exposed residents, but the investigators explicitly noted that the elevated rates were also consistent with co-exposure to naturally co-emitted mercury, and that available exposure data were inadequate to support a causal conclusion specific to H2S (Bates et al., 1998). By contrast, the more recent and more rigorously exposure-modeled Mt. Amiata residential cohort study found no positive association between chronic H2S exposure and cancer incidence or mortality of any type (Nuvolone et al., 2019; Nuvolone et al., 2020).

Reproductive/developmental — suggestive evidence, single study. A large retrospective cohort study of female petrochemical workers in Beijing found a statistically significant, roughly two-fold increased odds of spontaneous abortion among the subgroup of women whose occupational exposure was specifically categorized as H2S alone (Xu et al., 1998). RoseyFields characterizes this as suggestive rather than strong evidence because it is a single study, the facility-wide exposure assessment did not include quantitative H2S air concentrations, and the broader cohort had concurrent exposure to numerous other petrochemical agents.

Olfactory function — strong, consistent evidence, and directly concordant with the animal database. Workers with chronic occupational H2S exposure showed persistently decreased olfactory function on formal olfactory testing compared to unexposed controls (Hirsch and Zavala, 1999). RoseyFields treats this as an important corroborating human finding because it targets precisely the same tissue (olfactory epithelium) identified as the most sensitive endpoint in the animal toxicology database underlying the EPA reference concentration.

Gastrointestinal, hepatic, immune, renal, and endocrine systems — inadequate or no dedicated human epidemiological data identified. RoseyFields did not identify dedicated human epidemiological studies designed around these organ systems as primary outcomes, beyond the inclusion of “digestive diseases”

as a reference category in the Nebraska hospital-visit time-series study, which did not identify a significant H2S association for that category (Campagna et al., 2004).

3. Strong-Evidence Outcomes: Key Studies, Exposure Metrics, Effect Sizes, and Cross Population Consistency

Respiratory morbidity and mortality. The Mt. Amiata residential cohort study is the most rigorously exposure-modeled respiratory investigation available. Using individually assigned, dispersion-modeled H2S exposure estimates for a cohort of 33,804 residents followed over roughly 391,000 person-years (1998-2016), the study found that high H2S exposure was associated with increased hazard of respiratory disease mortality (hazard ratio approximately 1.34) and with increased hospitalization for chronic obstructive pulmonary disease and disorders of the peripheral nervous system, with the respiratory finding showing a coherent dose-response gradient across mortality and hospitalization outcomes and consistency across sexes (Nuvolone et al., 2019). A companion case-crossover analysis of short-term (day to-day) H2S exposure in the same geothermal region found a comparable increased hazard for respiratory mortality (approximately 1.12) and for hospitalization specifically for COPD and peripheral nervous system disorders following high-exposure days (Nuvolone et al., 2020). In the Nebraska community study, elevated ambient H2S/total reduced sulfur was associated with increased same-day and lagged pediatric hospital visits for respiratory disease at several monitoring stations (increases in the range of roughly 6- 40%), with a more limited association seen in adults (Campagna et al., 2004). The South Karelia pulp mill studies, using a longitudinal daily-diary design in adults living near the mill, found a graded increase in eye, respiratory, and central-nervous-system symptom intensity as ambient total reduced sulfur compound concentrations rose across three predefined exposure bands, and a cross-sectional comparison of the same exposed community against a matched reference community found a significantly increased rate of self-reported cough, respiratory infection, and headache, with an adjusted odds ratio for headache of approximately 1.7-1.8 (Marttila et al., 1995; Partti-Pellinen et al., 1996). Earlier Rotorua mortality data, while methodologically more limited (ecological, unadjusted for individual exposure), also identified statistically significant excess standardized mortality ratios for chronic obstructive respiratory disease and for pneumonia/influenza among geothermally-exposed residents (Bates et al., 1997).

Cross-population consistency. RoseyFields regards the respiratory finding as the most defensible H2S health effect in the epidemiological literature specifically because it recurs, in a directionally consistent way, across four independent populations (Italian, New Zealand, Finnish, American), three distinct exposure sources (geothermal, pulp mill, meatpacking/rendering), and multiple study designs (cohort, case-crossover, longitudinal diary, cross-sectional, time-series), using both mortality and morbidity/hospitalization endpoints. The consistency of the specific sub-diagnosis (COPD and chronic respiratory disease, rather than, for example, restrictive lung disease) across these very different populations further strengthens the biological plausibility of the association, given the established portal of-entry respiratory irritant mechanism of H2S.

Olfactory dysfunction. Hirsch and Zavala (1999) is, to RoseyFields’ knowledge, the principal published human study directly measuring olfactory function in chronically H2S-exposed workers, and it found a persistent decrement compared to unexposed referents. While this is a single study and therefore does not have the same cross-population replication as the respiratory finding, RoseyFields assigns it substantial weight because of its direct mechanistic concordance with the animal toxicology database (Section 3 of the companion toxicology page), in which nasal olfactory epithelial injury is the most sensitive and best-replicated animal finding.

4. Study Quality Considerations

RoseyFields’ confidence in each outcome category above is calibrated against the following recurring methodological limitations, which vary in severity across the literature:

Confounding. Ecological and cohort studies of geothermal and industrial communities face substantial risk of confounding by co-exposure to other pollutants released from the same source. This is most acute for the Rotorua cancer finding, where the study authors themselves attributed the elevated cancer incidence potentially to co-emitted mercury rather than to H2S specifically (Bates et al., 1998), and is a live concern for the Mt. Amiata cohort as well, given that the same geothermal district has an independent history of mercury mining and arsenic-contaminated groundwater (Nuvolone et al., 2019). The earlier Rotorua mortality analysis was also confounded by ethnicity and smoking-prevalence differences between the geothermally-exposed and reference populations, a limitation the original investigators explicitly acknowledged in concluding that their SMR findings did not amount to convincing evidence of a mortality effect (Bates et al., 1997).

Exposure misclassification. Most community studies rely on ambient monitoring networks or dispersion modeling to assign area-level or individually-modeled exposure, rather than direct personal or biological exposure measurement, because no validated H2S biomarker exists for retrospective exposure reconstruction (Section 1). Studies using more sophisticated individually-linked exposure reconstruction (residential/workplace/school geocoding combined with monitoring networks, as in the Rotorua and Mt. Amiata programs) substantially reduce, but do not eliminate, this concern relative to purely ecological (area-level) designs. The Xu et al. (1998) reproductive study is a clear example of the limitation at its most severe: H2S exposure was assessed only categorically (exposed to H2S alone versus other petrochemicals) based on workshop assignment, with no quantitative air concentration data at all.

Reverse causation. This is a less prominent concern for the respiratory and reproductive outcomes discussed above, which use prospective or retrospective cohort designs with exposure preceding outcome ascertainment, but is a more relevant caveat for cross-sectional symptom-survey designs (e.g., the South Karelia cross-sectional comparison and portions of the Rotorua asthma/lung-function work), where self selection into or out of a chronically malodorous community based on pre-existing respiratory sensitivity cannot be fully excluded.

Co-exposure to related reduced-sulfur compounds. Because H2S is essentially never emitted in isolation from industrial or pulp-mill sources — it is invariably accompanied by other malodorous reduced-sulfur compounds such as methyl mercaptan and dimethyl sulfide — several of the strongest community studies (the South Karelia series in particular) formally measured and modeled “total reduced sulfur” rather than H2S alone, meaning the observed symptom and respiratory associations are properly attributed to the reduced-sulfur compound mixture rather than isolated to H2S with certainty (Jaakkola et al., 1990; Marttila et al., 1995; Partti-Pellinen et al., 1996). RoseyFields treats this as a source of residual uncertainty specific to attributing effects to H2S individually, while noting that H2S is typically the dominant and most toxicologically active component of this compound class.

Small sample size and single-study reliance. The reproductive (Xu et al., 1998), ocular cataract (Bates et al., 2002), and peripheral neuropathy (Nuvolone et al., 2019; Goffeng et al., 2023) findings each currently rest on a comparatively narrow evidentiary base (one to two studies), in contrast to the multiply replicated respiratory finding, and RoseyFields’ confidence in these associations is calibrated accordingly as “suggestive” rather than “strong.”

Strength of the negative/null cognitive-function finding. RoseyFields gives particular methodological weight to the null finding in Reed et al. (2014) specifically because of its size (1,637 participants), its individually-modeled 30-year exposure reconstruction, and its use of an objective neuropsychological test battery rather than self-reported symptoms — design features that directly address the confounding and exposure-misclassification limitations present in the earlier, smaller, positive-finding community neurobehavioral studies (Kilburn and Warshaw, 1995; Kilburn, 1997, 1999).

5. Independent Expert Panel and Agency Weight-of-Evidence Conclusions

RoseyFields is not aware of any formal, independent multi-disciplinary science panel weight-of-evidence determination for H2S analogous to, for example, the “probable link” framework used by independent science panels evaluating some other chemicals. No such determination — for any human health outcome — has been issued by an independent expert panel for H2S. The closest available syntheses are: (a) a peer reviewed systematic review of the epidemiological literature on chronic low-level H2S exposure, which concluded that the human evidence for adverse chronic health effects at ambient community exposure levels remains inconsistent across outcome categories, with the clearest support for respiratory and irritant effects and weaker, more inconsistent support for neurobehavioral and cardiovascular outcomes (Lewis and Copley, 2015); and (b) periodic regulatory agency toxicological profile updates (which RoseyFields consults for literature-identification purposes but does not cite as primary evidentiary sources, consistent with RoseyFields’ practice of citing underlying peer-reviewed studies rather than secondary regulatory compilations). RoseyFields’ own weight-of-evidence characterization in Section 2 above (strong/suggestive/null by outcome) is derived directly from the primary peer-reviewed literature rather than from any single synthesizing panel determination, and RoseyFields recommends that any case specific expert report state plainly that no independent panel “probable link” determination exists for H2S, to avoid overstating the formality of the evidence base relative to chemicals that have received such a determination.

6. Use of Epidemiological Evidence, Alongside Toxicology, in Litigation Causation Analyses

General causation. RoseyFields’ general causation analyses for H2S rely on a weight-of-evidence synthesis combining the animal toxicology database (portal-of-entry nasal injury, and the systemic/mitochondrial mechanism at high concentrations; see the companion toxicology page) with the human epidemiological findings summarized above. For respiratory disease, RoseyFields regards general causation as reasonably well supported by the convergence of independent human cohort/case-crossover findings (Nuvolone et al., 2019; Nuvolone et al., 2020) with a mechanistically consistent, independently replicated animal irritant effect. For outcomes in the “suggestive” or “null/inconsistent” categories — cardiovascular disease, cancer, chronic-exposure cognitive/neurobehavioral effects, and reproductive effects — RoseyFields does not treat the current literature as supporting a general causation opinion without substantial additional case-specific analysis, and flags the specific limitation driving the “suggestive” or “inconsistent” characterization (single-study reliance, confounding, or a contradicting higher-quality study) in any report addressing these outcomes.

Specific causation. Consistent with standard toxic tort methodology, RoseyFields treats general causation support (an established capacity of H2S to cause a given health outcome in a relevantly exposed population) as necessary but not sufficient for a specific causation opinion regarding an individual plaintiff. Specific causation analyses additionally require case-specific dose reconstruction (see the companion risk assessment methodology page), a temporally appropriate exposure-to-onset relationship, differential diagnosis addressing alternative causes, and, where available, individual biomarkers or clinical findings (e.g., documented acute poisoning severity, or formal olfactory or pulmonary function testing) consistent with the toxicological and epidemiological mechanism relied upon.

Distinguishing acute and chronic causation frameworks. Because the acute poisoning case-series literature (strong, mechanistically well-understood) and the chronic low-level community exposure literature (more heterogeneous) rest on different exposure regimes and, in places, different outcome sets, RoseyFields maintains separate causation frameworks for acute, high-concentration exposure events and for chronic, community-level exposure claims, and does not apply findings from one exposure regime to support causation opinions in the other without an explicit, case-specific justification.

Rebuttal use. RoseyFields also uses this evidence base to evaluate opposing expert reports, in particular to test whether a report: (a) relies on outcome categories (e.g., cancer, cardiovascular disease, chronic cognitive impairment) for which the weight of the current epidemiological evidence is null or inconsistent, without disclosing the contradicting higher-quality studies identified above; (b) attributes a community symptom or health finding to H2S specifically where the underlying study measured only total reduced sulfur or a mixed exposure; or (c) invokes a “probable link” or comparable formal panel determination that, to RoseyFields’ knowledge, does not exist for this chemical.

RoseyFields updates this epidemiology page as new cohort follow-up data (particularly from the ongoing Mt. Amiata and Rotorua research programs) and new systematic reviews become available.

Bibliography