Hydrogen Sulfide Risk Assessment and Respiratory Effects

RoseyFields conducts H2S risk assessments using the following methodology, which draws on the peer reviewed toxicological literature underlying EPA’s Integrated Risk Information System (IRIS) inhalation reference concentration for hydrogen sulfide, and applies established dose-response and uncertainty factor methods to site- and case-specific exposure scenarios.

1. Chemical Identity, Regulatory Background, and Exposure Sources

Hydrogen sulfide (H2S; CASRN 7783-06-4) is a colorless, flammable gas with a characteristic “rotten egg” odor at low concentrations, commonly known by the synonym “sewer gas.” It has a molecular weight of 34.08 g/mol and is highly water-soluble and reactive, decomposing readily in biological tissue and in the atmosphere. Because of its reactivity and short atmospheric half-life, H2S exposure is typically evaluated as a localized, source-proximate hazard rather than a persistent regional pollutant.

H2S occurs naturally from the anaerobic decomposition of organic (sulfur-containing) matter and is released from a wide range of industrial and occupational sources that RoseyFields routinely encounters in litigation and regulatory support work, including:

• Oil and natural gas production, processing, and refining (“sour gas” operations), where H2S is a co-produced or process gas;
• Wastewater treatment plants, sewers, and lift stations, where anaerobic microbial activity generates H2S from sulfate-reducing bacteria;
• Concentrated animal feeding operations (CAFOs) and manure lagoons;
• Pulp and paper (kraft process) manufacturing;
• Landfills and solid waste facilities;
• Geothermal energy operations; and
• Certain mining and smelting operations.

Community and occupational exposures occur almost exclusively by inhalation; dermal and incidental oral exposure to H2S itself are not significant exposure pathways in most RoseyFields case work, and no validated oral toxicity value exists for chronic exposure (see Section 6). H2S has a low odor detection threshold (on the order of low ppb), but at high concentrations it paralyzes the olfactory nerve, so odor perception is an unreliable indicator of exposure at hazardous concentrations — a point RoseyFields flags in any dose-reconstruction relying on odor-complaint data. Acute, high-concentration exposures (500- 1,000 ppm) are immediately life-threatening, producing loss of consciousness, respiratory paralysis, and death within minutes (Allyn, 1931; Milby, 1962). RoseyFields’ methodology below addresses long-term, lower-level (chronic/subchronic) inhalation exposures, which are the exposure scenarios most often at issue in community and long-term occupational settings; acute lethality thresholds are treated as a separate, qualitative hazard screen.

There is no EPA-derived oral reference dose (RfD) currently available for H2S; the RfD previously listed on IRIS (3 x 10^-3 mg/kg-day, derived from a subchronic oral pig study) has been formally withdrawn (see Section 3), and RoseyFields does not use it in current assessments. The chronic inhalation reference concentration (RfC) is the operative toxicity value for H2S risk assessment and is the focus of Sections 3-6 below. No cancer slope factor or inhalation unit risk value has been derived for H2S (Section 7).

2. Point-of-Departure (POD) Selection

RoseyFields’ selection of the point of departure for the chronic inhalation RfC is based on the subchronic inhalation study of adult male Crl:CD rats conducted by Brenneman and colleagues (Brenneman et al., 2000). In that study, 10-week-old male rats (12 per group) were exposed by whole-body inhalation to 0, 10, 30, or 80 ppm H2S for 6 hours/day, 7 days/week, over a 10-week period, after which nasal passages were examined histologically at six standardized cross-sectional levels for lesions of the olfactory epithelium, graded on a 0 (normal) to 4 (severe) severity scale.

Brenneman et al. (2000) reported no treatment-related lesions at 10 ppm; at 30 and 80 ppm, animals developed multifocal, bilaterally symmetrical loss of olfactory neurons and basal cell hyperplasia in the dorsal medial meatus and the dorsal/medial ethmoid recess, with severity ranging from mild to severe and increasing with both dose and rostral-to-caudal position in the nasal cavity. On this basis, RoseyFields’ POD is a no-observed-adverse-effect-level (NOAEL) of 10 ppm (13.9 mg/m3) and a corresponding lowest observed-adverse-effect-level (LOAEL) of 30 ppm (41.7 mg/m3) for olfactory epithelial injury.

RoseyFields selects nasal (olfactory) lesions as the critical effect, and the Brenneman et al. (2000) NOAEL as the principal POD, for the following reasons, each grounded in the corroborating literature:

Reproducibility across independent studies. The same class of nasal lesion was reported independently by the Chemical Industry Institute of Toxicology’s 90-day inhalation bioassays in mice, Fischer 344 rats, and Sprague-Dawley rats (CIIT, 1983a; CIIT, 1983b; CIIT, 1983c), and again by Dorman and colleagues in a one generation reproductive/developmental study (Dorman et al., 2000). This cross-study, cross-strain, cross species concordance supports a robust and reproducible portal-of-entry effect rather than a strain-specific artifact.

Consistency with the known irritant mechanism of H2S. H2S is a direct-acting mucosal irritant, and portal of-entry injury to the nasal olfactory epithelium — the tissue with the greatest first-pass contact with inhaled H2S — is mechanistically the expected target organ, distinguishing it from more equivocal, indirect neurologic endpoints (discussed below).

Lower effect level than other subchronic bioassays. The Brenneman et al. (2000) NOAEL/LOAEL pair (10/30 ppm) is lower than the effective concentrations identified in the CIIT (1983a) 90-day mouse bioassay (in which nasal mucosal inflammation was seen only at 80 ppm), making the Brenneman study the more health-protective and more sensitive basis for the POD.

A weaker, non-adverse alternative endpoint exists at similar or lower exposure levels. A parallel body of literature reports biochemical and morphological changes in the developing rodent central nervous system at exposure concentrations in the same general range as the nasal lesion NOAEL/LOAEL — for example, altered neurotransmitter levels in postnatal rat pups exposed in utero and postnatally to H2S (Skrajny et al., 1992), and altered dendritic arborization of cerebellar Purkinje cells in pups exposed during gestation and early postnatal development (Hannah and Roth, 1991). RoseyFields treats these neurodevelopmental findings as important supporting/sensitivity information (see Sections 5 and 9) but not as the primary critical effect, because: (a) the causal link between the reported morphological/biochemical changes and a functional neurobehavioral deficit has not been established; (b) Dorman et al. (2000), examining a broader battery of functional endpoints (motor activity, acoustic startle, passive avoidance, physical/reflex developmental landmarks, and histopathology of nervous tissue) in pups exposed to the same concentration range, found no treatment-related functional or structural effects; and (c) the nasal lesion endpoint is unambiguously adverse on conventional pathology criteria, whereas the neurologic changes are characterized in the literature as “possible” rather than “clearly adverse” indicators of neurotoxicity. RoseyFields nonetheless carries these neurodevelopmental studies forward explicitly in the database uncertainty-factor discussion (Section 5) because the exposure levels associated with the two endpoint classes converge, which increases confidence that a protective value for the nasal effect is likely to also be protective of the neurologic findings — but this convergence is treated as a source of residual uncertainty, not as an independent line of evidence eliminating it.

RoseyFields does not use benchmark dose (BMD) modeling for this chemical, because a BMD/BMCL was not derived from the Brenneman et al. (2000) dataset in the underlying study or the supporting literature; the assessment instead uses the traditional NOAEL/LOAEL approach.

3. Pharmacokinetic Modeling and Dose Conversion (Animal-to-Human, Duration Adjustment)

Because Brenneman et al. (2000) exposed animals discontinuously (6 hours/day, 7 days/week) rather than continuously, and because H2S is a Category 1 (highly water-soluble, reactive) gas whose primary target is the extrathoracic (nasal) region, RoseyFields applies the standard two-step dosimetric adjustment set out in EPA’s 1994 inhalation dosimetry methodology to convert the animal NOAEL to a human equivalent concentration (HEC):

Step 1 — Duration adjustment. The experimental NOAEL is adjusted from an intermittent to a continuous (24-hour/7-day) exposure basis:

NOAEL(ADJ) = 13.9 mg/m3 x (6 hr/24 hr) x (7 day/7 day) = 3.48 mg/m3

Step 2 — Dosimetric (regional gas dose ratio, RGDR) adjustment. Because H2S is a reactive, water-soluble gas producing a portal-of-entry effect in the extrathoracic (nasal) region, the appropriate dosimetric adjustment uses the ratio of minute ventilation to extrathoracic surface area in the rat versus the human (the Regional Gas Dose Ratio for the extrathoracic region, RGDRET), rather than a blood:gas partition coefficient-based model (which would be used for a systemically-acting, respiratory-region gas). Using default physiological parameters (rat minute ventilation VE = 0.19 L/min, rat extrathoracic surface area SAET = 15 cm2; human VE = 13.8 L/min, human SAET = 200 cm2):

RGDRET = (VE/SAET)rat / (VE/SAET)human = (0.19/15) / (13.8/200) = 0.184

NOAEL(HEC) = NOAEL(ADJ) x RGDRET = 3.48 mg/m3 x 0.184 = 0.64 mg/m3

This HEC (0.64 mg/m3) is the dosimetrically-adjusted point of departure that RoseyFields carries forward into uncertainty factor application. RoseyFields notes that this approach assumes the RGDRET methodology (validated primarily for portal-of-entry respiratory irritants) reasonably approximates cross species differences in local nasal tissue dose for H2S; no compound-specific physiologically-based pharmacokinetic (PBPK) model has been published for H2S nasal dosimetry, which is treated as a source of model uncertainty distinct from the toxicological database uncertainty discussed below. No oral (ingestion) pharmacokinetic adjustment is performed because no current oral toxicity value exists for this chemical (Section 6).

4. Uncertainty Factor Derivation

RoseyFields applies a composite uncertainty factor (UF) of 300 to the NOAEL(HEC), consistent with the following factor-by-factor derivation:

Interspecies extrapolation (UF_A = 3, not the default 10). A partial (rather than full) interspecies uncertainty factor of 3 (10^0.5) is applied because the RGDR-based dosimetric adjustment performed in Section 3 already accounts for the toxicokinetic component of interspecies variability (differences in local delivered dose to the nasal epithelium). The reduced factor of 3 addresses only the remaining toxicodynamic uncertainty (species differences in intrinsic tissue sensitivity) that is not resolved by the dosimetric adjustment itself.

Intraspecies/sensitive-subpopulation variability (UF_H = 10, default). A full factor of 10 is applied to account for variability in human sensitivity, including potentially heightened susceptibility of children and other sensitive subgroups. This is treated as a screening-level default rather than a chemical-specific derivation, because the toxicological database does not provide adequate quantitative data (e.g., a validated physiologically based model of interindividual variability in nasal or CNS sensitivity) to replace the default with a data-derived value.

Subchronic-to-chronic extrapolation (UF_S = 10, default). Because the principal study (Brenneman et al., 2000) was 10 weeks (subchronic) rather than a lifetime/chronic bioassay, a full factor of 10 is applied to extrapolate to chronic (lifetime) exposure duration. RoseyFields notes no chronic (2-year) inhalation bioassay of H2S has been published, which is one of the principal database limitations discussed in Section 9.

LOAEL-to-NOAEL extrapolation (UF_L = 1; not applied). Because the POD is a NOAEL (10 ppm) rather than a LOAEL, no LOAEL-to-NOAEL factor is applied.

Database uncertainty/completeness factor (UF_D = 1; not applied as an additional multiplier). Although the H2S database lacks a conventional multigenerational reproductive bioassay, RoseyFields does not apply an additional database uncertainty factor, because: (1) a one-generation reproductive/developmental study is available and found no significant effects on fertility, reproductive performance, litter parameters, or postnatal growth and reflex development (Dorman et al., 2000); (2) developmental neurotoxicity has been specifically examined (Hannah and Roth, 1991; Skrajny et al., 1992; Dorman et al., 2000); and (3) subchronic inhalation toxicity has been characterized across three rodent species and both sexes (CIIT, 1983a; CIIT, 1983b; CIIT, 1983c). RoseyFields treats the absence of a full two generation study and the absence of a chronic bioassay as residual, qualitative uncertainties rather than as a basis for an additional numeric UF, consistent with the derivation reflected in the source toxicological review.

Modifying factor (MF = 1). No modifying factor is applied.

Composite UF = UF_A x UF_H x UF_S = 3 x 10 x 10 = 300.

RoseyFields flags this UF derivation explicitly whenever it is used in litigation or regulatory support, because the reduced interspecies factor (3 rather than 10) is contingent on the validity of the RGDR dosimetric adjustment in Section 3, and opposing experts may reasonably contest whether that adjustment fully captures interspecies toxicodynamic differences for a reactive, portal-of-entry-acting gas such as H2S.

5. Derivation of the Reference Concentration (RfC)

Applying the composite uncertainty factor to the human-equivalent NOAEL:

RfC = NOAEL(HEC) / UF = 0.64 mg/m3 / 300 = 2.13 x 10^-3 mg/m3, rounded to 2 x 10^-3 mg/m3 (approximately 1.4 ppb)

This is the chronic inhalation RfC that RoseyFields uses as the toxicity value for non-cancer H2S risk characterization. There is no corresponding oral RfD available for use in current assessments: the RfD previously listed on IRIS (3 x 10^-3 mg/kg-day), based on a NOAEL of 3.1 mg/kg-day and LOAEL of 15 mg/kg-day in an oral pig study (Wetterau et al., 1964), was withdrawn after reevaluation determined that the “diarrhetic digestive disorder” reported as the critical effect was not reproducible in that study and could not be reliably attributed to H2S exposure. RoseyFields does not reconstruct or rely on this withdrawn value in current case work, and any legacy risk assessment that references the 3 x 10^-3 mg/kg day oral RfD should be flagged for correction.

6. Carcinogenicity Classification and Quantitative Cancer Risk Values

No cancer slope factor (oral) or inhalation unit risk value is available for H2S, and none is derived or used in RoseyFields’ assessments. Under the applicable EPA cancer-guidelines framework, the available database was determined to be inadequate to classify H2S for carcinogenic potential, and no relevant chronic bioassay or epidemiological data addressing carcinogenicity were identified. Accordingly, RoseyFields’ H2S risk characterizations are limited to non-cancer hazard evaluation (Hazard Quotient/Hazard Index) and do not include an excess lifetime cancer risk calculation for this chemical. Where a case involves co-located chemicals with established cancer toxicity values (e.g., benzene co-emitted from the same oil and gas source), RoseyFields’ cancer risk analysis addresses those chemicals separately and does not attribute cancer risk to H2S itself.

7. Application in Practice: Hazard Quotient/Hazard Index, Dose Reconstruction, and Litigation Support

Exposure and dose reconstruction. RoseyFields’ first step in any H2S matter is to reconstruct a defensible estimate of the air concentration to which a plaintiff, worker, or community member was exposed, and over what averaging time and duration. Because H2S is short-lived, spatially variable, and often released episodically (e.g., intermittent lagoon agitation, tank venting, upset conditions), RoseyFields draws on multiple converging data streams where available: ambient or fence-line continuous monitoring data, regulatory agency complaint-response monitoring, stationary-source emissions inventories combined with atmospheric dispersion modeling (e.g., AERMOD) back-cast to the relevant exposure period, and, where litigated conduct involves odor complaints, an explicit caveat that odor perception is not a reliable proxy for exposure concentration because high concentrations can produce olfactory fatigue/paralysis (a point grounded in the CNS/olfactory toxicity literature discussed above). Time-activity assumptions (time spent outdoors, occupational shift patterns, residential proximity) are then combined with the reconstructed concentration to generate an exposure point concentration and averaging-time-adjusted chronic daily/annual exposure estimate.

Hazard Quotient and Hazard Index calculation. For non-cancer hazard characterization, RoseyFields calculates a chronic inhalation Hazard Quotient (HQ) as:

HQ = Exposure Point Concentration (chronic average, mg/m3) / RfC (2 x 10^-3 mg/m3)

An HQ at or below 1 is characterized as not indicating an appreciable risk of non-cancer effects over a lifetime of exposure at that concentration, consistent with the RfC’s definition as a threshold estimate with built-in uncertainty. An HQ exceeding 1 is characterized as indicating that the margin of exposure incorporated in the toxicity value (composite UF of 300 relative to the animal NOAEL) has been eroded, without itself quantifying a specific probability or severity of harm — RoseyFields is careful in expert reports and testimony to explain that HQ is a screening ratio relative to a conservative, uncertainty adjusted benchmark, not a linear predictor of disease incidence. Where H2S co-occurs with other chemicals sharing a common toxicological target (e.g., other portal-of-entry respiratory irritants or, per the developmental literature, other neurodevelopmental toxicants), RoseyFields calculates a Hazard Index (HI) as the sum of the relevant chemical-specific HQs, consistent with EPA’s mixtures guidance, and documents the toxicological basis for grouping (or not grouping) chemicals by common mechanism/target organ.

Acute versus chronic hazard screening. Because the litigated conduct in H2S matters frequently includes short-duration, high-concentration release events, RoseyFields’ methodology maintains a separate acute hazard screen (drawing on established acute exposure guideline levels and the case-report literature on acute CNS, respiratory, and neurobehavioral sequelae, e.g., Snyder et al., 1995; Tvedt et al., 1991a; Tvedt et al., 1991b; Wasch et al., 1989) that is distinct from, and not derived using, the chronic RfC methodology described above; the chronic RfC is not intended to be, and is not used by RoseyFields as, a benchmark for short-term or acute exposure events.

Litigation and regulatory support applications. RoseyFields uses the HQ/HI framework described above to support: (a) toxic tort causation analyses, in which the HQ is one component (alongside temporality, differential diagnosis, and case-specific medical/exposure history) of a broader causation opinion — RoseyFields does not treat an elevated HQ alone as establishing specific causation; (b) regulatory permitting and enforcement matters, where reconstructed ambient concentrations are compared to the RfC to evaluate whether a facility’s emissions present an appreciable chronic non-cancer hazard to nearby residents; (c) natural resource damage and property-value diminution matters, where chronic HQ exceedances support a claim of ongoing, unabated hazard; and (d) rebuttal of opposing expert reports that misapply the withdrawn oral RfD, that use acute lethality thresholds to characterize chronic community exposure (or vice versa), or that treat HQ as a quantitative probability of harm.

8. Key Uncertainties and Sensitive Populations

RoseyFields’ case-specific application of the H2S RfC accounts for the following uncertainties and sensitive-population considerations, each of which should be addressed explicitly in any expert report or regulatory submission:

Absence of a chronic bioassay. No lifetime (chronic, ~2-year) inhalation bioassay of H2S in any species has been published; the RfC is extrapolated from a 10-week subchronic study (Brenneman et al., 2000) using a default 10-fold subchronic-to-chronic UF. This is a significant source of residual uncertainty that RoseyFields discloses in every report using this value.

Reliance on a single-sex, single-species principal study. The Brenneman et al. (2000) study examined only adult male rats. While the nasal lesion finding is corroborated across species and both sexes in the CIIT subchronic bioassays (CIIT, 1983a; CIIT, 1983b; CIIT, 1983c), the quantitative POD itself is derived from a male-only dataset, and RoseyFields notes this when a case involves claimed sex-specific susceptibility.

Uncertainty in the dosimetric (RGDR) adjustment. As discussed in Section 3, the reduced interspecies UF (3, rather than the full default of 10) depends on the validity of the extrathoracic RGDR adjustment for cross-species scaling of local nasal tissue dose. RoseyFields treats this as an assumption subject to challenge, particularly in adversarial proceedings, and does not represent the RGDR methodology as eliminating interspecies toxicodynamic uncertainty entirely — hence the retained partial factor of 3 rather than 1.

Children and developmental sensitivity. A cluster of studies reports biochemical and morphological alterations in the developing rodent nervous system — altered frontal-cortex norepinephrine and serotonin levels in postnatally exposed rat pups (Skrajny et al., 1992) and altered dendritic arborization of cerebellar Purkinje cells in perinatally exposed pups (Hannah and Roth, 1991) — at exposure concentrations in a similar range to the nasal-lesion NOAEL/LOAEL. A companion study examining a comprehensive battery of functional developmental and reproductive endpoints in the same exposure paradigm found no treatment-related effects on fertility, litter outcomes, pup growth, reflex ontogeny, motor activity, acoustic startle, or passive avoidance, and no treatment-related histopathology of nervous tissue in pups (Dorman et al., 2000). RoseyFields treats the developmental neurologic literature as evidence supporting retention of the full 10-fold intraspecies/sensitive-subpopulation UF (rather than as a basis for an additional, separate database or child-specific UF), because the convergence of exposure levels across the nasal and neurologic endpoints suggests — but does not conclusively establish — that a value protective of nasal injury is also likely to be protective of the reported neurologic changes. Independent case-report and epidemiological literature (not incorporated into the RfC derivation itself) also raises the possibility of heightened pediatric susceptibility to acute, high-level H2S exposure, which RoseyFields addresses separately in any acute-exposure hazard screen for a case involving child plaintiffs.

Occupational and community cohort neurobehavioral findings. A body of literature reports persistent neurobehavioral, neurophysiological, and olfactory deficits in workers and community members following both acute high-level poisoning and, more controversially, intermittent or continuous lower-level exposure (Kilburn and Warshaw, 1995; Kilburn, 1997; Kilburn, 1999; Hirsch and Zavala, 1999). RoseyFields treats the acute-poisoning sequelae literature as well-supported for acute hazard characterization, but treats the “low-level chronic exposure” neurobehavioral literature as more equivocal — given the exposure-characterization challenges inherent in retrospective community studies — and does not rely on it to modify the RfC itself; it is instead addressed qualitatively as corroborating or context evidence in individual causation analyses where relevant symptomatology is alleged.

Odor as an unreliable exposure marker. Because high H2S concentrations can produce olfactory paralysis, community complaint records based on odor detection can understate true peak exposure concentrations; RoseyFields’ dose reconstructions do not treat the absence of odor complaints as evidence of the absence of hazardous exposure.

Oral exposure pathway. With the withdrawal of the oral RfD, RoseyFields does not perform quantitative oral (ingestion) risk assessment for H2S (e.g., from groundwater) and instead notes qualitatively that H2S is not persistent in water and rarely constitutes a significant oral exposure pathway; where an oral pathway is potentially relevant (e.g., sulfide-contaminated groundwater with anaerobic-decomposition odor complaints), RoseyFields recommends direct analytical characterization rather than reliance on a toxicity-value-based risk calculation.

Bibliography