Expert Witness for Cancer and Non-Cancer Risk Assessment for all Exposure Pathways

RoseyFields LLC — Paul E. Rosenfeld, Ph.D., Principal Forensic Environmental Chemist

Technical reference page for attorneys, judges, consultants, regulators, and scientists.

1. What Dr. Rosenfeld Does

Human health risk assessment converts environmental measurement into a statement about human consequence. It answers two quantitative questions: how much of a chemical entered a person’s body, and what does that dose imply for the probability of cancer and for the margin against non-cancer effects.

Dr. Rosenfeld performs that conversion for contaminated sites, industrial emissions, drinking water supplies, consumer products, and occupational settings. The inputs are environmental data — soil, groundwater, drinking water, indoor and ambient air, attic dust, sediment, fish tissue, produce, serum — together with the exposure history of the people at issue. The outputs are average daily dose in milligrams per kilogram per day, exposure concentration in micrograms per cubic meter, incremental lifetime cancer risk, hazard quotients and hazard indices by target organ, and comparisons against regulatory benchmarks and background.

The framework is EPA’s, not invented for the occasion. Risk Assessment Guidance for Superfund establishes the four-step architecture and the dose equations (U.S. EPA, 1989); Parts B through F supply preliminary remediation goals, remedial-alternative evaluation, standardized reporting, dermal methodology, and inhalation methodology (U.S. EPA, 1991a, 1991b, 2001a, 2004a, 2009). ATSDR’s public health assessment guidance supplies parallel exposure-dose methods for inhalation, water ingestion, and dermal contact, and a framework for chemical mixtures (ATSDR, 2018, 2020, 2023a, 2023b, n.d.). Where the exposure occurred in the past, the same architecture is applied retrospectively through dose reconstruction.

Three commitments govern the work, and they are the same three that govern the firm’s air dispersion modeling. Methods are the published, agency-endorsed methods, applied as written or with documented, defensible departures. Every input is sourced, and every uncertain input is bounded rather than buried — the deliverable is a range with identified drivers, not a single unqualified number. And the analysis is reproducible: the database, the statistical output, the exposure parameter table, and the risk calculation spreadsheets are retained and produced, so that an opposing expert can re-run the arithmetic.

Assignments include affirmative and rebuttal work, for plaintiffs, defendants, municipalities, and state agencies. Methodologies are evaluated under Daubert and analogous standards.

2. Credentials and Years of Experience

Dr. Rosenfeld holds a Ph.D. in Soil Chemistry from the University of Washington (1999), an M.S. in Environmental Science from U.C. Berkeley (1995), and a B.A. in Environmental Studies from U.C. Santa Barbara (1991). He has more than 25 years of professional experience in exposure assessment, historical dose reconstruction, environmental forensics, and air dispersion modeling, across hundreds of contaminated sites nationwide, and has prepared hundreds of exposure assessments over the course of his career.

He is founder and principal of RoseyFields LLC (2025–present) and was a founding partner and principal of Soil Water Air Protection Enterprise (SWAPE) from 2003 to 2025. He began his risk assessment career as a Risk Assessor at Bechtel (1999–2000) and subsequently served as Remediation Project Manager at Ogden (now Amec) and at Anteon Corporation, and as Senior Remediation Scientist at Komex H2O Science. As Remedial Project Manager for the U.S. Navy’s BRAC program at Treasure Island, he directed remediation and air monitoring to protect workers and adjacent communities. He is HAZWOPER-certified under 29 C.F.R. § 1910.120.

He taught Environmental Health Science 100 — the health effects of environmental contaminants — at the UCLA School of Public Health from 2003 through 2010, to undergraduates, physicians, public health professionals, and nurses, and held appointments as Lecturer and Assistant Researcher (2007–2011) and Adjunct Professor (2003–2006). He has also taught at San Diego State University and for the National Ground Water Association.

Peer-reviewed risk assessment publications. Dr. Rosenfeld is co-author of The Risks of Hazardous Waste (Rosenfeld & Feng, 2011) and of three volumes of the Handbook of Pollution Prevention and Cleaner Production (Cheremisinoff & Rosenfeld, 2009, 2010, 2011). His published quantitative risk work includes a cancer risk assessment for Camp Lejeune drinking water contamination spanning 1955 to 1987 (Rosenfeld et al., 2024), cancer risk from diesel exhaust exposure among railroad workers (Rosenfeld et al., 2022), PFAS exposure in firefighters (Rosenfeld et al., 2023), community health burden following refinery fires (Remy et al., 2019), and paired attic dust and human blood lipid studies near wood treatment and former PCB production facilities (Hensley et al., 2007; Feng et al., 2010; Gonzalez et al., 2010). He has published approximately forty peer-reviewed papers, chapters, and books and delivered more than fifty conference presentations and invited lectures, including on odor and health risk criteria (Rosenfeld et al., 2007) and on the toxicology and fate of PFOA, PFOS, brominated flame retardants, 1,4-dioxane, 1,2,3-trichloropropane, perchlorate, and tert-butyl alcohol.

Testimony experience. His curriculum vitae, current as of July 20, 2026, itemizes 114 listed testimony entries across more than 100 distinct matters in state and federal courts in over 30 states, comprising approximately 120 deposition appearances, 16 trial appearances, and testimony at a contested Daubert hearing, with the earliest listed matter in 2010 and the most recent in July 2026.

A current curriculum vitae with the complete publication list and testimony history is available for download.

3. The Four-Step Framework and Controlling Guidance

Risk assessment in the United States follows the four-step paradigm established by the National Research Council (1983, 2009) and implemented for contaminated sites by EPA (1989).

Hazard identification asks what chemicals are present, in what media, and what adverse effects they are capable of causing. Chemicals of potential concern are selected from the site database by comparison against screening levels, frequency of detection, and background, with the selection rationale documented rather than assumed.

Dose-response assessment establishes the quantitative relationship between dose and effect: a reference dose or reference concentration for non-cancer effects, and an oral slope factor or inhalation unit risk for carcinogens. Section 7 addresses how these values are selected and, where the science supports it, independently derived.

Exposure assessment identifies who contacted the contamination, through which pathways, at what concentration, and for how long, and converts that into dose. This is where most of the site-specific work resides, and where competing risk assessments most often diverge.

Risk characterization combines the two, expresses the result as cancer risk and hazard, compares it against the relevant benchmarks, and — critically — states the uncertainty. EPA’s guidance is explicit that risk characterization is not a number but a discussion (U.S. EPA, 2000).

The controlling guidance set is well defined. RAGS Part A supplies the baseline process (U.S. EPA, 1989). Part B supplies risk-based preliminary remediation goals (U.S. EPA, 1991a); Part C, evaluation of residual risk from remedial alternatives (U.S. EPA, 1991b); Part D, standardized planning, reporting, and review tables (U.S. EPA, 2001a); Part E, dermal risk assessment (U.S. EPA, 2004a); Part F, inhalation risk assessment (U.S. EPA, 2009). The Superfund Exposure Assessment Manual (U.S. EPA, 1988) and the Soil Screening Guidance and its technical background document (U.S. EPA, 1996a, 1996b, 2002a) supply pathway equations. Standard default exposure factors come from OSWER Directive 9285.6-03 (U.S. EPA, 1991c) and current values from the Exposure Factors Handbook (U.S. EPA, 2011). ATSDR’s exposure dose guidance provides parallel methods with age-specific intake rates and explicit central-tendency and reasonable-maximum estimates (ATSDR, 2020, 2023a, 2023b). California’s OEHHA guidance governs where state law applies, including Proposition 65 (OEHHA, 2015).

Where an assessment departs from default guidance — a site-specific dermal absorption fraction, a non-default exposure duration, an alternative toxicity value — the departure is identified as such, justified on the record, and tested by sensitivity analysis.

4. Data Evaluation and Exposure Point Concentrations

A risk assessment is only as sound as the data underneath it, and data evaluation is a substantive analytical step rather than a formality.

Usability. Analytical data are evaluated for fitness under EPA’s data usability guidance (U.S. EPA, 1992a): appropriate analytical methods and detection limits relative to the risk-based concentrations at issue, adequate quality control (blanks, surrogates, duplicates, matrix spikes), valid holding times, correct sample handling, and defensible data validation. Detection limits above the risk-based screening level are a recurring problem — a non-detect reported at a limit ten times the screening level supports no conclusion at all. Where laboratory data come from multiple consultants and multiple decades, methods and reporting conventions must be reconciled before the data can be pooled.

Non-detects. Censored observations are handled explicitly. Substitution at one-half the detection limit remains common practice for low-censoring datasets, but for datasets with substantial censoring, Kaplan-Meier and regression-on-order-statistics methods implemented in ProUCL are preferred, consistent with EPA guidance (U.S. EPA, 2015). Simply deleting non-detects biases the mean upward and is not defensible.

Exposure point concentration. The exposure point concentration is the concentration a receptor is assumed to contact over the exposure period, and EPA’s guidance is unambiguous that it is an estimate of the average concentration within the exposure unit, represented by the 95 percent upper confidence limit on the arithmetic mean rather than by the maximum detection (U.S. EPA, 1992b). The UCL is computed with ProUCL using the estimator appropriate to the distribution — normal, lognormal, gamma, or nonparametric — after formal goodness-of-fit testing, and where the 95% UCL exceeds the maximum detected value the maximum is used instead. Exposure units are defined on the basis of receptor behavior, not sampling convenience: a residential yard, a play area, a work zone, a water system service area.

Background. Distinguishing site-related contamination from natural and anthropogenic background is often the decisive issue, particularly for metals, PAHs, and dioxins. EPA’s background policy directs that background be characterized with its own dataset and compared statistically rather than by point comparison (U.S. EPA, 2002b), using two-sample tests appropriate to censored data. Anthropogenic background — urban fill, historical leaded gasoline, historical pesticide application, regional atmospheric deposition — is treated as distinct from natural background, because the two support different conclusions about attribution.

Spatial and temporal representativeness. Data are examined for whether they represent the exposure period. Groundwater and drinking water concentrations vary over time; soil concentrations at depth may not represent the surface soil a child contacts; post-remediation data do not represent pre-remediation exposure. Where the exposure period predates the data, concentrations must be reconstructed rather than assumed constant.

5. Exposure Pathways, Scenarios, and Receptors

A complete exposure pathway requires a source, a release mechanism, a transport medium, a point of exposure, and a route of entry into the body. If any element is missing the pathway is incomplete and contributes no dose. Pathway identification is therefore a factual exercise driven by site conditions and by how people actually used the property.

Pathways commonly evaluated. Incidental ingestion of soil and indoor dust; dermal contact with soil, dust, water, and sediment; inhalation of ambient air, indoor air, vapor intrusion, and resuspended particulate; ingestion of drinking water from public supply or private wells; dermal contact and inhalation during showering and other household water use; ingestion of home-grown produce, backyard eggs and poultry, milk and meat from locally raised animals, and fish and shellfish; and breast-milk transfer to nursing infants for persistent, lipophilic, or protein-binding compounds.

Receptors and scenarios. Receptors are defined by land use and behavior: resident adult and child, with age bins consistent with EPA’s early-life exposure guidance; indoor and outdoor worker; construction and utility worker; trespasser and recreational user; farmer or subsistence fisher where applicable. For each, EPA practice requires characterization of both the central tendency exposure and the reasonable maximum exposure — the high-end but plausible case, not the worst conceivable combination of parameters (U.S. EPA, 1989, 1992c). ATSDR’s guidance similarly directs paired CTE and RME estimates (ATSDR, 2023a).

Exposure parameters. Intake rates, exposure frequency, exposure duration, body weight, skin surface area, soil adherence, and averaging time come from the Exposure Factors Handbook (U.S. EPA, 2011) and standard default guidance (U.S. EPA, 1991c), with age-specific values applied for child receptors and site-specific values substituted where record evidence supports them. Where individual plaintiffs are at issue, exposure duration and frequency are drawn from residential and employment history — deposition testimony, deeds, utility records, school enrollment, employment files — rather than from a residential default, and each individual’s exposure profile is built year by year.

Time-activity and microenvironments. Where exposure varies across the day, time-weighted microenvironmental exposure is constructed from activity-pattern data, so that indoor, outdoor, occupational, and in-transit concentrations are weighted by actual time spent. This matters most for inhalation, where indoor concentrations may differ from modeled or monitored outdoor concentrations by a large factor in either direction.

6. Dose Equations by Route

Dose is computed route by route, and the equation form differs by route in ways that are frequently confused.

Oral. Average daily dose in mg/kg-day is the product of medium concentration, intake rate, exposure frequency, and exposure duration, divided by the product of body weight and averaging time. For non-cancer endpoints the averaging time is the exposure duration, giving an average daily dose over the period of exposure. For carcinogens the averaging time is a 70-year lifetime, giving a lifetime average daily dose. Using the wrong averaging time is a common and consequential error: it inflates or deflates cancer risk in direct proportion to the ratio of exposure duration to lifetime.

Dermal. Dermal dose is an absorbed dose, not an applied dose. For soil, absorbed dose per event is the product of soil concentration, adherence factor, exposed skin surface area, and a chemical-specific dermal absorption fraction; for water, a permeability coefficient and event duration govern (U.S. EPA, 1992d, 2004a). Because the resulting dose is absorbed rather than administered, it must be compared against a route-adjusted toxicity value derived from the oral value using the gastrointestinal absorption fraction. Skipping that adjustment overstates dermal risk. RAGS Part E is also explicit that the dermal pathway is quantitatively significant for only a limited set of compounds, and it identifies the screening criteria for when the pathway warrants full quantification.

Inhalation. Current EPA practice does not compute an inhalation intake in mg/kg-day. Under RAGS Part F, the metric is the exposure concentration in µg/m³ — the air concentration adjusted for exposure time, frequency, and duration and averaged over the appropriate period — which is compared directly against a reference concentration or multiplied by an inhalation unit risk (U.S. EPA, 2009). This reflects the fact that inhalation dosimetry depends on respiratory tract deposition and gas-blood partitioning rather than on simple body-weight-normalized intake. Legacy assessments that convert air concentrations to mg/kg-day using a default breathing rate and body weight are applying superseded methodology, and the substitution can change the result materially.

Multipathway and food-chain doses. Where deposition or discharge contaminates soil, produce, forage, animal tissue, surface water, or fish, dose is propagated through the standard multipathway equation set, with biotransfer factors, bioconcentration factors, and plant uptake factors (U.S. EPA, 2005c). For persistent bioaccumulative compounds, food-chain pathways frequently exceed direct contact pathways, and omitting them understates total dose.

7. Toxicity Value Selection and Chemical-Specific Derivation

Toxicity values are the second multiplier in every risk calculation, and they are contested at least as often as exposure.

Agency values and the hierarchy. Where an appropriate value exists, RoseyFields uses EPA’s Integrated Risk Information System, and where IRIS lacks a value, works down the recognized hierarchy: Provisional Peer-Reviewed Toxicity Values, ATSDR minimal risk levels, California OEHHA values, and other peer-reviewed sources, consistent with EPA’s Regional Screening Level framework (U.S. EPA, 2024). The selection and its basis are disclosed, and where agencies have adopted materially different values for the same compound — as they have for several PFAS, for hexavalent chromium, and for ethylene oxide — the divergence is identified and its effect on the result quantified rather than resolved silently by picking one.

Chemical-specific derivation. For compounds where the agency value is dated, absent, or not supported by the current literature, RoseyFields derives its own reference dose and cancer slope factor from primary study data, following EPA’s reference dose and carcinogen risk assessment methodology (U.S. EPA, 2002c, 2005a). The sequence is:

• Point-of-departure selection. Candidate endpoints are identified for each health outcome with substantial weight of evidence. Medium- and high-confidence human epidemiological studies are preferred to animal studies where the epidemiological database supports dose-response modeling, because human data eliminate interspecies extrapolation. Within the epidemiology, studies at exposure levels near the environmentally relevant range, with adequate control for confounding and co-exposure, are prioritized over high-exposure occupational cohorts.
• Benchmark dose modeling. The point of departure is derived as a benchmark dose lower confidence limit at a defined benchmark response, using EPA’s Benchmark Dose Software methodology (U.S. EPA, 2012), rather than as a NOAEL or LOAEL where the data support modeling.
• Pharmacokinetic dose conversion. Internal-dose points of departure are converted to human equivalent external doses using compound-appropriate toxicokinetic models rather than default body-weight scaling, including gestation and lactation modules where developmental endpoints govern.
• Uncertainty factor application. Interspecies, intraspecies, subchronic-to-chronic, LOAEL-to-NOAEL, and database factors are assigned on the basis of the actual critical study rather than by uniform default, and the composite factor is documented factor by factor.
• Cancer classification and slope factor. Weight of evidence is evaluated under EPA’s carcinogen guidelines, and the slope factor derived by low-dose linear extrapolation from the benchmark dose where a mutagenic or otherwise linear mode of action is indicated, or through a non-linear reference-dose approach where the mode of action supports it (U.S. EPA, 2005a).

The firm’s chemical-specific methodology pages document this derivation compound by compound — including PFOA, PFOS, PCE, TCE, vinyl chloride, methylene chloride, and 2,3,7,8-TCDD — so that the toxicity value used in any given matter is traceable to the underlying studies rather than to a single regulatory pronouncement.

8. Cancer and Non-Cancer Risk Characterization

Cancer. Incremental lifetime cancer risk is the lifetime average daily dose multiplied by the oral slope factor, or the lifetime-averaged exposure concentration multiplied by the inhalation unit risk. Risks are summed across pathways and across chemicals to a total incremental risk and compared against the CERCLA risk range of 10⁻⁶ to 10⁻⁴, within which 10⁻⁴ is generally the upper bound of acceptability and 10⁻⁶ the point of departure for remedial consideration (U.S. EPA, 1989, 1991a). Because the linear low-dose model is used, risks are additive and the arithmetic is straightforward; the interpretive question is what a given risk level means, and that is addressed in the risk characterization rather than left to the number.

Early-life exposure. For carcinogens acting through a mutagenic mode of action, age-dependent adjustment factors are applied — a factor of 10 for exposures from birth to age 2 and a factor of 3 for ages 2 through 16 — per EPA’s supplemental early-life guidance (U.S. EPA, 2005b). ADAFs are not applied to compounds without a mutagenic mode of action, and the mode-of-action determination is made and documented rather than assumed in either direction. For a residential scenario beginning at birth, the ADAF adjustment can increase calculated cancer risk by roughly a factor of two to three, which makes it a frequent point of dispute.

Non-cancer. The hazard quotient is the average daily dose divided by the reference dose, or the exposure concentration divided by the reference concentration. A hazard quotient at or below 1 indicates exposure below the level associated with appreciable risk of the critical effect; above 1 indicates that margin has been exceeded. Because the reference dose incorporates uncertainty factors, a hazard quotient above 1 is not a prediction that effects will occur — it is a statement that the protective margin has been consumed, and the interpretation depends on the magnitude of exceedance, the composite uncertainty factor, and the nature of the critical effect.

Hazard index and target-organ segregation. Hazard quotients are summed to a hazard index. Where the total hazard index exceeds 1, EPA guidance directs segregation into target-organ-specific and mechanism-specific indices rather than reliance on the aggregate, because summing quotients for chemicals affecting different organs by different mechanisms has no toxicological basis (U.S. EPA, 1989; ATSDR, 2018). Segregation frequently resolves an apparent exceedance — or fails to, which is itself informative.

Exposure duration and acute assessment. Chronic reference values are compared against chronic exposure; where the exposure was short-term, intermediate or acute minimal risk levels, acute reference exposure levels, AEGLs, or ERPGs are the appropriate benchmarks, and the averaging period of the exposure estimate must match the averaging period of the benchmark (ATSDR, 2020). Averaging a one-hour peak over a lifetime, or comparing a one-hour peak against a chronic value, are both errors and both occur.

9. Cumulative Risk, Mixtures, and Background

Real exposures are to mixtures from multiple sources, and cumulative assessment is where risk assessment is currently least standardized and most contested.

Default practice assumes dose additivity within a target organ and mechanism, which is the basis for summing hazard quotients and cancer risks. ATSDR’s framework for multiple chemicals and other stressors provides the structured approach for evaluating mixtures, including interaction profiles and the treatment of non-chemical stressors (ATSDR, 2018), and EPA’s mixtures guidance supplies the hierarchy: whole-mixture data where available, sufficiently similar mixture data next, and component-based additivity where neither exists (U.S. EPA, 2000b). Where toxicological interaction is plausible — synergism, antagonism, or shared saturable metabolic pathways — it is identified qualitatively even where it cannot be quantified.

Two mixture problems recur. The first is PFAS: co-occurring PFOA, PFOS, PFHxS, PFNA, and related compounds have overlapping toxicity profiles, and whether and how to sum them is an active scientific question. RoseyFields calculates a cumulative hazard index across co-occurring PFAS where case-specific data and current science support a shared endpoint and additive assumption, and treats the assumption as an area requiring explicit, case-specific justification rather than as settled. The second is dioxin-like compounds: dioxins, furans, and dioxin-like PCBs are summed as toxic equivalents using WHO toxic equivalency factors (Van den Berg et al., 2006), which is well established but sensitive to which TEF scheme is applied and to how non-detects are treated.

Comparison to background and to reference populations. Attribution requires context. Site-related dose is compared against dose from background media concentrations, and where biomonitoring data exist, measured serum or blood concentrations are compared against NHANES reference distributions for the relevant age, sex, and period (CDC, 2024). A measured concentration at the population median supports a different conclusion than one at the 95th percentile, and providing that comparison is part of a complete characterization.

10. Screening Levels, Cleanup Goals, and Proposition 65

Risk assessment runs in both directions: forward from concentration to risk, and backward from an acceptable risk level to a concentration.

Screening levels. EPA’s Regional Screening Levels are risk-based concentrations back-calculated at a target cancer risk of 10⁻⁶ and a target hazard quotient of 1, for residential and industrial scenarios (U.S. EPA, 2024). They are screening tools: exceedance triggers further evaluation rather than establishing risk, and a concentration below the RSL generally ends the inquiry for that chemical and pathway. Soil screening levels for the migration-to-groundwater pathway follow the Soil Screening Guidance methodology (U.S. EPA, 1996a, 1996b, 2002a).

Preliminary remediation goals and cleanup levels. RAGS Part B supplies the back-calculation methodology for developing risk-based preliminary remediation goals from a target risk level and site-specific exposure assumptions (U.S. EPA, 1991a), and Part C the evaluation of residual risk associated with remedial alternatives (U.S. EPA, 1991b). Cleanup levels ultimately reflect both the risk calculation and the nine CERCLA remedy-selection criteria, and the distinction between a risk-based goal and a selected cleanup level matters in litigation over adequacy of remediation.

California Proposition 65. The Safe Drinking Water and Toxic Enforcement Act requires warning for exposures to listed chemicals above safe harbor levels, and OEHHA’s regulations establish the derivation methodology: the No Significant Risk Level for carcinogens, set at the daily intake associated with a 10⁻⁵ lifetime cancer risk, and the Maximum Allowable Dose Level for reproductive toxicants, set at one one-thousandth of the NOEL for reproductive effects (OEHHA, 2015; Cal. Code Regs. tit. 27). Where OEHHA has not adopted a safe harbor number for a listed chemical, a defensible NSRL or MADL can be derived using the regulatory methodology from the primary toxicological literature, and RoseyFields has performed such derivations for compounds including PFOA and mercury. Proposition 65 work also requires exposure estimation for the specific product or setting at issue, since the comparison is between a reasonably anticipated daily intake and the safe harbor level.

11. Dose Reconstruction and Biomonitoring

Where the exposure is historical and the media data are absent or incomplete, dose must be reconstructed. The methodology has an established foundation in radiological dose reconstruction and transfers directly to chemical exposures (National Research Council, 1995; Till & Grogan, 2008).

Forward reconstruction. Historical media concentrations are reconstructed from facility operating and production records, emission or discharge inventories, historical monitoring, contaminant fate and transport modeling, groundwater flow and transport modeling, or air dispersion modeling, and then combined with the individual’s residential and occupational timeline to produce a year-by-year dose history. The Camp Lejeune assessment illustrates the approach across a multi-decade drinking water exposure period (Rosenfeld et al., 2024), as does the reconstruction of occupational diesel exhaust exposure from equipment inventories, duty cycles, and ventilation conditions (Rosenfeld et al., 2022).

Reverse reconstruction from biomonitoring. Where a measured internal dose exists — serum PFAS, blood lead, blood or adipose lipid dioxin and PCB concentrations — the historical intake consistent with that measurement can be back-calculated using compound-specific toxicokinetic models, accounting for elimination half-life, volume of distribution, age, sex, and pregnancy and lactation history. This is particularly powerful for persistent compounds, because body burden integrates cumulative historical exposure rather than reflecting only current intake, and it provides an independent physical constraint on a media-based reconstruction. The firm’s published work has used paired attic dust and blood lipid measurements in this way (Hensley et al., 2007; Feng et al., 2010).

Life-stage-specific dose. Reconstruction to the exposure window relevant to the alleged effect is what makes the analysis probative rather than merely descriptive. A birth-weight claim turns on first-trimester maternal internal dose; a childhood neurodevelopmental claim on early-life dose; a chronic hepatic or cardiovascular claim on cumulative adult dose; a cancer claim on cumulative lifetime dose with appropriate latency consideration. A single present-day estimate cannot answer any of these.

12. Uncertainty, Sensitivity, and Probabilistic Analysis

A risk estimate without an uncertainty analysis is incomplete as a matter of guidance, not merely of style (U.S. EPA, 1989, 2000).

Uncertainty is separated from variability. Variability is real heterogeneity in the exposed population — differences in intake, duration, body weight, behavior — and is addressed through the CTE and RME scenarios and, where warranted, through distributions. Uncertainty is limited knowledge about a fixed quantity — the true historical concentration, the true slope factor — and is addressed through bounding, sensitivity analysis, and disclosure.

Sensitivity analysis identifies which inputs actually drive the result. In practice a small number dominate: the exposure point concentration, the toxicity value, the exposure duration and frequency, and — for deposition-driven and food-chain pathways — the transfer factors. Reporting the risk estimate together with its response to plausible variation in those parameters is more useful to a decision-maker than a single point estimate, and considerably harder to impeach.

Where the data support distributions rather than point values, probabilistic assessment under RAGS Volume III replaces the deterministic calculation with Monte Carlo analysis, producing a risk distribution and identifying the percentile corresponding to the deterministic RME estimate (U.S. EPA, 2001b). Probabilistic analysis is a supplement to, not a replacement for, the deterministic calculation, and EPA’s guidance sets conditions on its use — including that input distributions be empirically supported rather than assumed.

Known limitations are stated affirmatively: reliance on default exposure factors where site-specific data are unavailable; uncertainty factors embedded in reference doses; interspecies and low-dose extrapolation in cancer slope factors; the additivity assumption in cumulative assessment; and the absence of toxicity values for compounds that are present but unquantifiable.

13. Daubert and Litigation Considerations

Federal Rule of Evidence 702, as amended effective December 1, 2023, requires the proponent to show more likely than not that expert testimony rests on sufficient facts or data, is the product of reliable principles and methods, and reflects a reliable application of those principles to the facts. Daubert (1993), Joiner (1997), and Kumho Tire (1999) supply the reliability framework, and the Reference Manual on Scientific Evidence (Federal Judicial Center & National Research Council, 2011) is the standard judicial reference for exposure and toxicology evidence. Methodologies are evaluated under Daubert and analogous standards.

Risk assessment maps onto the reliability factors well. The four-step paradigm originates with the National Research Council (1983, 2009). The equations, exposure factors, and risk benchmarks are promulgated federal guidance developed for regulatory purposes independent of litigation. Toxicity values are peer reviewed, and in the case of IRIS values, externally peer reviewed and publicly documented. The calculations are arithmetically transparent and fully reproducible from the produced database and spreadsheets.

The reliability disputes therefore concentrate on application, and in a recognizable set of places: exposure point concentrations based on maxima rather than upper confidence limits on the mean, or drawn from data that do not represent the exposure period; exposure durations and frequencies untethered to residential or employment history; averaging times mismatched to the endpoint; inhalation dose computed by superseded intake methodology rather than RAGS Part F; dermal absorbed dose compared against an unadjusted oral toxicity value; hazard indices summed across unrelated target organs; ADAFs applied to compounds without a mutagenic mode of action, or omitted where one exists; background and reference-population comparison omitted entirely; and toxicity values selected without disclosure of competing agency values.

Two further issues recur specifically in toxic tort work. The first is the distinction between a risk estimate and a causation opinion: a calculated cancer risk of 10⁻⁴ is a population-level probabilistic statement, and moving from it to specific causation for an individual plaintiff requires additional reasoning — dose relative to levels associated with effect in the epidemiology, latency, competing risk factors, and differential diagnosis — that must be supplied rather than assumed. The second is the difference between exceeding a regulatory screening level and demonstrating harm; screening levels are deliberately conservative and are not thresholds of injury. Both distinctions are stated explicitly in the work product, because an analytical gap between a defensible calculation and the opinion it is offered to support is the most common basis for exclusion.

Rebuttal work applies the same checklist in reverse, examining data usability, EPC derivation, statistical methods, exposure parameter selection, toxicity value selection, and pathway completeness, and re-running the calculations to test the sensitivity of the conclusions to the assumptions.

14. Representative Contaminants and Settings

Contaminants. Benzene, toluene, ethylbenzene, and xylenes; 1,3-butadiene; ethylene oxide; vinyl chloride; trichloroethylene, tetrachloroethylene, methylene chloride, and 1,1,1-trichloroethane; 1,4-dioxane; 1,2,3-trichloropropane; tert-butyl alcohol and MTBE; perchlorate; hexavalent and trivalent chromium; lead, arsenic, cadmium, mercury, manganese, nickel, and zinc; polycyclic aromatic hydrocarbons and creosote constituents; pentachlorophenol; polychlorinated biphenyls; 2,3,7,8-TCDD and other dioxins and furans; per- and polyfluoroalkyl substances including PFOA, PFOS, PFHxS, and PFNA; diesel particulate matter and diesel exhaust organics; respirable crystalline silica; asbestos; atrazine, glyphosate, and other pesticides and herbicides; brominated flame retardants; hydrogen sulfide and reduced sulfur compounds; carbon monoxide and carbon dioxide; PM₁₀ and PM₂.₅; microplastics and engineered nanoparticles.

Settings. Contaminated drinking water systems and private wells; petroleum refineries, petrochemical plants, and refinery fire and upset events; chemical manufacturing complexes; ethylene oxide sterilization facilities; secondary smelters and lead recycling; coke batteries and steel operations; wood treatment and creosoting sites; pulp, paper, and wood pellet operations; cement plants, quarries, and aggregate operations; foundries, metal finishing, and plating; municipal and hazardous waste landfills, incinerators, and transfer stations; composting facilities, biosolids land application, and anaerobic digesters; wastewater treatment works; railyards, rail corridors, and derailment sites; military installations, firing ranges, and airports with AFFF use; firefighter turnout gear and fire training grounds; frac sand mining and processing; oil and gas well pads and hydraulic fracturing; countertop and stone fabrication; consumer products under Proposition 65; and residential neighborhoods adjacent to any of the above.

15. Five Short Case Examples

Consistent with RoseyFields’ confidentiality practice, the following are described without client, counsel, facility, or location identifiers.

Multi-decade drinking water cancer risk assessment. For a population served by a contaminated water supply over a period exceeding three decades, historical concentrations of chlorinated solvents and related compounds were reconstructed from available monitoring records and water system operating history, resolved to annual exposure point concentrations. Ingestion, dermal, and inhalation-during-household-use pathways were quantified, lifetime average daily doses developed for age-specific exposure bins, and incremental cancer risks and hazard indices calculated and compared against the CERCLA risk range. The published methodology for an assessment of this type is documented in Rosenfeld et al. (2024).

Dioxin and metals risk assessment at a former municipal incinerator. Surface and subsurface soil data compiled from multiple consultants and multiple decades were evaluated for usability, reconciled to consistent methods and reporting limits, and pooled. Exposure point concentrations were developed as 95 percent upper confidence limits using ProUCL with distribution-appropriate estimators and censored-data methods, dioxin and furan congeners summed as toxic equivalents, and residential child and adult risks calculated across incidental ingestion, dermal contact, and dust inhalation, with site data compared against a separately characterized background dataset.

Occupational exposure reconstruction and cumulative dose. For workers with decades of employment history, exposure to diesel exhaust and associated constituents was reconstructed from equipment inventories, duty cycles, task assignments, ventilation conditions, and published emission factors, using mass-balance calculations for enclosed spaces. Cumulative dose was assembled year by year from job history and evaluated against the epidemiological dose-response literature, supporting both exposure and general causation analysis (Rosenfeld et al., 2022).

PFAS serum-based dose reconstruction. For individuals with measured serum PFAS concentrations, historical daily intake and drinking water concentrations were back-calculated using a one-compartment toxicokinetic model parameterized with compound-specific half-life and volume of distribution, accounting for age, sex, and pregnancy and lactation history. Reconstructed doses were compared against chemical-specific reference doses derived from the primary epidemiological literature, and measured serum concentrations were compared against NHANES reference distributions to characterize incremental exposure attributable to the source at issue.

Proposition 65 safe harbor derivation and exposure evaluation. For a listed compound without an adopted OEHHA safe harbor level, a No Significant Risk Level and Maximum Allowable Dose Level were derived from the primary toxicological literature following the methodology in OEHHA’s regulations, and reasonably anticipated daily intake from the product and use scenario at issue was estimated and compared against the derived levels.

16. Ten Frequently Asked Questions

1. What is the difference between a risk assessment and a causation opinion? A risk assessment estimates the probability of an adverse outcome in an exposed population, or the margin between a dose and a reference value. Specific causation for an individual requires additional reasoning: whether the reconstructed dose falls within the range associated with effect in the epidemiology, whether latency is consistent, and what competing causes exist. The risk number is an input to that analysis, not a substitute for it — and treating it as a substitute is a frequent basis for exclusion.

2. Does exceeding a screening level mean someone was harmed? No. Regional Screening Levels are back-calculated at a 10⁻⁶ cancer risk and a hazard quotient of 1 under conservative default assumptions, and are designed to identify what needs further evaluation (U.S. EPA, 2024). Exceedance triggers analysis; it does not establish injury. A concentration below the screening level, by contrast, generally does support ending the inquiry for that chemical and pathway.

3. Why use the 95% UCL on the mean rather than the maximum detected concentration? Because the receptor contacts an average concentration across the exposure unit over the exposure period, not the single highest point. EPA guidance directs use of the 95 percent upper confidence limit on the arithmetic mean as the exposure point concentration, with the maximum used only when the UCL exceeds it (U.S. EPA, 1992b). Using the maximum overstates dose and is not consistent with the guidance.

4. What does a hazard quotient above 1 actually mean? That the protective margin built into the reference dose has been consumed — not that effects will occur. The reference dose already incorporates uncertainty factors, often 10 to 1,000-fold. Interpretation depends on how far above 1 the quotient is, the size of the composite uncertainty factor, and the nature of the critical effect. A hazard quotient of 1.2 and one of 120 warrant very different discussion.

5. Can hazard quotients for different chemicals simply be added? Only within a common target organ and mechanism. Summing quotients across chemicals affecting unrelated organs by unrelated mechanisms has no toxicological basis, and where a total hazard index exceeds 1, EPA guidance directs segregation into organ- and mechanism-specific indices (U.S. EPA, 1989; ATSDR, 2018).

6. Which toxicity value should be used when agencies disagree? The divergence should be disclosed and its effect quantified, not resolved silently. RoseyFields uses the recognized hierarchy — IRIS, then PPRTVs, ATSDR MRLs, and OEHHA values — and where the agency values are dated, absent, or unsupported by current literature, derives a chemical-specific value from primary study data using benchmark dose modeling, pharmacokinetic dose conversion, and documented uncertainty factors (U.S. EPA, 2002c, 2005a, 2012).

7. How is site-related contamination distinguished from background? By characterizing background with its own dataset and comparing statistically rather than by point comparison, using tests appropriate to censored data (U.S. EPA, 2002b). Natural background and anthropogenic background — urban fill, legacy leaded gasoline, historical pesticide use, regional deposition — are treated separately, because they support different attribution conclusions. Chemical fingerprinting and congener or isotopic patterns can further separate sources.

8. Can dose be estimated when the exposure ended decades ago? Yes, within stated uncertainty, through dose reconstruction. Media concentrations are reconstructed from operating records, historical monitoring, and fate and transport or dispersion modeling, then combined with residential and employment history. Where biomonitoring data exist for persistent compounds, historical intake can also be back-calculated from measured body burden using toxicokinetic models, which provides an independent check (National Research Council, 1995; Till & Grogan, 2008).

9. Does a serum or blood measurement prove exposure to a particular source? It proves exposure. Attributing it to a source requires comparison against reference-population distributions such as NHANES for the relevant age, sex, and period (CDC, 2024), evaluation of alternative exposure pathways, and where possible pattern analysis — congener profiles, isomer ratios, or the mix of co-occurring compounds — that can be diagnostic of a source type.

10. What is needed to begin, and what is produced? Useful materials include all environmental analytical data with laboratory reports and validation, sampling plans, site and operating history, permit and regulatory files, any prior risk assessments, water system records where drinking water is at issue, biomonitoring results, and residential and employment history for the individuals at issue. The deliverable is a report documenting data evaluation, chemical selection, exposure pathways and parameters, exposure point concentration derivation with statistical output, toxicity value selection or derivation, dose and risk calculations, background comparison, and uncertainty and sensitivity analysis — with the database, ProUCL output, and calculation spreadsheets produced so the analysis can be independently reproduced.

References