environment · health · chemistry
ninety-eight percent of u.s. chemicals, never assessed
Only 1.8% of U.S. Commercial chemicals have a published human health assessment from EPA or ATSDR
Problem statement
More than 350,000 chemicals and mixtures of chemicals are registered for production and use worldwide — up to three times as many as previously estimated (Wang et al. 2020). Only a small fraction carry a completed human health assessment: among chemicals on the U.S. TSCA active inventory, just 1.8% have a published human health assessment from EPA or ATSDR, and developing one "typically takes at least 4 years" (EPA ETAP value-of-information case study, 2024). The conventional route is not scalable to the backlog — current approaches "rely on extensive animal testing, cost millions of dollars, and can take 2–3 years per chemical" (Judson et al. 2009). Newer high-throughput screening methods (in vitro assays, computational toxicology) flag molecular-level bioactivity but have not been shown to reliably predict chronic organism-level effects like endocrine disruption, developmental neurotoxicity, or carcinogenicity from low-dose, long-duration exposure — the exposure pattern most relevant to human health.
Why this matters
Pollution of all kinds remains responsible for an estimated 9 million premature deaths per year — one in six deaths worldwide (Fuller et al. 2022). Deaths from "modern" pollution risk factors — ambient air pollution and toxic chemical pollution such as lead — rose 7% between 2015 and 2019 and by more than 66% since 2000, while deaths from "traditional" pollution (household air pollution, unsafe water) fell. Data coverage runs in the opposite direction from where the gap is widest: because REACH registration deadlines took the highest-tonnage substances first, a published analysis of ECHA's registration data, which its authors call "the largest public dataset of in vivo and in vitro toxicity tests," is "clearly biased toward high-production volume substances" (Luechtefeld et al. 2016) — lower-volume chemicals get the least scrutiny of all. PFAS compounds circulated for decades before their persistence and bioaccumulation were characterized; similar latent risks likely exist among thousands of newer compounds.
What’s been tried and why it hasn’t worked
The EPA's ToxCast/Tox21 program has screened almost 10,000 substances using high-throughput in vitro assays aggregated from more than 20 assay sources, but these assays measure molecular-level endpoints (receptor binding, enzyme inhibition) rather than organism-level chronic effects. The most systematic test of whether they bridge that gap came back negative: a cross-validation of more than 600 ToxCast phase I assays against 60 in vivo endpoints using 84 different statistical classification methods found that, with the exception of chronic in vivo cholinesterase inhibition, "the overall predictive power of both the in vitro assays and the chemical descriptors was relatively low," the assays did no better than chemical structure descriptors alone, and prefiltering assay data outside the cross-validation loop — as earlier studies had done — "significantly biased estimates of model performance." The authors concluded the assays have "limited applicability for predicting in vivo chemical hazards using standard statistical classification methods" (Thomas et al. 2012). QSAR (quantitative structure-activity relationship) models trained directly on the Tox21 assay data perform better but vary widely and unpredictably by endpoint — AUC-ROC 0.55–0.87 across the Tox21 Challenge assays (Capuzzi et al. 2016) — and what they predict is an in vitro assay result, not an in vivo chronic effect. Organ-on-chip and organoid models offer better physiological relevance than cell-based assays but face the same challenge as all in vitro methods: no validated mapping exists from in vitro response to in vivo chronic effect at population scale. The European REACH regulation requires toxicity information for substances manufactured or imported above 1 tonne per year, with requirements rising by tonnage band — but a review of publicly available data for 9,801 registered substances found 5,551 of them (56.6%) "were missing explicit in vivo key experimental skin sensitization studies," registrants instead relying on read-across and study waivers.
What would unlock progress
An integrated approach combining high-throughput in vitro screening, QSAR prediction, and human biomonitoring data (exposome studies) could enable tiered screening that prioritizes the most concerning chemicals for deeper assessment. The key missing piece is a validated adverse outcome pathway (AOP) framework that connects molecular initiating events (measurable in vitro) to adverse outcomes in organisms. If even 10 well-characterized AOPs were validated for the most common toxicity endpoints, they could serve as bridges from high-throughput data to regulatory decisions.
Entry points for student teams
A student team could take an existing ToxCast dataset for a specific chemical class (e.g., organophosphate flame retardants) and attempt to build a predictive model linking in vitro assay results to known in vivo outcomes from the literature, evaluating which assay combinations are most predictive. Alternatively, teams could develop visualization tools for the ToxCast/Tox21 database that help regulators identify chemicals with concerning activity profiles that lack in vivo studies. Relevant disciplines: computational biology, toxicology, data science, environmental engineering, public health.
Genes
Structural cousins: stuck the same way, in other fields
Sources
Wang, Z., Walker, G. W., Muir, D. C. G. & Nagatani-Yoshida, K., "Toward a Global Understanding of Chemical Pollution: A First Comprehensive Analysis of National and Regional Chemical Inventories," Environmental Science & Technology, 54(5), 2575–2584, 2020, Accessed 2026-08-20; Thomas, R. S., Black, M. B., Li, L., Healy, E., Chu, T.-M., Bao, W., Andersen, M. E. & Wolfinger, R. D., "A Comprehensive Statistical Analysis of Predicting In Vivo Hazard Using High-Throughput In Vitro Screening," Toxicological Sciences, 128(2), 398–417, 2012, Accessed 2026-08-20; Judson, R. et al., "The Toxicity Data Landscape for Environmental Chemicals," Environmental Health Perspectives, 117(5), 685–695, 2009, Accessed 2026-08-20; Luechtefeld, T., Maertens, A., Russo, D. P., Rovida, C., Zhu, H. & Hartung, T., "Global Analysis of Publicly Available Safety Data for 9,801 Substances Registered under REACH from 2008–2014," ALTEX, 33(2), 95–109, 2016, Accessed 2026-08-20; U.S. EPA, "Value of Information Case Study on the Human Health and Economic Trade-offs Associated with the Timeliness, Uncertainty, and Costs of the Draft EPA Transcriptomic Assessment Product (ETAP)," July 2024, Accessed 2026-08-20; Capuzzi, S. J., Politi, R., Isayev, O., Farag, S. & Tropsha, A., "QSAR Modeling of Tox21 Challenge Stress Response and Nuclear Receptor Signaling Toxicity Assays," Frontiers in Environmental Science, 4:3, 2016, Accessed 2026-08-20; U.S. EPA, "Toxicity Forecasting (ToxCast)," Accessed 2026-08-20; ECHA, "Do I reach the one tonne a year threshold?", Accessed 2026-08-20; Fuller, R. et al., "Pollution and health: a progress update," The Lancet Planetary Health, 6(6), e535–e547, 2022, )00090-0/fulltext; accessed 2026-02-20, re-verified 2026-08-20 (supports the pollution-mortality figures only) go to source 1 ↗ go to source 2 ↗ go to source 3 ↗ go to source 4 ↗ go to source 5 ↗ go to source 6 ↗ go to source 7 ↗ go to source 8 ↗ go to source 9 ↗
Working notes
Notes made while writing and checking this brief, shown as written. They use working shorthand.
Related briefs: `environment-pfas-destruction-at-scale` (addresses destroying a specific known pollutant class, not the upstream screening problem); `water-emerging-contaminant-realtime-detection` (addresses detecting known contaminants in water, not determining whether chemicals are toxic in the first place). Source-bias note: the Lancet Commission frames this as a policy/investment failure; the binding constraints are also technical (no validated in vitro-to-in vivo extrapolation method) and economic (rodent testing at scale is prohibitively expensive). The `failure:not-attempted` tag reflects the Lancet Commission's finding that the vast majority of commercial chemicals have simply never been assessed — not that assessment was attempted and failed.
Reconciliation 2026-08-20: re-sourced after a source check found the brief's technical specifics were not in its sole original source. The Lancet Planetary Health pollution-burden update (Fuller et al. 2022) supports the mortality figures only; it carries nothing on ToxCast, bioassay costs, QSAR performance, or REACH. Corrections made: (1) the mortality sentence attributed all 9 million pollution deaths to chemical pollution — Fuller et al. attribute 9 million deaths to pollution of all kinds (one in six deaths worldwide), with the 66%-since-2000 rise belonging to "modern" risk factors defined as ambient air pollution plus toxic chemical pollution such as lead; the unsourced "more than malaria, HIV, and tuberculosis combined" comparison was removed. (2) "$2–6 million per chemical and takes 3–5 years" could not be verified in any primary source and was replaced with Judson et al. 2009's verified wording ("cost millions of dollars, and can take 2–3 years per chemical"). (3) "fewer than 5% have undergone comprehensive toxicological assessment" now rests on the EPA ETAP 2024 figure — 1.8% of TSCA active-inventory chemicals have a published human health assessment from EPA or ATSDR. The display title's "never tested" is a shorthand for that assessment gap, not a claim that no data of any kind exists: Judson et al. 2009 found roughly two-thirds of the ~10,000 chemicals they surveyed had at least limited toxicity summaries and about one-quarter appeared in at least one highly curated toxicology database. (4) "~700 in vitro assays" was replaced with EPA's own current framing (almost 10,000 substances, 20+ assay sources) plus Thomas et al. 2012's verified "more than 600 in vitro assays from the ToxCast phase I screening effort." (5) The "AUC typically 0.6–0.7 for developmental toxicity" figure could not be verified and is contradicted by the published range; it was replaced with the verified Tox21 Challenge span (AUC-ROC 0.55–0.87, Capuzzi et al. 2016), and the in-vitro-to-in-vivo failure claim now rests on Thomas et al. 2012, which is the direct evidence for it. (6) "The chemicals with the least toxicological data are often the ones produced in the largest volumes" was backwards and has been reversed to match Luechtefeld et al. 2016, whose REACH data analysis is "clearly biased toward high-production volume substances." (7) "compliance is low and penalties are weak" (unsourced) was replaced with the verified REACH dossier finding (56.6% of 9,801 substances missing explicit in vivo key skin-sensitization studies). The 1 tonne/year threshold is confirmed on ECHA's registration-obligations page. Genome tags untouched — but note that the `failure:not-attempted` rationale in the paragraph above attributes the never-assessed finding to the Lancet Commission; that attribution is wrong. The tag's evidentiary basis is now the EPA ETAP 2024 figure (1.8% of TSCA active-inventory chemicals with a published human health assessment) plus Judson et al. 2009, both of which describe assessment never having been undertaken at scale rather than having been attempted and failed. The failure-tag decision order (taxonomy.md §4) still applies on next review.
Reconciliation 2026-09-29: Fact repair after a site review read this write-up, including its titles, against the sources it cites. Read in full on 2026-09-29: Wang et al. 2020, from the Internet Archive copy of the publisher's PDF (https://web.archive.org/web/20230505051830/https://pubs.acs.org/doi/pdf/10.1021/acs.est.9b06379) because the ACS address blocks automated access; Judson et al. 2009 (https://pmc.ncbi.nlm.nih.gov/articles/PMC2685828/); Luechtefeld et al. 2016 (https://pmc.ncbi.nlm.nih.gov/articles/PMC5408747/); Capuzzi et al. 2016 (the cited Frontiers address); EPA's ToxCast page (the cited address); Fuller et al. 2022, from its open copy at https://pmc.ncbi.nlm.nih.gov/articles/PMC11995256/ because the Lancet address blocks automated access; and the EPA ETAP case study. For the ETAP study the cited NCBI address returned a robot check, so its Background chapter was read from the Internet Archive copy of the same page (https://web.archive.org/web/20251014000832/https://www.ncbi.nlm.nih.gov/books/NBK608021/) and the whole report was searched in EPA's own PDF of it (report EPA/600/X-23/082, downloaded from https://cfpub.epa.gov/si/si_public_record_Report.cfm?dirEntryId=361912&Lab=CCTE). The ECHA page was read from its Internet Archive copy dated 2024-05-26, because the live page refused the request. Thomas et al. 2012 sits behind a publisher block; only its abstract and citation record were read, from Europe PMC. (1) The main title said "Only ~5% of Commercial Chemicals Have Been Assessed for Chronic Low-Dose Toxicity." No source read gives a 5% figure for this. The EPA study says that among chemicals on the U.S. TSCA active inventory, "only 1.8% have a published human health assessment" from EPA or ATSDR. The title now states that figure, the inventory it is a share of, and the two agencies whose assessments it counts. (2) The short title said "Ninety-Five Percent of Chemicals, Never Tested." No source read gives 95%. The same EPA chapter says 15% of chemicals in commercial use have been evaluated with traditional animal-based toxicity tests, so "never tested" did not fit a figure above 85%. The short title now says "Ninety-Eight Percent of U.S. Chemicals, Never Assessed," which is the 1.8% figure turned around (the EPA figure shows 98.2% without an assessment). It means chemicals on the U.S. active inventory with no published human health assessment from EPA or ATSDR. It does not mean that no data of any kind exists for them. (3) The text called Luechtefeld et al. "the largest published analysis of registered safety data." The paper does not say that about itself. It says ECHA holds "the largest public dataset of in vivo and in vitro toxicity tests," and the sentence now says that. Confirmed against the sources and left as written: more than 350,000 chemicals and mixtures, up to three times earlier estimates (Wang et al.); 1.8% and "typically takes at least 4 years" (EPA); the Judson et al. quotation on cost and time; 9 million deaths a year, one in six deaths, and the rises of 7% since 2015 and more than 66% since 2000 for modern pollution (Fuller et al.); almost 10,000 substances and 20+ assay sources (EPA ToxCast page); more than 600 assays, 60 endpoints, 84 methods and all four quotations from Thomas et al. (abstract); AUC-ROC 0.55–0.87 (Capuzzi et al.); 9,801 substances, 5,551 missing the skin sensitization studies, which is 56.6% by division, and the "clearly biased" quotation (Luechtefeld et al.); the one tonne a year threshold (ECHA). The Source line entries match the fetched records for all nine sources; the EPA study's record also names its authors (Devito, Farrell, Hagiwara and others), which the Source line gives as U.S. EPA. NOT checked: the sentences on PFAS history, on organ-on-chip and organoid models, and on adverse outcome pathways carry no citation and were not found in the sources read in full; they may rest on general knowledge or on Thomas et al., which could only be read as an abstract, so they were left alone. The earlier note's phrase "the display title's 'never tested'" refers to the short title as it stood before this repair. Genes untouched.