Environmental toxins
The air you breathe and the everyday products you put on your skin show up in your bloodstream and on your epigenetic clock. The interventions that actually work are unglamorous — a bedroom air filter, fewer scented products, decent water filtration and, cheapest of all, a radon test — and while non-persistent chemicals clear within days, lead, cadmium and "forever chemicals" linger for years.
A large share — but not the overwhelming majority — of how long you live is environmental and modifiable rather than inherited. Twin studies put the heritability of lifespan at about 20–25% (reviews of the wider literature cluster around 20–30%),[1] with some large pedigree studies as low as 6%; a 2026 reanalysis argues these twin figures are deflated by accidents and infections and that heritability of intrinsic ageing is "above 50%" once such extrinsic deaths are stripped out (evidence: moderate).[2] Either way, environment and behaviour move a big, addressable fraction. Air pollution is the second-leading global mortality risk factor — about 7.9 million premature deaths a year, roughly one death in eight — and the dose-response runs through chronic disease, not acute toxicosis: 86% of those deaths are from noncommunicable disease.[3] Cadmium, lead, arsenic, phthalates, bisphenols, and perfluoroalkyl substances (PFAS) all show up in human blood or urine, and in US survey data cadmium and lead track with faster epigenetic-clock ageing. Two categories behave very differently, and conflating them is a common mistake: non-persistent compounds (phthalates, bisphenol A, parabens, most volatile organic compounds) clear within days to weeks once you stop the input, whereas bioaccumulative persistents — PFAS (serum half-life of years), cadmium (kidney half-life of decades), lead (stored in bone for years), and dioxins/PCBs — do not.
The exposome — environmental load as an aging input
The exposome is the cumulative environmental exposure an individual has from conception to death — air, water, chemicals in consumer products, dietary contaminants, and psychosocial stressors. It's now one of the dominant frameworks for explaining aging variance that genetics doesn't.[4]
The mechanism is mostly oxidative-stress-driven: pollutants raise reactive oxygen species, which damage DNA, accelerate telomere attrition, and impair autophagy. This shows up on epigenetic clocks before it shows up clinically.
What the epigenetic clocks pick up
Two of the more useful clocks for exposome work:
- GrimAge / GrimAge2 — trained to predict time-to-death from methylation-based stand-ins for seven plasma proteins plus a methylation estimate of smoking pack-years; GrimAge2 adds markers for C-reactive protein and long-term blood sugar. That built-in smoking component is worth remembering when it responds to tobacco-linked exposures like cadmium.[5]
- DunedinPACE — a "speedometer" measuring biological years aged per chronological year, built from 19 organ-system measures tracked across two decades, and reliable enough on repeat testing to detect change.[6] It is the clock most often used to test whether an intervention moved anything.[7]
A 2025 analysis of the US National Health and Nutrition Examination Survey (NHANES), covering 2,346 adults aged 50–84, found per-standard-deviation increases in some measured exposures associated with faster epigenetic ageing (the lead figure comes from a model that also adjusts for cotinine, a tobacco-smoke marker):[8]
| Exposure | Source | Epigenetic acceleration per SD |
|---|---|---|
| Cadmium (blood) | Tobacco smoke, polluted air, contaminated soils | +1.23 yr GrimAge, +1.27 yr GrimAge2 |
| Cotinine | Active or second-hand smoke | +1.40 yr GrimAge2 |
| Lead | Legacy industrial, paint, pipes, contaminated soil | +0.73 yr GrimAge, +0.67 yr GrimAge2 |
Read the cadmium row with its caveat: the association is essentially confined to current and former smokers — around +1.3 years in current smokers and +2.0 in former smokers, and null in people who never smoked — which is what you would expect if tobacco is doing most of the work.
A machine-learning analysis of 78 simultaneous chemical exposures in 1,693 US adults found the opposite of what you might expect: adding all 78 chemicals did not improve prediction of biological-age acceleration over demographics plus diabetes and body-mass index. Within the mixture, blood toluene, a urinary benzene metabolite and urinary benzophenone-3 (a sunscreen filter) contributed most — but the authors' own conclusion is that "single cross-sectional measurements of chemical exposures have limited utility for predicting individual aging rates."[9]
The single-substance regulatory model under-counts real-world risk. In the European HBM4EU biomonitoring data, 17% of children and adolescents across twelve countries carried a combined burden of five anti-androgenic phthalates above the level judged safe, against only 6% who exceeded the guidance value for any individual phthalate. The authors' own framing is the sharper one: 63% of those at mixture risk would have gone unnoticed in a single-substance assessment.[10]
Psychosocial exposome
The exposome includes social stressors too. A 2026 PNAS analysis found each additional chronic-stress relationship ("hassler") in someone's close social network was associated with ~1.5% faster biological aging by DunedinPACE.[11] See Stress and Purpose.
Air: filter what you breathe
People in industrialised countries spend roughly 90% of their time indoors — a large US activity survey found 87% in buildings and another 6% in vehicles[12] — and indoor levels of some pollutants, notably many volatile organic compounds (VOCs), can be several times higher than outdoors, while fine particulate matter (PM2.5) indoors can exceed outdoor levels whenever there are indoor sources — pollution pulled in from traffic and wildfire smoke, then added to by gas stoves, scented products, and off-gassing furniture.
PM2.5 is the clearest harm. Particles smaller than 2.5 µm cross the alveolar barrier into systemic circulation and have been recovered from coronary plaques, brain tissue, and placenta. The pooled European ELAPSE analysis put numbers on this: each 5 µg/m³ of long-term PM2.5 raised natural-cause mortality by about 13% — the high end of the published range, since re-analyses that harmonise age, exposure model and confounders across the European, US and Canadian cohorts land nearer 3–4% — and, crucially, the association persisted and even steepened in the subgroup exposed to less than 12 µg/m³ (the US limit at the time, and well below the EU's), so there is no clean "safe" threshold (evidence: strong).[13][14] A contrarian minority disputes the causal reading on statistical grounds, but causal-inference analyses support it. In US Medicare data — analysed with methods designed to approximate a causal contrast rather than a correlation — mortality ran about 9% higher at the current US limit of 9 µg/m³ than at the World Health Organization's guideline of 5 µg/m³ (rate ratio 1.088, 95% CI 1.064–1.113), across the 223 million person-years between those two levels, and no threshold was detectable down to 4 µg/m³.[15]
Air pollution and the ageing brain
The best-evidenced ageing endpoint for air pollution is dementia, not the lungs. A 2025 review of 32 studies found, in the 21 of them covering about 24 million people that could be pooled for fine particles, that each 5 µg/m³ of long-term PM2.5 raised incident-dementia risk by about 8% (HR 1.08, 95% CI 1.02–1.14). Nitrogen dioxide gave a smaller signal; black carbon (traffic soot) a steeper one, but on only six studies with a confidence interval that nearly touches no-effect. Nitrogen oxides, PM10 and ozone were all null. Certainty was moderate, limited by very high between-study heterogeneity and the possibility of residual socioeconomic confounding (evidence: moderate).[16] The 2024 Lancet Commission on dementia now lists air pollution among 14 modifiable risk factors, attributing roughly 3% of global dementia cases to it.[17] This is the mechanism that ties the HEPA and ventilation advice below to a healthspan outcome, not just a lung one.
A widely shared 2024 headline claimed wildfire smoke sharply raised dementia risk; it was retracted and replaced in 2025 after the authors found that people with Parkinson's disease or a secondary dementia diagnosis had been wrongly counted as dementia cases. Corrected, the wildfire-smoke association shrank from 18% to 12% higher odds and was no longer statistically significant (OR 1.12, 95% CI 0.98–1.28), while the association with non-wildfire PM2.5 held — ordinary science correcting itself (evidence: weak-preliminary).[18] Treat wildfire-specific dementia claims cautiously; the general PM2.5 signal above is the solid one, and an N95 respirator during heavy smoke or an air-quality index above 100 is a sensible precaution.
HEPA in the bedroom
The best-evidenced intervention for actually lowering your exposure is a single, properly-sized HEPA (high-efficiency particulate air) portable air cleaner running in the bedroom overnight. In a four-week double-blind crossover trial in 33 residents of a US senior housing facility beside a major highway, the active filter dropped bedroom PM2.5 from 8.3 to 1.3 µg/m³ and cut 24-hour personal exposure by about 60% (12.4 → 5.1 µg/m³). Blood pressure, though, did not move: home systolic averaged 127.0 mmHg on the real filter against 126.5 on the sham — a difference of half a millimetre of mercury, well inside the range expected from chance — with only non-significant trends toward benefit in older, heavier participants who started with higher pressure.[19] A 2025 review of 63 indoor-air-quality experiments is more encouraging, but only for intermediate markers: filtration improves heart-rate variability, blood pressure and pulmonary function. No filtration trial has yet shown a reduction in actual heart attacks, strokes or respiratory admissions; the review's authors go no further than suggesting that sustained indoor improvements "may decrease incidence rates of respiratory pathologies" (evidence: moderate for surrogate markers, absent for clinical events).[20]
True HEPA is rated to remove ≥99.97% of particles at 0.3 µm — its worst case, the size that penetrates most easily. Particles both larger and smaller, including the ultrafines from traffic and gas cooking, are captured more efficiently still.[21] It is not effective against gaseous pollutants — VOCs, ozone, NO₂. For gases, the only option in a portable unit is a dense activated-carbon bed; thin "carbon-coated" pre-filters in standard consumer units saturate quickly and provide little protection.
What about VOCs?
The intuition that you can "purify" gases away tends to fail in practice.
A 2024 chamber study at the US National Institute of Standards and Technology tested ten portable air cleaners across seven technologies and found the oxidation-based ones the worst offenders: photocatalytic-oxidation and germicidal-UV units had the highest aldehyde emission rates (202 and 243 µg/h), and three of the ten units put out more volatile organic compounds than they removed.[22] Ozone-generating "ionizers" and plasma units carry the same problem at the residential scale.
The honest VOC playbook is source control plus ventilation, not chemical destruction:
- Stop using aerosolised synthetic fragrances, plug-in air fresheners, and scented candles. Fragrance blends are undisclosed and can include phthalates (more on this below).
- Cook on an exhaust fan vented outdoors. Burning gas on a stove is a substantial NO₂ source.
- Open windows for cross-ventilation when outdoor air is reasonably clean.
- Use a HEPA filter for particulate matter and accept that VOC removal is a separate, harder problem.
The houseplant myth
A persistent claim, repeatedly debunked, is that houseplants meaningfully clean indoor air. The original 1989 NASA study used hermetically-sealed chambers; the dynamics don't translate to actual rooms.[23] Real-world ventilation dilutes VOCs faster than plant stomata can absorb them. A single potted plant delivers a median of about 0.023 cubic metres of cleaned air per hour. To match what ordinary air exchange with the outdoors already does for free, you would need on the order of 10–1,000 plants per square metre of floor.[24] Houseplants are nice. They are not air filters.
Endocrine-disrupting chemicals in personal care products
Endocrine-disrupting chemicals (EDCs) — phthalates, bisphenols, parabens, triclosan, certain glycol ethers — interfere with hormone synthesis, transport, and signalling at sub-toxicological doses, often violating the monotonic dose-response assumed in classical toxicology. They saturate cosmetics, deodorants, lotions, plastic food packaging, and stain-resistant or non-stick coatings. Daily use adds up: the young women in the French IRECO study described below used a median of 12 personal-care products a day before the intervention.[25]
One blood marker worth understanding here is sex hormone-binding globulin (SHBG) — a liver protein that sets how much testosterone and estradiol circulates in a usable form. Obesity, fat accumulation in the liver and the inflammation that goes with them suppress SHBG synthesis, releasing unbound androgens and estrogens; low SHBG tracks with metabolic syndrome and with the polycystic ovary syndrome (PCOS) phenotype.[26] Whether chemical exposure moves SHBG specifically is not something that review addresses, and no good biomarker of everyday endocrine-disruptor exposure is in routine clinical use.
What behaviour change actually does — the PERTH trial (Moderate)
The strongest interventional evidence is randomised. The PERTH trial (Nature Medicine 2026) gave participants combinations of plastic-free kitchenware, low-plastic personal-care products, and food sourced to minimise every plastic touchpoint from paddock to plate, against a no-intervention control, and measured urinary plastic-associated chemicals as the primary outcome.[27] Holding daily energy intake constant, the dietary intervention cut urinary levels of:
| Compound | Where it comes from | Reduction (95% CI) |
|---|---|---|
| Monobenzyl phthalate | Plasticiser, vinyl flooring, packaging | −53.5% (−72.7 to −20.6) |
| Bisphenol A (BPA) | Can linings, plastic food packaging, receipts | −59.7% (−82.5 to −6.9) |
| Mono-n-butyl phthalate | Plasticiser, some cosmetics | −37.5% (−55.6 to −12.0) |
(The bracketed 95% confidence interval is the range in which the true reduction most plausibly lies; where it stays well below zero, as here, a real drop is unlikely to be a chance finding.)
Two findings matter for what you actually do. First, food was the dominant lever — the arms given food with minimal plastic contact showed the broadest reduction, and in the study's accompanying observational cohort of 211 people, highly processed, plastic-packaged and canned foods were the important modifiable sources. Second, swapping personal-care products alone produced an independent drop in mono-n-butyl phthalate — so the cosmetics shelf is a real but smaller lever than the kitchen.
One result cuts against the headline, and it is the one to know. DEHP metabolites did not fall over the seven days — if anything they trended up — despite participants eating only food processed, stored and served in glass, steel or wood, and no packaged, canned or highly processed food at all. DEHP is the phthalate class most consistently tied to mortality, including in the survey data cited below. The authors offer three candidate explanations: DEHP already in the food rather than from its packaging, sources outside the diet, or slow washout from fatty tissue.
The point is still the timescale. These chemicals are non-persistent in the body and continuously replenished from daily use, so for most of them, cutting the input drops the urinary load within weeks. The trial is small — an exploratory pilot of 60 participants split into five arms of twelve, run for seven days — and measured biomarkers, not health outcomes; the three headline reductions above compare the low-plastic food-plus-kitchenware arm with the control. It proves the exposure is modifiable, not that modifying it extends life.
A parallel French intervention (IRECO) points the same way. For five days, 103 women aged 18–30 cut their daily personal-care products from a median of 12 to 7 and swapped the rest for supplied products free of eleven phenols (including parabens and bisphenol A), sixteen phthalates and two glycol ethers; over that horizon urinary phenoxyacetic acid fell 64%, bisphenol A 39%, methylparaben 30% and monoethyl phthalate 22%.[28] Five days is a short window and these are biomarkers rather than health outcomes — but the direction and the speed match PERTH closely. A 2026 review of dietary, behavioural and residential interventions against these non-persistent chemicals reaches the same overall conclusion — cutting the sources lowers the urinary markers — while judging the evidence that this improves health "limited and inconclusive."[29]
At the population level, higher urinary high-molecular-weight phthalate metabolites track with about 14% higher all-cause mortality in US survey data — that is per roughly a 2.7-fold rise in concentration, not per any everyday step — which the authors translate into roughly 90,000–107,000 excess deaths a year among Americans aged 55–64 (evidence: weak-moderate — this rests on single spot-urine snapshots, where reverse causation and residual confounding are hard to exclude; and, for balance against the chemical-industry conflicts flagged elsewhere on this page, the lead author sits on advisory boards for companies selling plastic and cosmetic alternatives, though this particular study was NIH-funded and declares no competing interests).[30] A larger 2025 analysis treating eight phthalates and bisphenol A as a single mixture puts the association higher still, at about 35% more deaths for each step up between thirds of exposure.[31]
EU regulators are catching up, and the sharpest move was on bisphenol A: in 2023 the European Food Safety Authority cut its tolerable daily intake roughly twenty-thousand-fold, to 0.2 nanograms per kilogram of body weight per day.[32] Mandatory classification of substances with endocrine-disrupting properties under the EU's labelling regulation applies to new substances from May 2025 and to those already on the market from November 2026.[33] The EU chemicals agency (ECHA) also added bisphenol AF and n-hexane to its candidate list of substances of very high concern in February 2026 — though on reproductive-toxicity and repeat-dose organ-toxicity grounds respectively, not as endocrine disruptors.[34] Until regulation reaches consumer products, individual mitigation is the only lever.
Substitution is not automatically progress. As BPA use fell, exposure to its replacement BPS rose — detected in 20% of a US population sample in 2000 and 60% by 2021 — and a 2026 scoping review of 34 human studies found it linked, in a handful of them, to outcomes including polycystic ovary syndrome and gestational diabetes.[35] The PERTH participants above carried BPS at twice US levels, in 97% of them.
Where to actually act
In rough order of effect size for a midlife adult:
- Drop synthetic fragrance. "Fragrance" or "parfum" on an ingredient list is a regulatory loophole: the individual components do not have to be declared, and diethyl phthalate is still a common fragrance solvent — although exposure to it has fallen substantially, with its urinary marker down about 42% in US survey data between 2001–02 and 2009–10.[36] Plug-in air fresheners, scented candles, fabric refreshers, perfumed laundry detergents, and most colognes are the highest-yield items to remove. Fragrance-free or essential-oil-only alternatives exist for every category.
- Get plastic out of food contact at heat or in fat. Phthalate and bisphenol leaching scales sharply with temperature and lipid content. Don't microwave in plastic, don't store fatty leftovers in plastic, don't drink hot liquids through plastic-coated cups. Borosilicate glass and stainless steel are inert. Cast iron, ceramic, and uncoated stainless replace PTFE/PFAS non-stick cookware; chipped or scratched non-stick pans should be retired.
- Read cosmetic and personal-care labels. Watch for parabens (ethyl-, methyl-, propyl-, butyl-), triclosan, oxybenzone (benzophenone-3), and the umbrella term "fragrance." Long-wear waterproof cosmetics frequently contain PFAS and are worth replacing first.
- Manage household dust. Persistent flame retardants and PFAS shed off treated furniture and carpeting and accumulate in dust — which then gets ingested via hand-to-mouth contact. A vacuum with true HEPA filtration plus ordinary handwashing with plain soap before meals does most of the work. Aerosolised antibacterial sprays and triclosan-containing soaps add to the load without measurable benefit.
Heavy metals — the strongest population evidence
If the air section has the largest headline death toll, heavy metals have the most solidly quantified one. Lead and cadmium are bioaccumulative persistents that stay in the body for years to decades; inorganic arsenic is mostly excreted in urine within days, but exposure through drinking water and rice is usually continuous. Either way the lever is preventing intake, never a "cleanse."
- Lead. There is no safe threshold. A 2023 modelling study estimated that lead exposure caused about 5.5 million adult cardiovascular deaths in 2019 — roughly six times the previous global estimate and on the same order as fine-particle air pollution — at a cost near US$6 trillion, about 6.9% of global GDP (evidence: strong for the CVD link; the exact count is genuinely uncertain — Larsen's own range runs from 2.3 to 8.3 million — but the mainstream estimate has moved toward it rather than away: the Global Burden of Disease study now attributes 3.5 million deaths in 2023 to lead, mostly cardiovascular, having adopted the same direct, non-blood-pressure pathway Larsen argued it was missing).[37][38] Lead drives hypertension, atherosclerosis, and — as the NHANES table above shows — epigenetic-age acceleration.
- Cadmium. A meta-analysis of 37 studies and about 350,000 people found roughly a third higher cardiovascular-disease risk and a 70% higher stroke risk in the most-exposed.[39] In US survey data, people at the high end of urinary cadmium (80th percentile) had about 74% higher cardiovascular mortality than those at the low end (20th percentile), and about two and a half times the heart-disease mortality.[40] A 2026 umbrella review that formally graded the credibility of 35 meta-analyses is more cautious, rating the cadmium–cardiovascular link only "weak" and the cadmium–stroke link "suggestive" (evidence: moderate).[41] Cadmium's kidney half-life is 10–30 years. The dominant sources are tobacco (the single biggest controllable one) plus organ meats, shellfish, leafy greens, and rice.
- Arsenic. A dose-response meta-analysis found each doubling of drinking-water arsenic raised cardiovascular-disease incidence by about 9% and coronary-heart-disease mortality by about 16%, with risk rising continuously from the WHO 10 µg/L guideline upward and no evident threshold — though the evidence below the guideline itself is suggestive rather than established; arsenic is also an International Agency for Research on Cancer (IARC) Group 1 human carcinogen for lung, skin, and bladder cancer (evidence: strong).[42] The main exposure is groundwater, especially private wells, plus rice. Arsenic is also the one metal where lowering the burden has been shown to change outcomes: among 10,977 adults in Bangladesh followed to 2022, community well-switching cut mean urinary arsenic from 283 to 132 µg/g creatinine, and those whose levels fell below the median had about 57% lower cardiovascular mortality than those who stayed high (adjusted HR 0.43), approaching the risk of people never highly exposed.[43] This was a high-exposure cohort — mean well-water arsenic around 100 µg/L, about ten times the WHO guideline — and the switching was community-driven rather than randomised, so it does not show the same reversibility at the sub-10 µg/L levels most readers face.
- Mercury. No distinctive cardiovascular signal; methylmercury from fish is chiefly a neurodevelopmental concern, so grade it lower for adult ageing endpoints.
A 2023 American Heart Association scientific statement reviewed lead, cadmium and arsenic as contaminant-metal cardiovascular risk factors — a literature review by a writing group rather than a formal classification, and one whose own per-metal evidence is uneven.[44] The mitigation is mundane and already on this page: filter your water (reverse osmosis for arsenic — though well water often carries arsenite, the form that needs pre-oxidation before treatment), test a private well, control household dust and soil, don't smoke — and, emphatically, not chelation (see below).
PFAS — the health-endpoint story
Water filtration for PFAS lives on the Water page; the health and regulatory picture belongs here. In 2023 IARC classified PFOA — one of the most-studied "forever chemicals" — as a Group 1 human carcinogen. The nuance matters: that top-tier rating rests mainly on animal and mechanistic data plus only limited human evidence (for testicular and kidney cancer), and a commentary by three authors at Gradient, an industry-linked consulting firm, argues the working group overstated the case;[45] the related compound PFOS was placed in the weaker "possibly carcinogenic" Group 2B (evidence: moderate, contested).[46] A 2022 US National Academies review found "sufficient evidence of an association" for reduced vaccine-antibody response, raised cholesterol, lower fetal growth, and kidney cancer; it judged the cardiovascular evidence insufficient.[47]
Because PFAS half-lives run to several years — pooled human estimates span about 1.5–5 years for PFOA, 3.4–5.7 for PFOS and 2.8–8.5 for PFHxS[48] — they are the clearest counterexample to the "clears in days" rule and the main reason source reduction (and water filtration) beats any attempt to flush them out. Two removal interventions now have randomised-trial support, and both are biomarker-only. In heavily exposed firefighters, blood or plasma donation lowered serum PFOS over a year — plasma donation roughly three times as effective as whole blood, and about a quarter off a starting level near 11 ng/mL (the donation trial was funded by the fire service that employed the participants).[49] Considerably more powerful are the bile-acid sequestrants: in adults from a contaminated area starting near 191 ng/mL, twelve weeks of cholestyramine cut serum PFOS by about 60%, and colesevelam by about 38% in a separate trial.[50][51] Neither approach has been shown to improve any health outcome, and the resins are prescription drugs with gastrointestinal and nutrient-absorption side effects. These are supervised medical options for the heavily exposed, not wellness routines.
Pesticides — the glyphosate question
Glyphosate, the world's most-used herbicide, is where regulators openly disagree. IARC rates it "probably carcinogenic" (Group 2A), driven by non-Hodgkin lymphoma, and a meta-analysis found about 41% higher lymphoma risk among the most heavily exposed — mostly agricultural workers, where recall and healthy-worker biases apply.[52] That pooled figure carries a wide range (1.13–1.75), rests on only six studies, and the largest of them — the prospective US Agricultural Health Study — was itself null. An independent 2026 meta-analysis of ten primary datasets, with no external funding and formal certainty grading, broadly replicated the signal: the highest exposure categories carried roughly 28–45% higher odds of lymphoma depending on the statistical model (OR 1.28, 95% CI 1.08–1.53, and 1.45, 1.06–1.98), while simply ever having been exposed showed only a modest association; the authors graded the overall certainty as moderate.[53]
The other pole of the split is Europe, not the US: the European Food Safety Authority reviewed more than 2,400 studies and identified no critical areas of concern in July 2023,[54] and the EU renewed glyphosate's approval to 2033.[55] The US position is currently unsettled: the US Environmental Protection Agency (EPA) concluded in 2020 that glyphosate is unlikely to be a human carcinogen, but a federal appeals court vacated that human-health finding in 2022, EPA withdrew the decision, and the reassessment it was ordered to redo is still not finished (evidence: moderate, contested).[56]
For a general adult the practical, low-grade step is choosing organic for high-residue items, and the benefit is modest. In a two-week randomised crossover in 39 pregnant participants, with samples analysed at a US Centers for Disease Control and Prevention (CDC) laboratory, median urinary glyphosate barely moved on an organic diet — 0.19 to 0.16 µg/L — and did not fall at all in people living near farmland.[57] An uncontrolled before-and-after switch in four families, run with an environmental campaign group, reported a much larger fall of about 70%, but with no control arm.[58] Washing produce is worth doing for surface residues — a plain rinse removes roughly a third and a baking-soda soak about half — but it does little for glyphosate, which is taken up into the plant tissue rather than sitting on the skin; those washing figures come from authors at the Environmental Working Group, an advocacy organisation.[59]
Radon — the cheapest high-yield test on this page
Radon is a colourless radioactive gas that seeps from soil into homes and is the second-leading cause of lung cancer after smoking, implicated in roughly 21,000 US deaths a year on EPA's modelling — a model-based figure with a wide range around it,[60] with pooled European residential studies showing about 16% higher lung-cancer risk per 100 becquerels per cubic metre of long-term average exposure (8% as measured, before correcting for the random uncertainty in radon measurements) and no threshold below which the risk disappears (evidence: strong).[61] For a never-smoker it is plausibly the single highest-yield environmental hazard on this page, and the fix is cheap: a test kit costs on the order of a restaurant meal, and remediation — where a test comes back high — is roughly the cost of a domestic appliance. "High" has a number: the US action level is 148 Bq/m³ (4 picocuries per litre); the World Health Organization recommends a lower reference level of 100 Bq/m³, and no more than 300 Bq/m³ where that cannot be reached.[62] Radon and smoking combine synergistically — more than additively, though by the US National Research Council's preferred model somewhat less than fully multiplicatively — so smokers benefit most of all.[63]
Water: a separate, larger story
PFAS, microplastics, heavy metals, and disinfection by-products are all consequential aqueous exposures, but the filtration story (RO vs. UF, the demineralisation trap, why activated-carbon pitchers are unreliable for PFAS, what to do about microplastics in bottled water) is a larger subject in its own right.
→ See Water for the full treatment.
The gut as a biotransformation organ
Persistent organic pollutants like PFAS were long assumed biologically inert — hence "forever chemicals." A 2025 Nature Microbiology paper from the University of Cambridge upended that. Specific human gut bacteria absorb PFAS molecules from the lumen and shuttle them out via faeces. Nine species took up 25–74% of the PFAS in their growth medium over 24 hours, and a wider screen of 89 strains found 38 of them high accumulators; the uptake itself happens within minutes. In germ-free mice colonised with a defined 20-strain human gut community, faecal PFNA ran roughly two to eight times higher than in germ-free controls.[64]
The same direction of effect appears in early work on dietary fibre, which interrupts the enterohepatic recirculation of PFAS — fibre traps biliary-secreted compounds for faecal excretion and feeds the bacterial species that do the sequestration. A Canadian study piloted a fibre-based method for lowering serum PFAS alongside its main cohort analysis,[65] and the same lead investigator's oat-fibre pilot in mice found only non-significant trends toward lower serum levels, for three of seven PFAS, against an inulin-fed comparison group.[66] Read this as mechanistically promising and Weak in humans: these are pilots and animal work, not trials showing that eating more fibre meaningfully clears your PFAS load.
Heavy metals do the opposite. In a systematic review of observational human studies, high arsenic, lead and mercury exposure went with a loss of beneficial commensal bacteria and more of the potentially harmful Collinsella, a pattern linked to a weaker gut barrier and more systemic inflammation; cadmium was the exception, and the authors judged the mercury and cadmium evidence too thin for a general conclusion.[67]
The honest read: this is mostly preclinical and pilot-level human data. It points in the same direction as everything else on the site that recommends fibre and a Mediterranean-pattern diet — and there's no downside to either. Don't take this as license to buy a "detox" probiotic stack. The dietary pattern carries the evidence; the specific strain cocktails do not yet.
→ See Dietary patterns and Fermented foods.
A practical environmental-toxin protocol
Ranked by evidence-to-effort:
- HEPA air filter in the bedroom, sized for the room (clean-air delivery rate, CADR, in m³/h of at least about five times the room's volume in m³ — roughly 12 × the floor area in m², the metric form of the US "2/3 rule" of CADR in cubic feet per minute ≥ 2/3 of floor area in square feet). Run it nightly. Skip plasma, ionizer, UV-PCO, and ozone-based "purifiers."
- Test your home for radon. A cheap test kit, then remediation only if it reads high — among the best evidence-to-cost trades on this list, especially for never-smokers.
- Reverse osmosis water filtration with remineralisation if your supply has any PFAS or industrial concern (reverse osmosis is also the fix for arsenic, but private-well water often carries arsenite, which needs pre-oxidation first — choose a system rated for what your test finds). Otherwise, a good ultrafilter is enough. Test a private well for arsenic and lead. Don't drink habitually from PET bottles.
- Eliminate synthetic fragrance from your household and personal care. Single biggest dermal/inhalational EDC reduction available.
- Remove plastic from food contact under heat or fat. Glass and stainless replace plastic for storage and reheating; cast iron and ceramic replace non-stick.
- Audit cosmetics and personal care for parabens, triclosan, oxybenzone, and PFAS in long-wear formulations.
- Source food with heavy metals in mind. Vary grains rather than leaning on rice (arsenic), go easy on organ meats and shellfish (cadmium), and prefer known-clean water for cooking.
- High-fibre dietary pattern — Mediterranean / MIND with legumes, vegetables, and whole grains. Independent of toxins, this is already what the rest of this site recommends.
- Vacuum and ventilate. True-HEPA vacuum, regular dusting, cross-ventilation when outdoor air is clean. Avoid antimicrobial sprays.
- Don't smoke. Avoid second-hand smoke. Tobacco is the single largest controllable source of cadmium and the highest-impact item on this list for anyone still exposed.
What's overrated
- Houseplants for air purification — pleasant, ineffective.
- Plasma / ionizer / UV-PCO purifiers — generate secondary pollutants; some studies show net VOC increase.
- Oral "detox" cleanses, juice fasts, chelation outside frank heavy-metal toxicity — no evidence base for healthy adults; some chelators are actively dangerous. The definitive test is now in: a 1,000-patient randomised trial of EDTA chelation in heart-attack survivors with diabetes — the subgroup in which an earlier trial had reported benefit, which is why it was chosen — found that blood lead fell by more than half and cardiovascular events did not budge: 35.6% on chelation against 35.7% on placebo (evidence: strong).[68] Chelation carries real risks — low blood calcium, kidney injury, and deaths reported to the CDC — and belongs to treating diagnosed poisoning, not wellness.[69] The hazard is specific: edetate disodium, the agent used in these deaths (TACT2 used it too, as slow, monitored infusions), strips calcium from the blood in a way that calcium disodium EDTA does not.
- Activated-carbon pitcher filters as a standalone PFAS solution — Testing by Duke University and North Carolina State University of 76 point-of-use filters found highly inconsistent PFAS removal by carbon.[70] In the published study, carbon filters removed roughly 60–70% of long-chain PFAS but only about 40% of the short-chain ones, while every under-sink dual-stage and reverse-osmosis filter tested showed near-complete removal; in four of eight whole-house carbon systems, PFAS levels in the filtered water actually rose.[71] A later review notes that improperly maintained carbon systems can raise PFAS levels in purified water.[72]
- Generic "anti-toxin" supplement stacks built around glutathione, NAC, milk thistle. There's a thin biomarker case for any of these in healthy adults; the evidence base is dietary pattern, not pills.
Cautions
- Personal biomonitoring (serum PFAS, urinary phthalate metabolites) is increasingly available commercially. The US National Academies (NASEM) guidance uses three bands, all for the sum of seven named PFAS (MeFOSAA, PFHxS, PFOA, PFDA, PFUnDA, PFOS and PFNA) rather than any single compound: below 2 ng/mL, adverse effects are not expected; 2 to under 20 ng/mL, there is potential for effects, especially in sensitive groups; 20 ng/mL or higher warrants further clinical follow-up.[73] A commercial panel that reports only one compound cannot be compared directly with these bands. For most healthy adults the test is more interesting than actionable: the corrective measures are the same as the protocol above whether or not you measured.
- Industrial occupational exposure (firefighting, semiconductor manufacturing, certain agriculture) is a different category — domestic mitigation does not substitute for occupational health monitoring.
Further reading
- Mesnage R. Environmental Health Is Overlooked in Longevity Research. Antioxidants (Basel) 2025.[74]
- Health Effects Institute / IHME. State of Global Air 2025.[75]
- Shenhar B et al. Heritability of intrinsic human life span is about 50% when confounding factors are addressed. Science 2026.[76]
- Strak M et al. Long-term exposure to low-level air pollution and mortality (ELAPSE). BMJ 2021.[77]
- Best Rogowski C et al. Long-term air pollution exposure and incident dementia: systematic review and meta-analysis. Lancet Planetary Health 2025.[78]
- Larsen B, Sánchez-Triana E. Global health burden and cost of lead exposure. Lancet Planetary Health 2023.[79]
- Chowdhury R et al. Environmental toxic metal contaminants and cardiovascular disease. BMJ 2018.[80]
- Lamas GA et al. Contaminant Metals as Cardiovascular Risk Factors (AHA scientific statement). J Am Heart Assoc 2023.[81]
- Lamas GA et al. EDTA chelation after myocardial infarction in diabetes (TACT2). JAMA 2024.[82]
- IARC. Carcinogenicity of PFOA and PFOS (Monographs Vol. 135). 2023.[83]
- Khodasevich D et al. Exposome-wide association study of environmental chemical exposures and epigenetic aging in the national health and nutrition examination survey. Aging (Albany NY) 2025.[84]
- Lee B et al. Negative social ties as emerging risk factors for accelerated aging. PNAS 2026.[85]
- Brook RD et al. Bedroom Air Filtration and Home Blood Pressures of Older Adults Living in a Residential Facility. J Clin Hypertens (Greenwich) 2026.[86]
- Link MF et al. Quantification of Byproduct Formation from Portable Air Cleaners Using a Proposed Standard Test Method. Environ Sci Technol 2024.[87]
- Harray AJ et al. Low-plastic diet and urinary levels of plastic-associated phthalates and bisphenols: the randomized controlled PERTH Trial. Nature Medicine 2026.[88]
- Cannarella R et al. Reducing exposure to non-persistent endocrine-disrupting chemicals and their impact on human health: Evidence from dietary, behavioral, and residential interventions. Ecotoxicol Environ Saf 2026.[89]
- Herkert NJ et al. Assessing the Effectiveness of Point-of-Use Residential Drinking Water Filters for Perfluoroalkyl Substances (PFASs). Environ Sci Technol Lett 2020;7(3):178–184.[90]
- Lindell AE et al. Human gut bacteria bioaccumulate per- and polyfluoroalkyl substances. Nat Microbiol 2025.[91]
- Lange R et al. Cumulative risk assessment of five phthalates in European children and adolescents. Int J Hyg Environ Health 2022.[92]
- Belsky DW et al. DunedinPACE, a DNA methylation biomarker of the pace of aging. eLife 2022.[93]