Most of the fourteen modifiable dementia risk factors the Lancet Commission names are things a person can act on or a clinician can treat — blood pressure, hearing, alcohol, activity, smoking. Air pollution is the one that mostly is not. What you breathe is set by the road outside your window, the heating your neighbours burn, and what your national regulator decided was tolerable. WHO puts it plainly: "most sources of outdoor air pollution are well beyond the control of individuals."
It is the awkward item on a list usually read as a personal to-do list, and the one item with nothing attached to sell — but the evidence has a shape worth knowing before anyone quotes a number at you.
PM2.5, and why particle size is the mechanism
PM2.5 means airborne particles 2.5 micrometres or smaller across — two and a half thousandths of a millimetre. It is a size class, not a substance — sulfates, nitrates, black carbon, dust and metals in varying proportions.
Size is the mechanism-relevant fact. WHO's position is that both PM10 and PM2.5 penetrate deep into the lungs, but PM2.5 can cross into the bloodstream — which is why its effects are cardiovascular and systemic rather than only respiratory. For the brain, two routes are proposed: directly along the olfactory nerve, and indirectly via the circulation across the blood-brain barrier, with the proposed damage being oxidative stress, neuroinflammation and neurovascular injury.
Worth noting: ultrafine particles and black carbon, the fractions most plausibly doing that work, are precisely the ones WHO says there is insufficient evidence to set guideline levels for.
What the cohort evidence actually shows
Three recent syntheses, and they do not agree as neatly as headlines suggest. The largest is a 2025 review in The Lancet Planetary Health: pooling 21 studies and 24 million people, it linked PM2.5 to dementia incidence at a hazard ratio of 1.08 (95% CI 1.02–1.14) per 5 µg/m³ — significant, modest, and carrying I² = 95%, near-total inconsistency between studies. NO2 came out at 1.03 (1.01–1.05) per 10 µg/m³ (I² = 84%) and black carbon at 1.13 (1.01–1.27) per 1 µg/m³ (I² = 97%); nitrogen oxides, PM10 and ozone showed no significant association. Study quality was mostly good, overall certainty of evidence moderate.
An earlier 2023 meta-analysis in the BMJ was less positive, and is the one most often misquoted. Across 14 poolable studies its estimate was 1.04 (0.99–1.09) per 2 µg/m³ PM2.5 — the confidence interval crosses 1.0. The higher figure sometimes lifted from it, 1.42, belongs to the seven studies that actively examined every participant, and its interval was 1.00–2.02 — touching the null exactly. The seven relying on medical records gave 1.03 (0.98–1.07). NO2 was 1.02 (0.98–1.06), not significant. The authors note that most included studies were at high risk of bias, and the figures "require interpretation with caution."
The third, in Nature Aging in 2025, graded evidence strength across 28 cohorts using the Burden of Proof framework. It found a significant non-linear relationship — a minimum 14% increase in dementia risk across the observed exposure range — and rated that evidence two stars out of five, meaning weak or inconsistent. Alzheimer's disease specifically scored three stars; vascular dementia was not significant. It also found evidence of publication bias, Egger's test p=0.03, with a funnel plot skewed toward positive results. Subgroup analyses gave one star for North American studies and no significant evidence in European ones. Read together: consistent in direction, inconsistent in size, with a documented lean toward positive findings. Not a settled effect size.
The confounding problem, stated honestly
This determines how much of the above to believe as causal. PM2.5 exposure is not randomly distributed: it tracks deprivation, traffic volume, traffic noise, occupation, housing quality and distance from green space, several of which are candidate dementia risk factors themselves. Good studies adjust for much of that. The Danish Nurse Cohort followed 25,233 nurses and found PM2.5 associated with dementia at 1.35 (1.15–1.59) per 2.6 µg/m³ after adjusting for lifestyle, socioeconomic status and road traffic noise — and found the noise association vanished once PM2.5 was controlled for (1.02, 0.93–1.11), which argues pollution is the operative exposure of the two.
But adjustment has a ceiling, and the Nature Aging authors say so directly: "it is not possible to account for all potential confounders." A residential address is a proxy for a lifetime of exposures, not a measurement of one. This is the trap we set out in our piece on inflammation and the brain: a plausible mechanism plus a consistent association is not a demonstrated cause.
The quasi-experimental evidence, which answers this better
Two studies partly escape that objection by using changes in exposure rather than levels. The stronger is an instrumental-variable analysis in the Review of Economic Studies in 2023. When the EPA began enforcing PM2.5 limits in 2004, cleanup pressure fell unevenly, producing quasi-random variation in what Medicare beneficiaries over 65 subsequently breathed. Across roughly a million people, a 1 µg/m³ increase in decadal PM2.5 raised the probability of a new dementia diagnosis by 2.15 percentage points (SE 0.846), with effects persisting below the regulatory threshold. The naive associational estimate in the same data was smaller, 0.63pp — here the causal design raised the estimate rather than dissolving it. Caveats: the outcome is a recorded diagnosis, and this is one design in one country.
A 2022 PNAS analysis of 2,239 older women found those whose local air improved most over the preceding decade had lower dementia risk (HR 0.80 per 1.78 µg/m³ of PM2.5 reduction, 95% CI 0.71–0.91) — alongside baseline PM2.5 itself at 1.16 (0.98–1.38), which did not reach significance. We are not reporting the positive half without the null half.
Guideline levels versus what people breathe
WHO's 2021 guidelines set PM2.5 at 5 µg/m³ annual mean and 15 µg/m³ over 24 hours (99th percentile), down from 10 and 25 in 2005; NO2 is 10 annual, PM10 15. Against that, WHO reports that in 2019 99% of the world's population lived where its guideline levels were not met. In Europe, where air is comparatively clean, the European Environment Agency found that in 2023 only 1.2% of monitoring stations breached the EU's legal annual limit while 92% exceeded the WHO annual guideline. The gap between legal and advisable is the story.
What an individual can actually do
Route choice, for the pollutants where it works. A randomised crossover trial in The Lancet had 119 older adults walk two hours on Oxford Street or in Hyde Park. The park walk improved lung function and reduced arterial stiffness for up to 26 hours; those gains were attenuated after the street. A synthesis of 41 roadside monitoring studies adds a nuance: ultrafine particles, carbon monoxide and NO2 fall off steeply with distance from the road, but PM2.5 mass showed no clear distance trend. A back-street route cuts traffic-specific exposure more than it cuts your PM2.5 number. Do not stop walking or cycling over this — see exercise against anything you could buy, and what green space does for attention.
Indoor sources, which you actually control. WHO's guidelines apply indoors as well as out. In the UK, domestic combustion produced 20% of national PM2.5 emissions in 2024, with burning wood indoors alone accounting for 12% — more than a stove's aesthetics are worth. Cooking and candles are smaller intermittent sources and gas hobs raise indoor NO2, so extraction and an open window while cooking are the obvious responses. Ventilating when the road outside is quiet rather than at rush hour follows the same logic, untested though it is.
Filtration — and be precise about what the trials measured. HEPA filtration reliably lowers indoor particle counts. In a randomised crossover trial in a Detroit senior residence, 40 older adults saw personal PM2.5 fall from 15.5 to 7.4 µg/m³ with true-HEPA units, alongside a 3.2 mmHg drop in systolic blood pressure (95% CI −6.1 to −0.2), though most secondary cardiovascular outcomes were not significantly improved. A 2024 meta-analysis of 15 filtration RCTs in 678 participants found a reduction in IL-6 (SMD −0.275, 95% CI −0.545 to −0.005, p=0.046) and no significant effect on TNF-α, CRP or any oxidative stress marker measured. So: particulate levels, yes. Short-term biomarkers, partly and inconsistently.
No filtration trial has measured cognition or dementia incidence. "Filters lower PM2.5" is true; "filters protect your brain" is untested. A blood-pressure effect is not nothing given how the vascular route runs, but it is the distinction we drew about hearing aids: a strong association does not mean treating the exposure changes the outcome. We are not recommending any device.
No supplement protects you from air pollution
This needs saying because the claim circulates, and because we sell an omega-3. The trials behind it are small and measure biomarkers: a JACC trial gave 65 healthy Shanghai students 2.5 g/day of fish oil and found benefit on 5 of 18 subclinical markers, and a B-vitamin study had ten participants in a chamber at 250 µg/m³. These are mechanistic probes, not protection — no cognitive outcomes, no clinical outcomes, and doses unrelated to a general-purpose capsule (ours provides 1,000mg of fish oil, 300mg combined EPA and DHA). Nothing we sell reduces the harm from air pollution, and we are not going to imply otherwise. If omega-3 is your actual question, read the dose-response evidence on its own terms. Marketing antioxidants as pollution protection sells a mechanism story, which is what does not clear our bar.
A last note on proportion: the Commission's own attributable estimate for air pollution came from a single epidemiological study, a detail the Nature Aging authors raise pointedly. The factors where individual action has the most evidence behind it are further up the same list. Air pollution is real, thinly evidenced at the individual level, and mostly a question for policy — which means the highest-leverage thing most people can do about it is not a purchase.
This article is information, not medical advice, and cannot assess anyone's individual risk. Nothing Proco sells treats or prevents dementia, or mitigates the effects of air pollution.
Sources: Best Rogowski CB, Bredell C, Shi Y, et al. Long-term air pollution exposure and incident dementia: a systematic review and meta-analysis. The Lancet Planetary Health. 2025;9:101266. doi:10.1016/S2542-5196(25)00118-4 · Wilker EH, Osman M, Weisskopf MG. Ambient air pollution and clinical dementia: systematic review and meta-analysis. BMJ. 2023;381:e071620. doi:10.1136/bmj-2022-071620 · Huang X, Steinmetz J, Marsh EK, et al. A systematic review with a Burden of Proof meta-analysis of health effects of long-term ambient fine particulate matter (PM2.5) exposure on dementia. Nature Aging. 2025;5:897-908. doi:10.1038/s43587-025-00844-y · Tuffier S, Zhang J, Bergmann M, et al. Long-term exposure to air pollution and road traffic noise and incidence of dementia in the Danish Nurse Cohort. Alzheimer's & Dementia. 2024;20:4080-4091. doi:10.1002/alz.13814 · Bishop KC, Ketcham JD, Kuminoff NV. Hazed and Confused: The Effect of Air Pollution on Dementia. Review of Economic Studies. 2023;90(5):2188-2214. doi:10.1093/restud/rdac078 · Wang X, Younan D, Millstein J, et al. Association of improved air quality with lower dementia risk in older women. PNAS. 2022;119:e2107833119. doi:10.1073/pnas.2107833119 · WHO global air quality guidelines: particulate matter (PM2.5 and PM10), ozone, nitrogen dioxide, sulfur dioxide and carbon monoxide. Executive summary. World Health Organization, 2021 · WHO fact sheet: Ambient (outdoor) air pollution, 24 October 2024 · European Environment Agency. Air quality status report 2025: particulate matter (PM2.5), 2023 data · Defra. Emissions of air pollutants in the UK — particulate matter (PM10 and PM2.5), 2024 data · Sinharay R, Gong J, Barratt B, et al. Respiratory and cardiovascular responses to walking down a traffic-polluted road compared with walking in a traffic-free area... a randomised, crossover study. Lancet. 2018;391:339-349. doi:10.1016/S0140-6736(17)32643-0 · Karner AA, Eisinger DS, Niemeier DA. Near-roadway air quality: synthesizing the findings from real-world data. Environmental Science & Technology. 2010;44:5334-5344. doi:10.1021/es100008x · Morishita M, Adar SD, D'Souza J, et al. Effect of portable air filtration systems on personal exposure to fine particulate matter and blood pressure among residents in a low-income senior facility: a randomized clinical trial. JAMA Internal Medicine. 2018;178:1350-1357. doi:10.1001/jamainternmed.2018.3308 · Ren J, Zhang Z, Cui Q, et al. The effect of indoor air filtration on biomarkers of inflammation and oxidative stress: a review and meta-analysis of randomized controlled trials. Environmental Science and Pollution Research. 2024;31:33212-33222. doi:10.1007/s11356-024-33414-7 · Lin Z, Chen R, Jiang Y, et al. Cardiovascular benefits of fish-oil supplementation against fine particulate air pollution in China. JACC. 2019;73:2076-2085. doi:10.1016/j.jacc.2018.12.093 · Zhong J, Trevisi L, Urch B, et al. B-vitamin supplementation mitigates effects of fine particles on cardiac autonomic dysfunction and inflammation: a pilot human intervention trial. Scientific Reports. 2017;7:45322. doi:10.1038/srep45322 · Livingston G, et al. Dementia prevention, intervention, and care: 2024 report of the Lancet standing Commission. Lancet. 2024;404:572-628. doi:10.1016/S0140-6736(24)01296-0