← Maths & Science

Air and Water Purification

The Thames below Westminster, Claude Monet
The Thames below Westminster, Claude Monet

Do air purifiers and water filters actually make you healthier? This page covers what the trials show for each, why the answer depends on how dirty the starting air or water is, the physics and arithmetic behind the marketing numbers, the gadgets that do nothing or harm, and the practices that are worth the money.

The answer in one minute

  • Water treatment is one of the best-proven health measures in history, where the water is contaminated. Filtration and chlorination of city water helped end typhoid and cholera in the rich world. Treating household water in poor countries still cuts child diarrhoea by a third to a half and, on the best estimate, child deaths by about a quarter.
  • On top of safe, regulated tap water, a home filter has no demonstrated general health benefit. Blinded trials found no fewer stomach illnesses. The exceptions are specific and real: lead pipes, PFAS (“forever chemicals”), private wells and severely weakened immune systems.
  • HEPA air purifiers reliably remove fine particles, which are a proven cause of disease. The health gains measured so far are real but modest. Short trials show a few points off blood pressure and fewer flare-ups of lung disease. No trial has been long or large enough to show fewer heart attacks or deaths. That benefit is inferred, plausibly, from the epidemiology of air pollution.
  • Much of the market is noise. Ionisers and ozone generators can make the air worse. Houseplants don't clean measurably. Alkaline, “structured” and bottled waters offer nothing that safe tap water lacks.

One principle governs both

The health value of purification is a product of two numbers: how much of a harmful exposure the device removes times how much harm each unit of that exposure does. Both terms shrink as the air or water gets cleaner, so the returns diminish steeply. Taking sewage out of drinking water saves lives by the hundreds of thousands. Taking the last traces out of water that already meets European standards may save none. The same purifier is a serious health tool in a Delhi winter or a wildfire season, and a marginal one in a ventilated house in clean countryside.

The claim “this device is good for your health” is also really two claims chained together: (1) the device makes the air or water cleaner, and (2) cleaner air or water, at that margin, improves health. Link (1) is usually easy to test and often true. Link (2) is hard to test and is where most marketing quietly cheats. It shows (1) and lets you assume (2). Keeping the two links apart is the whole skill of reading this literature.

Air: what is in it, and why it matters

Air pollution is a large health risk. The Health Effects Institute and IHME's State of Global Air 2024 attributes 8.1 million deaths in 2021 to it, making it the world's second-largest risk factor for death after high blood pressure. Most of that burden comes from PM2.5, particles smaller than 2.5 micrometres. They reach the deepest parts of the lung, and the finest of them cross into the bloodstream. Regulators treat the link between long-term PM2.5 exposure and cardiovascular and respiratory death as causal. In the large cohort studies, each extra 10 µg/m³ of long-term exposure goes with roughly 6–8% higher all-cause mortality. The meta-analysis behind the WHO's 2021 guidelines put it at 8%.

Indoors matters because that is where people are, about 90% of the time in rich countries. Indoor air is a mix of outdoor particles that leak in and pollution made inside: cooking (especially frying, and gas hobs, which also emit nitrogen dioxide), candles and incense, wood stoves, smoking, cleaning products and new furniture (VOCs, formaldehyde), mould, pet dander, pollen and respiratory viruses. A purifier deals with only part of this list. A particle filter does nothing for gases.

How a purifier works, and the arithmetic of sizing it

A HEPA filter is not a sieve. It is a mat of fibres that catches particles three ways. Large particles cannot follow the air as it bends around a fibre and slam into it (impaction). Mid-sized ones brush against fibres as they pass (interception). The smallest are jostled by air molecules into a random walk that sooner or later touches a fibre (diffusion). Diffusion weakens as size grows and impaction strengthens, so there is a most penetrating particle size in between, around 0.1–0.3 µm. That is why the standards are defined there: a US “true HEPA” filter must stop 99.97% of 0.3 µm particles, and the European H13 grade 99.95% at the most penetrating size. Counter-intuitively, a HEPA filter catches virus-sized particles better than 0.3 µm ones. “HEPA-type” and “HEPA-like” are marketing terms with no standard behind them.

The filter's efficiency is rarely the bottleneck. The bottleneck is how much of the room's air passes through it. That is the clean air delivery rate (CADR): the airflow multiplied by the efficiency, i.e. the volume of perfectly cleaned air the unit delivers per hour. Model a room as a well-mixed box of volume V. Outdoor air at concentration Cout leaks in at a air changes per hour, of which a fraction P of particles survives the building shell. Indoor sources emit S, particles settle onto surfaces at rate k, and the purifier removes CADR·C. Setting inflow equal to outflow gives the steady state:

C_indoor  =  ( a·P·C_out  +  S/V )  /  ( a  +  k  +  CADR/V )

With / without purifier  =  ( a + k )  /  ( a + k + CADR/V )

Take a 12 m² bedroom with a 2.5 m ceiling (30 m³), a typical home air-exchange rate a = 0.5 per hour, settling k = 0.2 per hour, and a purifier with a CADR of 150 m³/h, so CADR/V = 5 per hour. The ratio is 0.7 / 5.7 ≈ 0.12. In theory that is an 88% cut in fine particles. The standard sizing rule follows from the same formula. In the US it is written as “CADR in cubic feet per minute at least two-thirds of the floor area in square feet”. For an ordinary 8-foot ceiling that works out to exactly 5 air changes per hour. In metric: CADR (m³/h) ≥ 5 × room volume (m³).

Real homes in trials land nearer 30–60% than 88%. Units get switched off because they are noisy, doors stay open, air is not perfectly mixed, the rated CADR is measured on a fresh filter at top speed, and people spend their day in more than one room. The model says where the leverage is: a bigger CADR relative to the room, closed doors, and continuous running.

Air: what the trials actually show

Exposure (link 1): strong. Randomised and sham-controlled crossover trials consistently show portable HEPA units cutting indoor PM2.5 by about half. Examples are a 2015 double-blind trial in Shanghai student dormitories and a 2018 trial among low-income seniors in Detroit. Measured personal exposure falls less, because people leave the room.

Cardiovascular markers: consistent but small. Pooled across sham-controlled trials, filtration lowers systolic blood pressure by roughly 2–4 mmHg. One 2020 meta-analysis found −3.9 mmHg, a 2023 update −2.4 mmHg, rising to −3.6 mmHg when weaker studies are excluded. The effect on diastolic pressure is not significant. Trials also report better markers of inflammation and blood clotting. These trials last days to weeks. A few mmHg is trivial for one person and meaningful across a population, but nobody has yet shown that the effect persists for years.

Lung disease: encouraging, mixed. The best trial is CLEAN AIR (2022), which gave 116 former smokers with COPD real or sham HEPA units for six months. It missed its primary endpoint, a quality-of-life score. But moderate flare-ups were about two-thirds lower (rate ratio 0.32), rescue-inhaler use was about half (0.54), and participants who actually ran their units improved on the main score too. Severe exacerbations needing hospital care did not differ. A larger multicentre trial is now planned. In asthma, trials in children exposed to tobacco smoke found more symptom-free days. Allergy is a matter of particle weight. Dust-mite allergen is carried on heavy particles that settle within minutes, so an air purifier barely touches it, and a Cochrane review of dust-mite control measures found no benefit for asthma. Pet dander and pollen stay airborne longer, so filtration helps more.

Infections: plausible, not proven. HEPA filters demonstrably remove viral aerosols from air. But a 2025 scoping review in Annals of Internal Medicine covered 672 studies of air cleaning against airborne infection and found that fewer than one in ten measured anything in people. Most were chamber tests counting particles or harmless organisms. Whether a purifier in a home or classroom reduces actual infections is still largely untested, and the review found the by-products of ionising and plasma devices barely examined.

Hard outcomes: no direct evidence. No randomised trial has measured heart attacks, strokes or deaths. Such a trial would need many thousands of people over years, and nobody has funded one. So the mortality case for a home purifier rests on a chain: purifiers lower exposure (proven), and long-term exposure causes death (proven at population level). Concluding that purifiers save lives is a sound inference, but it is an inference, not a measurement.

A back-of-envelope: what might it be worth?

Suppose your exposure averages 15 µg/m³, typical of a European city, and a bedroom purifier halves it during 8 hours of sleep. Your time-weighted exposure falls by 0.5 × 15 × 8/24 = 2.5 µg/m³. Assume the cohort slope (8% per 10 µg/m³) applies to particles removed by a filter. That is a big assumption: the cohorts measured outdoor air, and particles from indoor sources may differ in toxicity. Under it, your mortality rate falls by about 2% at every age. Adult mortality follows the Gompertz law, μ(x) = Aebx, with b ≈ 0.09 per year (risk of death doubles about every 7.5 years). Multiplying the whole curve by (1 − δ) is the same as being younger by −ln(1 − δ)/b years:

Δ life expectancy  ≈  −ln(1 − δ) / b  =  −ln(0.98) / 0.09  ≈  0.22 years  ≈  2–3 months

That is two to three months of life expectancy from decades of nightly use: real, but not dramatic. Where air is several times dirtier the gain is several times larger, though the dose–response curve flattens at very high concentrations. For someone with heart or lung disease, the near-term benefits (fewer flare-ups) probably matter more than the long-run average. Treat the number as an order of magnitude, not a promise.

Air: what doesn't work, or harms

  • Ozone generators and ionisers. Ozone is itself a regulated lung irritant. A machine that “purifies” with it produces the pollutant it claims to fight. Indoors, ozone also reacts with scented products to form formaldehyde and ultrafine particles. California has capped the ozone output of air cleaners sold there at 0.050 ppm since 2010. Ionisers mostly charge particles so they stick to walls and floors instead of being captured, and some emit ozone too. Turn off any “ion” or “plasma” mode on a purifier.
  • Houseplants. The idea comes from a 1989 NASA chamber experiment. A 2019 review of the data found that matching the cleaning of ordinary building ventilation would take on the order of 100 to 1,000 potted plants per square metre of floor.
  • Salt lamps, “air-purifying” sprays and candles. There is no evidence for the lamps, and sprays and candles add VOCs and particles.
  • Consumer UV boxes and photocatalytic units. Upper-room ultraviolet germicidal lighting has a genuine track record in tuberculosis wards. Small consumer UV units give the air too little exposure time to matter. Some photocatalytic (PCO) units generate formaldehyde and acetaldehyde as by-products.

Water: one of the best-proven health interventions in history

John Snow traced London's 1854 cholera outbreak to the Broad Street pump, and in 1908 Jersey City began the first continuous chlorination of a US public supply. Within a generation, waterborne typhoid and cholera had all but disappeared from rich-world cities. How much of the period's mortality collapse water deserves is a live and instructive dispute. Cutler and Miller (2005) credited filtration and chlorination with nearly half of the decline in total mortality in major US cities from 1900 to 1936, three-quarters of the infant decline and two-thirds of the child decline. Anderson, Charles and Rees (2022) rebuilt the data and found a much smaller effect: filtration associated with an 11–12% fall in infant mortality, and no detectable effect from the other interventions. Cutler and Miller replied that after correcting data errors, filtration still explains 38% of the total decline. The size is contested; the sign is not.

The problem is far from solved worldwide. The WHO/UNICEF monitoring programme's 2025 report counts 2.1 billion people without safely managed drinking water, 106 million of them still drinking straight from rivers, lakes and ponds. The WHO estimates that safe water, sanitation and hygiene together could have prevented at least 1.4 million deaths in 2019, 395,000 of them children under five.

Water where it is unsafe: what the trials show

The largest meta-analysis of household water treatment in low- and middle-income countries (Wolf et al., The Lancet, 2022, 124 studies) estimates that childhood diarrhoea risk falls by 50% with filters (risk ratio 0.50), 37% with solar disinfection (0.63) and 34% with chlorination (0.66).

There is a well-known catch. Most of these trials were open: families knew they had the filter and reported their own children's diarrhoea. In the few trials that used a placebo, the effect largely vanished. A double-blind chlorination trial in Odisha, India, found diarrhoea prevalence of 1.69% with real chlorine and 1.74% with placebo. Researchers have argued for years over whether open trials overstate the effect through reporting bias, or whether blinded trials understate it because they were run in specific settings.

Death is the outcome nobody misreports, which is why a 2023 analysis by Kremer, Luby and colleagues matters. Pooling the child-mortality data recorded in 18 randomised water-treatment trials, they estimate about a quarter fewer deaths of children under five (odds ratio 0.76; 0.73 in a Bayesian model). An earlier draft's “30%” is the figure most often quoted. The cost per disability-adjusted life-year averted is in the tens of dollars, which puts chlorinating water among the cheapest health gains known, and charity evaluators have funded chlorine-dispenser programmes on that basis. There is a caveat: mortality was a secondary outcome in those trials, and critics question how much the result depends on analysts' choices. Large trials in Bangladesh, Kenya and Zimbabwe (WASH Benefits and SHINE) that combined water treatment with sanitation and handwashing found no effect on children's growth. Faecal germs travel by hands, food and flies as well as water, and treatment only works if families keep using it.

Water where it is already safe: the case for home filters

Regulated tap water in Europe and North America is among the most heavily tested things people consume. The question for most readers here is whether a filter on top of it helps, and that has been tested properly. In Melbourne, 600 families received either a real or a sham kitchen treatment unit. Their rates of stomach illness were identical (ratio 0.99). A sham-controlled trial in Davenport, Iowa found the same (0.98). A third, among older adults in Sonoma County, California, found a modest reduction of roughly 12–15%, so a small effect in some systems cannot be excluded. Overall, for germs, good tap water leaves little for a filter to do.

The cases where home treatment clearly earns its place are chemical and local:

  • Lead. There is no known safe level of lead, especially for children's brains. It leaches from old lead service pipes and plumbing, and these are still common in older buildings. The EU limit has been 10 µg/L since 2013 and tightens to 5 µg/L by 2036. In a building with old pipes, a filter certified for lead reduction (NSF/ANSI 53 or equivalent) is a sensible purchase. Run the tap after water has stood for hours, and use cold water for drinking and cooking.
  • PFAS (“forever chemicals”). These are linked to immune, thyroid, developmental and some cancer effects at very low doses. Since January 2026 EU drinking water must keep the sum of 20 PFAS below 0.1 µg/L. The US set a 4 parts-per-trillion limit for PFOA and PFOS in 2024. A May 2026 proposal would push compliance to 2031 and drop the limits for four other PFAS; it was not final at the time of writing. A 2020 Duke–NC State field study found that under-sink reverse-osmosis and two-stage filters removed PFAS almost completely, while pitcher filters removed only about half, variably.
  • Private wells. They are unregulated and can carry nitrate (dangerous for infants' blood oxygen), arsenic, uranium or bacteria. Test them, then treat what the test finds.
  • Severely weakened immunity (advanced HIV, transplants, some chemotherapy). Cryptosporidium resists chlorine. The US CDC advises boiling water or using a filter rated to 1 µm absolute or certified for cyst removal.

The arithmetic of “removes 99.9%”

Water engineers count in logs: the log reduction value is LRV = log₁₀(Cin / Cout). So 2-log means 99%, 3-log 99.9% and 4-log 99.99%. Two consequences follow.

LRV = log₁₀(C_in / C_out)      99.9% = 3-log      99.9999% = 6-log

Independent barriers in series:  LRV_total = LRV₁ + LRV₂ + …

First, a percentage is meaningless without the starting load. A 3-log filter fed water carrying a million bacteria per litre still delivers a thousand per litre. Second, because the surviving fractions multiply, independent barriers add in log space. That is why treatment plants stack coagulation, filtration and disinfection rather than trusting one perfect step, and why US rules demand 3-log removal of Giardia and 4-log of viruses for surface water. Portable “purifiers” for travel are held to a similar test: 6-log for bacteria, 4-log for viruses and 3-log for cysts. Most household pitcher filters claim none of this, because they are designed for taste and chemicals, not germs.

Water: trade-offs, side effects and myths

  • Neglected filters grow bacteria. Activated carbon strips out the chlorine that keeps tap water sterile, then gives bacteria a damp, food-rich surface. Old cartridges routinely put out water with more bacteria than the tap feeding them. These are mostly harmless species, but it defeats the purpose. Change cartridges on schedule.
  • Reverse osmosis strips everything. That includes calcium, magnesium and fluoride. The WHO has flagged that demineralised water is not ideal as a sole long-term supply. Observational studies tie magnesium-rich hard water to slightly lower cardiovascular mortality; the evidence is weak, but it is a reason not to strip water without a reason. Many units also send several litres down the drain per litre produced, though modern ones waste less.
  • Fluoride is now contested. A 2024 Cochrane update found the tooth-decay benefit of water fluoridation much smaller in the fluoride-toothpaste era than in older studies. A 2024 US National Toxicology Program review concluded with moderate confidence that fluoride above 1.5 mg/L, more than twice the level used in fluoridation, is associated with lower IQ in children. It found the evidence at normal fluoridation levels insufficient. Parents of young children on RO water may want to ask a dentist about fluoride.
  • Chlorination by-products are a trade worth making. Chlorine reacts with organic matter to form trihalomethanes and other compounds. Decades of exposure are associated in some studies with a modestly higher bladder-cancer risk, which is why they are regulated and why carbon filters, which remove them, have a defensible use. The WHO's rule is blunt: never compromise disinfection to reduce by-products, because the infection risk is far larger.
  • Microplastics: the exposure is real, the harm unknown. A 2024 Columbia study counted roughly a quarter of a million plastic particles per litre of bottled water, most of them nanoplastics. A 2024 NEJM study found that patients with microplastics in their artery plaque had higher rates of heart attack, stroke and death. It was observational, and confounding is plausible. Boiling hard tap water and pouring it through a coffee filter removed up to ~90% of spiked particles in a 2024 lab study, because the scale that forms traps them. In soft water the figure was only ~25%. Reverse osmosis removes most of them. Switching from tap to bottled water makes this exposure worse, not better.
  • Alkaline, “hydrogen” and “structured” water. Stomach acid has a pH of roughly 1.5–3.5 and the body holds blood pH within a hair of 7.4 regardless of what you drink. There is no credible evidence for any of these products.

Evidence scorecard

InterventionWhat it demonstrably doesHealth evidenceVerdict
Portable HEPA air purifierCuts indoor fine particles (PM2.5) by roughly 30–60% in real homes.Short trials: systolic blood pressure about 2–4 mmHg lower, better inflammation markers, fewer moderate COPD flare-ups. No trial yet of heart attacks or deaths.Proven to clean the air; modest, plausible health benefit
Ventilation and a cooker hood vented outdoorsRemoves what filters can't (CO₂, NO₂ from gas hobs, moisture, many VOCs) and dilutes the rest.Strong indirect evidence: gas cooking is linked to roughly 30–40% higher odds of childhood asthma.First line, before any gadget
Activated-carbon layerAdsorbs odours and some VOCs until it saturates. The thin layers in most purifiers saturate fast.None demonstrated.Nice to have
Ionisers, “plasma”, ozone generators, some PCO unitsLittle true particle removal. Can emit ozone, formaldehyde and other by-products.Ozone is a proven lung irritant.Avoid
HouseplantsYou would need on the order of 100–1,000 plants per m² of floor to match ordinary ventilation.None from air cleaning.Decorative only
Municipal filtration + chlorinationRemoves and kills faecal pathogens at city scale.Credited with between ~11% of the infant-mortality decline and ~40% of the total mortality decline in early-1900s US cities (estimates disputed).One of the best-proven health measures ever
Household treatment where water is unsafe (chlorine, filters, sunlight)Kills or removes pathogens at the point of use.Childhood diarrhoea down 34–50% in open trials (less in blinded ones). Under-five mortality down roughly a quarter in a meta-analysis.Proven, and extraordinarily cheap per life-year
Home filter on already-safe tap waterImproves taste. Certified models remove lead, some PFAS and chlorination by-products.Sham-controlled trials in Melbourne and Iowa found no fewer stomach illnesses; one in older Californians found a small reduction.No general benefit; real benefit for lead, PFAS and private wells
Reverse osmosisRemoves nearly everything: PFAS, nitrate, arsenic, but also fluoride and minerals.Benefit only where those contaminants are present.Right tool for a specific problem; wasteful otherwise
Bottled, alkaline, “hydrogen” or “structured” waterBottled water is not safer than regulated tap water and carries more plastic particles. Stomach acid erases any alkalinity.None.Marketing

Best practices: air

  1. Remove sources first. No smoking indoors. Use a cooker hood that vents outside whenever you cook, and always with a gas hob. Go easy on candles, incense and scented sprays. Keep a wood stove well maintained, or don't use one.
  2. Ventilate when outdoor air is clean. Opening windows dilutes the gases no filter catches. A cheap CO₂ monitor is a good proxy for stale air: outdoor air is about 420 ppm, and sustained readings above about 1,000 ppm mean a room needs more air.
  3. Filter where you spend the most time, which is usually the bedroom. Buy true HEPA (H13 or better, or “99.97% at 0.3 µm”) with a CADR of at least 5 × the room volume, or 2/3 of the floor area in ft² if the label is in cfm. Choose one large enough that its quiet setting does the job. A purifier that is off does nothing.
  4. Run it continuously with the door closed, and replace filters on schedule. Leave the ioniser off. Prefer models certified as ozone-free.
  5. With central air or heating ducts, fit a MERV 13 filter and run the fan. A do-it-yourself Corsi–Rosenthal box (four MERV 13 filters taped to a box fan) delivers clean air at a fraction of the cost of a commercial unit.
  6. Prioritise when it matters most: wildfire smoke, pollution episodes, homes near busy roads, and people with asthma, COPD or heart disease, as well as infants, pregnant women and the elderly.

Best practices: water

  1. Know your water before you buy anything. Read your utility's annual quality report, which is public in the EU and US. Find out whether your building has lead pipes, which is likely if it predates the mid-twentieth century and hasn't been re-plumbed. Test a private well every year.
  2. Match the device to the problem. For lead, use a filter certified for lead. For PFAS, use reverse osmosis or a two-stage under-sink carbon filter. For germs (travel, emergencies, boil notices), bring water to a rolling boil for 1 minute (3 minutes above 2,000 m), or use a purifier rated for bacteria, viruses and cysts. For taste alone, a pitcher is fine.
  3. Look for independent certification of the specific contaminant (NSF/ANSI 53 for health contaminants such as lead, 58 for reverse osmosis, 401 for emerging contaminants), not a generic “removes 99%”.
  4. Change cartridges on time, and clean pitchers and reservoirs. A spent filter can be worse than none.
  5. Don't replace safe tap water with bottled water for health. It costs hundreds of times more, is not safer, and carries more plastic particles.
  6. If you install reverse osmosis, keep calcium and magnesium in your diet and ask a dentist about fluoride for young children. Consider a remineralising stage.

The short version

Purification pays in proportion to the dirt it removes. Treating contaminated water is among the most powerful and cheapest health interventions known, both historically and today. Filtering water that already meets modern standards buys taste and peace of mind, plus real protection only against specific local problems: lead, PFAS, wells. Air is the reverse case. Fine particles are a large, proven cause of death even in rich countries, and HEPA purifiers measurably remove them. But the trials so far show modest short-term gains and no direct evidence on lifespan, so the long-run benefit rests on a sound but unmeasured inference. In both cases the best moves are unglamorous: remove the source, ventilate, match the device to a known problem, size it properly, and maintain it.

Sources

  • Health Effects Institute & IHME, State of Global Air 2024 — 8.1 million deaths attributable to air pollution in 2021. — stateofglobalair.org
  • Walzer et al., Hypertension, 2020, and Faridi et al., 2023 — meta-analyses of air-filtration trials and blood pressure (systolic −3.9 and −2.4 mmHg).
  • Hansel et al., “The CLEAN AIR study”, American Journal of Respiratory and Critical Care Medicine, 2022 — six-month sham-controlled HEPA trial in COPD.
  • Baduashvili et al., “Engineering infection controls to reduce indoor transmission of respiratory infections: a scoping review”, Annals of Internal Medicine, 2025. — pubmed
  • Cummings & Waring, “Potted plants do not improve indoor air quality”, Journal of Exposure Science & Environmental Epidemiology, 2019. — doi
  • WHO/UNICEF Joint Monitoring Programme, Progress on household drinking water, sanitation and hygiene 2000–2024 (2025). — data.unicef.org
  • Cutler & Miller, “The role of public health improvements in health advances”, Demography, 2005; Anderson, Charles & Rees, American Economic Journal: Applied Economics, 2022, and Cutler & Miller's comment in the same issue.
  • Wolf et al., “Effectiveness of interventions to improve drinking water, sanitation, and handwashing with soap on risk of diarrhoeal disease in children in low-income and middle-income settings”, The Lancet, 2022. — pubmed
  • Kremer, Luby, Maertens, Tan & Więcek, “Water treatment and child mortality: a meta-analysis and cost-effectiveness analysis”, NBER Working Paper 30835, 2023. — nber.org
  • Boisson et al., double-blind chlorination trial in Odisha, India, PLoS ONE, 2013; Schmidt & Cairncross, Environmental Science & Technology, 2009, on bias in unblinded water trials.
  • Hellard et al., “A randomized, blinded, controlled trial investigating the gastrointestinal health effects of drinking water quality”, Environmental Health Perspectives, 2001. — pmc
  • Colford et al., sham-device trials in Davenport, Iowa (American Journal of Epidemiology, 2005) and Sonoma County, California (American Journal of Public Health, 2009).
  • Herkert et al., “Assessing the effectiveness of point-of-use residential drinking water filters for PFASs”, Environmental Science & Technology Letters, 2020. — doi
  • Yu, Li, Zeng et al., boiling and filtering hard tap water to remove nano- and microplastics, Environmental Science & Technology Letters, 2024; Qian et al., nanoplastics in bottled water, PNAS, 2024; Marfella et al., microplastics in carotid plaque, NEJM, 2024.
How to read this page. An editorial science summary for orientation, not medical advice. If you have asthma, COPD, heart disease or a weakened immune system, or you are worried about your water supply, ask a doctor and your water utility or local health authority about your specific situation.