math scores after Los Angeles installed 1,800 filters in 18 schools, at $30–50 per student
Selected research
What the literature actually says.
Randomized trials, natural experiments, and cohort studies across many countries point the same direction: cleaner indoor air improves learning, attendance, and health. Filter by theme.
student absence in a randomized controlled trial, at roughly $12 per absence day avoided
COVID-19 incidence in Georgia schools that improved both ventilation and filtration
benefit-cost ratio for classroom air cleaning in the same randomized trial
-
A randomized controlled trial found classroom air purifiers cut student absenteeism 12.5%, with even larger effects for students who started with the highest absence rates. It worked out to roughly $12 per absence day avoided, a benefit-cost ratio of 9 to 1.
-
After a nearby gas leak, Los Angeles Unified installed 1,800 air filters across 18 schools in 2016. Comparing them with demographically similar schools, math scores were 0.20 standard deviations higher and English scores 0.18 higher, at an upfront cost of $30–50 per student. The effects persisted and grew over time.
Gilraine·Comparable filtration now costs as little as $1–5 per child per year
-
Higher daily air pollution significantly increases student absences, disciplinary referrals, and teacher absences, with Black students, Hispanic students, and those from lower-income households bearing the largest impacts.
-
Israeli students performed worse on high-stakes exams when air pollution spiked on the day of the test.
-
The same test-day pattern shows up in China.
-
Students attending schools downwind of major highways have meaningfully lower test scores, more behavioral incidents, and higher absenteeism than the same students attending otherwise similar upwind schools.
-
Improving ventilation and removing mold in schools produced gains in English and math scores as well as pass rates.
-
Just two hours of breathing filtered air improved executive function scores.
-
Across 156 elementary schools in Salt Lake County, each additional peak pollution day corresponded to an additional 2% of students testing below grade level in math — even after controlling for school disadvantage.
International Journal of Environmental Research and Public Health
-
When Swedish daycare centers installed air cleaners, illness-related absences dropped 55%.
-
A randomized trial in Lahore, Pakistan found purifiers initially cut PM2.5 by 25% and raised Grade 2 test scores 0.15 standard deviations after three weeks — but the gains disappeared once the purifiers stopped cleaning effectively. Maintenance is part of the intervention.
-
A double-blind trial among primary-school children in China found that purifying classrooms and bedrooms nearly halved personal PM2.5 exposure and improved Chinese-language and total academic scores by 0.22 and 0.20 standard deviations.
-
Across rural India, each roughly 1 µg/m³ increase in PM2.5 reduced reading outcomes by 1.11–2.39 percentage points and math outcomes by 0.53–1.90 — primarily by reducing school attendance.
-
A natural experiment across Tehran’s junior schools found academic performance at the most traffic-exposed schools fell 6.03 percentage points, about 0.51 standard deviations, relative to the least exposed.
-
Across nearly three million Brazilian university-entrance-exam observations, a 10 µg/m³ increase in exam-day PM10 reduced scores by 6.1 points, about 0.08 standard deviations.
Journal of the Association of Environmental and Resource Economists
-
Drops in air quality have been found to increase daily absence rates.
-
Children attending highly polluted schools showed 11% slower working-memory growth over a single year compared with peers in cleaner environments.
-
Both short-term classroom air pollution and long-term fine particulate exposure impair cognitive performance, including attention and visual processing speed. Harvard’s CogFx work found the same pattern in adults: slower response times and poorer cognitive function in worse air.
-
A systematic review and meta-analysis across nearly 5,000 children in North America, Europe, and Asia found each unit increase in PM2.5 was associated with about a quarter-point decline in IQ — broadly similar to what we see with lead.
-
Exposure to traffic-related pollutants reduces cognitive development and attention in school-aged children.
-
Neuroimaging links chronic exposure to fine and ultrafine particles with white-matter damage and cortical thinning in children. Nanoparticles have been found in brain tissue itself, supporting a plausible biological pathway for neuroinflammation.
-
PM2.5 at around 7.6 µg/m³ — below the current EPA standard, and a level millions of American children experience regularly — was associated with measurable differences in cortical surface area and in the volumes of brain structures critical for executive function, planning, and impulse control.
-
A 2025 NBER study found a 10 µg/m³ increase in PM2.5 causes a 28% increase in office discipline referrals, driven by violent behavior, defiance, and interpersonal conflict. Black students experience a 32% increase in violent-behavior referrals, and effects are strongest in low-income schools.
-
Schools with higher concentrations of PM2.5, NO₂, and CO have significantly more violent incidents and weapon-related referrals — even at pollution levels below EPA regulatory limits.
-
A 2024 meta-analysis found significant associations between PM2.5, PM10, and NO₂ exposure and increased ADHD risk in children, with postnatal exposure particularly implicated.
-
A longitudinal study found childhood air pollution exposure linked to poor mental health at age 18, including depression, anxiety, conduct disorder, and thought disorders — spanning all forms of psychiatric distress.
-
Prenatal influenza exposure has been associated with higher schizophrenia risk in offspring, with observational estimates of roughly 2.2 to 3 times in historical cohorts.
-
In a high-risk birth cohort, wheezing illness with human rhinovirus between birth and age three was associated with 9.8 times higher asthma risk by age six.
-
Filtering the air improves asthma control in children exposed to airborne particles from traffic.
-
Around 4 million new childhood asthma cases each year are attributable to traffic emissions.
-
Greater exposure to particulate matter and nitrogen oxides leads to significantly slower lung-function growth — reductions of about 5–11% that persist into adulthood and raise lifetime risk of chronic respiratory disease.
-
Exposure starts affecting health before birth. Meta-analyses show consistent associations between prenatal fine particulate exposure and reduced birth weight, as well as elevated preterm-delivery risk.
-
Early-life pollution exposure permanently reduces lifetime earnings: lower ambient pollution in the year of birth measurably raises earnings 30 years later.
Isen, Rossin-Slater & Walker·SSRN
-
Improvements in prenatal air quality have intergenerational effects on human capital. Prenatal pollution exposure can explain approximately 26% of the Black–white earnings gap for 1980s birth cohorts.
-
Severe prenatal and early-postnatal respiratory infections may have long-run consequences, with studies linking in-utero influenza exposure and early-life respiratory infection to lower adult educational attainment and earnings.
-
Measles causes a well-documented immune amnesia, erasing existing antibody memory for up to three years and increasing susceptibility to other pathogens.
-
Long Covid appears to have overtaken asthma as the most common chronic disease in children.
-
Even for otherwise healthy people, routine respiratory infection is a large recurring burden: estimates suggest we spend roughly 15–25 days a year, about 5% of life, sick with colds and influenza.
-
The global scale is enormous: 12.8 billion upper respiratory infection episodes worldwide in 2021, and more than 65 million serious lower-respiratory episodes a year attributed to viruses such as influenza and RSV. Respiratory infections account for around 7% of U.S. deaths from major causes.
-
Respiratory infection can be a short-term trigger for serious adult disease. Laboratory-confirmed influenza was associated with a 6.1 times higher risk of heart attack in the following seven days.
-
Severe respiratory infection may also be associated with later neurodegenerative disease: severe influenza was linked to 4.5–5 times dementia risk, and severe influenza and pneumonia to 2.6–4.1 times Alzheimer’s risk, in observational work.
-
Routine respiratory illness imposes a persistent economic burden even in non-pandemic years: an estimated 1–1.5% of annual productivity, roughly $600 billion globally or about 0.6% of global GDP, through absenteeism and presenteeism.
-
Across 169 Georgia elementary schools, COVID-19 incidence was 35% lower where ventilation was improved and 48% lower where both ventilation and filtration were improved.
-
A systematic review of 32 studies found HEPA filtration linked to reduced infection incidence (relative risk 0.4) in cohort studies — while noting strong publication bias and no consistent effect on symptom severity. The honest reading is promising but not settled.
-
Increasing ventilation or running a HEPA cleaner reduced airborne virus concentrations — notable because the study used real infected people rather than aerosolized surrogates.
-
A retrospective cohort study of more than 10,000 Italian classrooms found mechanical ventilation reduced COVID-19 incidence 68–80%, with stronger effects at higher ventilation rates and during high-transmission periods.
-
A Swiss school crossover study found portable HEPA filters cut classroom particle concentrations 77% and respiratory-infection absences from 22 to 13 — though the sample was too small to reach statistical significance.
-
A recent modeling study found indoor air quality interventions could meaningfully reduce population-level transmission: at 20% coverage and 80% efficacy, annualized incidence fell 7.2% for a SARS-CoV-2-like virus and 17.0% for an influenza-like virus. Targeting the highest-risk locations beat deploying at random.
-
Ventilation, filtration, and germicidal UV are proven for reducing airborne pathogens. Implementation barriers, not knowledge gaps, remain the primary obstacle.
-
Brief close-range contact does raise transmission risk, but the dominant driver of classroom spread is prolonged exposure to shared air. Cumulative time in a room with elevated CO₂ and multiple infectious classmates contributes far more than short-distance interactions.
-
In the homes of COVID-19 patients, viral RNA was detected in the air before but not after use of an appropriately sized portable air cleaner.
-
Adding HEPA portable air cleaners to hospital rooms was associated with a lower rate of air samples testing positive for SARS-CoV-2 RNA.
-
Vaccinating about 80% of children aged 3–15 in rural communities cut influenza infections among unvaccinated neighbours by roughly 61% — direct evidence that protecting children protects everyone else.
-
School-aged children experience earlier and higher infection peaks than adults during influenza epidemics, suggesting they play a central role in driving community transmission.
-
Japan’s national school-based influenza vaccination program substantially reduced excess mortality among older adults, highlighting the outsized epidemiological influence of children.
-
Surveillance via smart thermometers suggests children were key contributors to within-household transmission during COVID-19, with infection in children often preceding adult cases.
-
Analysis of U.S. state-level data found statewide school closures were associated with significant reductions in COVID-19 incidence and mortality — consistent with the role of schools in shaping respiratory disease spread.
No studies in that category yet.
This is a working library rather than a systematic review, and we try to include the inconvenient findings alongside the encouraging ones — null results, publication bias, and the trials where benefits disappeared once equipment stopped being maintained. If you think we have mischaracterised something or missed something important, please tell us.
Why this matters
Health, learning, and pandemic resilience — one intervention, three problems.
The case for clean airWhat we’re doing about it
A portfolio built to move from proof to policy to scale.
Our programsFound a gap?
We would rather hear about a study we missed or misread than keep it wrong.
Email the team