Two decades of nationwide data reveal contrasting patterns across birth outcomes, diabetes, and childhood ADHD as researchers probe whether the health effects associated with air pollution persist even as U.S. PM2.5 levels decline.
Study: Long-term air pollution exposure and childhood ADHD prevalence. Image Credit: Khosro / Shutterstock
A recent study in the journal Scientific Reports examined how annual and long-term exposure to fine particulate (PM2.5) air pollution relates to various health outcomes, including birth outcomes, diabetes prevalence, and childhood ADHD, using two decades of U.S. county-level data. The study further evaluated the strength of these associations after adjusting for temporal trends, socioeconomic and demographic factors, and spatial dependence.
Declining Fine Particulate Matter and Persistent Health Risks
PM2.5 consists of airborne particles with diameters ≤2.5 micrometers. Although PM2.5 concentrations in the United States have declined by nearly 50% since 2000, several health conditions linked to air pollution, including diabetes and childhood neurodevelopmental disorders, have not improved in corresponding ways.
Air pollution remains a significant health risk. Assessing whether long-term regional differences in exposure remain associated with geographic variation in health outcomes is critical, especially as average pollution levels decrease. This is particularly relevant for developmental and neurobehavioral outcomes, where environmental exposures can have lasting effects.
Early-life pollution exposure can harm both immediate and long-term childhood health. Adverse effects may begin prenatally, as PM2.5 contains metals and reactive compounds that can induce oxidative stress, inflammation, and disruptions in placental function. Prenatal pollution exposure has also been linked to low birth weight, epigenetic changes, and impaired fetal growth and neurodevelopment. Prenatal stress may further impact fetal development through increased glucocorticoid signaling, which has been associated with later cardiovascular, psychiatric, and neurodevelopmental disorders.
ADHD affects about 10% of U.S. children and is linked to early-life PM2.5 exposure, while low birth weight has separately been associated with subsequent ADHD symptoms. Early-life PM2.5 exposure is also associated with deficits in attention and working memory. While individual-level studies show a relationship between air pollution and ADHD, the impact of long-term regional differences in air pollution on geographic variation in ADHD prevalence remains poorly understood. Given regional variability and confounding socioeconomic and demographic factors, population-level analyses are needed to address this gap.
Tracking PM2.5 Exposure and Health Outcomes Across U.S. Counties
Annual PM2.5 concentrations were estimated from satellite-derived global gridded data that incorporated aerosol optical depth retrievals, chemical transport modeling, and ground-based measurements. These gridded values were aggregated to the county level using population-weighted averages for each U.S. county from 2000 onward.
Birth outcome data were sourced from national natality databases, with county-level birth weight counts downloaded for 2007–2024. The primary merged longitudinal dataset spanned 2007–2020 across 576 counties, while analyses incorporating annual socioeconomic and demographic covariates were restricted to 2010–2019. Low birth weight was defined as less than 2500 grams. Merging these with PM2.5 data by county and year produced a longitudinal dataset of 7,658 county-year observations, mainly from counties with larger birth counts.
For diabetes, annual age-adjusted prevalence estimates among adults were obtained from the United States Diabetes Surveillance System (USDSS), maintained by the Centers for Disease Control and Prevention. ADHD prevalence was estimated at the county level using small-area estimation of the National Survey of Children’s Health (NSCH) data for children aged 5–17 years. Socioeconomic and demographic information was obtained from the American Community Survey (ACS) and merged with health and PM2.5 data at the county and year levels.
Robust Association Between Long-term PM2.5 Exposure and Higher Childhood ADHD Prevalence
County-level analysis revealed that low birth weight averaged approximately 8.2% across reporting counties, while diabetes and childhood ADHD prevalence showed marked geographic variation across the United States. In models examining annual PM2.5 exposure and health outcomes from 2010 to 2019, the association between PM2.5 and low birth weight was positive but not statistically significant. After adjusting for socioeconomic and demographic variables, the association was further reduced, providing little evidence of an independent within-county association.
For diabetes prevalence, initial fixed-effects models showed a modest negative association with PM2.5, which remained modest across models adjusting for annual socioeconomic and demographic covariates. However, the apparent lag pattern became substantially attenuated and irregular after accounting for county-specific temporal trends. Lagged analyses without county-specific temporal trends showed that the temporal association diminished and eventually became positive after about 7 years, but this pattern was sensitive to differences in long-term trends across counties rather than indicating a consistent lagged relationship.
For childhood ADHD prevalence, cross-sectional models related county-level ADHD prevalence estimates from 2016 to 2018 to mean PM2.5 exposure over 2005–2015 across 3,106 counties. In unadjusted models, a 1 µg/m3 increase in PM2.5 was associated with a 0.523 percentage-point higher ADHD prevalence.
After adjusting for median household income, median age, educational attainment, and state fixed effects, this association was attenuated by approximately 42% but remained statistically significant at 0.301 percentage points per 1 µg/m3. The association persisted after excluding counties in the upper and lower 1% tails of PM2.5 exposure and was consistent across different exposure periods, although these periods cannot be interpreted as specific prenatal or childhood developmental windows.
Spatial analysis showed that while residual PM2.5 exposure exhibited pronounced geographic clustering, adjusting for residual spatial dependence, or unexplained similarities among neighboring counties, using a spatial error model did not materially affect the PM2.5-ADHD association. Residual spatial autocorrelation was no longer evident after this adjustment.
Conclusions
The current study found a robust ecological association between long-term PM2.5 exposure and childhood ADHD prevalence at the county level across the United States, which persisted after adjustment for selected socioeconomic and demographic factors, state-level differences, alternative exposure periods, and spatial dependence. The findings are consistent with the growing body of evidence linking air pollution to neurodevelopmental outcomes but do not establish that PM2.5 exposure causes ADHD in individual children.
Because ADHD prevalence estimates encompassed children aged 5–17 years across multiple birth cohorts, the study could not distinguish between prenatal and postnatal exposure mechanisms or identify a specific developmental exposure window. The aggregated county-level design may also mask differences within counties, while factors including smoking, obesity, healthcare access, population change, and other environmental or socioeconomic conditions could contribute to residual confounding.
Looking forward, future research should clarify the specific exposure windows and biological mechanisms underlying this association, ideally using individual-level longitudinal studies and designs better able to assess causal relationships. Such studies could help determine whether the observed county-level relationship reflects causal effects and identify when PM2.5 exposure may be most consequential during development.
Astrocyte changes may explain how stress affects mental health