Pregnant women living in areas with higher levels of microscopic soot and ash from wildfires may be more likely to have a child later diagnosed with autism, according to a nationwide study of nearly 900,000 births in the United States. The research, published in JAMA Pediatrics, indicates that exposure to wildfire smoke during the middle and later stages of pregnancy is associated with an elevated risk of autism spectrum disorder, though no meaningful link was found for attention-deficit/hyperactivity disorder (ADHD). As wildfires grow more frequent and intense, the findings point to the potential benefits of helping expectant mothers reduce their exposure to smoke-filled air.
Wildfire smoke is a complex mixture of gases and fine particulate matter, often referred to as PM2.5 because the individual airborne particles measure 2.5 micrometers or smaller in diameter. That is less than one-thirtieth the width of a human hair, making the particles small enough to bypass the body’s upper airways, penetrate deep into the lungs, and enter the bloodstream. In this study, researchers examined how prenatal exposure to these particles relates to two common neurodevelopmental conditions that begin in early childhood. Autism spectrum disorder is a condition characterized by differences in social communication, sensory processing, and repetitive behaviors, whereas ADHD involves persistent difficulties with sustaining attention, impulse control, and hyperactivity.
Rising temperatures, prolonged droughts, and human land use have caused wildfires to burn larger areas and produce heavier plumes of smoke across North America. Studies show that fine particles generated by wildfires tend to be more toxic than particles from other sources because of their distinct chemical makeup, smaller particle size, and frequent co-occurrence with high heat.
Once inhaled by a pregnant woman, these microscopic pollutants may affect fetal development through several biological pathways. The authors of a 2021 review argued that fine particles can cross the placenta and trigger brain inflammation and oxidative stress, an imbalance between cell-damaging molecules and the body’s ability to neutralize them, potentially disrupting how neurons and their supporting cells grow and function.
Brain imaging research has linked prenatal fine particle exposure to differences in newborn brain development, with a study of newborn brain imaging finding that higher prenatal exposure to general fine particulate matter was associated with reduced myelin, the protective fatty coating that insulates nerve fibers. Even so, epidemiological evidence linking air pollution to specific childhood diagnoses has been mixed. For instance, a 2020 meta-analysis of childhood studies found no consistent relationship between general particulate air pollution and an increased risk of ADHD. Only two prior studies had specifically investigated prenatal wildfire smoke and autism, both limited to California, and they reached differing conclusions about whether risk increases during the third trimester or only under extreme smoke conditions in areas with otherwise clean air.
“Wildfire smoke is an increasingly important public health concern, and its effects extend well beyond the communities where fires occur,” study lead author Stefania Papatheodorou, an associate professor of epidemiology in the Department of Biostatistics and Epidemiology at the Rutgers School of Public Health, told PsyPost. “Smoke can travel long distances, exposing pregnant individuals hundreds or even thousands of miles away. Although previous research has linked prenatal air pollution exposure with autism, much less was known about wildfire-related pollution specifically. We wanted to examine whether exposure to wildfire smoke during pregnancy was associated with subsequent autism diagnoses in a large national population.”
To examine how wildfire smoke relates to early brain development on a national scale, Papatheodorou and her colleagues analyzed health records from the U.S. Medicaid Analytic Extract database, which covers more than 40 percent of births nationwide. Their cohort included 893,212 mother-child pairs with pregnancies conceived on or after January 1, 2007, and delivered through 2014. To ensure accurate tracking of prenatal health and location, the researchers restricted the sample to women aged 12 to 55 who remained continuously enrolled in Medicaid from three months before pregnancy through one month after delivery and who had the same residential zip code at the start of pregnancy and at delivery.
On average, mothers in the study were about 25 years old, and the children were roughly evenly split between boys (50.6 percent, or 452,218 children) and girls (49.4 percent, or 440,994 children). Based on Medicaid enrollment records, 40.6 percent of the mothers were White, 30.5 percent were Black, 21.3 percent were Hispanic or Latino, 4.1 percent were Asian or Other Pacific Islander, and 3.5 percent were of unknown race or ethnicity. The researchers followed the children’s medical records through the end of 2018, resulting in an average follow-up period of 6.3 years for children diagnosed with autism and 7.1 years for those diagnosed with ADHD. Diagnoses were identified using validated algorithms that scan medical billing claims for specific diagnostic codes and age criteria.
“An important strength is the inclusion of a large population of Medicaid-insured families, who are often underrepresented in research,” Papatheodorou said. “Understanding environmental risks in these populations is essential to developing public health responses that serve all families.”
To estimate wildfire smoke levels where each mother lived, the team used a previously validated machine-learning framework that mapped daily concentrations of fine particulate matter across every square kilometer of the contiguous United States. The system subtracted background pollution from total particle levels that incorporated fire emissions, isolating the specific fraction of fine particulate matter originating from wildfires and aggregating those levels weekly for each mother’s zip code.
The researchers then used statistical survival models, which track how much time passes before a diagnosis occurs. They combined these with distributed lag models, a technique for estimating how exposure during each week of pregnancy relates to later risk, to evaluate how cumulative and weekly smoke levels across the first 33 weeks of pregnancy related to a child’s likelihood of being diagnosed with autism or ADHD. They stopped the primary prenatal tracking window at 33 weeks so that the analysis could cover most of pregnancy while still including many babies born prematurely.
Because air pollution levels and medical diagnosis rates vary across communities, the statistical models accounted for a wide range of individual and neighborhood factors. At the individual level, the authors adjusted for maternal age, race and ethnicity, number of previous births, season of conception, prenatal folic acid prescriptions, smoking, alcohol or substance abuse, and poor nutrition. They also controlled for average outdoor temperature during pregnancy, non-wildfire background particle pollution, and local socioeconomic measures, including zip-code median household income, high school graduation rates, population density, urban or rural status, county obesity rates, and distance to the nearest hospital.
Across the entire cohort of 893,212 children, 8,357 were diagnosed with autism and 22,302 were diagnosed with ADHD during the follow-up period, meaning the share of children diagnosed by age 8 (the cumulative incidence) was 1.9 percent for autism and 10.4 percent for ADHD. Depending on the year, wildfire smoke accounted for between 4.4 percent and 7.5 percent of the total fine particulate matter in the air. Pregnancy-long averages for wildfire-specific particles were relatively low at 0.55 micrograms per cubic meter of air, compared with about 9.4 micrograms per cubic meter from non-wildfire sources.
The researchers found that an interquartile range increase in cumulative wildfire fine particulate matter across pregnancy, defined as an increase of 0.75 micrograms per cubic meter (the spread between the 25th and 75th percentiles of exposure), was associated with a 30 percent relative increase in the risk of an autism diagnosis (a hazard ratio of 1.30, with the plausible range running from a 7 percent to a 59 percent increase). The relationship was not strictly linear, as risk estimates rose across low and moderate smoke levels before leveling off at the highest concentrations.
“This describes a relative difference in the rate of diagnosis during follow-up,” Papatheodorou explained. “Even a moderate association can matter at the population level when an exposure affects many pregnancies, but translating this result into an absolute increase in risk requires additional information.”
“One notable finding was that the association with autism was apparent in a national cohort rather than a population selected for proximity to major wildfires,” she added. “This reinforces the importance of considering transported smoke and cumulative exposure during pregnancy, rather than focusing only on people living near a fire.”
In contrast, the same increase in prenatal wildfire smoke exposure showed no meaningful association with ADHD, yielding a relative risk increase of just 3 percent that was not statistically significant (a hazard ratio of 1.03, with a plausible range from a 9 percent decrease to a 17 percent increase). Non-wildfire background particle pollution was associated with a smaller, 10 percent relative increase in autism risk (and was not associated with ADHD). When the team examined the timing of exposure week by week, they observed a gradual increase in risk from early to late pregnancy for both outcomes, with the strongest estimates for autism appearing during mid-to-late gestation, particularly after approximately week 15. The authors suggest this window may matter because late fetal development involves rapid brain growth, the formation of connections between neurons, and the maturation of neural circuits.
“We also did not observe a clear association with ADHD,” Papatheodorou noted. “That difference raises questions about whether the findings reflect differences in biological pathways, susceptibility, exposure timing, or how these conditions are identified, questions that require further investigation.”
When the researchers broke the data down by the child’s sex, the overall pregnancy-wide risk estimate for autism was 31 percent higher for both boys and girls, though the estimate for girls was less precise and its plausible range (from a 12 percent decrease to a 98 percent increase) included the possibility of no effect. In week-by-week analyses, boys showed elevated risk across most of gestation while increases among girls appeared mainly in mid-to-late pregnancy.
Autism estimates were elevated across all racial and ethnic groups, while regional analyses suggested higher but imprecise estimates in the South for both conditions, though the ranges overlapped with other regions. The authors noted that regional variation could reflect differences in fire behavior and smoke composition, climate and co-occurring heat, background pollution, socioeconomic factors, health care access, and local diagnostic practices.
The results also held up across sensitivity tests that extended the pregnancy window to 37 weeks or restricted the analysis to children diagnosed after age 5. To test how vulnerable the findings were to unmeasured confounding (hidden factors not captured in the data that might be related to both smoke exposure and autism), the authors calculated that such a factor would need to be linked to both higher wildfire smoke exposure and autism by a risk ratio of at least 1.92 each, conditional on the measured covariates, to fully explain away the observed association.
These findings align in part with other research linking airborne pollutants to neurological and psychiatric outcomes, including a study using genetic data to estimate long-term pollution exposure that connected particulate components like black carbon, rather than direct prenatal wildfire smoke, to higher risks of both autism and attention disorders, the latter sitting in some tension with the new study’s null ADHD finding. Focusing specifically on smoke, a study of pediatric emergency visits found that higher wildfire smoke exposure was associated with more emergency department visits for youth psychiatric crises such as anxiety, depression, and schizophrenia, suggesting that smoke exposure may carry short-term mental health risks for children as well.
At the same time, the lack of an association with clinical ADHD diagnoses partly echoes a study of school-age children that examined general industrial and traffic air pollution rather than wildfire smoke. That earlier investigation found that prenatal air pollution exposure was associated with modestly higher teacher-reported ADHD symptom scores, particularly in boys, even though it was not linked to a higher rate of formal clinical ADHD diagnoses.
“Our findings suggest that wildfire smoke exposure during pregnancy may have implications for children’s longer-term neurodevelopment, in addition to its better-known effects on respiratory and cardiovascular health,” Papatheodorou told PsyPost. “We observed an association with autism diagnoses, but our study does not establish that wildfire smoke causes autism. The broader message is that protecting people from wildfire smoke matters even far from an active fire. This is a public health challenge that requires cleaner indoor air, accessible information, and support for families—not a responsibility that should fall entirely on pregnant individuals.”
As with all research, there are some caveats to consider when interpreting these findings. Because smoke levels were assigned to each mother’s residential zip code, which was confirmed only at the start and end of pregnancy, the data could not capture moves in between, how much time pregnant women spent outdoors, or whether they used indoor air filters during smoky days. According to the authors, the leveling off of risk estimates at the highest smoke levels could reflect behavioral adaptations during severe smoke events, exposure misclassification, less precise estimates at the upper end of exposure, or selection related to early pregnancy loss. In line with that last possibility, higher wildfire smoke exposure early in pregnancy has been linked to pregnancy loss in a 2025 study of Colorado pregnancies led by one of the same authors.
“The most important caveat is that this was an observational study,” Papatheodorou cautioned. “We identified an association, and other factors may have contributed to the findings despite our efforts to account for them. Our exposure estimates represent outdoor pollution where participants lived, rather than the amount of smoke each individual actually inhaled.”
“We also identified autism through health care records, so access to evaluation and diagnostic services can influence which children receive a recorded diagnosis,” she added. “Autism has a complex, multifactorial basis. These findings should not be interpreted as identifying a single cause.”
There are also a few limitations related to the study population and time frame. Families who left the Medicaid program before their children were old enough to be evaluated could have been missed, though statistical weighting suggested that loss to follow-up did not skew the results. In addition, the pregnancies in this cohort occurred between 2007 and 2014 among low-income Medicaid beneficiaries, meaning the findings may not automatically generalize to higher-income populations or reflect the potentially heavier wildfire smoke conditions seen in more recent years. The authors noted that future studies using more recent birth cohorts are needed to test whether these associations persist under contemporary smoke conditions, as well as to evaluate the combined effects of wildfire smoke and other environmental and social stressors.
“Our next steps are to explore whether associations with wildfire smoke differ across autism subtypes and extend this work to other neurodevelopmental outcomes,” Papatheodorou said. “We also plan to examine how environmental and social conditions may shape susceptibility. Our long-term goal is to translate these findings into prevention by identifying when reducing exposure matters most, which populations need additional support, and which strategies effectively protect pregnant people and their children.”
“Our findings add to the reasons to take wildfire smoke seriously as a widespread environmental exposure,” she added. “They also underscore the need for further research to confirm the association and better understand the processes underlying it.”
The study, “Prenatal Exposure to Wildfire PM2.5 and Risk of ASD and ADHD in Children Enrolled in Medicaid,” was authored by Stefania Papatheodorou, Hayon Michelle Choi, Krista F. Huybrechts, Sonia Hernandez-Diaz, Yang Liu, Danlu Zhang, Matthew Shupler, Michael Leung, Wanyu Huang, Xinye Qiu, Yaguang Wei, Christopher J. McDougle, Evgenia Papatheodorou, Joel D. Schwartz, Brent A. Coull, Marc G. Weisskopf, and Antonella Zanobetti.