Three-quarters of US pregnancies since 2012 involved some wildfire smoke exposure

Wildfire smoke research has a geographic bias baked into it. Fires burn in a handful of western states, so that is where cohorts were assembled and where the evidence accumulated. Smoke, however, does not stay in the states that produce it.

A study published in June in the American Journal of Preventive Medicine measures how far it goes, using a dataset large enough to cover essentially every American birth over nine years.

The exposure map

Researchers combined satellite-based wildfire smoke plume data with restricted-use vital statistics microdata covering 34,440,915 live births in the contiguous United States from 2012 to 2020 [s1]. Prenatal exposure was defined as the number of days during gestation with medium or heavy wildfire smoke over the mother's county of residence [s1].

By that measure, 25.5 million infants — 74.2% — had some prenatal smoke exposure [s1].

The trend inside that number is the finding. Median exposure rose from 0.6 days in 2012–2015 to 2.0 days in 2016–2020, a statistically significant increase (p<0.001) [s1]. In the later period, 4.3 million births (22.7%) had at least seven days of exposure during gestation, also a significant increase over the earlier period (p<0.001) [s1].

Exposure was not distributed the way the fire maps would suggest. The highest prenatal exposures fell on non-Hispanic American Indian/Alaska Native infants, on infants in rural counties, and on states in the Upper and Western Midwest and the Rocky Mountains [s1]. The authors' conclusion is that prenatal wildfire smoke exposure is widespread, increasing, and not confined to regions near active fires [s1].

What this study is and is not

It is an exposure study. It counts smoke days over counties where pregnant people lived. It does not measure any birth outcome, and it does not establish that the exposure caused harm.

That distinction is easy to lose, because the framing invites the inference. The authors state the rationale plainly — the developing fetus is vulnerable to prenatal smoke exposure, and existing evidence is focused on a few western states [s1] — but the contribution here is the denominator, not a dose-response curve.

County of residence is also a coarse proxy. It does not capture how much time a person spent outdoors, how well their building sealed, or whether they had filtration. Those are exactly the factors that determine whether an outdoor smoke day becomes an inhaled dose.

The companion problem

Which brings up the more uncomfortable paper. A critical review published at the end of June in Current Environmental Health Reports assessed recent literature on whether wildfire smoke affects some populations differently from others, and whether wildfire-derived PM2.5 differs in effect from PM2.5 from other sources [s2].

Its finding on the first question is inconsistent evidence of differential exposure to and health risks from wildfire PM2.5 across population subgroups [s2]. The authors suggest one reason: how much wildfire PM2.5 gets into a building, and whether a person can take protective action, both vary with socioeconomic status — meaning the outdoor exposure measure used in most studies, including county smoke days, may not track the exposure that matters [s2].

On the second question, the review notes that some recent studies have documented larger health risks from wildfire-derived compared with non-wildfire-derived PM2.5, and then declines to endorse the conclusion. Further research is needed to evaluate whether those findings are confounded by the type of fuel burned, driven by methodological concerns, or true [s2].

Where the evidence is actually solid

The review is unusually direct about what is settled and what is not.

Respiratory health endpoints have been the most consistent and most commonly evaluated outcomes in wildfire smoke studies [s2]. That evidence base the authors regard as very well established — so much so that they explicitly encourage future research to move away from further documenting respiratory impacts [s2].

For non-respiratory outcomes, including cardiovascular disease, findings conflict and additional research is needed to resolve them [s2]. The same is true for smoke constituents other than PM2.5, and for differences between wildland-urban interface fires and wildland fires, where the material burning differs substantially [s2].

What to take from the pair

Read together, the two papers describe a research field where exposure measurement has outrun effect measurement. It is now possible to say with confidence that three-quarters of recent US pregnancies involved some wildfire smoke, that median exposure more than tripled between two four-year periods, and that the burden falls disproportionately on rural and American Indian/Alaska Native communities far from the fires [s1].

It is not yet possible to say with the same confidence what a given number of prenatal smoke days does, or whether wildfire smoke is more harmful per microgram than other particulate pollution [s2]. Coverage that pairs a hard exposure statistic with a soft outcome claim is combining the strong half of one paper with the weak half of another.

What to watch

The review points to the most useful next studies: outcomes beyond the respiratory system, constituents beyond PM2.5, and better handling of indoor infiltration and protective behaviour, which is where the exposure misclassification lives [s2]. Wildfire smoke is projected to continue increasing [s2], so the denominator this AJPM analysis established is likely to be the smallest one anyone reports for a while.