Background
Climate change is contributing to more frequent wildfires. Exposure to wildfire smoke during pregnancy increases the risk of adverse neonatal outcomes. Pregnant women with asthma and their newborns are particularly at risk of negative effects from wildfire smoke exposure and remain understudied. Determining critical windows of exposure during pregnancy could help inform mitigation strategies for this high-risk group.
Objective
This study aimed to assess the association of prenatal wildfire smoke exposure with adverse neonatal outcomes in mothers with asthma and to determine which period of pregnancy is most critical.
Study Design
This was a multicenter cohort study (Breathing for Life Trial) including mothers and newborns from 6 antenatal clinics across 4 cities in Australia (Sydney, Brisbane, Newcastle, Canberra). Pregnant women (n=1275) with asthma were recruited at 12 to 23 weeks of gestation. Chronic lung disease (other than asthma), miscarriages, and perinatal deaths were excluded. Prenatal wildfire smoke exposure days and daily wildfire-related particulate matter ≤2.5 μm concentrations were determined by residence. Exposure over pregnancy was assessed in 3 ways: (1) cumulative days of wildfire smoke (0, 1–4, 5–9, ≥10 days); (2) wildfire smoke–affected days and wildfire-related particulate matter ≤2.5 μm as continuous variables for each gestational month; and (3) a natural experiment comparing exposure during the 2019–2020 summer extreme wildfire period with historical controls. Neonatal outcomes included birthweight, low birthweight (<2500 g), gestational length, preterm birth (<37 weeks), neonatal intensive care unit admission, and cesarean delivery.
Results
The mean age of women was 30.7 years (SD, 5.5), 85% were White, 15% were smokers, and 69% were overweight or obese. The prenatal mean daily particulate matter ≤2.5 μm concentration (all sources) was 7.6 μg/m 3 (SD, 1.42); the median daily wildfire-related particulate matter ≤2.5 μm concentration was 0.1 μg/m 3 (range, 0–7.7 μg/m 3); and the median number of prenatal cumulative days of wildfire smoke exposure was 3 (range, 0–71 days). Prenatal exposure to ≥10 cumulative wildfire smoke days was associated with low birthweight (adjusted odds ratio, 4.2; 95% confidence interval, 1.2–13.9), preterm birth (adjusted odds ratio, 2.8; 95% confidence interval, 1.1–7.2), and neonatal intensive care unit admission (adjusted odds ratio, 5.0; 95% confidence interval, 1.4–17.8). Exposure to wildfire-related particulate matter ≤2.5 μm in the second and third gestational months was associated with preterm birth, small for gestational age, and neonatal intensive care unit admission. In the natural experiment, prenatal exposure was associated with low birthweight (adjusted odds ratio, 2.6; 95% confidence interval, 1.1–6.2), preterm birth (adjusted odds ratio, 2.5; 95% confidence interval, 1.2–5.2), and neonatal intensive care unit admission (adjusted odds ratio, 4.8; 95% confidence interval, 2.0–11.2).
Conclusion
In pregnant women with asthma, prenatal exposure to wildfire smoke early in pregnancy increases the risk of adverse neonatal outcomes. Measures to reduce wildfire smoke exposure early in pregnancy could be beneficial.
Video
Introduction
Climate change has led to a global increase in wildfires, , resulting in high levels of air pollution that impact respiratory outcomes such as asthma exacerbations and hospitalizations. ,,,,, Asthma affects up to 10% of pregnant women and increases the risk of adverse neonatal outcomes, particularly if the mother experiences an asthma exacerbation. , A recent position statement by The Thoracic Society of Australia and New Zealand on asthma during wildfires identified pregnant women with asthma as a vulnerable group ; however, the neonatal outcomes for this group are unreported. We have an established cohort of pregnant women with asthma, a group known to be susceptible to adverse fetal and neonatal outcomes, which offers the analytical advantage of not requiring recruitment of a large general population to assess differences in exposure. ,
AJOG at a Glance
Why was this study conducted?
Exposure to wildfire smoke during pregnancy increases the risk of adverse neonatal outcomes. Infants of pregnant women with asthma are particularly at risk, but this group has been understudied. Establishing critical timing of exposure could better inform mitigation strategies.
Key findings
Among women with asthma, exposure to ≥10 days of wildfire smoke over pregnancy increased the odds of low birthweight, preterm birth, and neonatal intensive care unit admission. Exposure to wildfire smoke in early pregnancy was associated with the highest odds ratio for adverse outcomes.
What does this add to what is known?
Extended exposure to wildfire smoke and exposure early in pregnancy increase the risk of adverse neonatal outcomes in newborns of women with asthma.
Exposure to air pollution during pregnancy impacts the development of the fetus. , Carbon particles from ambient air pollution cross the placental barrier and have been identified in fetal circulation and organs. , Wildfire smoke contains not only high levels of combustion-derived particulate matter ≤2.5 μm (PM 2.5 ) but also hazardous inorganic and organic substances. ,
Research to date suggests that prenatal wildfire smoke exposure is associated with increased risks of lower birthweight, preterm birth, and pregnancy loss. ,,, However, there is high heterogeneity in findings due to variation in wildfire exposure measures and analysis techniques. , A recent systematic review identified 13 studies that assessed the association between prenatal wildfire smoke exposure and birthweight. However, only 4 of these were considered sufficiently homogeneous for inclusion in a meta-analysis, which reported a negative (−16 g per 5-μg/m 3 increase in wildfire-related PM 2.5 over pregnancy) but inconclusive association. Several other neonatal outcomes have also been associated with prenatal wildfire smoke exposure, including neonatal intensive care unit (NICU) admission and cesarean delivery, ,,, but these results require replication and validation. Furthermore, establishing a critical window of exposure during gestation has been difficult, partly because of the short-term nature of wildfires.
This study aimed to use a high-risk (maternal asthma) cohort to assess the association of prenatal exposure to wildfire smoke days and wildfire-related PM 2.5 with neonatal outcomes. We applied 3 different methods of exposure classification to assess (1) the association between cumulative days of wildfire smoke exposure over pregnancy and neonatal outcomes, (2) the ascertainment of a critical window of vulnerability during pregnancy, and (3) the impact of a prolonged, intense fire season.
Materials and methods
Study design and population
Pregnant women with asthma were recruited from 6 Australian public hospital antenatal clinics in Sydney, Newcastle, Canberra, and Brisbane between March 2013 and October 2019 into the Breathing for Life Trial (n=1295). , Inclusion criteria included doctor-diagnosed asthma, symptoms of asthma and/or asthma medication use in the prior 12 months, age ≥18 years, and 12 to 23 completed weeks of gestation at randomization. Exclusion criteria included chronic lung disease other than asthma or concomitant chronic illness that may affect participation. The current study excluded cases of fetal or neonatal death and women who were lost to follow-up before birth (n=20). All women provided written informed consent. Ethics approval was obtained from the Hunter New England Human Research Ethics Committee (HREC; 12/10/17/3.04) and the Australian Capital Territory Health HREC (ETH.11.15.232).
Outcomes
Neonatal outcomes were identified using hospital birth records, and comprised birthweight (grams and standardized percentile), low birthweight (<2500 g), gestational length (weeks), preterm birth (<37 weeks), small for gestational age (<10 th percentile, calculated using GROW, version 8.0.6.2 [Gestation Network]), NICU admission, and cesarean delivery.
Exposure assessment
Air pollution PM 2.5 data, seasonal trend data, and statistical flags used to determine the wildfire PM 2.5 component were provided by the previously validated Centre for Safe Air, Australia ( www.safeair.org.au ). In summary, daily PM 2.5 concentrations were estimated at the 5-km grid cell corresponding to each woman’s residence for the duration of pregnancy using modeled data based on PM 2.5 observations from air pollution monitoring sites, combined with satellite, weather, and land use data via a random forest statistical model. The random forest model achieved an out-of-bag R 2 of 71.5% and a root-mean-square error of 4.5 μg/m 3. Seasonal-trend decomposition using LOESS (locally estimated scatterplot smoothing) methodology was then applied to determine seasonal and long-term trend components of daily total PM 2.5 . Wildfire smoke days were defined as days on which total PM 2.5 exceeded either the 95 th percentile or 2 standard deviations above the remainder PM 2.5 (total PM 2.5 minus seasonal and long-term trend) and for which a wildfire within 50 km of the grid area was confirmed by satellite. Wildfire-related PM 2.5 was defined as the remainder PM 2.5 on wildfire days after subtracting seasonal and long-term trend PM 2.5 from total PM 2.5 .
Exposure definitions
Cumulative exposure
The total number of wildfire smoke days in the period from 0 to 34 weeks of gestation was calculated, and pregnancies were categorized into groups of 0, 1 to 4, 5 to 9, or ≥10 cumulative days of wildfire smoke exposure. Because very little literature currently exists to inform a biologically based decision regarding exposure duration, the categories were determined according to the distribution of the data. The population was restricted to women (the “restricted cohort”) who gave birth after 34 weeks to ensure that the exposure and the outcome did not overlap and to minimize immortal time bias.
Monthly exposure
The number of wildfire days and wildfire-related PM 2.5 were determined for each month of pregnancy and treated as continuous measures for that month in the total cohort (n=1275).
Natural experiment
From July 2019 to February 2020, eastern New South Wales experienced multiple extensive, prolonged bushfires, which burned 6.7% of the state and 37% of the national parks. Pregnancies from Newcastle that covered at least 1 month of the most intense fire period (November 1, 2019, to January 31, 2020) were classified as exposed. The unexposed population included women in the study population from Newcastle who were pregnant for at least 1 month from November 1 to January 31 in the previous summers (2013–2019). Supplemental Figure 1 shows the mean daily PM 2.5 by month in the Newcastle subgroup over the 2011–2020 study period, with a dramatic peak during the 2019–2020 summer bushfires.
Covariates
Covariates identified as potential confounders from directed acyclic graphs were derived from data collected at randomization (randomization group, address, asthma control, body mass index, smoking in pregnancy, parity, maternal age, ethnicity) or from medical records at birth (infant sex). A sociodemographic variable, the Socio-Economic Indexes for Areas (SEIFA) based on disadvantage, was computed for each woman. SEIFA is a government-determined variable applied to all Australian residents by area (approximately 100 residences) and divided into quintiles from most to least disadvantaged. Although all recruited women had asthma, the propensity for asthma exacerbations and requirement for reliever medications can vary. Therefore, we also considered asthma control as a potential confounder, defined according to the 2016 Global Initiative for Asthma guidelines.
Statistical analysis
Generalized linear models were used to estimate the associations of cumulative days of wildfire smoke exposure (vs 0 exposure days), monthly incremental wildfire days, wildfire-related PM 2.5 exposure, and exposure during the 2019–2020 summer (vs unexposed) with each of the outcomes. Logistic regression was used for binary outcomes (low birthweight, preterm birth, NICU admission, and cesarean delivery), and linear regression was used for continuous outcomes (birthweight, birthweight percentile, and gestational age). Models were fitted using a generalized estimating equation (GEE) framework to account for clustering of infants within mothers and mothers within cities. For the GEE models, an exchangeable working correlation structure was assumed. The models for cumulative and monthly exposure were adjusted for season of conception, asthma control, smoking in pregnancy, SEIFA, and infant sex. To provide context for the wildfire PM 2.5 results, associations between total PM 2.5 per gestational month and perinatal outcomes were also assessed. The natural experiment included only a cluster term for infants within mothers and was adjusted for smoking in pregnancy, parity, and month of conception ( Supplemental Table 2 ). Given an unexpected disparity in multiple births between exposed and unexposed groups in the natural experiment, a post hoc stratified analysis was conducted for singleton and multiple births.
All analyses were performed using SAS software, version 9.4 (SAS Institute Inc, Cary, NC).
Results
Study population characteristics
In the total birth cohort (n=1275), the maternal mean age was 30.7 (SD, 5.5) years, 14.6% smoked during pregnancy, 68.7% were overweight or obese, and 33.9% had uncontrolled asthma at recruitment. The restricted cohort for the analysis of cumulative days of wildfire smoke exposure (n=1237) was similar to the total cohort, whereas the Newcastle subgroup for the natural experiment (n=744) were more obese, less advantaged, and more likely to smoke in pregnancy than the total cohort ( Table 1 ).
Table 1
Characteristics of the total birth cohort, restricted birth cohort (≥34 weeks’ gestation), and Newcastle natural experiment subgroup
| Characteristics |
Total cohort
N=1275 n (%) |
Restricted cohort
n=1237 n (%) |
Newcastle subgroup
n=744 n (%) |
|---|---|---|---|
| Infant sex | |||
| Female | 646 (50.7) | 622 (50.3) | 360 (48.4) |
| Male | 628 (49.3) | 615 (49.7) | 384 (51.6) |
| Multiple births | 60 (4.7) | 43 (3.5) | 43 (5.8) |
| Parity | |||
| 0 | 628 (49.3) | 609 (49.4) | 342 (46.0) |
| 1 | 379 (29.8) | 367 (29.7) | 221 (29.7) |
| ≥2 | 266 (20.9) | 258 (20.9) | 181 (24.3) |
| Season of conception | |||
| Summer | 290 (22.8) | 283 (22.9) | 132 (17.7) |
| Autumn | 349 (27.4) | 339 (27.4) | 220 (29.6) |
| Winter | 302 (23.7) | 293 (23.7) | 183 (24.6) |
| Spring | 333 (26.1) | 321 (26.0) | 209 (28.1) |
| Location | |||
| Newcastle | 812 (63.7) | 782 (63.3) | 744 (100) |
| Sydney | 303 (23.8) | 297 (24.0) | 0 |
| Canberra | 24 (1.9) | 23 (1.9) | 0 |
| Brisbane | 135 (10.6) | 134 (10.8) | 0 |
| Socio-Economic Indexes for Areas | |||
| 1 (most disadvantaged) | 137 (10.8) | 129 (10.5) | 103 (13.9) |
| 2 | 251 (19.8) | 243 (19.8) | 222 (29.9) |
| 3 | 415 (32.8) | 402 (32.8) | 310 (41.7) |
| 4 | 211 (16.7) | 208 (17.0) | 99 (13.3) |
| 5 (most advantaged) | 251 (19.8) | 244 (19.9) | 9 (1.2) |
| Missing | 10 | 11 | 1 |
| Randomization group | |||
| Intervention | 640 (50.2) | 618 (50.0) | 373 (50.2) |
| Control | 634 (49.8) | 617 (50.0) | 371 (49.8) |
| Ethnicity | |||
| White | 1056 (85.0) | 1026 (85.1) | 623 (86.9) |
| Other | 187 (14.7) | 143 (14.9) | 84 (13.1) |
| Missing | 32 | 31 | 27 |
| Maternal asthma control | |||
| Well controlled | 267 (21.7) | 264 (22.0) | 156 (22.1) |
| Partly controlled | 548 (44.4) | 533 (44.5) | 302 (42.8) |
| Uncontrolled | 418 (33.9) | 401 (33.5) | 247 (35.0) |
| Missing | 42 | 42 | 39 |
| Maternal age (mean, SD) | 30.7, 5.5 | 30.7, 5.5 | 29.8, 5.5 |
| Smoking in pregnancy | 184 (14.6) | 178 (14.5) | 125 (17.0) |
| Maternal BMI | |||
| Healthy weight | 379 (30.8) | 374 (31.3) | 181 (25.8) |
| Overweight | 377 (30.7) | 365 (30.6) | 213 (30.4) |
| Obese | 467 (38.0) | 448 (37.5) | 304 (43.4) |
| Missing | 45 | 44 | 43 |
| Neonatal outcomes | |||
| Birthweight (g) (mean, SD) | 3317 (603) | 3372 (517) | 3313 (587) |
| Birthweight percentile (mean, SD) | 46.8 (30.3) | 47.5 (30.1) | 47.7 (30.2) |
| Low birthweight <2500 g | 105 (8.2) | 68 (5.5) | 58 (7.8) |
| Gestational age (wk) (mean, SD) | 38.8 (2.1) | 39.0 (1.5) | 38.7 (2.0) |
| Preterm (<37 wk) | 133 (10.4) | 95 (7.7) | 85 (11.4) |
| Small for gestational age | 185 (14.6) | 164 (13.3) | 104 (14.0) |
| Neonatal intensive care unit admission | 78 (6.1) | 52 (4.2) | 40 (5.4) |
| Cesarean delivery | 444 (34.9) | 414 (33.5) | 266 (35.8) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
