Association of Seasonal Ambient Air Pollution With a Risk of Adverse Perinatal Outcomes in Full-term Pregnancies: A Retrospective Cohort Study in Wenzhou, China

Prenatal exposures to ambient air pollution have been proved to be associated with adverse perinatal outcomes in previous studies. However, few studies have examined the interaction between air pollution and season of conception on term low birth weight (TLBW) or macrosomia. Birth registry data of singleton live births in Wenzhou, China between January 2015 to December 2016 were accessed from the Wenzhou maternal and child health information management platform, and the ambient air pollutants in Wenzhou were obtained on the Chinese Air Quality Online Monitoring and Analysis Platform. Single/two-pollutant binary logistic regression models were used to assess the associations between ambient air pollutants (PM 2.5 , PM 10 , NO 2 , SO 2 , and O 3 ) and TLBW/macrosomia, further exploring if the season of conception interacts with air pollution to impact birth weight. Finally, 213,959 term newborns were selected, including 2452 (1.1 %) TLBW infants and 13173 (6.1 %) macrosomia. In single/two-pollutant models, we observed an increased risk of TLBW associated with maternal exposure to PM 2.5 , PM 10 , SO 2, and NO 2 during the entire pregnancy, especially in the the 2nd trimester. Maternal exposure to O 3 during the 1st trimester was associated with increased macrosomia risk, and O 3 exposure during the 3rd trimester was associated with increased TLBW risk contrarily. Pregnant women conceived in the warm season may undergo more adverse ambient air environment that is related to the risks of TLBW.


Introduction
Term low birth weight (TLBW) and macrosomia are two important adverse perinatal outcomes that indicate abnormal fetal intrauterine growth and nutrition. TLBW is interpreted as full-term (born at ≥37 completed weeks of gestation) newborns with birth weight less than 2500 g. Newborns with low birth weight (LBW) are not only associated with neonatal mortality and disease morbidity (Staff 2016), but also are linked with complications such as cardiovascular diseases and diabetes in later adulthoods (Osmond &Barker 2000). Commonly, macrosomia is de ned as a newborn whose birth weight is equal to or more than 4000 g, which has been proved to increase the risks of maternal complications (Beta et al. 2019a), such as postpartum hemorrhage and perineal trauma. Macrosomia newborns are also more likely to develop obesity and other metabolic disorders in their childhoods and adulthoods (Beta et al. 2019b).
An increasing number of recent studies have examined the potential effects of air pollution exposures to LBW, with results across studies varying widely (Bosetti et al. 2010, Li et al. 2019b, Liu et al. 2019, Stieb et al. 2012, Uwak et al. 2021). For instance, Stieb et al. (Stieb et al. 2012) summarized 62 studies and concluded that air exposure to NO 2 , PM 10, and PM 2.5 during pregnancy was in relation to newborns with LBW. But Bosetti et al. (Bosetti et al. 2010) pooled that maternal exposure to PM 2.5 and PM 10 during pregnancy had no independent associations with the risk of LBW. These inconsistencies may be due to the risk of bias and imprecision of the articles. Uwak et al (Uwak et al. 2021) systematically evaluated the quality of individual studies regarding the associations between prenatal exposure to PM air pollution and LBW, and found that the quality of evidence for coarse PM from single-pollutant models was rated as very low/low. Models that included multi-pollutant exposures may better represent gestational ambient air pollution exposure, but most pollutants were typically reported separately and failed to consider co-exposures.
Compared with LBW, associations of air pollution exposure with risks of macrosomia were less clear. Several epidemiological studies evaluated the effects of air pollution on the risk of macrosomia, with the inconsistent ndings over different trimesters. Shang et al. (Shang et al. 2021) reported that maternal exposure to PM 2.5 , PM 10 , SO 2 , and NO 2 in the 3rd trimester increased the term birth weight and increased the risk of macrosomia, but showed contrary effects during the 1st and 2nd trimesters, and the entire pregnancy. Another nationwide study in China found that maternal exposure to PM 2.5 increased the risk of macrosomia over the 1st, 2nd, 3rd trimesters and the entire pregnancy (Chen et al. 2020).
Season of conception is an important factor affecting the association between air pollution and birth weight. Most studies only adjusted the season as a confounding factor in the model (Consortium on et al. 2019, Li et al. 2019b, Shang et al. 2021, instead of treating it as a major research factor. It has been proposed that season of conception involved in air pollutant concentration and composition could explain part of the link between intrauterine fetal growth and risk of adverse perinatal outcomes, and the biological responsiveness of the human body to air pollution might vary with seasons (Kim et al. 2017). Furthermore, the meteorological factors, such as ambient temperature and humidity, form various speci c seasonal patterns which have also been related to birth weight (Schifano et al. 2013). Wang et al. (Wang et al. 2019a) found a seasonal pattern that signi cant associations were observed between air pollution and small for gestational age (SGA) for women who conceived during summer or fall. Other study reported that the seasonal patterns of air pollutants increased the risk of preterm birth in autumn and winter distinguishably (Zhao et al. 2020). To our knowledge, no published studies have assessed the interaction between air pollution and season of conception on birth weight in China. Therefore, we aimed to conduct a retrospective cohort study to assess the effects of prenatal exposure to air pollution (PM 2.5 , PM 10 , SO 2 , NO 2 , O 3 ) on TLBW/macrosomia at each trimester in Wenzhou using single/two-pollutant models, further examining if the season of conception interacts with air pollution to impact birth weight. We only included full-term infants in order to exclude the in uence of premature infants on the results.

Study area
Wenzhou (27°03′ 28°36′ N, 119°37′ 121°18′E) borders on the East China Sea, which is greatly in uenced by subtropical monsoon. Due to the signi cant alternating between winter and summer monsoons of Wenzhou, the in uence of the seasonal in uence between air pollution and birth outcomes will be more signi cant.
On the other hand, Wenzhou is known as the ever shoe capital of China, the rubber footwear industries can produce volatile organic compounds (VOCs) which are classi ed as major contributors to air pollution, especially O 3 ).

Study population
The data of participants were accessed from the Wenzhou maternal and child health information management platform, which covered 51 midwifery clinics and hospitals in this city. The total number of registered singleton live births from January 2015 to December 2016 was 226,721. We excluded a total of 12,762 (5.6%) of the births according to the following exclusion criteria: gestational period < 37 weeks, missing or outlier birth weight, the maternal residence address during pregnancy was not in Wenzhou, and maternal age < 13 y or > 50 y. Finally, a total of 213,959 mother-infant pairs were included for further analysis, and the selection owchart of the birth records was shown in Fig. 1. The maternal information (e.g. maternal age, maternal education, ethnicity, registered residence, parity, date of delivery, and gestational age at birth) and the newborn's information (e.g. birth weight, infant sex, date of birth) were collected. The study was approved by the ethics committee of the Second Hospital A liated to Wenzhou Medical University.
Ambient air pollution and meteorological data Ambient air pollutants in Wenzhou were obtained on the Chinese Air Quality Online Monitoring and Analysis Platform (https://www.aqistudy.cn/), which provides real-time and historical daily concentrations of pollutants from 17 monitoring site stations in Wenzhou. The data of this website originated mainly from the real-time data recorded by the Ministry of Ecology and Environment of the People's Republic of China (http://www.mee.gov.cn/).We collected the 24-h average of PM 2.5 , PM 10 , SO 2 , NO 2 , and the 8-h average of O 3 daily levels averaged for all of the 17 monitoring site stations from March 2014 to December 2016 for our study. The daily mean temperature data were collected from the China Meteorological Data Service Center (http://data.cma.cn). According to the monthly mean temperature, we classi ed the parturients into those conceived in the warm season (from May to October) and in the cold season (from November to April of the next year) (Jia et al. 2020, Wang et al. 2019b). In addition, the daily exposures were averaged over four exposure windows for each trimester and whole pregnancy: the 1st trimester (from pregnancy to gestational week 12), 2nd trimester (from gestational week 13-week 27), 3rd trimester (from gestational week 28 to delivery), and the entire pregnancy (from gestational week 1 to delivery). The de nition of the beginning of the rst trimester was calculated according to the last menstrual period.

Study outcomes
Our main outcomes were TLBW and macrosomia. We de ned TLBW as birth weight less than 2500 g (born at ≥37 completed weeks of gestation), and de ned macrosomia as birth weight equals to or more than 4000 g.

Statistical analyses
Differences in characteristics among TLBW, macrosomia, and normal birth weight infants (2500 g ≤ birth weight and < 4000 g) were examined with independent sample F-tests/student t-test or Wilcoxon-Mann-Whitney test for continuous variables and with chi-square tests for categorical variables. For descriptive analysis, the min, max, mean, standard deviation (SD), median, and interquartile range were calculated for air pollutants. The correlations between different air pollutants' concentrations were assessed by Pearson's correlation. We conducted multivariable unconditional logistic regressions to evaluate the relationships between air pollution exposures and TLBW/ macrosomia. The odds ratios (ORs) and 95% con dence intervals (CIs) were calculated for associations between per 10μg/m 3 increase of air pollution exposures (PM 2.5 , PM 10 , NO 2 , SO 2 , and O 3) during each exposure window (1st, 2nd, and 3rd trimesters, and the entire pregnancy) and the risk of TLBW/ macrosomia compared to normal birth weight. In addition, all models were constructed to adjust potential confoundings, such as gestational age (continuous), maternal age at delivery (< 20, 20-35, ≥ 35 years old), season of conception (warm season: from May to October, cold season: from November to April of the next year), maternal educational level (Junior high school or below, Junior high school above), parity (the numbers of delivery, once or more than once), ethnicity (Han or others), sex of infants (male or female). Single-pollutant and two-pollutant multivariate models were used to assess the effects before and after adjusting for their potential effects on each other.
To further evaluate the modi cation of the season of conception and meteorological conditions on the associations between air pollution and TLBW/ macrosomia in the whole pregnancy, we included a cross-product interaction term between season of conception and air pollution along with their main effect terms in multivariable unconditional logistic regressions.
In sensitivity analyses, we tted multivariable linear models to estimate the impact of air pollutants on birth weight as a continuous measure. The models all adjusted for the covariates described above.
All P values were two-sided, with P < 0.05 considered statistically signi cant. All analyses were performed using R 4.0.5.

Results
The neonate and relevant maternal characteristics of the parturients were shown in Table 1. From 1 January 2015 to 31 December 2016, a total of 213 959 newborns' birth records were included in our nal analysis. Of these, 2 452 (1.1 %) were TLBW, and 13 173 (6.1 %) were macrosomia. The mean gestational age of all included newborns was 39.01 ± 1.09 weeks, and 114 618 (54%) were male neonates. Gestational age, maternal age at delivery, maternal educational level, parity, ethnicity, and sex of infants had statistical differences between TLBW, macrosomia, and normal birth weight infants.

The distribution of air pollution
Descriptive statistics of air pollution concentrations during each exposure time window and their correlation were provided in Table 2. According to the new ambient air quality standards of China (GB 3095-2012, 35 μg/m 3 for PM 2.5 , 70 μg/m 3 for PM 10 , 40 μg/m 3 for PM 10 , 70 μg/m 3 for SO 2 , 40 μg/m 3 for NO 2 , 160 μg/m 3 for O 3 ), the average levels of PM 2.5 , PM 10 , and NO 2 were all higher than the limits, and the exposure levels of SO 2 and O 3 were lower than the limits in Wenzhou than those in national standard level . PM 2.5 , PM 10 , SO 2, and NO 2 were positively and strongly correlated with each other (Pearson's correlation coe cients ranged from 0.85 to 0.97). PM 2.5 , PM 10 , SO 2 and NO 2 were negatively correlated with O 3 (r = -0.10 to -0.67).
Considering the correlations among pollutants were strongly positively correlated except for O 3 , we evaluated the effect of air pollutants using two-pollutant models of PM 2.5 , PM 10 , SO 2, and NO 2 adjusting for O 3 . Then, we examined the O 3 adjusting for PM 2.5 , PM 10 , SO 2, and NO 2 respectively for the subsequent analysis.
In the two-pollutant models (Fig. 2B), after adjusting for O 3 , similar results were found during the 2nd trimester and the entire pregnancy compared to the single-pollutant models, with attenuated effects of the exposure during the 1st and 3rd trimesters on the risks of TLBW.
In the two-pollutant models (Fig. 3B), after adjusting for O 3 , similar results were found during the 2nd trimesters and the entire pregnancy compared to the single-pollutant models, while the effects of the maternal exposure to PM2.5, PM 10 , SO 2 , and NO 2 during the 3rd trimesters were attenuated, which were no longer statistically signi cant.
The effect of maternal exposure to O 3 on macrosomia during all the three trimesters were opposite to the other air pollutants described above (1st: OR = 1.05, 95%CI: 1.03-1.06; 2nd: OR = 1.04, 95%CI: 1.03-1.05; 3rd: OR = 0.94, 95%CI:0.93-0.95). The association between O 3 exposure and increased macrosomia risk during the 1st trimester was still robust in the two-pollutant model (Table S1). However, we did not nd any association between O 3 and macrosomia during the entire pregnancy in single/two-pollutant models ( Figure 3A, Table S1). Table 3 shows the summary statistics for air pollution and air temperature during the whole gestational period by the season of conception. Pregnant women who conceived in the cold season would experience relatively higher levels of PM 2.5 , PM 10 , SO 2 , NO 2 , and high ambient temperature, but relatively lower levels of O 3 , compared with those who conceived in the warm season during their pregnancy. The Pearson's correlation coe cients among these exposures in a different season of conception were shown in Table S2. The degrees of some correlations were different between warm and cold seasons. For instance, the correlations between air temperature and exposures were stronger in cold than warm seasons. Further, the correlation between O 3 and other exposures were positive in the warm season, but in the cold season, the correlations were negative. Table 4, we found interactions between air pollution (except for O 3 ) exposure and conception season both on TLBW and macrosomia. The associations between PM 2.5 , PM 10 , SO 2, and NO 2 exposure and TLBW/ macrosomia were stronger in women who conceived in the warm season compared with those who conceived in the cold season (interaction P value < 0.05). With per 10 µg/m 3 increase in PM 2.5 , PM 10 , SO 2, and NO 2 exposure during the entire pregnancy, the risk increased by 34%, 21%, 62%, and 38% for TLBW, and the risk decreased by 24%, 18%, 37% and 27% for macrosomia, respectively. We did not observe any interaction between O 3 exposure and the season of conception on TLBW / macrosomia.

Sensitivity analyses
Table S3 presents linear model results for the correlations of gestational exposure to air pollutants and birth weight in singlepollutant and two-pollutant models, which were consistent with the results in the logistic regression models. In the single/twopollutant models, per 10μg/m 3 increase in PM 2.5 , PM 10 , SO 2, and NO 2 during the 1st, 2nd trimester and the entire pregnancy were negatively associated with the increase in term birth weight. The effects of the PM 2.5 , PM 10 , SO 2, and NO 2 exposure on decreasing birth weight were stronger in women who conceived in the warm season compared with those who conceived in the cold season (Table S4).

Discussion
In this retrospective cohort study, increased TLBW risk was observed in maternal exposure to PM 2.5 , PM 10 , SO 2, and NO 2 during the entire pregnancy, espeacially in the 2nd trimester. Maternal exposure to O 3 during the 1st trimester was associated with increased macrosomia risk, and O 3 exposure during the 3rd trimester was associated with increased TLBW risk contrarily. The interactions between air pollution (PM 2.5 , PM 10 , SO 2, and NO 2 ) exposure and season of conception both on TLBW and macrosomia were observed.
The effects of PM 2.5 , PM 10 , SO 2, and NO 2 exposure on TLBW during the 2nd trimester, and the entire pregnancy observed in our study were consistent with some previous epidemiological studies. A recent meta-analysis including American, Asian, and European countries observed positive associations between air pollution (PM 2.5 , PM 10 , SO 2 , NO 2 ) and LBW in the entire pregnancy, and the pooled ORs for TLBW were 1.081,1.053,1.125, and 1.030, respectively, which were lower compared with our results ). These may be due to the high concentration of air pollution in Wenzhou. For instance, the average exposure concentration of PM 2.5 over the entire pregnancy in our study was around 42 μg/m 3 , while the the range of PM 2.5 concentrations over the observation period was 2.9-12.5 µg/m 3 in Rico, USA (Kirwa et al. 2019). Liang et al. found that exposure to PM 2.5 in the 1st and 2nd trimesters increased the LBW risk in China (Liang et al. 2019). However, Bosetti et al. (Bosetti et al. 2010) pooled that maternal exposure to PM 2.5 and PM 10 during pregnancy did not be associated with a risk of LBW. Our ndings added further evidence of the positive associations during the 2nd trimester and the entire pregnancy both in the single/two-pollutant models. The differences among these studies may be attributed to the population susceptibilities, exposure assessments, potential confounders' adjustments, and sample size.
The majority of studies reported positive or null associations between air pollution and LBW in the 3rd trimester (Bosetti et al. 2010.We observed that maternal exposures to air pollution (PM 2.5 , PM 10 , SO 2, and NO 2 ) in the 3rd pregnancy were negatively related to TLBW in single pollutant models, which were only consistent with one study (Shang et al. 2021).
Additionally, we found macrosomia risk increased for per 10μg/m 3 increase in PM 2.5 , PM 10 , SO 2 , and NO 2 during the 3rd trimester in single pollutant models. These results were similar to several previous studies. One study of Shaanxi Province in China reported that maternal exposure to PM 2.5 , PM 10 , SO 2, and NO 2 increased the term birth weight and increased the risk of macrosomia in the 3rd trimester (Shang et al. 2021). Another nationwide study in China showed that every 10 μg/m 3 increase of PM 2.5 concentration, the macrosomia risk increased by 3.3% over the 3rd trimester (Chen et al. 2020). In addition, Li et al. suggested that exposure to NO 2 in the 3rd trimester signi cantly increased birth weight (Li et al. 2019b). Nonetheless, these studies typically reported separately and failed to consider co-exposures. Interestingly, in our two-pollutant models, after adjusting for O 3 , the effects of the air pollution (PM 2.5 , PM 10 , SO 2, and NO 2 ) in the 3rd trimester both on TLBW and macrosomia were no longer statistically signi cant, but the correlations in the whole pregnancy were still robust. These suggested that the 2nd trimester may be the critical exposure periods for PM 2.5 , PM 10 , SO 2, and NO 2 . One study conducted by Zou et al. also found that the associations between NO 2 exposures and the higher odds of LBW and TLBW were generally stronger for early months than for later months of the gestation (Zou et al. 2021). Another study demonstrated that the strongest association for SGA was found for PM 2.5 in the 2nd trimester, but did not nd critical exposure periods for PM 10 and SO 2 ). Further studies are needed to add more evidences for "critical window" during pregnancy.
The associations between O 3 and adverse perinatal outcomes have yet been to be elaborated. Several studies reported positive association between LBW and O 3 exposure in during speci c trimesters and the entire pregnancy (Chen et al. 2018, Laurent et al. 2014, Michikawa et al. 2017 , and other studies found null association (Bosetti et al. 2010, Hannam et al. 2014. One study of 321,521 term newborns reported that O 3 exposure might increase birth weight and increase the risk of macrosomia during the 1st and 2nd trimester (Shang et al. 2021), the impact of exposure to O 3 on birth weight was contrary to other pollutants, and our ndings added further evidence to support these results.
The detailed underlying toxicological mechanisms of air pollution exposure during pregnancy affects birth weight in different pregnancy periods are still unclear. For the entire pregnancy, some hypotheses have suggested that PM exposure entered the blood through the respiratory tract, which would cause maternal systemic oxidative stress, lung and placental in ammation, DNA damage, coagulation and endothelial dysfunction, and hemodynamic changes (Kannan et al. 2006, Parker &Woodruff 2008. Other studies have shown that air pollution might cause histopathological changes in the placenta, vascular damage and dysfunction (Bolton et al. 2012, Clemente et al. 2016, Veras et al. 2008. The above-mentioned changes may interfere with the transfer of oxygen and nutrients to the mother's placenta, thereby affecting the growth and development of the fetus, resulting in abnormal birth weight of the newborn. O 3 is a secondary pollutant, forming from the atmosphere through a number of photochemical reactions involving VOCs. Chloe et al (Friedman et al. 2021) found positive associations between PM 2.5 during certain exposure periods and maternal IL-6 and TNFα, and negative associations were found between O 3 and maternal CRP, IL-6, and TNFα. Thus, patterns were inconsistent for associations between PM 2.5 and O 3 and cord blood in ammatory biomarkers. Another review reported that O 3 could interact at least theoretically with any organic molecule, activates the anti-oxidant response, promotes immune regulation towards antiin ammatory mechanisms, and leads to the ability to extensively inhibit pro-thrombotic events (Chirumbolo et al. 2021). The molecular mechanisms between O 3 and its interaction with other pollutants on birth weight need to be further studied and clari ed.
We also observed a multiplicative interaction effect between the season of conception and air pollution. The associations between PM 2.5 , PM 10 , SO 2, and NO 2 exposure and TLBW/ macrosomia were stronger in women who conceived in the warm season compared with those who conceived in the cold season. A retrospective cohort study based on birth records in Guangdong province reported that O 3 exposure was associated with LBW for the babies conceived in cold season (Wang et al. 2021). Wang et al. found that the associations between air pollution and SGA varied by season of conception. However, these studies did not explore any multiplicative interaction. Our interaction results may be explained by the fact that women who conceived in the warm season, experiencing higher exposure level of air pollutants during their subsequent pregnancy with relatively low ambient temperatures, would increase the risk of TLBW and reduce the risk of Macrosomia. Previous studies have reported that during low ambient temperatures, fetal thermoregulatory responses to cold include increased blood viscosity and vasoconstriction, and these may limit maternal blood ow to the placenta, thereby reducing fetal growth (Bruckner et al. 2014). What's more, cold weather conditions and heating may also accumulate to the detriment of the diffusion of pollutants, resulting in a higher concentration of air pollution in the cold environment (Song et al. 2017). However, the season may also be a surrogate for other exposures with seasonal patterns, including lifestyle, outdoor activities, and the biological responses to air pollution (Kim et al. 2017). These factors have also been related to birth weight. Further researches are needed to clarify the underlying mechanism of the seasonal association between air pollution and adverse perinatal outcomes.
This study had key strengths. First, to the best of our knowledge, it was the rst study to explore the multiplicative interaction between air pollution and season of conception on birth weight. Second, we considered co-exposures, which may better represent gestational ambient air pollution exposure. Third, after controlling for known confounding factors, we comprehensively evaluated the air pollution on TLBW/ macrosomia and birth weight for continuous at three trimesters and the entire pregnancy in a large sample size, adding new evidence to the current understanding of air pollution and birth weight.
This study also had some limitations. Due to the limited information collected on the birth registration system, our study did not consider some confounding factors that may be related to birth weight, such as pregnant women's diet, BMI, socioeconomic status, smoking, and nutritional status. However, previous studies reported that the effect estimates changed little with and without adjustment for the above risk factors (Brauer et al. 2008, Kashima et al. 2011, Schramm et al. 1996. In addition, there may be misclassi cations of exposure due to lack of information on pregnant women's activities and mobility of residence during pregnancy. But one study reported that the exposure misclassi cation is more likely to be non-differential, leading to attenuation of the associations (Wang et al. 2018).
In summary, we observed an increased risk of TLBW was associated with maternal exposure to PM 2.5 , PM 10 , SO 2, and NO 2 during the entire pregnancy, especially in the 2nd trimester both in single and two pollutant models. Maternal exposure to O 3 during the 1st trimester was associated with increased macrosomia risk, and O 3 exposure during the 3rd trimester was associated with increased TLBW risk contrarily. Pregnant women conceived in the warm season may undergo more harmful ambient air environment that is related to the risks of TLBW. Further studies are needed to explore the interacting mechanisms of air pollution and season change on fetal weight growth.
Funding The present study was supported by the Public Welfare Technology Research Project of Zhejiang Province (No. LGF20H260012); the Natural Science Foundation of Zhejiang Province (No. LQ22H260001) Availability of data and materials All the necessary data generated or analyzed during this study are included in this published article. However, the datasets for statistical analysis are available from the corresponding author on reasonable request.
Ethics approval and consent to participate The study was approved by the Medical Ethics Committee of the Second Hospital A liated to Wenzhou Medical University.

Consent for publication Not applicable.
Competing interests The authors declare no competing interests. Wang Q, Liang Q, Li C, Ren M, Lin S, Knibbs LD, Zhang H, Gong W, Bao J, Wang S, Wang X, Zhao Q, Huang C (2019b) Wang M, Lu C, Feng L, Ma H, Meng H, Qi M, Fan Q, Wang H, Zhou H, He J (2020): Seasonal response of the synergism of maternal comorbidities and long-term air pollution exposure on birth outcomes. Ecotoxicol Environ Saf 191, 110232 Zou Z, Liu W, Huang C, Cai J, Fu Q, Sun C, Zhang J (2021): Gestational exposures to outdoor air pollutants in relation to low birth weight: A retrospective observational study. Environ Res 193, 110354 Tables Note: a Difference in maternal and neonatal characteristics between TLBW and normal birth weight infants.
b Differences in maternal and neonatal characteristics between macrosomia and normal birth weight infants. a Units are ug/m 3 for PM 2.5 , PM 10 , NO 2 , SO 2 , and O 3 b All Pearson's correlation P < 0.05  Figure 1 The selection owchart of participants.

Figure 2
Association between PM 2.5 , PM 10 , SO 2 , NO 2, and TLBW by gestational period in single-pollutant and two-pollutant models. A Single-pollutant models included PM 2.5 , PM 10 , SO 2 , NO 2 respectively. B Two-pollutant models: included PM 2.5 , PM 10 , SO 2 , NO 2 adjusted for O 3 respectively. Multivariable unconditional logistic regression models were used to evaluate associations between air pollution (per 10μg/m 3 increase) and TLBW during each exposure windows (1st, 2nd, and 3rd trimesters, and the entire pregnancy). All models were adjusted for covariates, including Gestational age, maternal age at delivery, season of conception, maternal educational level, parity, ethnicity, sex of infants.

Figure 3
Association between PM 2.5 , PM 10 , SO 2 , NO 2, and macrosomia by gestational period in single-pollutant and two-pollutant models. A Single-pollutant models included PM 2.5 , PM 10 , SO 2 , NO 2 respectively. B Two-pollutant models included PM 2.5 , PM 10 , SO 2 , NO 2 adjusted for O 3 respectively. Multivariable unconditional logistic regression models were used to evaluate associations between air pollution (per 10μg/m 3 increase) and macrosomia during each exposure windows (1st, 2nd, and 3rd trimesters, and the entire pregnancy). All models were adjusted for covariates, including Gestational age, maternal age at delivery, season of conception, maternal educational level, parity, ethnicity, sex of infants.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. 1218Supplementarymaterial.docx