Next Article in Journal
Infection Spread and High-Resolution Detection of Close Contact Behaviors
Next Article in Special Issue
Spatial Variability of Heat-Related Mortality in Barcelona from 1992–2015: A Case Crossover Study Design
Previous Article in Journal
An Extension of the Failure Mode and Effect Analysis with Hesitant Fuzzy Sets to Assess the Occupational Hazards in the Construction Industry
Previous Article in Special Issue
Short-Term Associations of Fine Particulate Matter and Synoptic Weather Types with Cardiovascular Mortality: An Ecological Time-Series Study in Shanghai, China
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:

Prenatal Ambient Particulate Matter Exposure and Longitudinal Weight Growth Trajectories in Early Childhood

Department of Environmental Health, School of Public Health, Boston University, Boston, MA 02118, USA
Department of Pediatrics, School of Medicine, Boston University, Boston, MA 02118, USA
Biostatistics and Epidemiology Data Analytics Center, School of Public Health, Boston University, Boston, MA 02118, USA
Department of Biostatistics, T.H. Chan School of Public Health, Harvard University, Boston, MA 02115, USA
Department of Environmental Health, T.H. Chan School of Public Health, Harvard University, Boston, MA 02115, USA
Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA 02115, USA
Author to whom correspondence should be addressed.
Int. J. Environ. Res. Public Health 2020, 17(4), 1444;
Submission received: 6 February 2020 / Revised: 19 February 2020 / Accepted: 20 February 2020 / Published: 24 February 2020
(This article belongs to the Special Issue Air Pollution and Climate Change-Mediated Health Impacts)


Air pollution exposure during pregnancy has been associated with impaired fetal growth and postnatal weight gain, but few studies have examined the effect on weight growth trajectories. We examine the association between validated 1 km2 resolution particulate matter (PM2.5) concentrations, averaged over pregnancy, and sex-specific growth trajectories from birth to age six of participants in the Boston-based Children’s HealthWatch cohort (4797 participants, 84,283 measures). We compared weight trajectories, predicted using polynomial splines in mixed models, between prenatal PM2.5 above or below the median (9.5 µg/m3), and examined birth weight as an effect modifier. Females exposed to average prenatal PM2.5 ≥ 9.5 µg/m3 had higher weights compared to females exposed to < 9.5 µg/m3 throughout the study period (0.16 kg at 24 months, 0.61 kg at 60 months). In males, higher prenatal PM2.5 exposure was associated with significantly lower weights after 24 months of age, with differences increasing with time (−0.17 at 24 months, −0.72 kg at 60 months). Associations were more pronounced among low birth weight (<2500 g) females, but did not differ by birth weight status in males. Our findings demonstrate the complex association between air pollution exposures and childhood weight trajectories and emphasize the importance of sex-stratified analyses.

1. Introduction

Evidence is accumulating that weight growth trajectories in utero and during early postnatal periods are predictive of childhood overweight and obesity [1,2]. Investigating when the onset of childhood overweight and obesity occurs is of interest to understand the etiology of childhood and adult obesity and to identify critical periods for intervention [2]. Early-life overweight and obesity are associated with a range of chronic adverse health outcomes later in life, such as type 2 diabetes, coronary heart disease [3], and hypertension [4,5,6,7,8]. Although factors such as genetic susceptibility and nutrition are associated with overweight and obesity [9,10,11], the rapid rise of obesity implicates environmental risk factors as contributors to this trend [12]. Specific early-childhood growth trajectory phenotypes, such as the “thrifty phenotype” whereby catch-up growth follows fetal growth restriction, have been linked to adverse cardiometabolic outcomes in adulthood [13,14].
The vast majority of epidemiological studies to date associating prenatal ambient particulate matter with an aerodynamic diameter of 2.5 microns (PM2.5) and weight have focused almost exclusively on weight outcomes (birth weight, raw weight, body mass index (BMI), adiposity) measured cross-sectionally [15,16]. Modelling weight as a longitudinal outcome—growth trajectories—is a more informative measure than weight modelled as a cross-sectional outcome to understand steps on the causal pathway between early-life air pollution exposure and morbidities later in life. Few studies have investigated the link between prenatal ambient air pollution exposure and infant and early-childhood growth trajectories [17,18,19,20,21,22].
Using electronic health records (EHRs) and surveys administered to obtain detailed maternal and child demographic information, we investigated the association between prenatal PM2.5 exposure and weight growth trajectories from birth to age six years in the Boston-based Children’s HealthWatch (CHW) cohort. We assessed exposure using concentrations from spatially and temporally resolved PM2.5 predictions at 1 km2 resolution at maternal residence. Based on previous evidence associating prenatal PM2.5 and traffic exposure with low birth weight (LBW), we hypothesize a significant association between prenatal PM2.5 exposures and weight growth rates in our study population.

2. Materials and Methods

2.1. Study Population

We identified the study population from participants who enrolled in the CHW at the Boston Medical Center between 1 January 2008 and 31 December 2015. The CHW is an ongoing, sentinel surveillance study that gathers clinical and interview data from primary care sites or emergency department (ED) visits [23]. Institutional review board (IRB) approval (protocol number H-34069) was obtained from Boston University Medical Campus IRB prior to data collection and participants provided informed consent.
At primary care or ED visits, trained CHW interviewers surveyed caregivers accompanying children younger than 48 months in a private setting. The survey covers multiple domains, including demographic and socioeconomic characteristics, breastfeeding practices, smoking status, child health status, and information about material hardship—including housing, food, and energy insecurity. Respondents were excluded if the interviewee was not the primary caregiver, if they did not speak English or Spanish, were not knowledgeable about the child’s household, had been interviewed previously that year, lived out of state, or did not consent to participate. Caregivers of critically ill or injured children were not approached.

2.2. Linking EHR and CHW Survey Data

As shown in Figure 1 and Figure S1, we linked CHW survey data to EHRs and to PM2.5 predictions. We matched CHW surveys to the EHR based on date of the child’s CHW interview, gender, date of birth, and medical record number. EHR data coverage included births until 31 December 2015, resulting in an EHR data range from 2005 to 2015. Birth weight and weight (kg) at each visit were extracted from the EHR, as well as address at each visit, age at each visit (months), gestational age (weeks), primary diagnosis for admission (ICD-10 code), child sex, child date of birth, and visit type (inpatient or outpatient). Missing EHR gestational age (42%) and birth weight (42%) data were imputed using CHW survey data. Correlation coefficients between the two data sources were 0.97 for birth weight and 0.95 for gestational age. We used the EHR value if CHW survey and EHR values differed.

2.3. Weight Outcome Data and Analytical Sample Selection

Of 100,712 visits, 84,283 visits included weight measurements. Only participants with two or more weight measurements over the study period were included in the study. We excluded visits with missing weight data, if exact weight measurements were repeated between visits, if an address was missing, could not be geocoded, or Boston Medical Center was listed as the address. Biologically implausible weight values, defined as a sex-specific weight-for-age z score of less than -6 or more than 5, were also dropped from analyses, as recommended by the Centers for Disease Control and Prevention [24]. This process yielded a final analytical sample of 4797 caregiver/child dyads with over 70,369 visits (Figure S1).
In the final sample, 28%, 64%, and 8% of weight measurements were from ED, outpatient, and inpatient visits, respectively. We used weight measurements from all visit types as our dependent variable, and performed a sensitivity analysis excluding inpatient weights.

2.4. Exposure Assessment

Geocoding. We geocoded addresses recorded on the medical record to the corresponding residential parcel [25]. Of all addresses listed in the EHR, 0.8% were either missing or listed as a P.O. Box. Of the remaining addresses that were included in the geocoding process, 98% were matched to a residential address. We linked geocoded addresses at each visit to predicted PM2.5 data and several other spatial covariates, discussed in more detail below.
Prenatal ambient PM2.5. Details of the validated 1-km2 resolution PM2.5 prediction model can be found in Kloog et al. (2014). Briefly, this modeling approach used a combination of aerosol optical depth (AOD) satellite data retrieved using the multi-angle implementation of atmospheric correction (MAIAC) algorithm, land use and meteorological variables, and daily ambient monitor PM2.5 concentrations to calculate daily PM2.5 predictions on a 1-km2 grid between 2000 and 2015 [26]. The daily ambient PM2.5 monitoring data were obtained from the U.S. Environmental Protection Agency (EPA) Air Quality System (AQS) database, which includes ground monitoring sites located in urban, suburban, and rural areas across the United States, and from the Interagency Monitoring of Protected Visual Environments (IMPROVE) network, which includes monitors located in national parks and other federal lands.
We assigned PM2.5 to the geocoded addresses at birth using the closest 1 km2 grid centroid. We calculated average PM2.5 concentration over the prenatal period, which we estimated using date of birth and gestational age (in weeks). We categorized prenatal PM2.5 as a bivariate variable above and below the median (9.5 µg/m3) and as tertiles in a sensitivity analysis because of the non-linear association with weight.

2.5. Covariates

We assigned block group-level covariates from American Community Survey (ACS) 2006–2010 5-year summary data to each geocoded address. Linked covariates included median block group household income and percent with less than a high school degree (both continuous).
Individual covariates collected from the EHR included number of moves within the study period (continuous), child’s birth weight (binomial categorized at > 2500 grams), child’s gestational age (continuous and binomial categorized at > 37 weeks), child sex, and birth date.
We obtained the following covariates from the CHW survey: year of enrollment, caregiver BMI (underweight, normal weight, overweight, obese), caregiver race/ethnicity (non-Hispanic white, non-Hispanic black, Hispanic, other), breastfed during pregnancy (yes/no), caregiver smoking status in the past five years (yes/no), caregiver immigration status (U.S.-born, yes/no) and caregiver educational attainment (no schooling or some high school, high school, postsecondary), a composite measure of food, energy, and housing insecurity (referred to herein as “cumulative hardship”) [27], and mother’s age at birth (derived from child date of birth and mother’s age at CHW enrollment date).

2.6. Growth Trajectories Model

To estimate associations between average prenatal PM2.5 and growth trajectories, we applied a mixed effects model. We ran all models separately for males and females because of evidence of sex differences in growth trajectories and air pollution susceptibility [11,28,29]. In the model, we adjusted for several covariates a priori known to be biologically related to childhood postnatal weight, or as confounders of the association between weight and PM2.5. They include: caregiver race/ethnicity, cumulative hardship (categorical), child’s gestational age (categorical), block group median household income (continuous), and caregiver immigration status. Our main model also included a random intercept for child and a random slope for age to account for repeated measurements within subject and to adjust for heterogeneity in trends over time [30,31]; this model allows for correlated repeated weight measurements and varying number of measures per child. This modeling approach has been shown to produce good model fit in this and several other cohorts [9,32,33,34,35,36,37].
To model the growth trajectories we first applied generalized additive mixed effect models, controlling for covariates described above, and used penalized splines for age to visually inspect the shape of the relationship between weight and age defining sex-specific growth curves, and weight and PM2.5 [38].
To create a model that allows for a close approximation of the true growth function, we then used linear spline models with cubic polynomial terms for age. We used an iterative process to test combinations of one, two, three or four knot points at knot placements 3, 6, 9, 12, 18, 24, 36, and 40 months (number of participants with weight observations at different age ranges is shown in Supplemental Table S1). We explored two- and three- degree polynomial functions by adding linear, quadratic, and cubic terms to the model at the aforementioned knot points. We assessed model fit by comparing Akaike Information Criterion (AIC) and log-likelihood values between non-nested and nested models, respectively. Models for both sexes were the same and included a quadratic term, and cubic term, with knots at 6 and 12 months (Supplemental Table S2). We evaluated interactions between age terms and covariates using a likelihood ratio test.
We found that including interactions between covariates and age terms in the model provided a significantly (p-value < 0.05) better fit than simply including the covariates as main effects, suggesting that the effect of covariates varies over age. We also tested year of enrollment to test for time trend, but found that it did not improve model fit. In addition, we included in the model interaction terms between prenatal PM2.5 and the age terms to allow the shape of the curve to differ between age groups.
The final model was:
Yij = B0 + B1ageij + B2ageij2 + B3ageij3 + B4(ageij − 6months)2 + B5(ageij − 12months)2 + B6exposurei + B7−ncovariatesi + (Bnageij*(exposurei + covariatesi)) + (Bnageij2*(exposurei + covariatesi)) + (Bnageij3*(exposurei + covariatesi)) + ((Bn(ageij − 6months)2)*(exposurei + covariatesi)) + (Bn(ageij − 12months)2*(exposurei + covariatesi)) + b0i + b1iageij + eij,
where Yij is weight for the ith subject at time j, exposurei is average prenatal PM2.5 for the ith subject; covariatesi are non-time-varying covariates for the ith subject, and b0i and b1i are the subject-specific random intercept and slope, respectively, for the ith subject.
To optimize interpretability of the model, we estimated predicted weight and associated 95% confidence interval at specified ages for the two levels of PM2.5. We then computed the differences in weight between levels of PM2.5 over the growth trajectory at the specified ages.
We tested effect modification by birth weight (< 2500 grams), as growth trajectories may differ by birth weight. The phenotype of low birthweight followed by catchup growth has been associated with a range of cardiometabolic outcomes [13,14,39].

2.7. Sensitivity Analyses

We performed a sub-analysis including mother’s BMI (underweight, normal, overweight, obese) at study enrollment as a model covariate given literature showing that maternal and paternal weight are associated with childhood growth and obesity outcomes [9,10,11]. This analysis was conducted within a subset of the study population (63%) as the remaining participants were missing biological mother’s BMI. We also performed a sensitivity analysis excluding all subjects born <37 weeks gestation and a separate analysis excluding inpatient weights [40].
We report results as mean difference in weight (kg) between above and below median prenatal PM2.5 concentrations using t-tests, reported separately for males and females. All statistical tests were 2-tailed and a p-value of < 0.05 is used to denote statistical significance. Statistical analyses were conducted using R version 3.3 (R Foundation for Statistical Computing, Vienna, Austria).

3. Results

We present characteristics of the study population stratified by child sex in Table 1. The CHW cohort is an ethnically diverse, low-income population. Cohort participants were 50% non-Hispanic black and 35% Hispanic, 43% had a post-secondary degree, and 42% of mothers enrolled were immigrants to the United States. The average block group median household income was $43,792 and $43,442 (United States Dollar) for males and females respectively, which was lower than the state average of $70,114. The median mother’s age at delivery was 27 years (± 6.3). Average prenatal PM2.5 was approximately normally distributed across the population. The mean prenatal PM2.5 concentrations were similar between males (9.6 ± 1.2, range: 6.5–14.0 µg/m3) and females (9.5 ± 1.2, range: 6.3–14.1 µg/m3).
Supplemental Figure S2 displays the plot of the penalized spline model with a smooth term for model age in the model for weight, for males and females. As expected, both males and females have an exponential rate of growth in the first few months of life, which slows and becomes linear around 12 months of age, consistent with U.S. growth curve trends [41].
Supplemental Table S3 presents observed weights and estimated weights in the study population compared to the U.S. reference population weights from ages 0 to 72 months. Overall, models produced values close to observed weights. Cubic polynomial models slightly overestimated weight between 12 and 24 months for males, and slightly underestimated weights after 48 months for females. Observed weights in the study population were lower than the U.S. population during the early infancy period (0–3 months) but were higher at all other ages.
Figure 2 shows estimated childhood weight trajectories by levels of prenatal PM2.5 exposure below the median and above the median, and the weight trajectory of the general U.S. population for comparison, from birth through 72 months. Among males, above-median prenatal PM2.5 exposure results in growth trajectories that are significantly lower compared to the below-median prenatal PM2.5 exposure group from 2–6 years of age. The model predicts a 0.17 kg lower weight at 24 months and 0.72 kg lower weight at 60 months. We see an association in the opposite direction among females, where above-median prenatal PM2.5 is associated with significantly higher weights at all ages, with the exception of birth weight. The greatest difference in weight was at 72 months, where high exposure groups had weights 0.64 kg higher than low exposure groups (Table 2).
In a sensitivity analysis, mother’s BMI at enrollment was associated with weight in all male prenatal PM2.5 models. Male babies of underweight mothers have statistically significantly lower weights compared to those from normal BMI mothers (−0.47, p = 0.0013), but not for female babies. The direction and magnitude of effects were similar when compared to a model that omitted BMI. We found similar results to our main analysis when we categorized PM2.5 into tertiles, with similar growth curves between the second and third tertiles of PM2.5 and lower rates of growth for those at or below the first tertile (data not shown, but available upon request). We restricted to full-term births (>37 weeks) and to outpatient weight measurements only, finding similar magnitude and direction of effect for male and female models (data not shown, but available upon request).
Results for polynomial age models for prenatal PM2.5, stratified by low birth weight (LBW, <2500 g) and non-low birth weight (>=2500 g), are presented in Figure 3, and Tables S4 and S5. As with the unstratified sample, estimated weights for males were significantly higher for the lower prenatal PM2.5 group in the non-LBW group, after 24 months. Absolute weight differences were more pronounced in the non-LBW group as compared to the LBW group (e.g., 0.82 kg in non-LBW vs. 0.39 kg in LBW at 60 months). Among females, differences between high and low PM2.5 exposure groups were greater among LBW compared to non-LBW females. Among LBW females, the above-median prenatal PM2.5 group had 0.24 kg greater weight at 6 months of age and 3.31 kg greater weight at 72 months compared to below-median PM2.5.

4. Discussion

We found differences in weight growth trajectories between levels of prenatal PM2.5 exposure in a racially and ethnically diverse population. Males exposed prenatally to PM2.5 greater than the median (9.5 µg/m3) had lower weights after 24 months of age compared to less-exposed groups. In contrast, over the growth trajectory (birth to age 6), female weights were greater for higher PM2.5 prenatal exposure, which was driven by the positive relationship between PM2.5 and weight in LBW females.
Exposure to ambient air pollution such as particulate matter is a ubiquitous and modifiable risk factor. Inhalation of PM2.5 during pregnancy can interfere with fetal growth via oxidative stress (OS), intrauterine inflammation, endothelial function, and altered mitochondrial function [42,43,44]. These biological processes may alter trophic mechanisms that control growth throughout the life course [4].
The existence of sex-specific differences in the association between prenatal PM2.5 and growth trajectories is consistent with the broader literature on air pollution and birth outcomes, albeit with considerable variation in the magnitude and direction of effect. Overall, more studies have reported increased susceptibility to in utero PM2.5 exposure among males compared to females [28,45,46,47]. In a systematic review, females were more commonly found to be at higher risk of LBW, but in a re-analysis of data from four studies, males were at higher risk of LBW in the presence of high prenatal PM2.5 [45]. Ebisu and Bell (2012) reported a 3.2% (95% CI: 0.8, 5.6%) lower relative risk of LBW per IQR increase of PM2.5 elemental carbon in females compared to males [46]. The opposite effect was found in a pregnancy cohort located in Krakow: males had 188.6 g lower birth weight in the fourth compared to the first quartile of prenatal PM2.5 [47].
The literature is sparse with reference to prenatal outdoor ambient air pollution exposure and sex-specific differences in measures of weight later in childhood. Chiu et al. (2017) found that 1 µg/m3 increase in prenatal-average PM2.5 was associated with a 0.36 kg (95% CI: 0.12-0.68) increase in fat mass for males, but not females, and an increase in waist to hip ratio in females at four years of age [48]. Animal and human studies have demonstrated that prenatal PM2.5 exposure can induce sex-specific epigenetic modifications in leptin methylation, which is associated with adult metabolic disorders [49,50]. Sex-specific differences in energy metabolism and increased OS vulnerability in males have been found in animal studies, and may explain sex-specific differences found here [51].
Beyond the sex-specific effects, our findings are broadly consistent with a growing literature linking air pollution exposures with childhood growth [19,21,22,52]. In a Boston-area pregnancy cohort, Fleisch et al. (2015) found increased odds of weight-for-length > 95th percentile at 6 months of age in fourth quartile third-trimester PM2.5 and distance to roadway < 50 m compared to the referent groups, though estimates were not statistically significant. In a follow-up study, children whose mothers lived < 50 m from a major roadway at the time of delivery had 2.1 kg (95% CI: 0.8, 3.5) greater total fat mass compared to children (median 7.7 years of age) living >=200 m [52]. Inconsistent with our findings, this same study found each interquartile range increase in one-year average PM2.5 concentrations prior to each measurement occasion was associated with lower BMI-z score and total and truncal fat mass in mid-childhood (average 8 years of age) [52]. Investigators also examined trimester-specific associations, finding no association between third trimester PM2.5 and BMI outcomes [21].
A Massachusetts birth cohort with similar demographic characteristics to our study found an increased risk of overweight (BMI z-score >= 85th percentile) and obesity (BMI z-score > = 95th percentile) at ages 2–9 years in the highest versus lowest quartile of average prenatal PM2.5 exposure (OR = 1.3 (95% CI: 1.1, 1.6)) and postnatal PM2.5 in the first two years of life (OR = 1.2 (95% CI:1.1, 1.5)) [18]. Jerrett et al. (2014) assessed associations between traffic density within 150 m of the home and longitudinal sex-specific BMI growth trajectories between the ages of 5 and 11 years of age, finding no significant association [15].
The exposure and outcome assessments and study population sociodemographic characteristics may explain some of the discrepancies between our study and other studies that have assessed prenatal PM2.5 with weight trajectories. For instance, we used 1-km2 PM2.5 predictions, while Chiu et al. 2017 used measures from the nearest monitor, which may have decreased exposure variability and increased exposure measurement error. Other studies used BMI z-score and physiological measures of adiposity as their outcomes [18,19,20,48]. In the present study, we used raw-weights rather than z-scores to examine the true shape of the growth trajectory. Though McConnell et al. (2015) and Jerrett et al. (2014) found positive associations between postnatal near roadway pollution and traffic density with rates of BMI growth, their exposure metrics incorporated traffic density and meteorological conditions, averaged over the year of each measurement, whereas our study only considered PM2.5.
Further, the CHW study population was more ethnically diverse and the prevalence of multiple hardships was higher as compared to the more ethnically homogenous and high-income Boston-area Project Viva cohort [19,21,52]. Inconsistencies may also be explained by our weaker measure of smoking (ascertained at the time of survey by asking whether caregiver smoked in the last 5 years), which is a predictor of weight. Greater cigarette smoke exposure and has been shown to have a synergistic effect with air pollution in increasing growth rates [20]. Further investigation is required to examine the role of other social determinants of health, such as housing conditions or food insecurity, acting independently or as modifiers of the association between ambient pollutants and weight gain. Using data from the CHW Boston site, we found that homelessness during pregnancy and average PM2.5 during the second trimester were marginally associated with reduced birthweight, while participating in Women, Infants, and Children (WIC) programs was associated with increased birthweight [53]. A deeper comprehension of these stressors can help identify root causes and potential solutions to the childhood obesity challenge, and why obesity may persist into adulthood [54].
There are a number of limitations in this study that may also explain our findings. We did not have measures of maternal smoking during pregnancy, nutrition, and maternal pre-pregnancy BMI, which are risk factors for childhood weight gain [9,55]. We attempted to control for these measures using a variable measuring smoking status during the past five years and maternal BMI at study enrollment. CHW does not collect information on diet, physical activity, and other environmental exposures that may be jointly associated with air pollution and weight gain, so there may be residual confounding in our analysis [56]. To address this concern, we controlled for both a measure of hardship, including food insecurity, and block group median income, both of which were associated with child growth in our models.
There are some limitations inherent in the use of an EHR for data ascertainment. Health providers may inconsistently record weight measurements, resulting in non-differential outcome misclassification. However, weight measurements recorded in EHRs were found to be prone to < 0.7% error from 0 to 5 years of age according to a large prospective cohort study using EHRs [57]. Lastly, many of the visits in early childhood are routine checkups, whereas visits later in childhood may be comprised of children with poorer health outcomes, and thus differential susceptibility to the effects of PM2.5 exposure later in life.
In spite of these limitations, our study has several strengths. This study included a large sample size: 4797 participants, with 70,369 weight measurements. The analytical method used to model the appropriate function of weight for age allowed us to explore longitudinal differences in trajectories by levels of ambient air pollution exposure. This method further allowed us take full advantage of the EHR containing measurements collected at varying time points and frequencies, and to estimate change in slope during specified growth time periods. Using an EHR for epidemiological analyses is a novel and relatively inexpensive, source of longitudinal data ascertainment, allowing for many measurements and limited potential for recall and participation bias [58]. The CHW cohort is ethnically diverse, making our results more representative of vulnerable populations, which are understudied to date. Survey data collected by CHW also provided rich information for covariates on multiple hardships, immigration status and caregiver characteristics. Up-to-date parcel-level reference data used in the geocoding process also strengthened confidence in exposure assessment.
Our findings, and those of other studies examining early childhood weight trajectories, have multiple important public health implications. The associations in females are consistent with other risk factors implicated in the “thrifty phenotype” of low birthweight followed by rapid weight gain [2]. This phenotype has been linked to several morbidities in adulthood, including obesity, metabolic syndrome, type 2 diabetes, and cardiovascular disease [14]. Although the differences in weights between PM2.5 levels were small, the ubiquity of air pollution exposure across the population implies that even low levels of PM2.5 may shift population prevalence of excess weight gain over the life course. Differential weight trajectories, as noted here, have been shown to persist into late childhood and adulthood [11]. Our results point to a potentially susceptible period during which introduction of interventions known to promote healthy growth during the early childhood period could reduce the potential impacts of PM2.5 on growth [59].

5. Conclusions

In conclusion, we found in our study that low-birth weight females may be at increased susceptibility for weight gain in early childhood when exposed to higher prenatal PM2.5, with a significant inverse association among males. Studying growth trajectories, rather than attained measures of birth weight and BMI, provides an opportunity to understand susceptible phenotypes and periods of potential interventions. Because of the unique risk patterns found in the CHW population, additional studies are needed in a variety of different study populations and geographies to replicate our findings and to further explore the sex-specific differences found in this study. Future studies should also consider extending the follow-up period through adolescence and adulthood and the implications of specific growth trajectory phenotypes on adult morbidities, such as cardiovascular disease and diabetes.

Supplementary Materials

The following are available online at Table S1. Number of weight observations and subjects within each age range enrolled in the Boston, Massachusetts based Children’s HealthWatch Cohort, 2008–2015; Table S2. Estimates of Akaike Informaiton Criterion (AIC) obtained by mixed models to test optimal growth trajectory fit; Table S3. Observed and predicted weights (kg) in the study population compared to general U.S. population growth standards; Table S4. Estimated weight (kg) by PM2.5 categories, males, stratified by birth weight; Table S5. Estimated weight (kg) by PM2.5 categories, females, stratified by birth weight; Figure S1. Analytical sample selection from linked Children’s HealthWatch survey and electronic medical record data; Figure S2. Cubic regression splines illustrating weight (kg) growth by age generated from generalized additive models.

Author Contributions

Conceptualization, A.S.R., M.P.F., S.E.d.C., M.S., J.I.L., and A.Z.; methodology, A.S.R., M.P.F., M.S., B.A.C., and A.Z.; software, S.C.; validation, A.Z. and J.E.H.; formal analysis, A.S.R.; investigation, A.S.R., M.P.F., M.S., S.E.d.C., and A.Z.; resources, S.C. and M.S.; data curation, S.C.; writing—original draft preparation, A.S.R.; writing—review and editing, M.P.F., S.E.d.C., M.S., S.C., J.I.L., B.A.C., J.E.H., and A.Z.; visualization, A.S.R..; supervision, M.P.F., J.I.L., J.E.H., and A.Z.; project administration, S.E.d.C., M.S., and A.Z.; funding acquisition, J.I.L. All authors have read and agreed to the published version of the manuscript.


This research was funded by National Institutes of Health grants (T32 ES014562, P50 MD010428, P30 ES000002, and R01 ES024332); and Environmental Protection Agency grants (RD83615601 and RD83587201). Its contents are solely the responsibility of the grantee and do not necessarily represent the official views of the USEPA. Further, the USEPA does not endorse the purchase of any commercial products or services mentioned in the publication.


The authors thank the families and children who participated in the Children’s HealthWatch survey. The authors thank Joel Schwartz, from the Harvard T.H. Chan School of Public Health for providing exposure data, Howard Cabral for providing statistical advice, and Jacopo Vanoli and Kevin Griffith for coding assistance. We would like to thank Linda Rosen at the Boston Medical Center Clinical Data Warehouse for her invaluable assistance in clinical data collection and organization. We thank Na Wang at the Biostatistics and Epidemiology Data Analytics Center.

Conflicts of Interest

The authors declare no conflict of interest.


  1. Baird, J.; Fisher, D.; Lucas, P.; Kleijnen, J.; Roberts, H.; Law, C. Being big or growing fast: Systematic review of size and growth in infancy and later obesity. BMJ 2005, 331, 929. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  2. Stettler, N.; Zemel, B.S.; Kumanyika, S.; Stallings, V.A. Infant Weight Gain and Childhood Overweight Status in a Multicenter, Cohort Study. Pediatrics 2002, 109, 194–199. [Google Scholar] [CrossRef] [PubMed]
  3. Barker, D.J.P.; Osmond, C.; Forsén, T.J.; Kajantie, E.; Eriksson, J.G. Trajectories of growth among children who have coronary events as adults. N. Engl. J. Med. 2005, 353, 1802–1809. [Google Scholar] [CrossRef] [PubMed]
  4. Barker, D.; Eriksson, J.; Forsen, T.; Osmond, C. Fetal origins of adult disease: Strength of effects and biological basis. Int. J. Epidemiol. 2002, 31, 1235–1239. [Google Scholar] [CrossRef] [Green Version]
  5. Dennison, B.A.; Edmunds, L.S.; Stratton, H.H.; Pruzek, R.M. Rapid infant weight gain predicts childhood overweight. Obesity (Silver Spring) 2006, 14, 491–499. [Google Scholar] [CrossRef]
  6. Matthews, E.; Wei, J.; Cunningham, S. Relationship between prenatal growth, postnatal growth and childhood obesity: A review. Eur. J. Clin. Nutr. Adv. 2017, 1, 919–930. [Google Scholar] [CrossRef] [Green Version]
  7. Reilly, J.J.; Armstrong, J.; Dorosty, A.R.; Emmett, P.M.; Ness, A.; Rogers, I.; Steer, C.; Sherriff, A. Early life risk factors for obesity in childhood: Cohort study. Br. Med. J. 2005, 330, 1357. [Google Scholar] [CrossRef] [Green Version]
  8. The GBD 2015 Obesity Collaborators Health; Afshin, A.; Forouzanfar, M.H.; Reitsma, M.B.; Sur, P.; Estep, K.; Lee, A.; Marczak, L.; Mokdad, A.H.; Moradi-Lakeh, M.; et al. Effects of Overweight and Obesity in 195 Countries over 25 Years. N. Engl. J. Med. 2017, 377, 13–27. [Google Scholar] [CrossRef]
  9. Linabery, A.M.; Nahhas, R.W.; Johnson, W.; Choh, A.C.; Towne, B.; Odegaard, A.O.; Czerwinski, S.A.; Demerath, E.W. Stronger influence of maternal than paternal obesity on infant and early childhood body mass index: The Fels Longitudinal Study. Pediatr. Obes. 2013, 8, 159–169. [Google Scholar] [CrossRef] [Green Version]
  10. Parsons, T.J.; Power, C.; Manor, O. Fetal and early life growth and body mass index from birth to early adulthood in 1958 British cohort: Longitudinal study. BMJ 2001, 323, 1331–1335. [Google Scholar] [CrossRef] [Green Version]
  11. Giles, L.C.; Whitrow, M.J.; Davies, M.J.; Davies, C.E.; Rumbold, A.R.; Moore, V.M. Growth trajectories in early childhood, their relationship with antenatal and postnatal factors, and development of obesity by age 9 years: Results from an Australian birth cohort study. Int. J. Obes. 2015, 39, 1049–1056. [Google Scholar] [CrossRef] [PubMed]
  12. Gillman, M.W. Developmental Origins of Health and Disease. N. Engl. J. Med. 2005, 353, 1848–1850. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  13. Hales, C.N.; Barker, D.J. The thrifty phenotype hypothesis. Br. Med. Bull. 2001, 60, 5–20. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  14. Vaag, A.A.; Grunnet, L.G.; Arora, G.P.; Brøns, C. The thrifty phenotype hypothesis revisited. Diabetologia 2012, 55, 2085–2088. [Google Scholar] [CrossRef] [Green Version]
  15. Jerrett, M.; McConnell, R.; Wolch, J.; Chang, R.; Lam, C.; Dunton, G.; Gilliland, F.; Lurmann, F.; Islam, T.; Berhane, K. Traffic-related air pollution and obesity formation in children: A longitudinal, multilevel analysis. Environ. Health 2014, 13, 49. [Google Scholar] [CrossRef] [Green Version]
  16. Zheng, T.; Zhang, J.; Sommer, K.; Bassig, B.A.; Zhang, X.; Braun, J.; Xu, S.; Boyle, P.; Zhang, B.; Shi, K.; et al. Effects of environmental exposures on fetal and childhood growth trajectories. Ann. Glob. Health 2016, 82, 41–99. [Google Scholar] [CrossRef] [Green Version]
  17. Malmqvist, E.; Liew, Z.; Källén, K.; Rignell-Hydbom, A.; Rittner, R.; Rylander, L.; Ritz, B. Fetal growth and air pollution -A study on ultrasound and birth measures. Environ. Res. 2017, 152, 73–80. [Google Scholar] [CrossRef] [Green Version]
  18. Mao, G.; Nachman, R.M.; Sun, Q.; Zhang, X.; Koehler, K.; Chen, Z.; Hong, X.; Wang, G.; Caruso, D.; Zong, G.; et al. Individual and Joint Effects of Early-Life Ambient PM2.5 Exposure and Maternal Pre-Pregnancy Obesity on Childhood Overweight or Obesity. Environ. Health Perspect. 2016, 125, 067005. [Google Scholar] [CrossRef]
  19. Fleisch, A.F.; Rifas-Shiman, S.L.; Koutrakis, P.; Schwartz, J.D.; Kloog, I.; Melly, S.J.; Coull, B.A.; Zanobetti, A.; Gillman, M.W.; Gold, D.R.; et al. Prenatal Exposure to Traffic Pollution: Associations with Reduced Fetal Growth and Rapid Infant Weight Gain. Epidemiology 2015, 26, 43–50. [Google Scholar] [CrossRef]
  20. McConnell, R.; Shen, E.; Gilliland, F.D.; Jerrett, M.; Wolch, J.; Chih-Chieh, C.; Lurmann, F.; Berhane, K. A Longitudinal Cohort Study of Body Mass Index and Childhood Exposure to Secondhand Tobacco Smoke and Air Pollution: The Southern California Children’s Health Study. Environ. Health Perspect. 2015, 123, 360–366. [Google Scholar] [CrossRef] [Green Version]
  21. Fleisch, A.F.; Aris, I.M.; Rifas-Shiman, S.L.; Coull, B.A.; Luttmann-Gibson, H.; Koutrakis, P.; Schwartz, J.D.; Kloog, I.; Gold, D.R.; Oken, E. Prenatal Exposure to Traffic Pollution and Childhood Body Mass Index Trajectory. Front. Endocrinol. 2019, 9. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  22. Kim, J.S.; Alderete, T.L.; Chen, Z.; Lurmann, F.; Rappaport, E.; Habre, R.; Berhane, K.; Gilliland, F.D. Longitudinal associations of in utero and early life near-roadway air pollution with trajectories of childhood body mass index. Environ. Health 2018, 17. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  23. Cutts, D.B.; Meyers, A.F.; Black, M.M.; Casey, P.H.; Chilton, M.; Cook, J.T.; Geppert, J.; De Cuba, S.E.; Heeren, T.; Coleman, S.; et al. US housing insecurity and the health of very young children. Am. J. Public Health 2011, 101, 1508–1514. [Google Scholar] [CrossRef] [PubMed]
  24. Centers for Disease Control Growth Charts -WHO and CDC Child Growth Standards. Available online: (accessed on 14 October 2017).
  25. MassGIS MassGIS Data -Master Address Data. Available online: (accessed on 23 September 2017).
  26. Kloog, I.; Chudnovsky, A.A.; Just, A.C.; Nordio, F.; Koutrakis, P.; Coull, B.A.; Lyapustin, A.; Wang, Y.; Schwartz, J. A new hybrid spatio-temporal model for estimating daily multi-year PM2.5 concentrations across northeastern USA using high resolution aerosol optical depth data. Atmos. Environ. 2014, 95, 581–590. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  27. Frank, D.A.; Casey, P.H.; Black, M.M.; Rose-Jacobs, R.; Chilton, M.M.; Cutts, D.B.; March, E.; Heeren, T.; Coleman, S.; Ettinger de Cuba, S.; et al. Cumulative Hardship and Wellness of Low-Income, Young Children: Multisite Surveillance Study. Pediatrics 2010, 125, e1115–e1123. [Google Scholar] [CrossRef] [Green Version]
  28. Lakshmanan, A.; Chiu, Y.H.M.; Coull, B.A.; Just, A.C.; Maxwell, S.L.; Schwartz, J.; Gryparis, A.; Kloog, I.; Wright, R.J.; Wright, R.O. Associations between prenatal traffic-related air pollution exposure and birth weight: Modification by sex and maternal pre-pregnancy body mass index. Environ. Res. 2015, 137, 268–277. [Google Scholar] [CrossRef] [Green Version]
  29. Clougherty, J.E. A growing role for gender analysis in air pollution epidemiology. Environ. Health Perspect. 2010, 118, 167–176. [Google Scholar] [CrossRef] [Green Version]
  30. Fitzmaurice, G.M.; Laird, N.M.; Ware, J.H. Applied Longitudinal Analysis, 2nd ed.; Wiley: Hoboken, NJ, USA, 2011; ISBN 0470380276. [Google Scholar]
  31. Howe, L.D.; Tilling, K.; Matijasevich, A.; Petherick, E.S.; Santos, A.C.; Fairley, L.; Wright, J.; Santos, I.S.; Barros, A.J.; Martin, R.M.; et al. Linear spline multilevel models for summarising childhood growth trajectories: A guide to their application using examples from five birth cohorts. Stat. Methods Med. Res. 2016, 25, 1854–1874. [Google Scholar] [CrossRef] [Green Version]
  32. Patel, R.; Tilling, K.; Lawlor, D.A.; Howe, L.D.; Bogdanovich, N.; Matush, L.; Nicoli, E.; Kramer, M.S.; Martin, R.M. Socioeconomic differences in childhood length/height trajectories in a middle-income country: A cohort study. BMC Public Health 2014, 14, 932. [Google Scholar] [CrossRef] [Green Version]
  33. Tilling, K.; MacDonald-Wallis, C.; Lawlor, D.A.; Hughes, R.A.; Howe, L.D. Modelling childhood growth using fractional polynomials and linear splines. Ann. Nutr. Metab. 2014, 65, 129–138. [Google Scholar] [CrossRef]
  34. O’Keeffe, L.M.; Kearney, P.M.; Greene, R.A.; Zuccolo, L.; Tilling, K.; Lawlor, D.A.; Howe, L.D. Maternal alcohol use during pregnancy and offspring trajectories of height and weight: A prospective cohort study. Drug Alcohol Depend. 2015, 153, 323–329. [Google Scholar] [CrossRef] [PubMed]
  35. Chirwa, E.D.; Griffiths, P.L.; Maleta, K.; Norris, S.A.; Cameron, N. Multi-level modelling of longitudinal child growth data from the Birth-to-Twenty Cohort: A comparison of growth models. Ann. Hum. Biol. 2014, 41, 168–179. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  36. Lourenço, B.H.; Villamor, E.; Augusto, R.A.; Cardoso, M.A. Determinants of linear growth from infancy to school-aged years: A population-based follow-up study in urban Amazonian children. BMC Public Health 2012, 12, 265. [Google Scholar] [CrossRef] [PubMed]
  37. Grajeda, L.M.; Ivanescu, A.; Saito, M.; Crainiceanu, C.; Jaganath, D.; Gilman, R.H.; Crabtree, J.E.; Kelleher, D.; Cabrera, L.; Cama, V.; et al. Modelling subject-specific childhood growth using linear mixed-effect models with cubic regression splines. Emerg. Themes Epidemiol. 2016, 13, 1. [Google Scholar] [CrossRef] [Green Version]
  38. Wood, S.N. Generalized Additive Models, 2nd ed.; Chapman and Hall/CRC: New York, NY, USA, 2017; ISBN 9781315370279. [Google Scholar]
  39. UNICEF; WHO. Low Birthweight: Country, Regional and Global Estimates; UNICEF: New York, NY, USA, 2004. [Google Scholar]
  40. Reddy, U.M.; Bettegowda, V.R.; Dias, T.; Yamada-Kushnir, T.; Ko, C.-W.; Willinger, M. Term pregnancy: A period of heterogeneous risk for infant mortality. Obstet. Gynecol. 2011, 117, 1279–1287. [Google Scholar] [CrossRef]
  41. Ogden, C.L.; Kuczmarski, R.J.; Flegal, K.M.; Mei, Z.; Guo, S.; Wei, R.; Grummer-Strawn, L.M.; Curtin, L.R.; Roche, A.F.; Johnson, C.L. Centers for Disease Control and Prevention 2000 growth charts for the United States: Improvements to the 1977 National Center for Health Statistics version. Pediatrics 2002, 109, 45–60. [Google Scholar] [CrossRef] [Green Version]
  42. Kannan, S.; Misra, D.P.; Dvonch, J.T.; Krishnakumar, A. Exposures to airborne particulate matter and adverse perinatal outcomes: A biologically plausible mechanistic framework for exploring potential effect modification by nutrition responses to PM exposures. Environ. Health Perspect. 2006, 114, 1636–1642. [Google Scholar] [CrossRef]
  43. Janssen, B.G.; Janssen, B.G.; Byun, H.; Gyselaers, W.; Lefebvre, W.; Andrea, A. Placental mitochondrial methylation and exposure to airborne particulate matter in the early life environment: An ENVIRONAGE birth cohort study to airborne particulate matter in the early life. Epigenetics 2016, 10, 536–544. [Google Scholar] [CrossRef] [Green Version]
  44. de melo, J.O.; Soto, S.F.; Katayama, I.A.; Wenceslau, C.F.; Pires, A.G.; Veras, M.M.; Furukawa, L.N.S.; de Castro, I.; Saldiva, P.H.N.; Heimann, J.C. Inhalation of fine particulate matter during pregnancy increased IL-4 cytokine levels in the fetal portion of the placenta. Toxicol. Lett. 2015, 232, 475–480. [Google Scholar] [CrossRef]
  45. Ghosh, R.; Rankin, J.; Pless-Mulloli, T.; Glinianaia, S. Does the effect of air pollution on pregnancy outcomes differ by gender? A systematic review. Environ. Res. 2007, 105, 400–408. [Google Scholar] [CrossRef]
  46. Ebisu, K.; Bell, M.L. Airborne PM2.5 chemical components and low birth weight in the northeastern and Mid-Atlantic regions of the United States. Environ. Health Perspect. 2012, 120, 1746–1752. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  47. Jedrychowski, W.; Perera, F.; Mrozek-Budzyn, D.; Mroz, E.; Flak, E.; Spengler, J.D.; Edwards, S.; Jacek, R.; Kaim, I.; Skolicki, Z. Gender differences in fetal growth of newborns exposed prenatally to airborne fine particulate matter. Environ. Res. 2009, 109, 447–456. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  48. Chiu, Y.H.; Hsu, H.H.; Wilson, A.; Coull, B.A.; Pendo, M.P.; Baccarelli, A.; Kloog, I.; Schwartz, J.; Wright, R.O.; Taveras, E.M.; et al. Prenatal particulate air pollution exposure and body composition in urban preschool children: Examining sensitive windows and sex-specific associations. Environ. Res. 2017, 158, 798–805. [Google Scholar] [CrossRef] [PubMed]
  49. Chen, M.; Wang, X.; Hu, Z.; Zhou, H.; Xu, Y.; Qiu, L.; Qin, X. Programming of mouse obesity by maternal exposure to concentrated ambient fine particles. Part. Fibre Toxicol. 2017, 1–13. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  50. Calderón-Garcidueñas, L.; Torres-Jardón, R.; Kulesza, R.J.; Park, S.-B.; D’Angiulli, A. Air pollution and detrimental effects on children’s brain. The need for a multidisciplinary approach to the issue complexity and challenges. Front. Hum. Neurosci. 2014, 8, 613. [Google Scholar] [PubMed] [Green Version]
  51. Mauvais-Jarvis, F. Sex differences in metabolic homeostasis, diabetes, and obesity. Biol. Sex Differ. 2015, 6, 14. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  52. Fleisch, A.F.; Luttmann-Gibson, H.; Perng, W.; Rifas-Shiman, S.L.; Coull, B.A.; Kloog, I.; Koutrakis, P.; Schwartz, J.D.; Zanobetti, A.; Mantzoros, C.S.; et al. Prenatal and early life exposure to traffic pollution and cardiometabolic health in childhood. Pediatr. Obes. 2016, 12, 48–57. [Google Scholar] [CrossRef] [Green Version]
  53. Rhee, J.; Patricia Fabian, M.; de Cuba, S.E.; Coleman, S.; Sandel, M.; Lane, K.J.; Sade, M.Y.; Hart, J.E.; Schwartz, J.; Kloog, I.; et al. Effects of maternal homelessness, supplemental nutrition programs, and prenatal PM2.5 on birthweight. Int. J. Environ. Res. Public Health 2019, 16, 4154. [Google Scholar] [CrossRef] [Green Version]
  54. Cunningham, S.A.; Datar, A.; Narayan, K.M.V.; Kramer, M.R. Entrenched obesity in childhood: Findings from a national cohort study. Ann. Epidemiol. 2017, 27, 435–441. [Google Scholar] [CrossRef]
  55. Suzuki, K.; Sato, M.; Zheng, W.; Shinohara, R.; Yokomichi, H.; Yamagata, Z. Childhood growth trajectories according to combinations of pregestational weight status and maternal smoking during pregnancy: A multilevel analysis. PLoS ONE 2015, 10, 1–12. [Google Scholar] [CrossRef] [Green Version]
  56. Jerrett, M.; McConnell, R.; Chang, C.C.R.; Wolch, J.; Reynolds, K.; Lurmann, F.; Gilliland, F.; Berhane, K. Automobile Traffic around the Home and Attained Body Mass Index: A Longitudinal Cohort Study of Children aged 10–18 Years. Prev. Med. 2010, 50, S50–S58. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  57. Smith, N.; Coleman, K.J.; Lawrence, J.M.; Quinn, V.P.; Getahun, D.; Reynolds, K.; Chen, W.; Porter, A.H.; Jacobsen, S.J.; Koebnick, C. Body weight and height data in electronic medical records of children. Int. J. Pediatr. Obes. 2010, 5, 237–242. [Google Scholar] [CrossRef] [PubMed]
  58. Casey, J.A.; Schwartz, B.S.; Stewart, W.F.; Adler, N.E. Using Electronic Health Records for Population Health Research: A Review of Methods and Applications. Annu. Rev. Public Health 2016, 37, 61–81. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  59. Braun, K.V.; Erler, N.S.; Kiefte-de Jong, J.C.; Jaddoe, V.W.V.; van den Hooven, E.H.; Franco, O.H.; Voortman, T. Dietary Intake of Protein in Early Childhood Is Associated with Growth Trajectories between 1 and 9 Years of Age. J. Nutr. 2016, 146, 2361–2367. [Google Scholar] [CrossRef]
Figure 1. Data sources linked to create final analytical dataset for growth trajectory analysis. CHW: Children’s HealthWatch; PM2.5: particulate matter with an aerodynamic diameter of 2.5 microns.
Figure 1. Data sources linked to create final analytical dataset for growth trajectory analysis. CHW: Children’s HealthWatch; PM2.5: particulate matter with an aerodynamic diameter of 2.5 microns.
Ijerph 17 01444 g001
Figure 2. Predicted weight (kg) over age (months) by levels of average prenatal PM2.5. Note: Models adjusted for: age, age2, age3, quadratic spline terms at 6 and 12 months, gestational age, ethnicity, education, U.S.-born, cumulative hardship, and block group median income.
Figure 2. Predicted weight (kg) over age (months) by levels of average prenatal PM2.5. Note: Models adjusted for: age, age2, age3, quadratic spline terms at 6 and 12 months, gestational age, ethnicity, education, U.S.-born, cumulative hardship, and block group median income.
Ijerph 17 01444 g002
Figure 3. Predicted weight (kg) over age (months) by levels of average prenatal PM2.5, stratified by birth weight categories for (a) males and (b) females. Note: Models adjusted for: age, age2, age3, quadratic spline terms at 6 and 12 months, gestational age, ethnicity, education, U.S.-born, cumulative risk, and block group median income
Figure 3. Predicted weight (kg) over age (months) by levels of average prenatal PM2.5, stratified by birth weight categories for (a) males and (b) females. Note: Models adjusted for: age, age2, age3, quadratic spline terms at 6 and 12 months, gestational age, ethnicity, education, U.S.-born, cumulative risk, and block group median income
Ijerph 17 01444 g003
Table 1. Characteristics of 4797 children and their caregivers enrolled in the Boston, Massachusetts based Children’s HealthWatch Cohort, 2008–2015.
Table 1. Characteristics of 4797 children and their caregivers enrolled in the Boston, Massachusetts based Children’s HealthWatch Cohort, 2008–2015.
Child Characteristics
Males (n (%))Females (n (%))
Total2603 (100)2194 (100)
Birth weight (g) (% missing: 1.6 males, 1.1 females)
<2500292 (11.4)289 (13.3)
>=25002269 (88.6)1881 (86.7)
Gestational age (% missing: 0.4 males, 0.4 females)
<37 weeks411 (15.9)309 (14.1)
>= 37 weeks2181 (84.1)1876 (85.9)
Breastfed During Pregnancy (% missing: 0.8 males, 0.6 females)
Yes1999 (77.4)1688 (77.4)
No584 (22.6)492 (22.6)
Cumulative hardship (% missing: 12.7 males, 11.6 females) a
0 hardships806 (35.5)655 (33.8)
1–3 hardships1288 (56.7)1121 (57.8)
>3 hardships179 (7.9)163 (8.4)
Number of overall visits (inpatient and outpatient)
mean ± SD14.8 ± 14.914.5 ± 14.1
Block group median income ($)
mean ± SD43,792.4 ± 22,003.743,442.3 ± 22,424.0
Self-reported caregiver characteristics
Marital status (% missing: 0.5 males, 0.4 females)
Married933 (36.0)782 (35.8)
Not married1657 (64.0)1403 (64.2)
Ethnicity (% missing: 1.2 males, 1.0 females)
Hispanic919 (35.7)766 (35.3)
Black, non-Hispanic1294 (50.3)1099 (50.6)
White, non-Hispanic222 (8.6)191 (8.8)
Other137 (5.3)116 (5.3)
Education (% missing: 0.5 males, 0.3 females)
Less than high school611 (23.6)524 (24.0)
High school graduate853 (33.0)723 (33.0)
Post-secondary 1125 (43.5)941 (43.0)
Country of birth b (% missing: 0.8 males, 0.3 females)
U.S.-born1484 (57.5)1266 (57.9)
Not U.S.-born1099 (42.6)921 (42.1)
Smoked in last 5 years (% missing: 4.1 males, 2.9 females)
Yes618 (24.8)562 (26.4)
No1879 (75.3)1568 (73.6)
Age at child’s birth
Mean ± SD26.8 ± 6.327.0 ± 6.3
Note: SD, standard deviation; a Refers to biological mother; b Score derived from questions about housing, energy, and food insecurity.
Table 2. Mean predicted weight (kg) by prenatal PM2.5 category.
Table 2. Mean predicted weight (kg) by prenatal PM2.5 category.
Prenatal PM2.5 GroupBirth3 Months6 Months12 Months18 Months24 Months36 Months48 Months60 Months72 Months
Males (n = 2244; weight measurements = 32,405)
<9.5 µg/m3 a3.006.17 8.09 10.25 11.86 13.3315.98 18.55 21.44 24.99
(2.94, 3.06)(6.11, 6.22)(8.03, 8.15)(10.17, 10.32)(11.78, 11.96)(13.22, 13.44)(15.81, 16.14)(18.34, 18.78)(21.16, 21.72)(24.62, 25.36)
≥9.5 µg/m3 a3.026.17 8.1410.29 11.77 13.1615.7118.16 20.72 23.60
(2.95, 3.09)(6.11, 6.23)(8.07, 8.21)(10.21, 10.37)(11.68, 11.87) (13.04, 13.28) (15.54, 15.88)(17.94, 18.38)(20.44, 21.00)(23.25, 23.94)
p-value b0.700.990.300.410.<0.00001
Females (n = 1931; weight measurements = 27,148)
<9.5 µg/m3 a2.95 5.58 7.36 9.52 11.05 12.48 15.11 17.71 20.52 23.79
(2.88, 3.02)(5.52, 5.64)(7.30, 7.42)(9.45, 9.60)(10.96, 11.15)(12.35, 12.60)(14.93, 15.30)(17.45, 17.97)(20.19, 20.83)(23.35, 24.23)
≥9.5 µg/m3 a2.98 5.71 7.55 9.70 11.21 12.64 15.41 18.18 21.13 24.42
(2.91, 3.06)(5.64, 5.78)(7.48, 7.61)(9.62, 9.79)(11.11, 11.30)(12.51, 12.76)(15.22, 15.58)(17.93, 18.41)(20.83, 21.43)(24.05, 24.81)
p-value b0.500.0060.00010.0010.
Note: All estimates are from polynomial mixed models adjusted for: age, age2, age3, quadratic spline terms at 6 and 12 months, gestational age, ethnicity, education, U.S.-born, cumulative risk and block group median income; ∆ = absolute difference in weight between low and high exposure categories (kg). a Values are mean estimated weights in kg (95% CIs); b p-values for difference between low and high exposure categories.

Share and Cite

MDPI and ACS Style

Rosofsky, A.S.; Fabian, M.P.; Ettinger de Cuba, S.; Sandel, M.; Coleman, S.; Levy, J.I.; Coull, B.A.; Hart, J.E.; Zanobetti, A. Prenatal Ambient Particulate Matter Exposure and Longitudinal Weight Growth Trajectories in Early Childhood. Int. J. Environ. Res. Public Health 2020, 17, 1444.

AMA Style

Rosofsky AS, Fabian MP, Ettinger de Cuba S, Sandel M, Coleman S, Levy JI, Coull BA, Hart JE, Zanobetti A. Prenatal Ambient Particulate Matter Exposure and Longitudinal Weight Growth Trajectories in Early Childhood. International Journal of Environmental Research and Public Health. 2020; 17(4):1444.

Chicago/Turabian Style

Rosofsky, Anna S., M. Patricia Fabian, Stephanie Ettinger de Cuba, Megan Sandel, Sharon Coleman, Jonathan I. Levy, Brent A. Coull, Jaime E. Hart, and Antonella Zanobetti. 2020. "Prenatal Ambient Particulate Matter Exposure and Longitudinal Weight Growth Trajectories in Early Childhood" International Journal of Environmental Research and Public Health 17, no. 4: 1444.

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop