Ambient Airborne Particulates of Diameter ≤1 μm, a Leading Contributor to the Association Between Ambient Airborne Particulates of Diameter ≤2.5 μm and Children’s Blood Pressure

This study was nested within the larger SNEC study, a cross-sectional study in Liaoning Province, China, that investigated the relationship between air pollution exposure and children’s health. The study population and design have been fully described elsewhere.³⁴ Briefly, from April 2012 to June 2013, we selected 7 of 14 cities in Liaoning Province as the study sites (Liaoyang, Dalian, Benxi, Anshan, Fushun, Dandong, and Shenyang). We selected these 7 cities in effort to maximize the intercity and intracity air pollution variabilities. We selected all urban districts in each city for the study, including 5 Shenyang districts, 4 districts each from Dalian and Fushun, 3 districts each from Anshan, Benxi, and Dandong, and 2 districts from Liaoyang. Within each geographic district, we used a random selection algorithm to choose one elementary and one middle school, yielding a total of 62 schools for the study. We then chose 1 or 2 classrooms randomly from each grade in the selected schools. Children residing in the study district for at least 2 years were eligible to participate in this study (Figure S1 in the online-only Data Supplement). The study protocol complied with the World Medical Association Declaration of Helsinki-Ethical Principles for Medical Research Involving Human Subjects and was approved by the Human Studies Committee of Sun Yat-sen University. Consent was obtained from the parent/guardian of each participant before data collection.

All study personnel were trained according to the American Academy of Pediatrics protocol for BP measurements and passed a qualifying examination.³⁵ We asked participants not to consume substances that might affect BP (such as alcohol, tea, coffee, and cigarettes) and not to exercise for at least 30 minutes. Participants rested for 5 minutes in a quiet temperate-controlled room, after which trained nurses used a standardized mercury sphygmomanometer to measure BP on the upper right arm in a sitting position, according to a standardized procedure. An appropriately sized cuff was selected for each child, which was placed ≈2 cm above the crease of the right elbow. The BP measurement was performed in 3 successive pairs at 2-minute intervals. Systolic BP (SBP) was determined by the onset of the Korotkoff sound (K1), and the fifth Korotkoff sound (K5) determined diastolic BP (DBP). The average of these 3 BP values was used for data analysis. We defined hypertension as SBP or DBP ≥95th percentile for sex, age, and height of children based on the framework of The Fourth Report on the Diagnosis, Evaluation, and Treatment of High Blood Pressure in Children and Adolescents.³⁵

We used fixed stadiometers to measure participants’ height (to the nearest 0.1 cm) and electronic scales to measure weight (to the nearest 0.1 kg). During the height and weight measurements, participants were required to wear light garments and to remove their shoes. Body mass index (BMI) was calculated as weight divided by height squared (kg/m²). Overweight was defined as BMI >85th age–sex-specific percentile and obese as BMI >95th age–sex-specific percentile, based on US Centers for Disease Control and Prevention growth charts.³⁶

We operationalized parental education as the highest completed level of education by either parent (high school or higher/middle school or lower). A child was defined as having passive tobacco smoke exposure if a household member smoked cigarettes daily in the home. Annual family income was defined as the Chinese Yuan (RMB) earned. We calculated average weekly exercise as the average hours of physical activity per week. Finally, in house coal use for heating or cooking defined indoor exposure to coal combustion emissions.

Daily PM₁ and PM_2.5 concentrations were estimated at a spatial resolution of 10 km×10 km during 2009 and 2012, based on satellite remote sensing, meteorologic data, and land use information. The details have been previously published.^37,38 In brief, we obtained daily ground PM₁ and PM_2.5 measurements from 77 China Atmosphere Watch Network stations, and we developed a generalized additive model linking the ground monitored PM data with combined Moderate Resolution Imaging Spectroradiometer Collection 6 aerosol optical depth data and other spatiotemporal predictors (eg, meteorologic data, land cover data, vegetation data, active fires data, and elevation data).

We used 10-fold cross-validation to assess the predictive ability and error of the model. The coefficient of determination (R²) and root mean square error was 71% and 13.0 μg/m³ for monthly predicted PM₁, 59% and 22.5 μg/m³ for daily predicted PM₁, 75% and 15.1 μg/m³ for monthly predicted PM_2.5, and 83% and 18.1 μg/m³ for daily predicted PM_2.5, respectively. We geocoded each participant’s home address and superimposed the predicted daily PM₁ and PM_2.5 concentration grids. We then averaged the daily PM₁ and PM_2.5 concentrations for each participant over the 4-year period from 2009 to 2012 to assign exposure.

We characterized distributions of all variables and assessed normality. We assessed crude associations among the covariates using the χ² tests for categorical variables and Spearman rank correlations for continuous variables, including air pollutants.

We used generalized linear mixed models to investigate the associations between PM and BP. We used a 2-level logistic regression model to investigate the associations between ambient air pollution and hypertension, in which children comprised the first level and schools comprised the second level (details in the online-only Data Supplement), as previously described.^38,39 Effects are expressed as odds ratios and their corresponding 95% CI, for a 10 µg/m³ or an interquartile range (ie, IQR=75th% tile to 25th% tile) difference in particulate concentrations.

We constructed 2 adjusted regression models. First, we adjusted only for participant age. Second, we selected covariates for adjustment a priori, based on the literature,^{11,13,15,40–42} as factors antecedent to the air pollution exposure and risk factors for elevated BP. We then retained those covariates that were distributed unequally between exposure groups in our study population, including: age (years), sex (male or female), annual income (≤30 000 RMB or >30 000 RMB), BMI (kg/m²), parental education (high school or higher/middle school or lower), passive tobacco smoke exposure (living with daily cigarette smoking in the home; yes/no), home coal use (yes/no), and exercise time per week (hours) in the second model. All covariates were assigned fixed effects with the exception of schools, which was a random effect.

We evaluated the robustness of our results using sensitivity analyses. First, we estimate the PM-BP association excluding one city at a time in the second adjusted model. We then estimated the PM-BP associations by individually excluding participants with a family history of hypertension, participants exposed to tobacco smoke, and participants exposed to coal combustion emission in house. Furthermore, to assess the impact, we used the lowest of 3 BP values measured in succession for data analysis.

In addition, the PM-BP associations could differ among population subgroups. To test for potential effect modification, we conducted stratified analyses by the median age of the study population (>11 versus ≤11 years), sex (female versus male), family income per year (≤30 000 versus >30 000 RMB), exercise time per week (≤7 versus >7 hours), birthweight status (normal birthweight versus low birthweight), and children’s weight status (normal weight versus overweight and obese). Furthermore, the statistical significance of their cross-product terms entered into the second adjusted model was tested.

We used R software (version 3.6.1, R foundation for Statistical Computing, Vienna, Austria) to analyze the data in R studio (version 1.1.453). We defined the statistical significance of main effects and interactions as P<0.05 for a 2-tailed test.

We randomly selected 10 428 children, aged 5 to 17 years, from the study sites. We excluded 861 children who failed to complete the questionnaire and physical examination or resided for <2 years in the study district (n=213). A total of 9354 participants were included in the final analysis (response rate: 91.7%). Characteristics of participants, stratified by BP status, are demonstrated in Table 1. The mean age of the participants was 10.9 years (SD=2.3 years). Four thousand seven hundred seventy-one (51.0%) were male, 48% were passively exposed to tobacco smoke, and 9.5% were exposed to coal combustion at home. Mean SBP was 111.0 mm Hg and mean DBP was 64.5 mm Hg. A total of 1289 (13.8%) children were classified as hypertensive.

Table 2 summarizes the distributions of air pollutants in 24 districts from 2009 to 2012 and their pairwise correlations. The median (interquartile range) of PM₁ and PM_2.5 concentration was 44.36 μg/m³ (13.24 μg/m³), and 51.85 μg/m³ (9.98 μg/m³), respectively. PM₁ levels were positively correlated with PM_2.5 levels (r_s=0.91). The ratio of PM₁/PM_2.5 in this study ranged from 0.80 to 0.96.

In the second adjusted model, we found that PM₁ was positively associated with SBP and hypertension (Table 3). Each 10 μg/m³ increment of PM₁ concentration was significantly associated with a 2.56 mm Hg (95% CI, 1.47–3.65) higher SBP and 61% greater odds of prevalent hypertension (odds ratio, 1.61 [95% CI, 1.18–2.18]). Similar associations were also observed for PM_2.5, albeit with smaller effect sizes, but no significant association was detected for PM_1–2.5 (Table S8). The above associations were not substantially different when we excluded one city at a time (Tables S2 and S3), when we excluded participants with a family history of hypertension (Table S4), when we excluded participants exposed to tobacco smoke (Table S5), when we excluded participants exposed to indoor coal combustion emissions in the home (Table S6), when we used the lowest of 3 BP values measured in succession for data analysis (Table S7), when we incorporated one additional pollutant (PM₁₀, SO₂, NO₂, or O₃) into the adjusted model (data not shown), or when we repeated our second adjusted model using BMI Z-score in lieu of BMI (data not shown). In addition, each 0.01 increment in the PM₁/PM_2.5 ratio was significantly associated with a 0.36 mm Hg (95% CI, 0.18–0.54) higher SBP and 0.24 mm Hg (95% CI, 0.06–0.42) higher DBP and 7% greater odds of prevalent hypertension (odds ratio, 1.07; 95% CI, 1.02–1.12). When we estimated the relationship of SBP, DBP, and hypertension with quartiles of PM₁ and PM_2.5 exposure (Figure), we found significant trends in SBP and hypertension, where the adjusted β coefficients and odds ratios increased with increasing PM quartiles.

In the stratified analysis, we found statistically significant interactions with age group (P<0.01). There were stronger associations in children aged ≤11 years than in those aged >11 years. For example, each 10 μg/m³ increment of PM₁ concentration was associated with a 4.36 mm Hg (95% CI, 2.78–5.94) increase in SBP among children ≤11 years of age, while the increase was 2.39 mm Hg (95% CI, 0.85–3.93) among those aged >11 years of age (Table 4). Similar relationships were observed for PM_2.5 (Table S1). Significant effect modification by weight was detected in SBP, with stronger associations among overweight and obese children. No statistically significant modifying effects were present for sex, income, exercise time, or birthweight for either PM₁ or PM_2.5 (Table 4; Table S1 in the online-only Data Supplement).

In this first study of long-term PM₁ exposure and children’s BP, we found significant associations between greater PM₁ exposure and higher SBP and odds of hypertension. We detected analogous associations for PM_2.5, but no significant associations were observed for PM_1–2.5. In addition, we found that younger children (≤11 years) and overweight/obese children may be more susceptible to the hypertensive effects of PM₁ and PM_2.5 exposure.

No prior study has investigated the associations between ambient PM₁ pollution and childhood BP, but 2 epidemiological studies have evaluated this question in adults. A study of 63 Australian adults assessed short-term exposure to indoor PM₁ exposure, although no significant association was detected.⁴³ In our prior work, we found that a 10-μg/m³ greater PM₁ concentration increment was associated with a 0.57 mm Hg higher SBP and 0.19 mm Hg higher DBP, as well as 5% higher odds of hypertension among Chinese 24 845 adults.²² In addition, some studies have also explored associations between smaller ultrafine particles (<0.1 μm) and BP. The results of these studies generally supported positive associations between PM_0.1 and BP.^44–53 Nevertheless, evidence on long-term ambient PM₁ exposure and BP in children remains scarce, and so further investigation is required to confirm the associations.

Another important finding is the similar associations between long-term PM₁ and PM_2.5 exposure with children’s BP but without an association for PM_1–2.5. The ambient PM₁/PM_2.5 ratio was also positively associated with BP. Our study sites are highly urbanized and industrialized, with substantial combustion emissions leading to more PM₁ generation than PM_1–2.5 generation (PM₁/PM_2.5 ratios >80%).²⁰ Thus, we speculate that PM₁ might be responsible for the observed associations between PM_2.5 exposure and BP in Chinese children. Many developed countries might experience a similar situation, as the mostly motor vehicle traffic-related air pollution contribute more PM₁ than PM_2.5.^19,20 However, evidence on this topic is scarce in developed countries. Similarly, in our previous study of Chinese adults, the associations of PM_2.5 could be largely attributed to PM₁.²² Our findings on the relative effect of different PM sizes are also consistent with several previous Chinese studies of PM exposure and other health outcomes.^21,23,54,55 Specifically, Hu et al²³ observed that PM₁ had greater impacts on mortality from cardiovascular and respiratory diseases than PM_2.5 or PM₁₀ in province of Zhejiang, China. They speculated that PM_2.5 and PM₁₀ related deaths were mainly attributable to PM₁. Similarly, Lin et al²¹ found that PM₁ had a stronger effect on cardiovascular mortality than PM_2.5 or PM₁₀ in Guangzhou City, China. In another large epidemiological study, greater short-term ambient PM₁ and PM_2.5 exposures were significantly associated with a greater frequency of emergency hospital visits, with most of the PM_2.5 association having been attributed to PM₁.⁵⁵ Additionally, we recently found very similar associations for greater exposures to PM₁ and PM_2.5 with higher odds of asthma-related symptoms among children living in northeastern China.⁵⁴ Although the previously published literature focused on health outcomes other than BP and hypertension, it provides indirect support for our hypothesis that the observed PM_2.5 associations may be attributable to PM₁.

We detected stronger associations between BP and ambient PM₁ and PM_2.5 exposure among younger children (aged ≤11 years) and those with abnormal weight (ie, overweight or obese). To our knowledge, no prior studies have examined the modifying effects of age on the PM-BP association in children. Children experience vigorous growth and development, with greater rates of cellular mitosis than for older children or adults, comprising an important window of vulnerability.⁵⁶ Most (≈80%) alveoli develop postnatally and the lung continues to develop throughout adolescence.⁵⁶ Immature metabolic pathways may further enhance vulnerability to environmental insults.²⁵ In addition, children tend to be very physically active, leading to much greater air intake into less developed lungs relative to older children and adults,⁵⁷ further increasing younger children’s vulnerability to air pollution relative to older children.

There is little evidence available on synergistic effects for body weight and air pollution exposure on children’s BP. However, investigators have hypothesized that overweight/obesity and air pollution exposure can induce systemic inflammation, which is associated with the cause of high BP.^6,40 We previously reported stronger associations of ambient PM₁₀ and O₃ exposure and BP and hypertension among overweight/obese children participating in the SNEC study.⁴⁰ Similarly, Li et al⁴¹ also observed a synergistic effect of overweight and PM₁₀ on children’s BP in China. However, Zhang et al⁵⁸ did not observe modification by overweight and obesity on associations between PM_2.5 and PM₁₀ with BP and hypertension in Chinese children and adolescents. Although our results provide new evidence in support of body weight as a modifier of air pollution-BP/hypertension associations in Chinese children, additional studies are necessary to more definitively characterize the potential interactions between air pollution and overweight/obesity on BP/hypertension.

The underlying mechanisms of the relationship between PM exposure and hypertension remain unclear. It is hypothesized that in the short-term, PM causes autonomic nervous system imbalances and direct local oxidative/inflammatory vascular actions and that these changes combine with chronic secondary induced systemic inflammation and oxidative stress in the long-term to potentially elevate BP.^7,9,59–61 PM diameter is an important determinant of health effects, as finer PM may enter the blood stream directly leading to secondary proinflammatory responses.^62–64 PM₁ is smaller in size than PM_2.5. It penetrates deeper than PM_2.5 and deposits into the smaller branches of the respiratory tree, even penetrating down to the air-blood barrier where it enters the circulatory system.^65,66 In addition, the surface area to mass ratio is larger for PM₁ than for PM_2.5, and hence a greater surface area is available for adsorption of toxic agents,⁶⁷ which are able to elicit localized oxidative and inflammatory reactions directly.⁶⁸ Therefore, PM₁ may have a dominant contribution to PM-BP effects, especially considering that PM₁ constitutes a large proportion in PM.

This study has several strengths. First, the large sample size (n=9354) and high response rate (91.7%) are strengths. Second, we selected participants based on a multistage random sampling procedure across 7 northeastern cities, providing a generalizable and representative of population sample, and several sensitivity analyses showed that the PM-BP associations were robust. Third, unlike most previous studies, which focused on adults in western countries, our study provides evidence of PM-BP associations among children in a developing country. Fourth, we report associations between PM₁ and children’s BP for the first time, providing new information to help assess the risks and indicating that policy makers should pay closer attention to the potential adverse health impacts of PM₁.

Our study also has several limitations. First, we adjusted for several important potential confounders, yet there may still be unmeasured confounding factors (eg, noise, fasting blood glucose levels, and lipid profiles) which affected the effect estimates but were unavailable in the current study.⁶⁹ Second, the cross-sectional nature of the study design precludes temporality, in that it is unclear if long-term ambient PM exposure preceded children’s BP and hypertension. Still, we think it unlikely for children’s BP to have impacted the distribution of ambient air pollution around their residential area. Finally, we used a mercury sphygmomanometer to measure BP, which may be prone to operator measurement errors. However, before the measurement, we extensively trained the field staff on a standardized measurement protocol to minimize the potential for this bias to the greatest extent practical.

Our study provides novel evidence on PM-BP associations. Higher BP and prevalent hypertension were associated with greater long-term ambient PM₁ and PM_2.5 exposures in Chinese children, particularly in younger children (≤11 years of age) and overweight or obese children. PM₁ might play a leading role in the observed associations between PM_2.5 and BP and appeared to be responsible for a clinically significant hypertension burden associated with PM_2.5 Chinese in children. As such, we believe that policy makers should pay closer attention to the potential adverse health impacts of PM₁ and that well designed longitudinal investigations are urgently needed to confirm our findings.