High blood pressure (HBP) in children was once considered a rare disease, but it is now a public health concern worldwide (1). Today, childhood HBP affected over 1.13 billion people worldwide as it significantly increases the risks of heart attack, stroke, and other complications (2). In recent years, the incidence of HBP has increased constantly in developing countries. Environmental and lifestyle changes might contribute to its prevalence (3, 4).

Evidence indicated that air pollutants sulfur dioxide (SO₂) and carbon monoxide (CO) might be important risk factors for hypertension (5, 6) and other CVDs in adults (7, 8). Animal experiments also showed that short- and long-term exposure to SO₂ could cause functional damage to the cardiovascular system (9, 10). Its possible mechanisms could be explained by oxidative stress, alterations in the autonomic nervous system, or ion concentration change in body fluids (9, 11, 12). However, this evidence was limited (13) compared to similar studies about PM₁₀, PM_2.5, ozone, and NO₂, especially in cohort studies and large populations of children and adolescents in developing countries (14–19). Moreover, the limited evidence supported inconsistent conclusions. For example, 5-day short-term exposure to SO₂ was reported to have no association with BP increase (20). The long-term exposure investigation, however, observed a strong correlation between SO₂/CO exposure and hypertension (16). Greenness plays another important role in cardiovascular health, according to the assessment of the beneficial impact of greenspace on a dozen health outcomes, including BP (21). Cross-sectional and retrospective studies mostly focused on the potential benefits of plants to attenuate the respiratory and cardiovascular health risks of PM₁₀, PM_2.5, and NO₂ pollution in adults or seniors (22–26). Few of them focused on SO₂ and CO. A cross-sectional study in Northern China found a 10-μg/m³ increase in SO₂ was responsible for a 2.43% increase in mean arterial pressure among adults. These identified harmful effects of SO₂ mainly occurred among people who lived in low-greenness environments (27). Subgroup analysis of greenness may provide a concise illustration of the role of greenness on health outcomes. Although the association may be non-linear, higher greenness is probably to have a more positive effect on hypertension (28). In terms of biological mechanisms, exposure to better green space can increase immunoregulation, lowering the risk of inflammatory and cardiovascular diseases. Greenness can also improve physical activities and reduce noise effects (29) to counteract the negative effects of environmental pollution by improving stress resilience (30). In addition, children and adolescents of different genders, ages and BMIs have various physiological, psychological, and socio-environmental differences. The blood pressure in different subgroups may be influenced by growth and development characteristics, for example, sex-related differences in hormone secretion, age-related differences in puberty timing (31), obesity-related differences in endocrine regulation/insulin resistance, and environment sensitivity (32, 33). Previous findings on these subgroups' sensitivity to SO₂, CO and greenness exposure are limited (18, 27) and therefore further validation and supplementation are still needed.

In this multicenter, longitudinal, prospective open cohort study in China, we hypothesized that SO₂ and CO were risk factors for HBP in school-aged children and adolescents, while greenness protected blood pressure. We also aimed to investigate whether a high level of greenness can reduce this risk, and whether different BMIs groups had different sensitivities.

There were 219,956 children and adolescents without HBP at the baseline included in the analysis, 46,652 in Beijing and 173,304 in Zhongshan. The characteristic of the study subjects is displayed in Table 1. The mean age was 12.9 (SD: 2.89), and the mean NDVI was 0.30 (SD: 0.08). The average time of follow-up was 3.09 years. Mean SO₂ and CO concentrations were 12.27 (SD: 3.93) μg/m³ and 0.97 (SD: 0.14) mg/m³. Among the participants without HBP at baseline, 52,515 were identified with new HBP during the follow-up, the cumulative incidence was 23.88%, and an incidence density was 7.72 per 100 person-year. The levels of NDVI exposure, BMI, and SO₂ and CO concentrations were higher, but the average age, height, and weight were lower in the HBP group than in the non-HBP group (p < 0.001). The location of survey schools in Beijing and Zhongshan, their SO₂ and CO concentration distribution and the NDVI levels in 2014 are shown in Figure 1.

We analyzed the non-linear correlation of SO₂ and CO with the risk of HBP after adjusting age, sex, BMI, greenness, and city. We found that the risk of HBP in children and adolescents increases with the concentration of SO₂ and CO (Figures 2A1, A2), Cox model with RCS. Meanwhile, the association of HBP risk with SO₂ exposure was lower in high-greenery areas than in low-greenery areas, but this was not observed for CO exposure (Figures 2B1, B2), Cox model with RCS. We also explored a relationship between greenness and HBP, that risk of HBP decreases with the increase of NDVI (Figure 2A3).

Based on the above preliminary results, we further analyzed the quantitative association of SO₂, CO, and NDVI with BP levels and HBP risks, as illustrated in Figure 3. We found that the concentration of SO₂ and CO positively correlated with SBP and DBP levels and HBP risks. It also had a higher effect on SBP: increased SO₂ and CO exposure were significantly associated with SBP levels [β = 1.30 and 0.78, 95% CI: (1.26, 1.34) and (0.75, 0.81)] and were less associated with DBP levels [β = 0.81 and 0.46, 95% CI: (0.79, 0.84) and (0.44, 0.48)]. The HR of HBP was 1.58 and 1.42 respectively [95%CI: (1.57, 1.60) and (1.41, 1.43)]. NDVI was negatively correlated with SBP and DBP. Therefore, it could protect HBP: an increase in NDVI was significantly associated with SBP [β = −0.53, 95% CI: (−0.56, −0.49)] and DBP levels [β = −0.65, 95% CI: (−0.67, −0.63)], and the HR of HBP was 0.90 [95% CI: (0.89, 0.91)].

AF values were used to evaluate the theoretical benefits of HBP reduction through improving SO₂ and CO among children and adolescents. The results showed that eliminating SO₂ and CO exposure around schools could get a considerable theoretical benefit, but the benefit could vary by the levels of greenness. Improvement of SO₂ and CO exposure benefits more in the low-level group of NDVI than in a high-level group of NDVI. The lower greenness group had a higher risk of HBP induced by SO₂ and CO pollution. Figure 4 shows that AF values of HBP risks attributable to SO₂ and CO in the low NDVI group of areas were 26.31% and 20.04% respectively, which were higher than those in a high NDVI group of areas (13.90% and 17.81%).

We also did a stratified analysis by BMI. Those who live in low-level green areas with normal nutritional status obtain great theoretical benefits. For obese participants, there was no significant difference between low and high-level greenness in the AFs of SO₂ and CO. As shown in Figure 4, SO₂ and CO-induced HBP risk was higher for normal-weight children and adolescents with a low-level greenness, the AFs of which were 30.90% and 22.64%, while only 14.41% and 18.65% for those living with a high-level greenness. However, the risk of HBP attributable to SO₂ and CO was not as high as expected for obese children in a low-level greenness (AFs = 10.64% and 8.61%), nor was it significantly lower in the high-level greenness (AFs = 9.60% and 10.72%).

Previous studies indicated that HBP in childhood or adolescence was a significant predictor for adult HBP and cardiovascular diseases (40, 41). However, the association of greenness and long-term SO₂/CO exposure with childhood BP has yet to be studied in depth, particularly in a longitudinal study with potential causal inference. As far as we know, this was the first study that used a longitudinal, two-center, dynamic cohort in China to confirm that greenness can alleviate the damage effects of SO₂/CO exposure on the risks of HBP among children and adolescents. The study also approved that improving individual exposure to SO₂/CO can significantly benefit those with low-level greenness and normal weight. Our findings supported the reduction of the HBP risks among children and adolescents by improving greenness and reducing gaseous pollution around schools by a variety of measures.

Our findings were consistent with most previous relevant studies (6, 27, 39). For instance, a cross-sectional study of around 10,000 children from seven northeastern cities in China found that higher greenness levels around schools significantly lower the risk of childhood HBP (42). Another cross-sectional study found that high levels of SO₂ and CO increased arterial blood pressure and HBP among children aged 5–17 in northeastern China (16). In addition, a two-decade population-based study in Tehran revealed that diastolic blood pressure was more sensitive to CO while SBP was more sensitive to SO_2. They found that adults exposed to higher SO₂ pollution had a significantly higher risk of HBP than those exposed to CO (14). Combined with our study, we suggested that children might be more sensitive to CO air pollution than adults. Another study on the interaction between obesity and air pollution on BP in Chinese children confirmed that children's obesity amplified the effects of long-term air pollution on BP (15). The result suggested that the BP effects of air pollution might be harder to eliminate in obese children than in normal-weight children. This further confirmed the plausibility of our findings. However, several other findings are inconsistent with ours. A nationwide cross-sectional investigation in China found that increases in greenness were associated with reductions in SBP and DBP (39), but not in our study. The reasons for that might be due to the sample distribution, and greenness measurement differences. Our study filled a gap in the pre-cardiovascular hazard caused by specific gaseous pollutants in cohort studies of 7–17 years-old children and adolescents compared to previous studies. We further found that improving air pollution according to greenness level is a specific healthy measure, and theoretical benefits vary with children of different body sizes.

Viewpoints vary on the mechanism underlying HBP or BP affected by gaseous pollutants. One theory posited that gaseous (NO_x, SO₂, CO, and O₃) and particle (PM₁₀ and PM_2.5) pollution triggers pulmonary oxidative injury and systemic inflammation, leading to oxidative stress, with consequences of endothelial injury, vasoconstriction, thrombosis, and changes in blood pressure (43). Another study suggested that pollutants activate respiratory sensory nerves, affecting airway receptors, baroreceptors, and chemoreceptors, thereby modifying the autonomic nervous system control of BP (44). There were also several perspectives on the effect mechanisms of greenness on BP, such as reducing stress, improving physical activity, and reducing respiratory diseases (45–48). We hypothesized that greenness could also control BP by changing pollution levels. Through analyzing individual air quality components or different geographical regions in China, we discovered that greenness had a prominent role in improving urban air quality, especially in northern China. Its contribution can reach 16.2% (49). Studies demonstrated that trees could direct filter air pollutants, such as SO₂, which are absorbed mainly through stomata. Absorption is not the only way in which plants improve air quality. The higher the tree canopy cover, the better the barrier effect, and the more pollutant mixtures from high in the air can be limited. Tree species, plants cover rate, length of the leaf season, pollution concentration, and precipitation in different cities will all affect the air purification of greenness, thus bringing about different levels of health benefits (49–53). However, for obese children, the function of greenness is very limited in eliminating the pollutants-related HBP risk attributed to pollutants. These findings and mechanisms indicated that targeted measures should control the individual pollutants in different areas and for children with different BMIs.

In addition to the above findings, secondary results are shown in the above figures and Supplementary Appendix. We found that greenness had a protective effect on participants' blood pressure. The mechanism follows: green plants absorb SO₂ and CO primarily via leaf stomata and reduce air temperatures by transpiration, influencing microclimate, which can then promote physical activity and social engagement, and link to mental health benefits (50, 54, 55), thus bringing physical health benefits. Our study also investigated the differences in results between boys and girls, in different ages, as shown in Figures 2–4 and Supplementary Figure A2. There was an obvious difference between boys and girls on the nonlinear associations of SO₂, CO, and greenness with the risk of HBP, indicating that girls are more subjected to pollutants and greenness differences than boys. The participants were grouped into four age groups every 3 years as an interval and analyzed in subgroups. The group of 7–9 years old was more sensitive than the other groups. The association between CO and HBP outcomes was weaker in the 13–15 years old group, their association of NDVI with HBP was even the opposite. The result might be because of the indoor and outdoor activity time among children with academic pressure. However, no detailed age sensitivity to gaseous pollutants has been reported in previous studies. In addition, two typical northern and southern cities were included in our study to reflect China's overall situation in and to improve the result scalability. Considering the differences between cities, we also tested the data for both cities separately. Although the two cities may have some differences in detail, the overall results appear to be consistent with the combined results after we adjusted for the city in our main results, as shown in Supplementary Figures A7, A8. The blood pressure hazards caused by SO₂, and CO pollution were similar in Zhongshan. Still, the greenery protective effects were more stable across genders in Zhongshan than in Beijing, probably due to pollutant proportions and the plant species in the two cities. Therefore, we also did some sensitivity analysis, as shown in Supplementary Figures A3–A8.

This study has the following strengths: first, we used a dynamic longitudinal cohort covering almost all school-age students for 12 years. Second, this study focused on northern and southern cities in China, making the results more representative and balanced. Third, we used the restricted cubic spline analyses to explore the non-linear association between exposure and ending, which might be more relevant to reality. At last, in addition to exploring the independent association between gaseous pollutants and green space with BP separately. We also assessed the benefits of improving a single pollutant under different greenness conditions to provide theoretical ideas for policy development.

There were still some limitations: firstly and most importantly, we used a convenient sample to develop a cohort study and only collected physical examination and environmental data. No family background or lifestyle habits were included. As a result, our study did not investigate some confounders, like salt intake, parents' education level, temperature, and traffic noise, which might lead to fluctuation in the theoretical benefits of greenness (42, 56, 57). Secondly, we adopted annual mean air pollutants concentrations and NDVI at the school level. The inherent limitations of NDVI did not allow us to distinguish between plant types, which might also affect health differently. And the home addresses collected in our study were imprecise, but the school addresses were accurate to latitude and longitude coordinates. Although students may spend even time at school and home, the local climatic and geographic conditions were similar. As a result, we used the school environment as individual exposure. Thirdly, we assumed that students included in our study did not change schools. Individuals may still develop HBP after exposure, but we did not follow up on this situation, so we might underestimate the incidence of HBP in our study. In addition, a one-time BP value is not sufficient to confirm the HBP, three repeated visits on different occasions would be more accurate. Fourth, our study did not consider children who dropped out of school; therefore, it could not represent the entire children and adolescent population. As our study was based on a dynamic open cohort of a natural population, selection bias still exists. It excluded more obese children and included children with lower BMI and weaker effects on HBP, while the effect of SO₂ and CO on blood pressure was still observed.

In summary, our study supported a positive correlation between air pollution and HBP risks among children and adolescents, and a negative correlation between greenness and HBP risks. Meanwhile, the greenness reduces the risks of HBP attributed to SO₂ and CO exposure among students, but its benefit was more effective for normal-weight participants than those obese. It was suggested that some targeted measures should be taken to reduce the specific gaseous pollutants. According to the characteristic of various districts, improving green space construction and preventing childhood obesity could reduce the burden of childhood HBP and subsequent cardiovascular disease risks.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the Biomedical Ethics Committee of Peking University Health Science Center. Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.

YZ and SC conceived the study design, performed the analysis, interpreted the findings, and wrote the manuscript. LC, YW, and JW prepared, analyzed, cleaned the data, interpreted the findings, and helped with manuscript preparation. TM, MC, and QM cleaned the data, interpreted the findings, and helped with manuscript preparation. JL, LW, WL, XL, JZ, and XW helped with manuscript preparation. JM and XG contributed to the conception of the work. YX and JZ worked on study design and data collection and helped with manuscript preparation. YD contributed to the conception of the work and helped with manuscript preparation. All authors read and approved the final manuscript.