Associations of residential greenness with hypertension and blood pressure in a Chinese rural population: a cross-sectional study

Limited epidemiological literature identified the associations between residential greenness and hypertension in low-/middle-income countries. A random sampling strategy was adopted to recruit 39,259 residents, ≥ 18 years, and from 5 counties in central China. Blood pressure was measured based on the protocol of the American Heart Association. Hypertension was defined according to the 2010 Chinese guidelines for the management of hypertension. The satellite-derived normalized difference vegetation index (NDVI) and enhanced vegetation index (EVI) were applied to estimate the residential greenness. Mixed logit model and mixed linear model were utilized to explore the relationships of residential greenness with hypertension and blood pressure. Higher residential greenness was associated with lower odds of hypertension and blood pressure levels. For instance, an interquartile range (IQR) increase in NDVI500m was linked with lower odds of hypertension (OR = 0.92, 95%CI 0.88 to 0.95), a decrease of −0.88 mm Hg (95% CI −1.17 to −0.58) and −0.64 mm Hg (95% CI −0.82 to −0.46) in SBP and DBP, respectively. The effect of residential greenness was more pronounced in males, smokers, and drinkers. Long-term exposure to residential greenness was linked with lower odds of hypertension. More prospective studies are needed to verify the hypothesis.


Introduction
Cardiovascular diseases (CVDs) drive huge deaths worldwide, and hypertension is the major cause of CVDs Collaborators GBDRF (2018). In 2017, over 10 million global deaths were due to high systolic blood pressure (SBP) Collaborators GBDRF (2018). Along with rapid urbanization, the hypertension prevalence was increasing, reflecting inadequate control in China (Lu et al. 2017). Apart from hereditary, behavioral, and other recognized hypertension-related factors, environmental impacts have been demonstrated in the published studies (Vienneau et al. 2017). The underlying greenspace-health mechanisms include weakening injury (e.g., air pollution-buffering, noise-buffering, heat-buffering), capability recovery (e.g., alleviating anxiety and depression), and improving capability (e.g., offering breathing spaces, increasing exercise, promoting social-interaction activities) (Markevych et al. 2017).
With rapid urbanization in China, traditional agricultural work has shifted (i.e., overuse of chemical fertilizers and pesticides) and the human health of rural-dwelling faced a major challenge due to worsened natural environment (i.e., air pollution, forest degradation) (Yang 2013). It is of public significance to assess the effect of greenness on hypertension in rural residents in China. Here, this study performed population-based analyses in Chinese rural dwellers. We hypothesized that high residential greenness was linked with lower odds of hypertension. Also, the potential effect modifiers in the relationships were examined.

Study population
Rural cohort is a prospective population-based investigation designed to obtain the epidemic of chronic noncommunicable diseases in Chinese rural-dwelling (Liu et al. 2019b). Between July 2015 and September 2017, the baseline survey was developed in Henan province which is in central China. Henan province is a major agricultural province (coverage area 167,000 km 2 ) and has a large permanent resident population of about 94.02 million (rural-dwelling proportion = 79%). Additionally, the air pollution in the study area was serious and the concentration of fine particulate matter (PM 2.5 ) was up to 73.4 (standard deviation = 2.6 μg/m 3 ) (Liu et al. 2019a). Detailed cohort design was described in the prior study (ChiCTR-OOC-15006699) Liu et al. 2019b). Briefly, a random sampling was applied to select permanent dwellers, 18-79 years, and living in 5 rural zones in central China (Figs. 1 and 2). Overall, 39,259 participated in the study. Subjects without SBP and diastolic blood pressure (DBP) were excluded in the final analyses. The signed informed consent was received before the survey. This study was approved by the Zhengzhou University review board.
Outcome SBP/DBP was measured thrice utilizing OMRON HEM-770A according to the American Heart Association guideline (Pickering et al. 2005). The protocol was published previously ). Prior to measurement, smoking, drinking, and physical exercise were forbidden within half an hour. In addition, Fig. 1 The flowchart of participant recruitment talking was forbidden during the measurement. According to the 2010 Chinese guideline for the management of hypertension , participants were diagnosed with hypertension if they had a diagnosis history of hypertension or use of the antihypertensive drug or SBP ≥ 140 mm Hg or DBP ≥ 90 mm Hg.

Greenness exposure
The NDVI and enhanced vegetation index (EVI) databases were downloaded from the National Aeronautics and Space Administration to estimate the individual's exposure to residential greenness at a 500-m buffer. The assessment was described elsewhere (Xie et al. 2020). The two vegetation indexes could effectively characterize vegetation states and processes. NDVI has been used widely for environmental applications (Jia et al. 2018). Compared with NDVI, EVI could minimize canopy-soil variations and improved sensitivity over dense vegetation conditions Qiu and Yang (2018). To compare with previous studies (Dzhambov and Markevych 2018;Vienneau et al. 2017;Xiao et al. 2020;Yang et al. 2019), residential greenness was defined as mean NDVI/EVI values in the 500-m buffer around the individual's home address. In addition, the average of NDVI/EVI in the 1000-m/3000-m buffer was included in sensitivity analyses. Long-term exposure to greenness was defined as the 3-year average of NDVI/ EVI prior to the survey. The data processing was performed using ArcGIS 10.7 (ESRI, Redlands, CA).

Statistical analysis
Mixed logit model and mixed linear model were employed to explore the associations between residential greenness and hypertension and blood pressure. Long-term exposure to greenness was utilized in the main model. The models were widely used in recent studies (Xiao et al. 2020;Yang et al. 2019). Given the possible influence of regional clustering, the region was controlled as a random effect in the analyses (Xie et al. 2020). Adjustment of fixed effects was as follows: (1) crude model: non; (2) adjusted model: adjusted for age, sex, marriage, education, income/month, smoking, drinking, fatrich diet, exercise, family history of hypertension, BMI, and PM 2.5 (Dzhambov and Markevych 2018;Lane et al. 2017). To test effect modifications of behavioral factors in the greenness-hypertension/blood pressure associations, the interaction analysis was performed by putting the interaction items in the adjusted model. Considering the possible mediation in the greenness-hypertension/blood pressure pathways (e.g., air pollution, exercise) (de Keijzer et al. 2019;Xie et al. 2020), serial mediation analyses were carried out. To verify the reliability of conclusions, sensitivity analyses were conducted: (1) The average of 1-year and 2-year NDVI 500m /EVI 500m prior to the survey was employed to test long-term impacts of greenness on hypertension and blood pressure; (2) the greenness-hypertension/blood pressure relationships were investigated utilizing different buffers (NDVI 1000m /EVI 1000m , NDVI 3000m /EVI 3000m ); (3) to eliminate the causal effect of hypertension on blood pressure, subjects with hypertension and taking the antihypertensive drug were excluded.
Analyses were conducted in R 4.0.2 loading "lme4" and "mediation" packages.
An IQR increment in vegetation index was related to 8-11% lower hypertension prevalence, 0.79-1.18 mmHg decreased SBP, and 0.64-1.11 mm Hg decreased DBP (Fig.  3). For example, each IQR (0.08 units) increase in NDVI 500m was related to 8% lower odds of hypertension, a reduction of 0.88 mmHg in SBP, and 0.64 mmHg in DBP in the adjusted model. Interaction analysis concluded that the effect of residential greenness on hypertension/blood pressure was more pronounced in males, smokers, and drinkers (P interaction < 0.05) (Fig. 4). For instance, high NDVI 500m was related to 13% lower odds of hypertension among males, whereas 5% among females (P interaction = 0.005). In addition, there was a decrease of 1.30 mm Hg in SBP among males, compared with 0.63 mmHg among females (P interaction = 0.002).
PM 2.5 , exercise, and BMI partially mediated the relationships between residential greenness and hypertension and blood pressure (Table S3). The proportion of mediation for PM 2.5 , exercise, and BMI varied between 3.7-21.6%, 0.7-4.4%, and 23.6-36.7%, respectively. Compared with main analyses, using the 1-year/2-year average NDVI 500m / EVI 500m and NDVI 1000m /EVI 1000m showed similar results (Fig. S1). The enhanced association was detected using NDVI 3000m /EVI 3000m . For example, each IQR increase in NDVI 3000m was related to 19% lower odds of hypertension. Changes in SBP/DBP remained similar when participants taking the antihypertensive drug were excluded. After excluding hypertensive patients, the relationships were attenuated but remained significant. Each IQR increase in NDVI 500m was linked to 0.40 mmHg lower SBP and 0.32 mm Hg lower DBP.

Discussion
The studies of the greenness-hypertension pathways are still in the initial stage and lack evidence from low-/middle-income countries. The current study was one of few large epidemiological studies assessing the effect of greenness on hypertension in Chinese rural-dwelling. We found that elevated residential greenness was associated with 8-11% decreased hypertension prevalence, 0.79-1.18 mmHg lower SBP, and 0.64-1.11 mm Hg lower DBP. In addition, sex, smoking, and drinking could further modify the relationships. Given increasing environmental pollution accompanied by the accelerated urbanization process and high hypertension prevalence in China as well as other low/middle-income countries, our findings suggested the importance of greenness on human health and it was necessary to protect the green environment in the rural region.
In the present study, each IQR increment in NDVI 500m was related to 8% lower odds of hypertension. Similar conclusions were drawn in the published studies (Brown et al. 2016;de Keijzer et al. 2019;Dzhambov and Markevych 2018;Jia et al. 2018;Vienneau et al. 2017;Xiao et al. 2020;Yang et al. 2019). For instance, Brown et al. (2016) indicated that 13% (95% CI 10.6-16.6%) lower odds of hypertension among elderly Americans (≥65 years) attributed to exposure to greenness. Additionally, each 0.1-unit increment in NDVI 500m was related to 24% (95% CI 18-31%) lower hypertension prevalence in Chinese children (Xiao et al. 2020). However, Jendrossek et al. (2017) revealed no significant association among German adults (OR = 0.889, 95% CI 0.561-1.409). A meta-analysis of 4 studies concluded no significant greenness-hypertension association (OR = 0.99, 95% CI 0.81-1.20) Twohig-Bennett (2018). One of the possible causes for the difference in the studies mentioned above was the difference in the study population, study design, and statistical analysis (i.e., adjustment of variables). Additionally, the difference in the greenness between urban-dwelling and rural-dwelling may be one of the reasons for the inconsistent results of previous studies. A study conducted in Germany found only the significant greenness-SBP/DBP relationship in the urban region, but no significant relationship in the rural region (Markevych et al. 2014). The study explained that the urban residents lacking greenspace may be more susceptible to exposure to greenness, compared with rural-dwelling. Fig. 4 Interaction analyses in the relationships between residential greenness and hypertension and blood pressure. Abbreviations: NDVI, normalized difference vegetation index; EVI, enhanced vegetation index.
Note: Adjusted for age, sex, marriage, education, income, smoking, drinking, fat-rich diet, exercise, family history of hypertension, BMI, and PM 2.5 However, a nationwide study conducted in China concluded that the low availability of greenspace partly restricted the health effect of greenness in urban-dwelling and urban residents may make less profit from greenness than rural-dwelling (Huang et al. 2021). This study also reported that rural participants exposure to more greenness than urban participants (NDVI 0.54 vs. 0.27 units) (P-value <0.001). In our study, the median NDVI 500m was 0.4990 (IQR 0.0840) units in the study area and a study conducted in 3 urban areas in Liaoning province, China, reported that NDVI 500m was 0.29 (IQR 0.17) units (Yang et al. 2019). It showed that the greenness in the study area was better than Liaoning province in China. To date, there is a limited evidence from rural-dwelling in low-/ middle-income countries especially in China; therefore, a comprehensive greenness-hypertension pathway remained deeply studied. In addition, the meta-analysis identifying those relationships needs to be updated. Elevated residential greenness was linked with lower SBP/ DBP in the current analysis. In line with our result, Lane et al. (2017) found 4.3 mmHg lower SBP and 1.2 mmHg lower DBP per IQR increment in NDVI 250m among Indian adults. However, inconsistent results existed in several previous studies (Bijnens et al. 2017;Bloemsma et al. 2019;Dzhambov and Markevych 2018;Madhloum et al. 2019;Markevych et al. 2014;Xiao et al. 2020;Yang et al. 2019). For instance, the significant NDVI 500m -SBP association was observed, whereas no significant NDVI 500m -DBP association in Chinese adults (Yang et al. 2019). In addition, Madhloum et al. (2019) drew the opposite conclusion indicating no significant greenness-SBP relationship, whereas the significant greenness-DBP relationship in newborns. No significant association was found between greenness and SBP/DBP in adolescents (Bloemsma et al. 2019). The study population mentioned above (e.g., adult, newborn, children) has different characteristics that may modify the effect of greenness. Besides, differences in study design and greenness assessment may explain the inconsistency. Thus, the greenness-SBP/DBP pathways remained unclear.
One of the mechanisms by which long-term exposure to residential greenness reduces the risk of hypertension could be explained by immunological pathways (Rook 2013). Specifically, people residing in green space are exposed to more diverse microbes beneficial to the host immune system (improving immune regulation) and less noise and air pollution (reducing inflammatory response) are more likely to increase physical activity (strengthening immune and nervous system), promote the exchange of microbiota (increasing social interactions), and promote metabolism (sunlight helping in the synthesis of vitamin D). In this study, we found that exercise partially mediated the greenness-hypertension relationships and intermediary proportion ranged from 0.7 to 4.4%. It indicated that the natural environment (i.e., forest, farmland) in the rural region could increase physical activity. Huang et al. (2021) also observed the slight indirect effect of exercise (1.27%) on the greenness-hypertension relationships among the rural population. Second, decreased levels of cortisol induced by the restorative process linked to greenness are also believed to be important mechanisms Twohig-Bennett (2018). Third, high residential greenness ease tension and depression by increasing social interactions (Banay et al. 2019).
Previous epidemiological studies indicated cigarette use, and alcohol consumption could increase the risk of hypertension and sex may further modify these relationships (Bowman et al. 2007;Lui et al. 2020;Sesso et al. 2008). For instance, Sesso et al. (2008) observed a J-shaped relationship of drinking with the risk of hypertension in females and a positive relationship in males. Our interaction analysis showed that males, smokers, and drinkers tended to be more susceptible to the effect of greenness. Whether sex modifies the greenness-hypertension relationship remained unclear in previous studies (Dzhambov and Markevych, 2018;Jia et al. 2018;Xiao et al. 2020;Yang et al. 2019). For example, Jia et al. (2018) reported OR male = 0.14 (95% CI 0.11-0.48) and OR female = 0.45 (95%CI 0.36-0.63) and it was consistent with our finding. However, Yang et al. (2019) found no significant association in males, whereas a significant association in females. In addition, Xiao et al. (2020) observed no significant sex interaction in NDVI 500m -hypertension relationship (OR male = 0.81 vs. OR female = 0.72, P interaction = 0.279).
Smoking and drinking could trigger inflammatory response, oxidative stress, and metabolic disorders, which further affect systemic vascular resistance (Joseph et al. 2017). Long-term exposure to greenness could provide diverse microbes, some of which are important inducers of the immunoregulatory pathways, and activation of the immune regulatory system consequently reduced chronic inflammation (Rook 2013). Thus, the effect of greenness on hypertension and blood pressure was more pronounced in smokers and drinkers.
Several limitations existed in the present study. First, as an inherent drawback of cross-sectional design, causal associations should be treated with caution. Second, demographics, behavioral factors, and health status were collected using the questionnaire, which may introduce recall bias. Third, some potential confounders including salt consumption, noise, the walkability of the community, and indoor exposure to greenness were not adjusted because of unavailability (Dzhambov and Markevych, 2018;Xiao et al. 2020).

Conclusion
The current study has evaluated the associations of residential greenness with hypertension in Chinese rural-dwelling. We found that long-term exposure to residential greenness was correlated with decreased odds of hypertension. Furthermore, males, smokers, and drinkers were more susceptible to the effect of greenness. Our findings suggested the health effect of green environment (e.g., forest, farmland) on rural population, and it is necessary to protect the natural environment. Given the limitations of design, prospective studies are warranted in future analysis.