UEF//eRepository
DSpace https://erepo.uef.fi
Rinnakkaistallenteet Luonnontieteiden ja metsätieteiden tiedekunta
2020
Association between community greenness and obesity in
urban-dwelling Chinese adults
Huang, WZ
Elsevier BV
Tieteelliset aikakauslehtiartikkelit
© Elsevier B.V.
CC BY-NC-ND https://creativecommons.org/licenses/by-nc-nd/4.0/
http://dx.doi.org/10.1016/j.scitotenv.2019.135040
https://erepo.uef.fi/handle/123456789/7861
Downloaded from University of Eastern Finland's eRepository
Journal Pre-proofs
Association between community greenness and obesity in urban-dwelling Chinese adults
Wen-Zhong Huang, Bo-Yi Yang, Hong-Yao Yu, Michael S. Bloom, Iana Markevych, Joachim Heinrich, Luke D. Knibbs, Ari Leskinen, Shyamali C.
Dharmage, Bin Jalaludin, Lidia Morawska, Pasi Jalava, Yuming Guo, Shao Lin, Yang Zhou, Ru-Qing Liu, Dan Feng, Li-Wen Hu, Xiao-Wen Zeng, Qiang Hu, Yunjing Yu, Guang-Hui Dong
PII: S0048-9697(19)35032-6
DOI: https://doi.org/10.1016/j.scitotenv.2019.135040
Reference: STOTEN 135040
To appear in: Science of the Total Environment Received Date: 18 September 2019
Revised Date: 14 October 2019 Accepted Date: 16 October 2019
Please cite this article as: W-Z. Huang, B-Y. Yang, H-Y. Yu, M.S. Bloom, I. Markevych, J. Heinrich, L.D.
Knibbs, A. Leskinen, S.C. Dharmage, B. Jalaludin, L. Morawska, P. Jalava, Y. Guo, S. Lin, Y. Zhou, R-Q. Liu, D. Feng, L-W. Hu, X-W. Zeng, Q. Hu, Y. Yu, G-H. Dong, Association between community greenness and obesity in urban-dwelling Chinese adults, Science of the Total Environment (2019), doi: https://doi.org/10.1016/
j.scitotenv.2019.135040
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
© 2019 Elsevier B.V. All rights reserved.
Association between community greenness and obesity in urban-dwelling Chinese adults.
Wen-Zhong Huang1,†, Bo-Yi Yang1,†, Hong-Yao Yu1,†, Michael S. Bloom1,2, Iana Markevych3, Joachim Heinrich3, Luke D. Knibbs4, Ari Leskinen5,6, Shyamali C. Dharmage7, Bin Jalaludin8, Lidia Morawska9, Pasi Jalava10, Yuming Guo11, Shao Lin2,3, Yang Zhou1, Ru-Qing Liu1, Dan Feng1, Li-Wen Hu1, Xiao-Wen Zeng1, Qiang Hu12,*, Yunjing Yu13,*, Guang-Hui Dong1,*
1Guangdong Provincial Engineering Technology Research Center of Environmental Pollution and Health Risk Assessment, Department of Occupational and Environmental Health, School of Public Health, Sun Yat-sen University, Guangzhou 510080, China
2Department of Environmental Health Sciences and Epidemiology and Biostatistics, University at Albany, State University of New York, Rensselaer, New York 12144, USA
3Institute and Clinic for Occupational, Social and Environmental Medicine, University Hospital, LMU Munich, Ziemssenstraße 1, 80336 Munich, Germany
4School of Public Health, The University of Queensland, Herston, Queensland 4006, Australia
5Finnish Meteorological Institute, Kuopio 70211, Finland
6Department of Applied Physics, University of Eastern Finland, Kuopio 70211, Finland
7Allergy and Lung Health Unit, Centre for Epidemiology and Biostatistics, School of Population & Global Health, The University of Melbourne, Melbourne, VIC 3010 Australia
8Centre for Air Quality and Health Research and Evaluation, Glebe NSW 2037, Australia
9International Laboratory for Air Quality and Health, Queensland University of Technology, Queensland 4001, Australia
10Department of Environmental and Biological Sciences, University of Eastern Finland, Kuopio,
FI 70211, Finland.
11Department of Epidemiology and Preventive Medicine, School of Public Health and Preventive Medicine, Monash University, Melbourne VIC 3004, Australia
12Department of Pediatric Surgery, Weifang People's Hospital, Weifang 261041, China
13State Environmental Protection Key Laboratory of Environmental Pollution Health Risk Assessment, South China Institute of Environmental Sciences, Ministry of Environmental Protection, Guangzhou, 510655, China
*Address correspondence to:
Qiang Hu, MD, PhD, Professor, Department of Pediatric Surgery, Weifang People's Hospital, Weifang 261041, China. Email: DHQ7936@163.com
Yunjiang Yu, MD, PhD, Professor, State Environmental Protection Key Laboratory of Environmental Pollution Health Risk Assessment, South China Institute of Environmental Sciences, Ministry of Environmental Protection, Guangzhou, 510655, China. Email:
yuyunjiang@scies.org.
Guang-Hui Dong, MD, PhD, Professor, Guangdong Provincial Engineering Technology Research Center of Environmental Pollution and Health Risk Assessment, Department of Occupational and Environmental Health, School of Public Health, Sun Yat-sen University, 74 Zhongshan 2nd Road, Yuexiu District, Guangzhou 510080, China. Phone: +862087333409; Fax:
+862087330446. Email: donggh5@mail.sysu.edu.cn; donggh512@hotmail.com
†These authors contributed equally to this work and should be listed as the first author.
Abstract
Living in greener places may protect against obesity, but epidemiological evidence is inconsistent and mainly comes from developed nations. We aimed to investigate the association between greenness and obesity in Chinese adults and to assess air pollution and physical activity as mediators of the association. We recruited 24,845 adults from the 33 Communities Chinese Health Study in 2009. Central and peripheral obesity were defined by waist circumference (WC) and body mass index (BMI), respectively, based on international obesity standards. The Normalized Difference Vegetation Index (NDVI) was used to quantify community greenness.
Two-level logistic and generalized linear mixed regression models were used to evaluate the association between NDVI and obesity, and a conditional mediation analysis was used also performed. In the adjusted models, an interquartile range increase in NDVI500-m was significantly associated with lower odds of peripheral 0.80 (95% confidence interval [CI]: 0.74–
0.87) and central obesity 0.88 (95% CI: 0.83–0.93). Higher NDVI values were also significantly associated with lower BMI. Age, gender, and household income significantly modified associations between greenness and obesity, with stronger associations among women, older participants, and participants with lower household incomes. Air pollution mediated 2.1%-20.8%
of the greenness-obesity associations, but no mediating effects were observed for physical activity. In summary, higher community greenness level was associated with lower odds of central and peripheral obesity, especially among women, older participants, and those with lower household incomes. These associations were partially mediated by air pollutants. Future well-designed longitudinal studies are needed to confirm our findings.
Keywords: green space; obesity; mediation analysis; cross-sectional study; Chinese adults.
1. Introduction
Over the past four decades, the prevalence of obesity worldwide has nearly tripled (WHO, 2018). The obesity pandemic has become one of the gravest threats to human health to date (Swinburn et al., 2019). China has also witnessed a significant increase in the prevalence of obesity (Swinburn et al., 2019), and now has the largest number of overweight population worldwide (Di Cesare et al., 2016). Effective intervention strategies are needed to help combat this trend. The etiology of obesity is complex, including both environmental and genetic factors (Hinney et al., 2015; Lachowycz and Jones, 2011). From the perspective of public health, identifying obesogenic environmental factors may be particularly important, as many can be modified through changes in individual lifestyles or changes in population-level government policies.
Exposure to nature, in particular defined by greenness, has been found to be protective against several adverse health outcomes (Twohig-Bennett and Jones, 2018). Greenness is normally considered to benefit health by reducing exposures to environmental unfavorable factors (e.g., heat, noise and air pollution), relieving stress and promoting physical activities (Markevych et al., 2017). These beneficial responses also involve the pathophysiologic pathways of obesity (Archer et al., 2018; Foraster et al., 2018; Guo et al., 2018; Schreier et al., 2013; van der Valk et al., 2018). Accumulating epidemiological evidence indicates an association between greenness and obesity (Table S1). Overall, however, the results are somewhat inconsistent.For example, four studies found no significant association between greenness and obesity (Coombes et al., 2010; Mowafi et al., 2012; Paquet et al., 2014; Potestio et al., 2009),while two
reported positive associations between the two (Cummins and Fagg, 2012; Wilhelmsen et al., 2017). In addition, these studies are mainly from developed countries and little information is available from developing countries, such as China.
Most previous studies concerning greenness-adiposity associations focused on only body mass index (BMI) or general obesity. Waist circumference (WC), a proxy for abdominal fat mass, is associated with the risk of premature mortality in adults, independent of BMI (Pischon et al., 2008). BMI and WC represent different adipose tissue distributions and can vary in their associations with different non-communicable diseases (Fox et al., 2007). However, little evidence is available regarding the associations between greenness and abdominal obesity.
In this study, we aimed to investigate whether residing in greener space was associated with reduced prevalence of obesity and indicators of adiposity, and to evaluate air pollution and physical activity as mediating factors. We analyzed data from the 33 Chinese Communities Health Study, which is a large-scale population study conducted in northeast of China. During the past four decades, China has experienced the rapid urbanization (National Bureau of Statistics, 2016), attendant with the lifestyle changes (Su et al., 2017), a dramatic increase of obesity (Jiang et al., 2015), a sharp decline of green space (Zhou et al., 2014), and greater air pollution (Yang et al., 2017a). Therefore, China provides a suitable setting for exploring the effects of greenness on obesity.
2. Methods
2.1. Study settings
Between April 1 and Dec 31, 2009, the 33 Chinese Communities Health Study (33CCHS) was conducted in Liaoning province, which is located in northeastern China. There are more than 20 million permanent residents in Liaoning Province, of whom more than 64% are urban dwellers. The overall population obesity prevalence in Liaoning province is 17.3%, which is 5.4 percentage higher than the national average (Chinese Center For Disease Control And Prevention, 2014).
2.2. Study design and participants
The 33CCHS design, recruitment, and inclusion criteria were previously described in detail (Yang et al., 2018; Yang et al., 2017b). Briefly, a representative general population sample was enrolled from the population base using a random number generator based four-stage stratified clustering sampling strategy (Fig. 1). First, three cities (Shenyang, Anshan, and Jinzhou) were randomly selected out of 14 provincial cities in Liaoning province. Shenyang is comprised of five districts and Jinzhou and Anshan comprise three districts each, yielding a total of 11 districts. Second, we randomly selected three communities (area ranging from 0.25 to 0.64 km2) from each of the 11 districts. Third, 700–1000 households were randomly selected from each study community. Fourth, one adult was randomly selected from each household. Participants had to meet all of the following criteria: (1) aged 18–74 years; (2) a minimum of five years
residence in the study community; (3) no current severe illness (e.g. cancer); and (4) not pregnant.
We invited 28,830 potential participants, 24,845 of whom enrolled in the study (response rate=
86.2%). Participants completed a standardized research questionnaire to provide information about demographic characteristics (e.g., age, sex and annual household income), lifestyle (e.g., exercise and dietary habit) and current health problems (e.g. diagnosed heart failure or coronary heart disease). Prior to data collection, written informed consent was obtained from all participants, and all study procedures had been approved by the Human Studies Committee of Sun Yat-sen University.
2.3. Outcome assessment
The measurements of body stature and body weight have been described in detail in a previous study (Dong et al., 2013). Briefly, measurements of height, weight, and waist circumference (WC) were obtained from each participant using standardized protocols during a clinical examination (WHO, 2000). Body mass index (BMI, kg/m2) was calculated from body weight (kg) divided by the square of height (m). Peripheral obesity was defined as BMI ≥30kg/m2, and central obesity was defined as WC ≥102 cm in males and ≥88 cm in females (WHO, 2003).
Additionally, to ensure comparability with other studies and to test the robustness of our results, we also used the criteria for peripheral and central obesity recommended by the Working Group on Obesity in China (peripheral obesity: BMI ≥28 kg/m2; central obesity: WC ≥85 cm for men and ≥80 cm for women) (Zhou et al., 2002).
2.4. Exposure assessment
Community greenness was quantified by the Normalized Difference Vegetation index (NDVI;
Tucker, 1979) and the Soil Adjusted Vegetation Index (SAVI; Huete, 1988), which were derived from Landsat 5 Thematic Mapper satellite images at 30m×30m resolution (http://earthexplorer.usgs.gov). Both NDVI and SAVI are calculated using the land surface reflectance of near-infrared (NIR) and visible red (RED) light, and their values range from −1 to +1, with higher values indicating more greenness. Compared to NDVI, SAVI also included a soil adjustment factor to reduce the impact of background optical characteristics. We obtained two cloud-free Landsat 5 Thematic Mapper images taken in August 2010, as the greenest month in northeastern China and time closest to our health data collection. The mean value of NDVI or SAVI in 500m and 1000m circular buffers around each study community's centroid was defined as greenness. Communities were approximately 0.25–0.64 km2 and were separated by approximately 1.5 km within each study district. Thus, we focused on NDVI500-m as the primary measure of exposure although we also investigated NDVI1000-m in sensitivity analyses (Markevych et al., 2014; Swinburn et al., 2019). Greenness exposure was calculated in ArcGIS 10.4 (ESRI, Redlands, CA, USA).
2.5. covariates and candidate mediators
We defined potential confounding variables as common causes of greenness exposure and obesity, but not being on the pathways between greenness and obesity (Greenland et al., 1999).
We used a directed acyclic graph (DAG), constructed with DAGitty v2.3 software (Textor et al., 2016) to choose a minimum sufficient set of covariates to adjust for confounding and to
identify intervening variables between exposure and outcome based on the literature (Fig. S1).
We retained age (years), gender (woman vs. man), race (Han vs. others), and annual household income (<10,000 Yuan vs. ≥10,000 Yuan) as covariates in our main models, and air pollution (i.e., NO2 and PM2.5) and physical activity as candidate mediators.
There is a detailed description of the estimates of PM2.5 and NO2 exposure in our previous study (Yang et al., 2018; Yang et al., 2017b). In brief, we collected two types of daily aerosol optical depth (AOD) data with a spatial resolution of 0.1°×0.1° from the Aqua Atmosphere Level 2 Product Collection 1 (Deep Blue [DB] and Dark Target [DT]). We combined DB and DT AOD data using an inverse variance weighting method. Next, a generalized additive model was developed to link AOD data with ground-level PM2.5 measurements, land use information, meteorological data, vegetation data, and other spatial predictors (i.e., calendar month, elevation). A 10-fold cross-validation procedure indicated the adjusted R2 and root mean squared error were 75% and 15.1 μg/m3, respectively, for predicted PM2.5. We linearly interpolated PM2.5 values from the four nearest grid cells to accommodate pollutant migration.
As a result, different communities within the same grid cell had various predicted value of PM2.5
(i.e., depending on levels of the nearby grid cells). Measurement of NO2 was using chemiluminescence and reported per hour by air-monitoring stations located in each study district (Yang et al., 2017b). Three-year (2006–2008) average concentrations of PM2.5 and NO2
were then calculated for each study community. Physical activity information was collected from the study questionnaire (yes: exercised ≥180 min per week vs. no: exercised <180 min per week).
2.6. Statistical analysis
We characterized distributions for all variables and identified outlying observations for further examination. Spearman rank correlations were used to test the pairwise correlations among NDVI, SAVI and air pollutants. Based on recent published studies (Persson et al., 2018;
Shanahan et al., 2015), we hypothesized a linear relationship between greenness and obesity.
We used generalized linear mixed models to assess the associations of greenness exposure with BMI, WC and obesity prevalence (PROC GLIMMIX in SAS), with a random intercept to account for clustering within communities (Yang et al., 2018; Yang et al., 2017b). The results were presented as regression coefficients (β) for BMI and WC and odds ratios (ORs) for obesity prevalence, and their corresponding 95% confidence intervals (CIs), respectively. We used crude, unadjusted models and main models adjusted for the confounders selected by DAGs (Fig.
S1 and Fig. S2).
We conducted several sensitivity analyses to evaluate the robustness of the results. First, we repeated the analysis using NDVI1000-m and SAVI as alternative greenness indicators. Second, we used the definition of central and peripheral obesity as proposed by the Working Group on Obesity in China. Thirdly, we tested for non-linear relationships between greenness and obesity, employing NDVI500-m as categorical tertiles. Finally, we estimated the greenness-adiposity association after excluding participants who were underweight (BMI ≤ 18.5 kg/m2), participants who had cardiovascular diseases (defined as self-reported heart failure, coronary diseases or myocardial infarction) and participants who regularly adopted a low-calorie diet, to assess the potential impact of existing disorders and individual behavior.
The associations of greenness with obesity might differ in population subgroups (Fong et al., 2018), so we evaluated potential effect modification by age, gender, and household income level. Age was stratified as younger (<55 years) and older (≥55 years), and household income level was categorized as low-income (<10,000 Yuan/year) and high-income (≥10,000 Yuan/year). We incorporated cross-product terms into regression models (i.e., greenness ∗ age or greenness ∗ income) to identify important interactions and stratified the effect estimates for interpretation.
Our DAG and previous work (Markevych et al., 2017) suggested physical activity and air pollution as intervening variables linking greenness and human health, and so we incorporated these factors into mediation analyses. We used the mediation analysis macro for SAS which was extended from Baron and Kenny’s traditional statistical approach based on counterfactual framework and estimated the mediating effects of air pollutants and physical activity, along with their bias-corrected 95% confidence intervals of indirect paths (i.e., the mediating effect) with delta method (Valeri and Vanderweele, 2013). The proportion of the effect mediated was calculated as: (βindirect effects / βtotal effects) ×100%. PM2.5, NO2 and physical activity were tested one-at-a-time. Mediation analysis with a random intercept of community was used to account for the multi-level nature of the data. However, multi-level modelling for the mediator of NO2
failed to converge because of the feature of our data and it will probably lead to inaccurate estimates, we thus adjusted for community as fixed effects to partially offset the issue of clustering for NO2’s mediation effect.
All statistical analyses were conducted in SAS v9.4 (SAS Institute, Inc. Cary, NC, USA). A
two-tailed p-value less than 0.05 was considered statistically significant.
3. Results
3.1. Population characteristics
The mean age of participants was 45.6 years and nearly half were women (49.0%) (Table 1).
Most participants (76.8%) had a household income of ≥10,000 Yuan per year. The majority (94.5%) of the study population were of Han ethnicity and 30.8% of participants exercised regularly. BMI and WC were normally distributed with a mean value of 24.40 kg/m2 and 83.24 cm (an average of WC value of 86.18 cm in men and 80.18 cm in women), respectively.
Approximately, 5.8% had peripheral obesity and 14.0% had central obesity. Obese participants were more likely to be younger or have higher household income than normal weight participants (Table 1).
3.2.Greenness exposure
The distribution of greenness indicators in the 33 study communities is shown in Table S2.
There was obvious spatial heterogeneity in greenness values among different communities. For example, NDVI500-m values ranged from 0.18 to 0.80. In addition, the correlation between NDVI and SAVI were positive and strong (correlation coefficient ranged from 0.89 to 0.98), while the correlations between greenness marker and air pollutants (PM2.5 and NO2) were relatively weak and negative (correlation coefficient ranged from -0.05 to -0.43; Table S3).
3.3. Greenness and adiposity indicators
The associations between NDVI500-m and NDVI1000-m with indicators of adiposity are shown in
Table 2. In crude models, higher NDVI500-m and NDVI1000-m were consistently and significantly associated with lower BMI and lower WC, as well as with lower prevalence of peripheral and central obesities. The inverse associations remained significant for NDVI500-m and NDVI1000-m
with BMI and peripheral and central obesities after adjustment for age, gender, ethnicity, and income, but not for WC. More specifically, each IQR (0.17 unit) increase in NDVI500-m was associated with 0.18 kg/m2 (95% CI: −0.24 to −0.11) lower BMI, 20% (95% CI: 26%-13%) lower odds for peripheral obesity, and 12% (95% CI: 17%-7%) lower odds for central obesity.
3.4. Sensitivity analyses
In sensitivity analyses, the associations were similar for NDVI in 1000m buffer (Table 2) and for SAVI in both buffers (Table S4). The results were also consistent with those from the main models when we used the Working Group on Obesity in China definition of obesity (Table S5).
In addition, when NDVI500-m was categorized as tertile, we found significant trends between greenness level and obesity outcomes, with the effect estimates decreasing as NDVI500-m levels increased (Fig. 2). Further, estimates did not differ substantially after excluding underweight participants (BMI ≤ 18.5 kg/m2) or those with cardiovascular disease or on a low-calorie diet (Table S6).
3.5.Effect modification
In stratified analysis by age, we found stronger associations for NDVI500-m with peripheral obesity and central obesity among older participants (≥55 years) compared with younger participants (<55 years), but without heterogeneous effects for BMI and WC (Table 3). Gender
modified associations between greenness and BMI, WC, peripheral and central obesities where associations were protective and stronger for women although slightly unprotective, and close to the null for men. Significant interactions were also found for household income with NDVI500-m on BMI, WC, peripheral and central obesities, with stronger and protective effects among those with lower income (≤ 10,000 Yuan/year) than among those with higher income (>
10,000 Yuan/year).
3.6. Mediation analyses
Table 4 shows the results of our mediation analysis to quantify the proportion of NDVI500-m- obesity associations attributed to air pollutants. We found that ambient PM2.5 and NO2
significantly mediated 7.2% and 2.1% of the estimated association between greenness and BMI, respectively, and 11.1% and 20.8% of the estimated association between greenness and waist circumference, respectively. No significant mediating effects were observed for physical activity.
4. Discussion
4.1. Key findings
Our large population-based epidemiological study shows exposure to higher levels of community greenness were significantly associated with lower BMI and lower prevalence of peripheral and central obesities. Additionally, we found stronger beneficial associations among women, older participants, and among participants with lower household incomes. Air pollutants, not physical activity, were found to mediate the associations between greenness and obesity with a ratio of 15-23%. To the best of our knowledge, this is one of very few large population-based studies to have assessed the associations between community greenness and obesity in the developing countries.
4.2.Comparison with previous studies
We found that greater greenness exposure was significantly associated with lower BMI and lower prevalence of peripheral obesity.Our results are similar to those from 11 of 18（the 18 studies are detailed in Table S1）previously published studies focusing on associations between greenness and BMI among adults (Nielsen and Hansen, 2007; Paquet et al., 2014;
Pereira et al., 2013; Persson et al., 2018; Prince et al., 2011; Rubin et al., 2005; Sarkar, 2017;
Tilt et al., 2007; Toftager et al., 2011; Tsai et al., 2016; Villeneuve et al., 2018), all of which reported a significant protective association between greenness and BMI or peripheral obesity (Table S1). However, contrary to our results, two studies observed an inverse pattern for greenness and general obesity (Cummins and Fagg, 2012; Prince et al., 2011). Two previous
studies reported non-linear relationships between greenness exposure and obesity, in which moderate level greenness exposure was associated with higher obesity prevalence but high- level greenness exposure was associated with lower obesity prevalence (James et al., 2017;
Klompmaker et al., 2018). Further, three studies reported no association between greenness and BMI or peripheral obesity (Coombes et al., 2010; Mowafi et al., 2012; Persson et al., 2018).
We also detected inverse associations of greenness with central obesity prevalence, while there are fewer studies in this domain. In line with our results, a prospective cohort study of 5,126 Swedish adults reported inverse associations of greenness with central obesity (Persson et al., 2018). However, another cohort study found no significant association between greenness and WC-defined central obesity among 3,205 Australian adults (Paquet et al., 2014).
Collectively, our results are consistent with most of previously published studies from other, mostly developed, nations. Discrepancies in study results might be attributed in part to differences in study populations (e.g., race, gender proportion, age, and lifestyle factors), various exposure assessment strategies (e.g., NDVI, distance to green space and percentage of green space), and the presence of absence of other co-exposures (e.g., air pollutants, noise, and psychosocial stress). Still the weight of evidence supports an association between greater greenness exposure and less adiposity (specifically BMI), peripheral obesity, and central obesity. However, most previous studies have been cross-sectional to date and therefore further comprehensive longitudinal studies are needed to confirm our findings.
4.3.Susceptible populations
The inverse associations between greenness exposure and obesity were strongest among older participants and women, with modest positive or null associations among young participants and men. However, the evidence for effect modification by age, gender and household income levels on association between greenness and obesity is limited and inconsistent. Most previous studies generally incorporated effect modifiers as covariates and only reported their adjusted effects. For example, a cross-sectional study of more than 333,000 participants in the UK found a stronger inverse association between greenness and obesity among middle aged (51-60 years) participants compared to older (> 60 years) and younger (≤ 50 years) participants (Sarkar, 2017) and an earlier cross-sectional study of 77,000 adult UK participants found no differences in the greenness-obesity associations by age (Cummins and Fagg, 2012).
Several previous investigations have suggested stronger greenness-adiposity associations among women than among men especially the pronounced beneficial effects that we identified with respect to BMI, WC and obesity (Astell-Burt et al., 2014; Sarkar, 2017; Wen and Maloney, 2011). Despite this, our findings are not unexpected. In China, retired urban residents and women may more likely to use green spaces than the younger working population and men. For example, walking or square dance is one of the favorite recreational activities for middle-aged- to-old adults, especially for women (Gao et al., 2016). They tend to perform these activities in nearby green space such as parks. Older residents or women in China may also spend more time in surrounding green area for childcare obligations (Tamosiunas et al., 2014).
We also found that the inverse associations of greenness exposure with obesity were more pronounced among participants with lower household incomes than participant with higher
household income. An study of 333,183 British participants reported similar results (Sarkar, 2017). However, a cross-sectional study of more than 97 million resident of 496 U.S. cities did not detect modification effects on associations between greenness and city-level ecological obesity data by income levels (Browning and Rigolon, 2018), nor did the aforementioned cross- sectional investigation of 77,000 UK adults (Cummins and Fagg, 2012). Effect modification by income may be due to the fact that less affluent individuals are prone to be less mobile and therefore more dependent on their neighborhood greenspace compared to more affluent individuals (Maas et al., 2008; Schwanen et al., 2002).
4.4.Underlying mechanism
Although the mechanisms through which green space may benefit health are still unclear, a series of potential biopsychosocial pathways have been suggested by previous study (Markevych et al., 2017). First, greenness can reduce the level of ambient air pollution (Hirabayashi and Nowak, 2016), greater exposure to which was related to higher risk of obesity (Wei et al., 2016), and greener areas may, in some cases, have fewer pollutant sources (e.g., parks and forests). Consistent with this hypothesis, we found that ambient PM2.5 and NO2
concentrations partially mediated greenness-obesity associations. Second, the availability of green spaces (e.g., parks), has been associated with physical activity such that greater proximity was correlated to more physical activity (Lachowycz and Jones, 2011), also be a protective factor for obesity. Yet, the evidence for the mediating effects of physical activity was not detected. Two previously published large cross-sectional studies also reported associations between quantity of green space or green space proximity and adult BMI were independent of
physical activity (Astell-Burt et al., 2014; Cummins and Fagg, 2012). However, a large cross- sectional study of land characteristics reported that approximately 32% of the association between greenness and obesity was mediated by physical activity (Villeneuve et al., 2018).
Possible explanations for this discrepancy may be: (1) different ways to assess and evaluate the level of physical activity might contribute to the difference of its results of mediating effect.
Both our and Cummins and Fagg’s large cross-sectional studies quantified physical activity as categorical variable (e.g. low, moderate or high). While physical activity was presented as sessions per week calculated with a sum of moderate and vigorous forms of physical activity and their respective time of duration (Astell-Burt et al., 2014) and proxied with metabolic equivalent task hours per week (Villeneuve et al., 2018); (2) the type and quality of available green space may be an important factor in determining the exercise behavior which could further influence weight status (Coombes et al., 2010), for the characteristics of greenness could vary across different study sites. For instance, high-volume roads lined with trees are captured by satellite-derived greenness, but offer little opportunity for physical activity and may in fact reduce its mediation effect; (3) although physical activity is an important factor when evaluating the association between greenness and obesity, the level of healthy exercise motivated by greenness in our study (e.g. a short walk in the park every morning) may not result in significantly increased energy expenditure (Heneberg, 2014). Third, evidence suggests that greenness is also associated with greater social cohesion, reduced exposure to noise and decreased mental and physical stress (Markevych et al., 2017; Rook, 2013). As no mediation effect was found for physical activity, the calming effects of being proximity to greenness might play an important protective role on the pathway linking greenness with obesity (Pun et al.,
2018; van der Valk et al., 2018). However, due to the lack of these data in the study, we are thus unable to investigate them as potential mediators on the link between greenness and obesity.
Further well-captured variables of physical activity and stress are needed in future study to validate our hypothesis.
4.5.Strengths and Limitations
Our study has four main strengths. We enrolled a large and representative sample of Chinese urban-dwelling adults with a high response rate (86.2%), which allowed sufficient statistical power to detect moderate effects and to reduce potential selection-related biases. Apart from examining general obesity like in most prior studies, we further evaluated the beneficial effects of greenness on abdominal obesity, and found that the results were consistent. Additionally, a parsimonious, yet comprehensive panel of covariates was adjusted to minimize confounding in our results without introducing further bias (Schisterman et al., 2009). Finally, we conducted moderation and mediation analyses with factors selected a priori based on potential to modify or mediate associations between greenness and obesity. We employed a conditional mediation analysis approach to quantify the proportion of effects mediated by air pollutants, which tends to provide more precise estimates of mediation effects than the traditional approach of Baron and Kenny (Baron and Kenny, 1986).
There are also several limitations in our study and our results should be interpreted with those in mind. First, due to the cross-sectional design of our study, we could not determine if the found relationships are causal, including the inability to evaluate temporality between
more active individuals chose to reside in greener communities in our study (Boone-Heinonen et al., 2011). However, physical activity was not a mediator in our study in our study so we believe that any impact of reverse causality will be modest. Second, greenness exposure was measured at the community level leading to 33 unique data points, which may have misclassified personal exposure for some study participants. However, any such misclassification will be non-differential and will bias our effect estimates towards the null (Hutcheon et al., 2010), which indicates our estimates are more likely to be conservative. Third, Physical activity information was dichotomous (whether exercise > 180 minutes per week). The lack of detailed data about duration, time, and site (e.g., inside or outside the house) of physical activity and sedentary behaviors prevents us from evaluating the mediation effect of physical activity more precisely. Additional investigation capturing a more nuanced profile of physical activity is necessary to more clearly define the role as an intervening variable in greenness- obesity associations in China. Fourth, NDVI and SAVI are sensitive to season and do not informative about the quality, structure, and composition of greenspaces and vegetation. We were therefore unable to identify the associations between the different types of greenspace and obesity and unable to evaluate how different types of greenspace may differentially affect obesity-related behaviours. In addition, we did not exclude blue space from the NDVI assessment, which might have confounded the effects of green space on obesity. Finally, given the wide range of factors affecting obesity, there is still possibility for residual confounding by other unmeasured factors, such as environmental noise.
5. Conclusion
In summary, greater community greenness levels were associated with lower BMI and peripheral and central obesity prevalence in Chinese adults. In particular, the impacts appeared to be most substantial especially among women, older individuals, and for those with lower household incomes. Air pollution, but not physical activity, was found to be partially mediated the associations. Our findings highlight the importance of integrating preventive health strategies into urban designs to mitigate the growing obesity epidemic in China, and to target prevention efforts in vulnerable subpopulations.
Declaration of interests
None
Acknowledgements
The research was funded by the National Natural Science Foundation of China (No.81872582;
No.81872583; No.81703179; No.81673128); the Guangdong Provincial Natural Science Foundation Team Project (2018B030312005); the Science and Technology Planning Project of Guangdong Province (No.2018B05052007; 2017A050501062); and Science and Technology Program of Guangzhou (201807010032; 201803010054). The authors acknowledge the cooperation of participants in this study who have been very generous with their time and assistance.
References
Archer, E.; Lavie, C.J.; Hill, J.O. The Contributions of 'Diet', 'Genes', and Physical Activity to the Etiology of Obesity: Contrary Evidence and Consilience. Prog Cardiovasc Dis 2018;61:89-102
Astell-Burt, T.; Feng, X.; Kolt, G.S. Greener neighborhoods, slimmer people? Evidence from 246,920 Australians. Int J Obes (Lond) 2014;38:156-159
Baron, R.M.; Kenny, D.A. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol 1986;51:1173-1182
Boone-Heinonen, J.; Gordon-Larsen, P.; Guilkey, D.K.; Jacobs, D.R., Jr.; Popkin, B.M.
Environment and Physical Activity Dynamics: The Role of Residential Self-selection.
Psychol Sport Exerc 2011;12:54-60
Browning, M.; Rigolon, A. Do Income, Race and Ethnicity, and Sprawl Influence the Greenspace-Human Health Link in City-Level Analyses? Findings from 496 Cities in the United States. Int J Environ Res Public Health 2018;15
Coombes, E.; Jones, A.P.; Hillsdon, M. The relationship of physical activity and overweight to objectively measured green space accessibility and use. Soc Sci Med 2010;70:816-822
Chinese Center For Disease Control And Prevention: Chinese Center For Disease Control And Prevention report: Monitoring data of chronic diseases and their risk factors. 2014
Cummins, S.; Fagg, J. Does greener mean thinner? Associations between neighbourhood greenspace and weight status among adults in England. Int J Obes (Lond) 2012;36:1108-1113
Di Cesare, M.; Bentham, J.; Stevens, G.A.; Zhou, B.; Danaei, G.; Lu, Y., et al. Trends in adult body-mass index in 200 countries from 1975 to 2014: a pooled analysis of 1698 population-based measurement studies with 19.2 million participants. Lancet 2016;387:1377-1396
Dong, G.H.; Qian, Z.M.; Xaverius, P.K.; Trevathan, E.; Maalouf, S.; Parker, J., et al.
Association Between Long-Term Air Pollution and Increased Blood Pressure and Hypertension in China. Hypertension 2013;61:578-584
Fong, K.C.; Hart, J.E.; James, P. A Review of Epidemiologic Studies on Greenness and Health:
Updated Literature Through 2017. Curr Environ Health Rep 2018;5:77-87
Foraster, M.; Eze, I.C.; Vienneau, D.; Schaffner, E.; Jeong, A.; Heritier, H., et al. Long-term exposure to transportation noise and its association with adiposity markers and development of obesity. Environ Int 2018;121:879-889
Fox, C.S.; Massaro, J.M.; Hoffmann, U.; Pou, K.M.; Maurovich-Horvat, P.; Liu, C.Y., et al.
Abdominal visceral and subcutaneous adipose tissue compartments: association with metabolic risk factors in the Framingham Heart Study. Circulation 2007;116:39-48
Gao, L.; Zhang, L.; Qi, H.; Petridis, L. Middle-aged Female Depression in Perimenopausal Period and Square Dance Intervention. Psychiatr Danub 2016;28:372-378
Greenland, S.; Pearl, J.; Robins, J.M. Causal diagrams for epidemiologic research.
Epidemiology 1999;10:37-48
Guo, C.; Wang, H.; Feng, G.; Li, J.; Su, C.; Zhang, J., et al. Spatiotemporal predictions of obesity prevalence in Chinese children and adolescents: based on analyses of obesogenic environmental variability and Bayesian model. Int J Obes (Lond) 2018;
Heneberg, P. Energy expenditure of hunter-gatherers: when statistics turns to be unreliable.
Endocrine, metabolic & immune disorders drug targets 2014;14:152-158
Hinney, A.; Herrfurth, N.; Schonnop, L.; Volckmar, A.L. [Genetic and epigenetic mechanisms in obesity]. Bundesgesundheitsblatt Gesundheitsforschung Gesundheitsschutz 2015;58:154-158
Hirabayashi, S.; Nowak, D.J. Comprehensive national database of tree effects on air quality and human health in the United States. Environmental Pollution 2016;215:48-57
Huete, A.R. A soil-adjusted vegetation index (SAVI). Remote Sensing of Environment 1988;25:295-309
Hutcheon, J.A.; Chiolero, A.; Hanley, J.A. Random measurement error and regression dilution bias. BMJ 2010;340:c2289
James, P.; Kioumourtzoglou, M.A.; Hart, J.E.; Banay, R.F.; Kloog, I.; Laden, F.
Interrelationships Between Walkability, Air Pollution, Greenness, and Body Mass Index. Epidemiology 2017;28:780-788
Jiang, Y.; Xu, Y.; Bi, Y.; Wang, L.; Zhang, M.; Zhou, M., et al. Prevalence and trends in overweight and obesity among Chinese adults in 2004–10: data from three nationwide surveys in China. The Lancet 2015;386:S77
Klompmaker, J.O.; Hoek, G.; Bloemsma, L.D.; Gehring, U.; Strak, M.; Wijga, A.H., et al.
Green space definition affects associations of green space with overweight and physical activity. Environ Res 2018;160:531-540
Lachowycz, K.; Jones, A.P. Greenspace and obesity: a systematic review of the evidence. Obes Rev 2011;12:e183-189
Maas, J.; Verheij, R.A.; Spreeuwenberg, P.; Groenewegen, P.P. Physical activity as a possible mechanism behind the relationship between green space and health: A multilevel analysis. BMC Public Health 2008;8:206
Markevych, I.; Schoierer, J.; Hartig, T.; Chudnovsky, A.; Hystad, P.; Dzhambov, A.M., et al.
Exploring pathways linking greenspace to health: Theoretical and methodological guidance. Environ Res 2017;158:301-317
Markevych, I.; Tiesler, C.M.T.; Fuertes, E.; Romanos, M.; Dadvand, P.; Nieuwenhuijsen, M.J., et al. Access to urban green spaces and behavioural problems in children: Results from the GINIplus and LISAplus studies. Environ Int 2014;71:29-35
Mowafi, M.; Khadr, Z.; Bennett, G.; Hill, A.; Kawachi, I.; Subramanian, S.V. Is access to neighborhood green space associated with BMI among Egyptians? A multilevel study of Cairo neighborhoods. Health Place 2012;18:385-390
National Bureau of Statistics, 2016. Statistical Communiqué of China on 2015 National Economic and Social Development.
Nielsen, T.S.; Hansen, K.B. Do green areas affect health? Results from a Danish survey on the use of green areas and health indicators. Health Place 2007;13:839-850
Paquet, C.; Coffee, N.T.; Haren, M.T.; Howard, N.J.; Adams, R.J.; Taylor, A.W., et al. Food environment, walkability, and public open spaces are associated with incident development of cardio-metabolic risk factors in a biomedical cohort. Health Place 2014;28:173-176
Pereira, G.; Christian, H.; Foster, S.; Boruff, B.J.; Bull, F.; Knuiman, M., et al. The association between neighborhood greenness and weight status: an observational study in Perth Western Australia. Environ Health 2013;12:49
Persson, A.; Pyko, A.; Lind, T.; Bellander, T.; Ostenson, C.G.; Pershagen, G., et al. Urban residential greenness and adiposity: A cohort study in Stockholm County. Environ Int 2018;121:832-841
Pischon, T.; Boeing, H.; Hoffmann, K.; Bergmann, M.; Schulze, M.B.; Overvad, K., et al.
General and abdominal adiposity and risk of death in Europe. N Engl J Med 2008;359:2105-2120
Potestio, M.L.; Patel, A.B.; Powell, C.D.; McNeil, D.A.; Jacobson, R.D.; McLaren, L. Is there an association between spatial access to parks/green space and childhood overweight/obesity in Calgary, Canada? Int J Behav Nutr Phys Act 2009;6:77
Prince, S.A.; Kristjansson, E.A.; Russell, K.; Billette, J.M.; Sawada, M.; Ali, A., et al. A multilevel analysis of neighbourhood built and social environments and adult self- reported physical activity and body mass index in Ottawa, Canada. Int J Environ Res Public Health 2011;8:3953-3978
Pun, V.C.; Manjourides, J.; Suh, H.H. Association of neighborhood greenness with self- perceived stress, depression and anxiety symptoms in older U.S adults. Environ Health 2018;17:39
Rook, G.A. Regulation of the immune system by biodiversity from the natural environment: an ecosystem service essential to health. Proc Natl Acad Sci U S A 2013;110:18360-18367
Rubin, G.J.; Brewin, C.R.; Greenberg, N.; Simpson, J.; Wessely, S. Psychological and behavioural reactions to the bombings in London on 7 July 2005: cross sectional survey of a representative sample of Londoners. BMJ 2005;331:606
Sarkar, C. Residential greenness and adiposity: Findings from the UK Biobank. Environ Int 2017;106:1-10
Schisterman, E.F.; Cole, S.R.; Platt, R.W. Overadjustment bias and unnecessary adjustment in epidemiologic studies. Epidemiology 2009;20:488-495
Schreier, N.; Moltchanova, E.; Forsen, T.; Kajantie, E.; Eriksson, J.G. Seasonality and ambient temperature at time of conception in term-born individuals - influences on cardiovascular disease and obesity in adult life. Int J Circumpolar Health 2013;72:21466
Schwanen, T.; Dijst, M.; Dieleman, F.M. A Microlevel Analysis of Residential Context and Travel Time. Environment and Planning A: Economy and Space 2002;34:1487-1507
Shanahan, D.F.; Fuller, R.A.; Bush, R.; Lin, B.B.; Gaston, K.J. The Health Benefits of Urban Nature: How Much Do We Need? Bioscience 2015;65:476-485
Su, C.; Jia, X.F.; Wang, Z.H.; Wang, H.J.; Ouyang, Y.F.; Zhang, B. Longitudinal association of leisure time physical activity and sedentary behaviors with body weight among Chinese adults from China Health and Nutrition Survey 2004-2011. Eur J Clin Nutr 2017;71:383-388
Swinburn, B.A.; Kraak, V.I.; Allender, S.; Atkins, V.J.; Baker, P.I.; Bogard, J.R., et al. The Global Syndemic of Obesity, Undernutrition, and Climate Change: The Lancet Commission report. Lancet 2019;393:791-846
Tamosiunas, A.; Grazuleviciene, R.; Luksiene, D.; Dedele, A.; Reklaitiene, R.; Baceviciene, M., et al. Accessibility and use of urban green spaces, and cardiovascular health:
findings from a Kaunas cohort study. Environmental Health 2014;13:20
Textor, J.; van der Zander, B.; Gilthorpe, M.S.; Liskiewicz, M.; Ellison, G.T.H. Robust causal inference using directed acyclic graphs: the R package 'dagitty'. Int J Epidemiol 2016;45:1887-1894
Tilt, J.H.; Unfried, T.M.; Roca, B. Using objective and subjective measures of neighborhood greenness and accessible destinations for understanding walking trips and BMI in Seattle, Washington. Am J Health Promot 2007;21:371-379
Toftager, M.; Ekholm, O.; Schipperijn, J.; Stigsdotter, U.; Bentsen, P.; Gronbaek, M., et al.
Distance to green space and physical activity: a Danish national representative survey.
J Phys Act Health 2011;8:741-749
Tsai, W.L.; Floyd, M.F.; Leung, Y.F.; McHale, M.R.; Reich, B.J. Urban Vegetative Cover Fragmentation in the U.S.: Associations With Physical Activity and BMI. Am J Prev Med 2016;50:509-517
Tucker, C.J. Red and photographic infrared linear combinations for monitoring vegetation.
Remote Sensing of Environment 1979;8:127-150
Twohig-Bennett, C.; Jones, A. The health benefits of the great outdoors: A systematic review and meta-analysis of greenspace exposure and health outcomes. Environ Res 2018;166:628-637
Valeri, L.; Vanderweele, T.J. Mediation analysis allowing for exposure-mediator interactions and causal interpretation: theoretical assumptions and implementation with SAS and SPSS macros. Psychological methods 2013;18:137-150
van der Valk, E.S.; Savas, M.; van Rossum, E.F.C. Stress and Obesity: Are There More Susceptible Individuals? Curr Obes Rep 2018;7:193-203
Villeneuve, P.J.; Jerrett, M.; Su, J.G.; Weichenthal, S.; Sandler, D.P. Association of residential greenness with obesity and physical activity in a US cohort of women. Environ Res 2018;160:372-384
Wei, Y.J.; Zhang, J.F.; Li, Z.G.; Gow, A.; Chung, K.F.; Hu, M., et al. Chronic exposure to air pollution particles increases the risk of obesity and metabolic syndrome: findings from a natural experiment in Beijing. Faseb J 2016;30:2115-2122
Wen, M.; Maloney, T.N. Latino residential isolation and the risk of obesity in Utah: the role of neighborhood socioeconomic, built-environmental, and subcultural context. J Immigr Minor Health 2011;13:1134-1141
World Health Organization, WHO Technical Report Series NO. 894, 2000. Obesity: preventing and managing the global epidemic.
World Health Organizations, Geneva, 2003. WHO Technical Report Series NO. 916. Diet, nutrition and the prevention of chronic diseases.
World Health Organizations website. http://www.who.int/en/news-room/fact- sheets/detail/obesity-and-overweight.
Wilhelmsen, C.K.; Skalleberg, K.; Raanaas, R.K.; Tveite, H.; Aamodt, G. Associations between green area in school neighbourhoods and overweight and obesity among Norwegian adolescents. Prev Med Rep 2017;7:99-105
Yang, B.Y.; Liu, Y.; Hu, L.W.; Zeng, X.W.; Dong, G.H. Urgency to Assess the Health Impact of Ambient Air Pollution in China. Adv Exp Med Biol 2017a;1017:1-6
Yang, B.Y.; Qian, Z.M.; Li, S.; Chen, G.; Bloom, M.S.; Elliott, M., et al. Ambient air pollution in relation to diabetes and glucose-homoeostasis markers in China: a cross-sectional
study with findings from the 33 Communities Chinese Health Study. Lancet Planet Health 2018;2:e64-e73
Yang, B.Y.; Qian, Z.M.; Vaughn, M.G.; Nelson, E.J.; Dharmage, S.C.; Heinrich, J., et al. Is prehypertension more strongly associated with long-term ambient air pollution exposure than hypertension? Findings from the 33 Communities Chinese Health Study. Environ Pollut 2017b;229:696-704
Zhou, B.F.; Cooperative Meta-Analysis Group of the Working Group on Obesity in, C.
Predictive values of body mass index and waist circumference for risk factors of certain related diseases in Chinese adults--study on optimal cut-off points of body mass index and waist circumference in Chinese adults. Biomed Environ Sci 2002;15:83-96
Zhou, D.C.; Zhao, S.Q.; Liu, S.G.; Zhang, L.X. Spatiotemporal trends of terrestrial vegetation activity along the urban development intensity gradient in China's 32 major cities. Sci Total Environ 2014;488:136-145
Figure legends
Fig.1. Sampling procedure for the 33 Communities Chinese Health Study.
Fig.2. Associations between tertiles of NDVI500-m and obesity metrics in the 33 Chinese Communities Health Study. A, peripheral obesity; B, central obesity; The associations were adjusted by age, gender, ethnicity and household income level. (Q1: quartile 1-reference category; Q2: quartile 2; Q3: quartile 3 with boxes representing the effect estimate of each quartile and whiskers representing the 95% confidence interval).
Abbreviations: CI, confidence interval; OR, odds ratio.
Fig.3. Associations of per 0.17-unit increase in NDVI500-m with adiposity and obesity indicators, stratified by age, gender, and household income level. A, BMI; B, WC; C, peripheral obesity defined as BMI ≥30 kg/m2; D, central obesity defined as WC >102 cm in men and >88 cm in women.
Abbreviations: CI, confidence interval; OR, odds ratio; β, unstandardized regression coefficient.
*Interaction is statistically significant (P <0.05).
Table 1 Characteristics of study participants from the 33 Chinese Communities Health Study, by peripheral obesitya
Obesitya Non-obesitya Total
Characteristics (n = 1435) (n =23,410) p-value (n = 24,845) Age (n, %)b
<55 years 1007 (70.2) 17,691 (75.6) <0.001 18,698 (75.3)
≥55 years 428 (29.8) 5719 (24.4) 6147 (24.7)
Sex (n, %)
Men 738 (51.4) 11,923 (50.9) 0.715 12,661 (51.0)
Women 697 (48.6) 11,487 (49.1) 12,184 (49.0)
Nationality (n, %)
Han 1358 (94.6) 22,112 (94.5) 0.774 23,470 (94.5)
Others 77 (5.4) 1298 (5.5) 1375 (5.5)
Household income levels per year (n, %)b
<10,000 Yuan 394 (27.5) 5367 (22.9) <0.001 5761(23.2)
≥ 0,000 Yuan 1041 (72.5) 18,043 (77.1) 19,084(76.8) Regular exercise (n, %)
No 1010 (70.4) 16,188 (69.2) 0.326 17,198 (69.2)
Yes 425 (29.6) 7222 (30.8) 7647 (30.8)
BMI, kg/m2 (mean ±
SD)b 32.8 ± 3.83 23.9 ± 3.02 <0.001
24.4 ± 3.70 WC, cm (mean ± SD)b 99.8 ± 9.73 82.2 ± 9.78 <0.001 83.2 ± 10.61 Abbreviations: BMI, body mass index; WC, waist circumference; SD, standard deviation.
aObesity defined as BMI ≥30 kg/m2
bP <0.05 for difference between obese and non-obese groups.
Table 2 Associations between an IQRa increase in NDVI and adiposity and obesity metrics from the 33 Chinese Communities Health Study (n=24,845)
β (95% CI) OR (95% CI)
Model BMI WC Peripheral obesityb Central obesityc
NDVI500-m
Crude -0.31 (-0.37, -0.24) -1.24 (-1.43, -1.05) 0.79 (0.73, 0.86) 0.98 (0.93, 1.03) Adjustedd -0.18 (-0.24, -0.11) -0.14 (-0.32, 0.04) 0.80 (0.74, 0.87) 0.88 (0.83, 0.93) NDVI1000-m
Crude -0.34 (-0.41, -0.28) -1.19 (-1.38, -1.00) 0.77 (0.71, 0.83) 0.99 (0.94, 1.04) Adjustedd -0.21 (-0.28, -0.15) -0.11 (-0.30, 0.07) 0.78 (0.72, 0.85) 0.89 (0.84, 0.95) Abbreviations: BMI, body mass index; CI, confidence interval; IQR, interquartile range; NDVI, normalized difference vegetation index; OR, odds ratio; β, unstandardized regression coefficient;
WC, waist circumference.
aIQR = 0.17-unit for NDVI500-m, and 0.15-unit for NDVI1000-m.
bPeripheral obesity defined as BMI ≥30 kg/m2.
cCentral obesity defined as WC > 102 cm in men and > 88 cm in women.
dAdjusted for age, gender, ethnicity and household income.
Table 3 Associations between an IQRa increase in NDVI500-m and obesity metrics by age, gender and household income from the 33 Chinese Communities Health Study (n = 24,845)
β (95% CI)b OR (95% CI)b
Group BMI P for
interaction WC P for
interaction
Peripheral obesityd
P for
interaction Central obesitye P for interaction
Age 0.018 0.560 0.065 0.345
<55 years -0.12 (-0.19, -0.04)c 0.05 (-0.15, 0.26) 0.84 (0.76, 0.93) 0.90 (0.84, 0.96)
≥55 years -0.34 (-0.49, -0.19)c -0.63 (-1.03, -0.22) 0.76 (0.64, 0.90) 0.86 (0.77, 0.97)
Gender <0.001 <0.001 <0.001 0.002
Men 0.06 (-0.04, 0.16)c 0.66 (0.40, 0.92)c 0.92 (0.83, 1.03)c 0.98 (0.88, 1.10)c
Women -0.25 (-0.34, -0.17)c -0.46 (-0.70, -0.22)c 0.74 (0.65, 0.84)c 0.88 (0.82, 0.94)c
Household
income 0.001 0.023 0.010 0.010
<10,000 Yuan -0.48 (-0.66, -0.30)c -1.02 (-1.46, -0.58)c 0.58 (0.46, 0.72)c 0.72 (0.63, 0.82)c
≥10,000 Yuan -0.08 (-0.15, -0.01)c 0.21 (0.02, 0.41)c 0.85 (0.78, 0.93)c 0.94 (0.88, 1.00)c
BMI, body mass index; CI, confidence interval; IQR, interquartile range; NDVI, normalized difference vegetation index; OR, odds ratio; WC, waist circumference
aIQR was 0.17-unit for NDVI500-m.
bAdjusted for age, gender, ethnicity and household income levels.
cP <0.05 for interaction between NDVI500-mand covariate.
dPeripheral obesity defined as BMI ≥30 kg/m2.
eCentral obesity defined as WC >102 cm in men and > 88 cm in women.
Table 4 Indirect effects linking greenness (NDVI500-m) to BMI and WC from the 33 Chinese Communities Health Study (n = 24,845)
Indirect path β (95% CI)a Indirect/Total effects (%)a
BMI WC BMI WC
PM2.5 -0.0136 (-0.0256, -0.0001)b -0.0079 (-0.0425, 0.0300) 7.2 2.1 NO2 -0.0197 (-0.0293, -0.0102)b -0.0450 (-0.0706, -0.0193)b 11.1 20.8
PA -0.0003 (-0.0015, 0.0001) -0.0040 (-0.0108, 0.0001) 0.1 2.5
Abbreviations: BMI, body mass index; NO2, nitrogen dioxide; NDVI, normalized difference vegetation index; PA, physical activity; PM2.5, particles with aerodynamic diameter ≤ 2.5 µm; β, unstandardized regression coefficient; WC, waist circumference.
aAdjusted for age, gender, ethnicity and household income levels.
bP <0.05.
Highlights
New evidence on the associations between greenness and obesity from developing world.
The study was conducted in 24845 adults in northeastern China in 2009.
Greenness was beneficially associated with both central and peripheral obesity in China.
The beneficial associations were stronger among women, older participants, and those with lower household incomes.
Air pollution partially mediated the associations between greenness and obesity.
Graphical abstract
