Impact of built environment on physical activity and obesity among children and adolescents in China:A narrative systematic review

Ruopeng An*,Jing Shen,Qiuying Yng,Yn Yng

a Department of Kinesiology and Community Health,University of Illinois at Urbana-Champaign,Champaign,IL 61820,USA

bCollege of Education,Beijing Sport University,Beijing 100084,China

c College of Physical Education and Health,East China Normal University,Shanghai 200241,China

Abstract Background:Neighborhood built environment may profoundly influence children’s physical activity(PA)and body weight.This study systematically reviewed scientific evidence regarding the impact of built environment on PA and obesity among children and adolescents in China.Methods:A keyword and reference search was conducted in Active Living Research,Cochrane Library,PubMed,and Web of Science.Studies that met all of the following criteria were included in the review:(1)study designs—experimental studies,observational studies,and qualitative studies; (2) study subjects—Chinese children and/or adolescents aged ≤17 years; (3) exposures—neighborhood built environment; (4)outcomes—PA and/or body weight status; (5) article type—peer-reviewed publications; (6) time window of search—from the inception of an electronic bibliographic database to May 31,2018;(7)country—China;and(8)language—articles written in English.Results:A total of 20 studies,including 16 cross-sectional studies,3 longitudinal studies,and 1 descriptive study,met the predetermined selection criteria and were included in the review. A total of 13 studies adopted subjective built environment measures reported by parents and/or children,2 adopted objective measures(e.g.,geographical information system,field observations),and 5 adopted both objective and subjective measures.PA behaviors included PA,physical inactivity,sedentary behavior,active/passive commuting from/to school,and park visits.Among the 16 studies that provided some quantitative estimates of the influence of built environment on PA and body weight status,all reported a statistically significant relationship in the expected direction.Availability and accessibility in proximity to greenspaces,parks,recreational facilities,and sidewalks were found to be associated with increased PA levels, reduced sedentary behaviors, and/or active commuting among Chinese children and adolescents.In contrast,the absence of bike lanes and living in higher density residential areas were associated with increased likelihood of childhood overweight and obesity.Conclusion:Neighborhood built environment plays an important role in Chinese children’s PA engagement and weight outcomes.Building new exercise facilities and enhancing the accessibility of existing facilities hold the potential to enhance PA engagement among Chinese children and adolescents.In addition,urban designs that incorporate sidewalks,bike lanes,walking paths,less motorized traffic,and lower residential density are likely to promote PA and prevent childhood obesity in China.2095-2546/© 2019 Published by Elsevier B.V.on behalf of Shanghai University of Sport.This is an open access article under the CC BY-NC-ND license.(http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Body weight;Chinese;Exercise;Literature review;Local environment;Neighborhood environment;Physical environment edEn

1. Introduction

Obesity is a leading cause of morbidity and premature mortality worldwide.1-4Childhood obesity is linked to various immediate and long-term adverse health outcomes such as sleep apnea,hypertension,heart disease,stroke,type 2 diabetes,osteoarthritis,and certain types of cancer,and it leads to social and psychological problems such as stigmatization and poor selfesteem.5,6In 2010, 43 million children worldwide (35 million in developing countries) were estimated to be overweight or obese.7By 2020, the prevalence of childhood overweight/obesity is projected to reach 9%,or 60 million children worldwide.8Based on the 2017 Report on Childhood Obesity in China, the prevalence of childhood overweight increased from 2.1% in 1985 to 12.2%in 2014 among Chinese children aged ≥7 years,and the prevalence of childhood obesity increased from 0.5%to 7.3% during the same time period, resulting in an increase in overweight/obese children from 6.2 million to 35.0 million nationwide.9Based on the Physical Activity and Fitness in China—the Youth Study,the prevalence of obesity among Chinese school children and adolescents aged 9-17 years reached 12%,and approximately 37%of them did not meet screen time viewing recommendations.10It is projected that the prevalence of childhood overweight/obesity will reach 28%,or 49.5 million children,in 2030.9

An active lifestyle helps to decrease the risk of obesity in both adults and children.11In contrast, physical inactivity is a leading risk factor for morbidity and mortality.12-14A vast majority of children and adults in both developed and developing countries fall short of the meeting guidelinesrecommended levels for physical activity(PA).15The relationship between built environment and PA among children in general has been extensively documented.16,17Accumulating evidence suggests that the availability of parks, playgrounds,gyms, and other facilities, as well as other neighborhood features such as walkability and safety, profoundly shape children’s daily PA patterns.18,19Roemmich et al.20found that residential locations closer to parks were associated with higher PA levels in young children than those farther away from parks. Reimers et al.21reported that adolescent girls residing far away from a gym were less likely to engage in indoor sports activities than those residing near a gym.Molnar et al.22found that poorer neighborhood safety was negatively associated with PA among youth aged 11-16 years.

The majority of studies on the relationship between built environment and PA focus on adults and children in developed countries(e.g.,the United States),16-19whereas relevant studies on the child population in developing countries remain limited.The characteristic and quality of various types of neighborhood built environment can noticeably differ between developed and developing countries,23which may differentially impact the PA and obesity patterns among local residents.In the coming decades, China is expected to undergo an unprecedented urbanization and urban expansion.24This circumstance creates a critical opportunity for urban planners to design smart cities and environments conducive to an active lifestyle among residents,including children and adolescents.24However,this grand goal would not be accomplished without a profound understanding of the relationship between built environment and PA. Moreover, cultural norms in China may play an important role in children’s PA patterns.25For example, Confucian ideology regards sport and PA participation as predominantly masculine in nature,which results in Chinese women engaging in low levels of PA during childhood and adolescence.25Owing to the potential mediating effect of Chinese culture, the relationship between the built environment and PA in China might not resemble that in Western countries and thus warrants targeted research.One study that reviewed the built environmental correlates of PA in China, primarily among working-age and older adults, documented increased levels of PA in relation to proximal nonresidential locations, pedestrian-friendly infrastructure,aesthetics,and recreational amenities.26

This study is,to our knowledge,the first that systematically reviewed the existing literature regarding the impact of neighborhood built environment on PA and body weight status among children and adolescents in China. Findings from this review can be informative to policymakers and stakeholders in their efforts to design and modify certain features of the neighborhood built environment that promote PA, decrease sedentary behavior, and contribute to healthy weight among the Chinese child population.The study also identified limitations and gaps in this field that warranted future research.

2. Methods

1. Study selection criteria

Studies that met all of the following criteria were included in our review: (1) study designs—experimental studies (e.g., randomized,controlled trials or pre-post studies),observational studies (e.g., longitudinal studies, case-control studies, or crosssectional studies),and qualitative studies;(2)study subjects—Chinese children and/or adolescents aged ≤17 years;(3)exposures—neighborhood built environment (e.g., sports facilities, parks,greenspaces, bike lanes, and sidewalks); (4) outcomes—PA and/or body weight status; (5) article type—peer-reviewed publications;(6)time window of search—from the inception of an electronic bibliographic database to May 31, 2018; (7) country—China;and(8)language—articles written in English.

Studies that met any of the following criteria were excluded from the review: (1) studies that incorporated no outcome pertaining to PA and/or body weight status;(2)studies that evaluated an area substantially larger than a neighborhood,such as a city or province;(3)studies that evaluated a resident’s transportation mode but that had no evaluation of the neighborhood’s transportation system; (4) studies that did not include human participants; (5) controlled experiments conducted in manipulated rather than naturalistic settings; (6) articles not written in English;(7)letters,editorials,study/review protocols,or review articles;or(8)studies involving Chinese participants who lived outside the Mainland of China, Hong Kong, China, or Macao,China.

2. Search strategy

A keyword search was performed in 4 electronic bibliographic databases:Active Living Research(ALR),Cochrane Library, PubMed, and Web of Science. Administered by the University of California, San Diego, ALR was founded in 2001 by an interdisciplinary team with research-and-practice expertise in public health, transportation, planning, parks and recreation,school activity programs,behavioral science,and obesity prevention. Over the past 2 decades, ALR has established a rich database of articles covering community designs that promote PA, physical/mental health, economic vitality, and environmental sustainability. The Cochrane Library is a collection of databases in medicine and other health specialties.

The search algorithm included all possible combinations of keywords associated with the 4 groups: (1) “environment”,“environments”, “environmental”, “neighborhood”, “neighbourhood”, “neighborhoods”, “neighbourhoods”, “place”, “places”,“residence”, “residential”, “area”, “areas”, “space”, “spaces”,“spatial”, “greenspace”, “land”, “lands”, “landscape”,“landscaping”, “park”, “parks”, “community”, “communities”,“municipal”, “garden”, “gardens”, “wood”, “woods”, “forest”,“forests”,“facility”,“facilities”,“gym”,“gyms”,“gymnasium”,“gymnasiums”, “path”, “paths”, “trail”, “trails”, “resource”,“resources”, “amenity”, “amenities”, “urban”, “city”, “cities”,“town”, “towns”, “county”, “counties”, “walkability”, “built”,“building”,“buildings”,“street”,“streets”,“streetscape”,“road”,“roads”, “highway”, “highways”, “freeway”, “freeways”,“expressway”, “expressways”, “sidewalk”, “sidewalks”,“footway”, “footways”, “footpath”, “footpaths”, “track”,“tracks”, “walkway”, “walkways”, “passage”, “passages”,“passageway”, “passageways”, “block”, “blocks”, “transport”,“transportation”, “transit”, “geography”, “geographies”,“geographic”, or “geographical”; (2) “motor activity”, “motor activities”, “sport”, “sports”, “physical fitness”, "physical exertion”, “physical activity”, “physical activities”, “physical inactivity”, “sedentary behavior”, “sedentary behaviors”,“sedentary behaviors”, “sedentary behaviors”, “sedentary lifestyle”, “sedentary lifestyles”, “inactive lifestyle”, “inactive lifestyles”, “exercise”, “exercises”, “active living”, “active lifestyle”,“active lifestyles”,“play”,“outdoor activity”,“outdoor activities”,“step”,“steps”,“walk”,“walking”,“run”,“running”,“bike”, “biking”, “bicycle”, “bicycling”, “cycle”, “cycling”,“stroll”, “strolling”, “active transport”, “active transportation”,“active transit”, “active commuting”, “travel mode”, “mode of travel”, “physically active”, “physically inactive”, “obesity”,“obese”,“adiposity”,“overweight”,“body mass index”,“BMI”,“body weight”, “waist circumference”, “waist to hip”, “waistto-hip”,or“body fat”;(3)“China”or“Chinese”;and(4)“child”,“children”, “childhood”, “juvenile”, “pubescent”, “pubertal”,“puberty”, “adolescent”, “adolescents”, “adolescence”, “youth”,“teen”,“teens”,“teenage”,“teen-age”,“teenaged”,“teen-aged”,“teenager”,“teenagers”,“teen-ager”,“teen-agers”,“kid”,“kids”,“youngster”, “youngsters”, “minor”, “minors”, “student”,“students”, “preschooler”, or “preschoolers”. The MeSH terms“environment design”, “exercise”, “overweight”, “obesity”,“child”, “adolescent”,“China”, and “Chinese” were included in the PubMed search. All keywords in PubMed were searched with the “[All fields]”tag,which are processed using Automatic Term Mapping.27The search function TS = Topic was used in Web of Science, which launches a search for topic terms in the fields of title, abstract, keywords, and Keywords Plus®.28Titles and abstracts of the articles identified through the keywords search were screened against the study selection criteria. Potentially relevant articles were retrieved for evaluation of the full text. Two coauthors of this review (JS and QY) independently conducted title and abstract screening and identified potentially relevant articles. Inter rater agreement was assessed using the Cohen’s kappa (κ =0.74). Discrepancies were resolved through discussion with a third coauthor(RA).

A reference list search(i.e.,backward reference search)and a cited reference search (i.e., forward reference search) were conducted based on the full-text articles meeting the study selection criteria that were identified from the keyword search.Articles identified from the backward and forward reference searches were further screened and evaluated using the same study selection criteria. The reference search was repeated on newly identified articles until no additional relevant articles were found.

3. Data extraction and preparation

A standardized data extraction form was used to collect the following methodologic and outcome variables from each article included authors, publication year, country, study design,sample size, age range, proportion of female participants,sample characteristics, statistical model, nonresponse rate,geographical coverage, setting,type of built environment measure, detailed measure of built environment, type of PA measure, detailed measure of PA, type of body weight status measure, detailed measure of body weight status, estimated effects of built environment on PA or body weight status,and key findings on the relationship between built environment and PA or body weight status.

4. Data synthesis

A tabulation of extracted data by 2 coauthors of this review(YY and RA)revealed that no studies used the same measures for built environment,PA,and body weight to provide quantitative estimates for the impact of neighborhood built environment on PA and/or body weight status.This finding precluded a meta-analysis.Therefore,in this article we have summarized the common themes and findings of the included studies narratively.The data extraction,theme identification,and narrative summarization were independently conducted by 2 coauthors of this review (JS and QY). Discrepancies were resolved through discussion with a third coauthor(RA).

5. Study quality assessment

We used the National Institutes of Health’s Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies to assess the quality of each included study.29This assessment tool rates each study based on 14 criteria(Table 1).For each criterion, a score of 1 was assigned if “yes” was the response,whereas a score of 0 was assigned otherwise(i.e.,if an answer of“no”, “not applicable”, “not reported”, or “cannot determine”resulted).A study-specific global score ranging from 0 to 14 was calculated by summing the scores across all criteria. The study quality assessment helped measure the strength of scientific evidence but was not used to determine the inclusion of studies.Two coauthors of this review (JS and QY) independently conducted the study quality assessment,with discrepancies resolved through discussion with a third coauthor(RA).

3. Results

1. Identification of studies

assessment.Table 1 Study quality 20Y Y YY N Y Y Y N Y N N NN 8 19Y Y YY N N N Y N N N N YY 7 18Y Y YY N N N Y Y N N N YN 7 17Y Y YY N N N N N N Y N YY 7 16Y Y YY N N N Y Y N Y N YY 9 15Y Y YY N N N Y N N N N YN 6 14Y Y YY N N N Y Y N N N YY 8 13Y Y YY N N N Y N N N N YY 7 12Y Y YY N Y Y Y N Y N N NY 9 y ID 11Y Y YY N N N N N N N N YY 6 Stud 10Y Y YY N N N N N Y N N YY 7 9Y Y YY N N N N N N N N YN 5 8Y Y YY N N N N N N N N YY 6 7Y Y YY N N N N Y N N N YY 7 6Y Y YY N Y Y N N Y N N NY 8 5Y Y YY N N N Y Y N N N YY 8 4Y Y YY N N N Y Y N N N YY 8 3Y Y YY N N N N N N N N YY 6 2Y Y YY N N N N N N N N YY 6 1Y Y YY Y N N N N N N N YY 7 aper of red %?bjective in this p e tim criteria for being easu n clearly specified and 4.Were all the subjects selected or recruited from thee er description,or(including the sam r level,did iformly to all ere the exposure(s)osure as f exposure,eing ariables)nce over ent v ented ories o t variables)ented les m to the exposure sionun plem e ≤20%?t so that one could reaplem variab,pow sonably expect to see an association between exposure more than o res(dependen t o vary in amoun utcome(s)?ding e different levels of the exp d o and exclulied an un ple size justification me (e.g., categ and app vided?easured as continuous variable)?depend ates pro me(s)b ses in this pap uestion or o er,w?2.Was the study populatio easures(in dy participants?im assessed ined,valid,reliable,and easu dy participants?for their impact on the im osures that can me if it existed w-up after baselin efore the outco utcome assessors blinded tential confo Were inclusionied 1.Was the research q clearly stated?ilar populations ined,valid,reliable,and eframe sufficien po in 3.Was the participation rate of eligible persons ≥50 etween exposure(s)variance and n articipants?easured b related to the outco same or sim particip interest m osure m period)?Criterio defined?in the study prespecif 5.Was a sam ants?effect estim r the analy measured?7.Was the tim utco osure m exam 9.Were the exp nsistently across all stu co 10.W 6.Fo utcome m an8.Fo d o r exp the study or exp clearly def as the exposure(s)time?ere the o 11.W clearly def nsistently across all stu co 12.W ere the o status of p as loss to follo 13.W ere key 14.W d adjusted statistically an relationship b tal score To was,a score of 1 4 criteria.Study quality assessdies.28 Fo r each criterion tu al S ection scores across all 1 up ross-S ming ort and C y sum oh al C bservation ool for O 0 to 14,was calculated b from t T en ging ssessm uality A was included in the review.ealth’s Q ic global score,ran ecif ational Institutes ostud f H y-sp hether a study etermine w m the N ned otherwise.A as adopted fro as assig f 0 w as not used to d t tool w ereas a score o tific evidence,but w en This study quality assessm nse,wh was the respo f scien gth o es.=y no;Y easure stren N=tes:iations:No assigned if“yes”ment helped m brev Ab

Fig.1. Study selection flowchart.

Fig. 1 shows the study selection flowchart. We identified 8286 articles total through the keywords and references search.After removing duplications, 7922 unique articles underwent title and abstract screening, after which 7879 articles were excluded. The full texts of the remaining 43 articles were reviewed against the study selection criteria. Of these,23 articles were excluded. The primary reasons for exclusion were lack of neighborhood environment measures,no outcome reported on PA and childhood obesity, and adult sample instead of children and/or adolescents.The remaining 20 studies that examined the relationship between built environment and children’s PA and/or body weight status were included in our review.30-49

2. Basic characteristics of the included studies

Table 2 summarizes the basic characteristics of the 20 included studies.A total of 5 studies were conducted in Hong Kong,China,and the remaining studies were conducted in the Mainland of China. All included studies were published in or after 2006 — 1 each in 2006, 2008, 2009, 2011, 2012, and 2016; 2 each in 2010 and 2015; 4 in 2017; and 6 in 2014. A total of 14 studies exclusively focused on PA,2 studies exclusively focused on body weight status, and the remaining 4 studies examined both PA and body weight status. A single study adopted a qualitative study design, 3 adopted a longitudinal study design, and the remaining 16 studies adopted a cross-sectional study design. Sample sizes were generally large but varied substantially across studies.A total of 4 studies had a sample size between 20 and 99; 5 had a sample size between 100 and 999; 9 had a sample size between 1000 and 9999; and the remaining 2 studies had a sample size of＞10,000. The mean and median sample sizes were 7498 and 1359, respectively, with a standard deviation of 18,545 and a range from 20 to 80,928. The majority of studies (n=12)recruited adolescents aged 11-17 years, 4 recruited children aged 6-10 years, and the remaining 4 recruited both children and adolescents. All included studies recruited both girls and boys, with the 2 sexes largely equally distributed in the samples. A variety of statistical models were applied across studies, including generalized estimating equation, logistic regression, hierarchical logistic regression, mixed-effects logistic regression,ordinal logistic regression,generalized linear model, hierarchical linear regression, multilevel Poisson regression, and multilevel path modeling. The majority of studies (n=15) adjusted for some individual sociodemographics in the statistical analysis.A total of 9 studies adjusted for certain environment and/or PA characteristics(e.g.,greenspace and sedentary behavior)in the statistical analysis.

Table 3 summarizes the measures for built environment,PA, and body weight status used in the included studies. The majority (n=13) of the studies adopted subjective built environment measures reported by parents and/or the children themselves, 2 studies adopted objective built environment measures, and 5 studies used both objective and subjective built environment measures. Specific measures of the built environment varied substantially across studies.Parent-and/or self-reported subjective built environment measures included both general questions (e.g., “Whether or not you have sports facilities in your neighborhood?”or“How long does it take to walk to the nearest exercise facility?”)and more specific questions pertaining to the built environment (e.g., accessibility and availability of sports/recreational facilities, greenspace,parks, playgrounds, and video game shops, as well as questions about neighborhood aesthetics and attractive natural sights in the neighborhood). Among the subjective built environment measures, only 2 studies used measures known to be validated in previous research (e.g., the Neighborhood Environment Walkability Scale).39,40Objective built environment measures (e.g., population density, road density, street intersection density, and distance to nearest park and/or recreational facility) were constructed based on geographical information systems, whereas other built environment measures such as walking paths, playgrounds, jogging/cycling lanes,transportation,schools,and parks were tabulated in field observations by trained observers in the community under study.

g Settin Urban,burban Urban,su and rural Urban Urban Urban rural Urban and Urban rural Urban and rural Urban and an burb Su Urban Urban rural Urban and Urban and rban subu ical raph Geog verage co usehold,unity in a unity in a eigh-Ho school and mm cocity usehold,Ho school and mm cocity me and n Ho borhood in a city -Residential den -green sity and space in a city Residential den sity in a city usehold and unity in 9 ces Ho mm co provin eighme and n Ho borhood in a city roximity in a to an exercise 10ity and me p in ces un orhood Ho facility p rovin useho ld and Ho neighborhood in a citymm Co park in a city Family and neighborhood in 2 citieshb Neig city ity in 10 un ity and mm ces un Co provin mm Co park in a city Nonresponse)(%rate Statistical model 0.9 Hierarchical logistic regression 0.7 Hierarchical logistic regression 15.3 n Logistic regressio 10.7 istic Mixed effects log regression 10.7 istic Mixed effects log regression 3.9 at-Generalized estim ing equations 7.6 ear Hierarchical lin regression N A ariate Multiv regression NA NA 15.1 n Logistic regressio 10.3 n Logistic regressio 22.7 Generalized linear del mo 21.5 n Logistic regressio 32.9 ear Hierarchical lin regression le Samp characteristics ool nior high sch ts Ju studen ool nior high sch ts Ju studen Grades 7-1 3 ts studen Grades 7-9 junior high school Grades 7-9 jun io r ts studen high school ts studen ildren an d Ch adolescents Grades 4-6 prischool ts mary studen Residents Grades 5-6 prischool ary tsunity Grades 7-1 residentsprim ol-aged 2 secstudenanen ondary school t ts mm studen Grade 3ts mary Co scho studen Perm residentsunity mm Co residents Female)(%49.8 49.8 56.0 53.8 53.8 Urban:49.0 Rural:46.0 52.8 51.4 50.0 50.8 48.3 63.2 51.2 51.6 e(year)11-1 7 A g le Samp size 1787 11-1 7 1792 Boys:14.5±0 14.6.1 Girls:.1±0 13-1 5 29,139 2375 13-1 5 2375 6-18 6935 Boys:11.1±0.911.2 Girls:.9±08＜1 280 487 10-1 1 34 15-1 9 1528 8-10 497±1.7.0 14 9993 8＜1 463 11-2 0 29 f the studies included in the review.Table 2 Basic characteristics o y Stud desig n R egion First author(year)y Stud ID Crosssectional ainland of e M Th China Li et al.(2006)30 1 Crosssectional ainland of e M Th China Li et al.(2008)31 2 Crosssectional o ng,Hong K C hina et al.Wong(2010)32 3 Crosssectional ainland of e M Th China et al.Xu (2010)33 4 Crosssectional ainland of e M Th China et al.Xu (2010)34 5 inal ngitud Lo ainland of e M Th China Cui et al.(2011)35 6 Crosssectional o ng,Hong K C hina ang et al.Hu (2013)36 7 Crosssectional ainland of e M Th China et al.An (2014)37 8 e Qualitativ ong,Hong K China He et al.(2014)38 9 Crosssectional Crosssectional ainland of e M ainland of Th China e M Th China Jia et al.(2014)39 Li et al.(2014)40 1011 inal ngitud Lo ong,Hong K C hina et al.Wong(2014)41 12 Crosssectional Crosssectional ainland of e M ainland of Th China e M Th China o et al.Gu (2014)42 Liu et al.(2015)43 1314(continued on next page)

Setting ical Geograph coverage nresponse)(%No rate Statistical model Sample characteristics Urban and an suburb Urban Urban Urban and an suburb Urban and rural Urban,suburban,and rural Green space in a neighd in a city unity and comcity Home and borhoo School in a city Commin a citycompark School and munity nationwide School and munity nationwide.3 586.06.7.5 61.9 32NA Ordinal logistic ear oisson regression Generalized lin el od mixed m ltilevel P Mu model Hierarchical linear regression ltilevel path Mu modeling gistic regression Lo rit anen unity Perm residents Grades 3-5 p mary school students School students Comm residents Grades 4-12 school-aged students School-age students).ontinued(C Table 2 ale em)Age(year) F(%53.0 46.0 15-20 8-12 Sample size 751189 Study design Crosssectional Crosssectional Region ainland of ong,The M China ng K Ho China r First autho ear)g et al.(y Zhan 015)44(2 Wong et al.016)45(2 y Stud ID 1516 46.9 51.5 50.9 9-11 11-20 55228 1 3.71±2.94 2080,9 Crosssectional Crosssectional Crosssectional ainland of The M China ainland of The M China ainland of The M China mes et al.(2Wang 017)46 Liu et al.017)47et al.Go (2017)48(2 171819 47.1 6-17 9487 Longitudinal ainland of The M China Yang et al.017)49(2 20 t applicable.no A=Abbreviation:N

PA behaviors measured in the studies included PA(n=13),physical inactivity (n=2), sedentary behavior (n=2), active/passive commuting from/to school (n=2), and park visits(n=1). Most studies (n=17) measured PA levels using questionnaires administered to parents and/or children, whereas 1 study adopted an objective measure by having participants wear an accelerometer, 1 study adopted an objective measure through both a parent-reported questionnaire and participant use of a wearable accelerometer,and 1 study collected qualitative data from a focus group of children. Parent- and/or self-reported PA questionnaires included both standardized questionnaires (e.g., the International Physical Activity Questionnaire (IPAQ)) and general questions (e.g., “How often do you do exercises continuously for at least 30 min?”). Among the studies that adopted subjective PA measures,8 used measures known to be validated in previous researches(i.e.,the Adolescent Physical Activity Recall Questionnaire, Physical Activity Questionnaire for Children, IPAQ, IPAQ Short Form, and Godin Exercise Leisure-time Questionnaire). A total of 18 studies included body weight status measured as body mass index(n=12), overweight (n=4), and obesity (n=2). Body mass index (BMI)-related measures were based on both objectively measured(n=7)and self-reported(n=5)height and weight.

3. Key findings

Table 4 summarizes the estimated effects of built environment on PA and/or body weight status among Chinese children and adolescents. Among the 16 studies that provided some quantitative estimates of the relationship between built environment measures and PA, all of them reported a statistically significant relationship in the expected direction.A lack of recreational facilities in a community was found to be positively associated with physical inactivity among local children.30In contrast,perceived availability of sport facilities in a neighborhood was positively associated with leisure time overall PA level32,41as well as moderate-to-vigorous PA level.36,48Commuting time to the nearest sport facility played an essential role in PA engagement.Longer commuting time was found to be negatively associated with the likelihood of meeting the PA recommendation42and using walking paths.39In contrast,children living within a 10-min walking distance and/or 500 m of an exercise facility tended to engage in more leisure-time PA.37,47

In addition to exercise facilities, other environmental characteristics were found to be associated with PA. For instance,adolescents living in residential neighborhoods without sidewalks or without vacant fields were 30%and 70%more likely to be physically inactive,respectively.30The proportion of residential greenspace was found to be negatively correlated with monthly park visits.43Adolescents living in a neighborhood with a higher residential density were more likely to spendless time on PA and more time on sedentary behaviors such as television viewing.33,36Among the 3 studies that estimated the longitudinal relationship between the built environment and body weight status,2 reported a statistically significant association, whereas the remaining study reported no association.Boys living in places lacking bike lanes were 60%more likely to be overweight or obese.31Adolescents living in a higher residential density area were 87%more likely to be overweight.34In contrast, no association was found between neighborhood mixed land use and childhood overweight.40

Table 3 Measures of built environment,PA,and body weight status in the studies included in the review.

(continued on next page)

Table 3(Continued).

iew.ded in the rev dies inclu t status in the stu eigh body w A and ironment on P f built env Table 4 Estimated effects o f study eight status indings o Body w Main f PA f built environment eight status dy w Estimated effects o Bo PA First author(year)Study ID urban areas,adolescents eight or verw and sub rban ore likely to be o In u were m ese.ob alks and facilities in the associated mups in the com A in ith increased P e ositiv A.ere significantly 1.Perceived availability of sport facilities in e a p ay cancel ysical inactivity.et m pportunities in as associated w f active o ay hav d.f sidew ity w game sho puter/Intern borhoo 1.Lack o un comm ph with 2.Having video nity w boys.eighborhood m e n th im pact on adolescents’level of P out the effects o the neigh-2.Having a com besity in ado lace of resise living in rban CI:verw olescents in suburban areas e lanes ik eight/obese.eight and o eight ith p as associated w ith tho or u mparison w 1.8-3.4)verw)had ithout b CI:2.7-6.0 ses w(95%es ely to be o verw I:5%ou e risk of o more lik=4.0,9 co lescents w dence.In a rural area,ad 2.5,95%C g in h in areas(OR ere 1.6 tim nearby w.4 greater risks of being o 2.Boys liv 0)-2(O R=1.04 and obese.,.4 ere ithearity as.26)e s=Rb =1R=ith inactivrh lic facilities(OR ore likely to lt)wity.ood w CI:1.0-1.6).ith-I:3.Adolescent boys living in surroundings wI:oderate and O times(95%C=1 e shops n ith inactiv 1.Perceived availability of sport facilities w 95%C ith being sufficiently active.f peravailability of neighborhood sport le on thend as fou.2=0.049 for eo g6,ility of sport facilities in hysicallyam id =1f spo 1.2-2.4 for difficu e conso (O ORgirls=0.001).ndrt es m irls CI:1.15-1.56,p ＜sole being physically active in the=1.3,95%ere 1.3 tim om6,p=2.A significant interaction effect o ere 1.71.5,(O ields w s=video gam ysically active w oy ideo gam Rg ailability of v ed availability o ith the odds of being p game con 1.0-1.9 for m Rboy R=ph oys and p nly in the absence of v 0.001;ere fou-1(O ys alks w sitively associated w for b facilities andf v 1.7,95%C out sidew I:out vacant f as associated w as positively associ-46 he access to pub 0%e(ORbo borhoo 1.10 e w in boys by 5.0%C be inactive(OR=0 more likely to be inactive.Unav eigh I:consoles in the h facilities and 1.2-2.5)their hom odds of being 1.1-2.1).I:positively 1.17,O 4.No significant associations w associated w 2.Adolescents living in a neighbo ed%C een perceiv ce o ideo Li et al.(2006)30 1.T95po ceiv 4.Li et al.(2008)311.Th(p girls).3.Perceived availabd w .4 the n ated w active o 951.34,95%betw presen et al.ng Wo (2010)32 1 2 3(continued on next page)

f stud y gs o findin Main eight status dy w Bo PA tial populaore likely den-Students in the higher residen ere also m ith th eir pared w pulation po tion density tertile w eight com e overw to b un terparts in the low co sity areas.egative association between A for al P recreation density and ly expanding urban ay be a n hina.ere m f C T h pulation po adolescents in a rapid area o ositively assourb an w as p ensity eight am ong Residential d ith overw ciated w ese adolescents.in Ch ential density A e on P rs than tial io av igher resid less tim tary beh est residen ts in the h n seden e o nterparts in the low Studen tertile spent significantly ore tim and m eir cou th density areas.ot in the local m ute as n whose school w ely to co m o re lik ere m Childrenun ity w comm passively.s her ere no significant association mental varig ith h e es.ood environ d w ore tim g e-gam A.playin borh eighborhoo eigh ensity reported m There w een n ys living in a n 1.betw iables and children’s P 2.Bo population d using the Internet and(continued on next page)).tinued on(C Table 4 ment s uilt environ weight statu Estimated effects of b dy Bo P A r(year)First autho y ID Stud dle tertiles of C I:(95%overweight,a 1.87-fold id CI:1.12-(95%1-fold ith those in the pared w 2.61)higher likelihood of being respectively,com-2.85)and 1.7 1.23 lower tertile.ORgirls ases sityA).er ential tertile had fording e ther (aOR in e on P.94;as not munity.th ith residential density had 1-0.73=0I:%C 0.63 eing%C greenspace.I:95)of b ential den less tim for o entary ool w early twice C I:5%,p1,p=mute to ariables inclu e and tertile,students in the ere children hose school rts facilities in her resid hose school w ere nearly 4 tim as associatedde hose sch behaviors time in the cru es playingI:4-1.21=0.64,95 am.4-13,ere n .0 CI:-0.05 to 2.10,p 3.73,9=1.16,95%C e o 0.81 CI:0 ensity w as in the local com sity as positively associatedin=0.8 founding v io an areas,after adjusting assively com ildren w(β).ts in the hig I:middle resid munity,w R=tertile spent significantly mute as w munity w nd in spo n sed munity.as likely to.8 CI:0 I:d in the univariable analyith students from the low r tim to p ere children w 5%rhoo ositive associations w V viewing 1.02,95%se and e-g odel 0.05 0.59-1.18,respectively potential con%C pared w 5%0.42-0.97;1.01,96,95 OR 0.01).,p ＜lower odds(OR 6)bo were fou higher P A category,after adjusting cofactors,ch in the com ore tim sedentary behav higher and residential den 1.42-2.66)0.37,9=1.94,95%C ical m whose sch 2.In rural areas,children w 8,p ＜-2 t in the com ely coml w oo more likely(aO school than w=1.0 PA was in the com 2.Studen t m om bu(p ＜1.For girls,p 2.81.9 high the neigh Internet u ses(β=r boys,higher d-4 0.05).er T model(β=.C no ＜0.1),density w 2.Fo MV hierarch passiv.0 with with 0.24 0)3 31 et al.(201 010)34Students in the higher and m Xu et al.(2 Xu Cui et al.(2011)35 1.In urb ang et al.Hu (2013)36 4 5 6 7

ed).ontinu(C Table 4 f study eight status indings o Body w Main f PA f built environment eight status dy w Estimated effects o Bo PA First author(year)Study ID s for association envi-’s rs ith orhood childrenio s of sex,school nd w ere fou boreigh as found hb neig besogenic behav ility of n evidence w erceived s and o ht statu dy tal characteristics and.eight statu een p icant interactions w d w e,an en erceived availab sport facilities.No betw ronm weig in this stu signif No grad the pod ho nde t-e as fou actors mu ith leisure tim A facilitators included nel,conveay station,recreath ark d sport mendato oo ity to an exercise facility w ded hard toany ith air ps.F o m borh aintain or ely associated w A.alls and een the com hting,bridge or tun istance to a p erfume sho A recom A barriers inclu bw eigh e P malls w ds.opping as P perceived better roads betw ore likely to have used ore such facilities ere m conditioning,and p lescents to m re-tim earest sport facility and people in recreatio uilding m rted a d ho eeting the P ositiv elp active ado nient transportation,su icient lig grounds,sh e to the n of m ho be p od.suff repoose w tion th was found.perceived as P Proxim ose w find toilets in shopping mun toPA Th n gro＜1 km and and aesthetics w walking paths.Increasing awareness of n facilities or b may h crease their leisu in An inverse association g tim in likeliho tion eigh t(OR 0.05).childhood overw alking paths as 1.22,95% a relationship between land-use-=ith those living further away,indialking distance to an ere CI:0.85-1.75,p ＞rt ere signifie istance 1 km w 0.01)e alking 1.078,nly ref spo ith w.017 mutf CI:(O or every nificantly correw-up p-rted the d res w(95%).ark ＞0.530-0.933).res ied by baselin correlated wCI:1 ales,the roads and aesthetics sco A at follow-up 1.068,time,the R=in in com 0.01in.F e P 5%muting less likely to use the w,9 I:6-1.12 se ere 6.79%ore likely to have som 44 10 m as m repo to a p ositively and sig ed availability o CI:1.011-1.047,p ＜alking p had engaged in leisuremix access and eek(β=CI:0.65-0.98)opared mendation com use(OR sitively ithin 10-min w .0 ho es a w6,p ＜1.022-1.11=1 2.Participan facilities from baseline to follo orhood exercise facility w 3.67-10.01) mA.A recom ath u ts w 0.80,95%3 tim po hb 3).ho spent ＜viduals w ith w leisure-time P I:cantly odif 0.703,95%C more leisure-tim 1.03 Increased perceiv ing tim(O oads and aesthetics sco pared w path 1.072)..029,95%A ＞neig ith a significant effect observed o R=were p He et al.(2014)38Factors perceived ho%C I:lated w e to the nearest sport facility had 80%from ,w om ong those w Participants w(O 3.In m R=in increment in com 95 dicted =1erall.This effect w am ho spent ≥30 m%C ds -m An et al.(2014)37 C 4)39 1.R (βovPA time P 95od meeting the P with those w 10 Jia et al.(201 Li et al.(2014)40There w et al.Wong(2014)41 Guo et al.(2014)42 8 9 10111213(continued on next page)

f study gs o indin Main f eight status Body w PA-ace,reen n of residential greensp earest park,and dis-iththe rade increasing w te between were negareen spaces and monthly park an g ility of g ark)reen e residents’-rb th ith the g residence and the nearest p use of u with etation,accessib e to the n affected ility of g rtio walking tim (w f the shortest rou eg A.po ro rade rrelated The p tance g istance o co the d tively visitation.Quality of v spaces,and availab s significantly park satisfaction levels and spaces for P f the neighborhoodere f trees)w as ese w earest dists w aren education.ce o of being risk ob ith n eighborhood,ho ith p dary 1.The aesthetics o t(presen uced d of being ciated w ren w en oo asso vironm elih nly in child pleted a secon en related to a red obese.e lik 2.Th positively tance to a park in the n but o had com ore tith m ou d posieighborhood w atural sights spent nasium ely associan .pliance.ositiv A com PA uildings or n Children living in a n as p ours w as significantly VP ren’s M accessibility wwith ility of a school gym residents’A.attractive b e in P more tim e availab f school h ith child correlated Th side o ated w Park tively ility of ith a A ns of availab convenient ere associated w VP in M Children’s perceptio organizations and A facilities w articipation clubs and access to P level of p high(continued on next page)ed).tinu on(C Table 4 f built environment eight status Estimated effects o Body w PA First author(year)Study ID eighborhood farther er trees 0.863,,e obese.(β=.3 45ith less ith few -0 0.210 to-1.517;β=-0.655 to-0.035).rhood variables ildren living in a n ark and w f neighbo ore likely to b parents.ay from a p I:I:luence o 95%C ed to exist only for children w were m e inf aw Th 2.95%C seem educated e higher rhood 4,an=0.0 ace on citizens’0.020),w .9 n citizens’=0park rb rt--0.42,SE n level alk-1.Ch=1 n citizens’f the vegetation,9,7).visbo(β e to reach a park,th el-0.17,p=ark o =0.042).earest park o.4 A(β=ere decreased by 6%(OR 0.010),sho relative importance of protial greensp earest p f greensp n level o aces and hts in the neigh 0.101,SE f use of u ith objectively assessed her the satisfactio ore availability o.99).0.010).0.88-0=0.09).e hig efficient=-0.12,p=e to n n(β=istance to n f residen 9,SE est road d .1her the quality o e m ig=0 I:her the satisfaction lev-0.11,p=rtion o ds w ig park visitatio isitatio n(β==0.09).VP A(co od95%C Liu et al.(2015)43 Aspo sociations and park v ing tim itation(β=1.The less tim the satisfactio greenspace for P 2.Th parks,th(β 3.The h the h SE Attractive natural sig were associated w%M g et al.et al.Zhan 015)44(2 Wong(2016)45 141516 outside were nasium ith the as assom w,ns izatio venient access 1)0.00 ort high levels with access to a gym ore w=0.1 plied m 1.14).ithin 500 1.2,St.βh scores on availrgan ig and con the school hours com(R R=5.278,p ＜es elin arks w A(β=0.001)more likely to rep Children guid PAum ber of p ith P 6).who reported h A.MVhe n ciated w 0.04icantly=4.489,p ＜Children p=ability of sports clubs and o(βto P A facilities(β=ifVP sign of M mes et al.7)47 T Go (2017)46 Liu et al.(201 Wang et al.(2017)48 171819

ed).ontinu(C Table 4 f study eight status indings o Body w Main f PA f built environment eight status dy w Estimated effects o Bo PA First author(year)Study ID=risk ratio;RR hysical activity;=p PA ith ity;rrent wer lipo .ere living in a mort A ren living a farth etro TS was concu rous physical activ TS ely to rep e decrease of A f child ho w ere less lik moderate-to-vigo 1.Th e increase o th distance from school.ildren w 2.Ch tan area w A=VP s ratio;M dd=o OR idence interval;conf I=ol;C gly associated scho se relationship 0.001).2 m iles-1ere.5(0from e in lived ＞18.6-36.4,0)..6,95%I:.4 ely to engagCI:hoho 40.3 as stron ay from school,th ith children w ool w miles),the second closest R=lived the clos-.0 iles)w ose w(O a dose-respon CI:6 closest(1-2 m 5%Distance to sch =26.0,95%C3-10 with times as lik OR =active travel to and TS TS d 8.7-6R=and statistical significance(p ＜ared w.58.10,9 with A Comp 6,an mile),and the third est(＜0 41AT aw ,2S,respectively 7.0;24 and O dds ratio;A sted o Yang et al.ju ad(2017)49R=aO iations:ard error.20brev Ab =stand SE

4. Study quality assessment

Table 1 reports criterion-specific and global ratings from the study quality assessment.The included studies on average scored 7 out of 14 (range: 5-9). All studies included in the review clearly stated the research question/objective,specified and defined the study population, had a participation rate of≥50%,recruited participants from the same or similar populations during the same time period, and prespecified and uniformly applied inclusion and exclusion criteria to all potential participants. Most studies measured and statistically adjusted key potential confounding variables for their impact on the relationship between exposures and outcomes(n=16),had an attrition rate of ≤20% (n=17), and examined different levels of the exposure in relation to the outcome(n=10).In contrast,none of the studies had the outcome assessors blinded to the exposure status of the participants. Only 1 study provided a sample size justification using power analysis. Exposures of interest(e.g.,built environment and other environment characteristics) before the outcomes were measured in 3 studies. A total of 3 studies had a reasonably long follow-up period that was sufficient for changes in outcomes to be observed,4 studies assessed the exposures more than once during the study period,and 6 studies implemented valid and reliable exposure measures.

4. Discussion

This analysis systematically reviewed scientific evidence regarding the influence of neighborhood built environment on PA and body weight status among Chinese children and adolescents. A total of 20 studies, including 16 cross-sectional, 3 longitudinal, and 1 descriptive, met the selection criteria and were included in this review.A total of 13 studies adopted subjective built environment measures reported by parents and/or children, 2 adopted objective measures (e.g., geographical information systems, field observations), and 5 adopted both objective and subjective measures.PA behaviors included PA,physical inactivity, sedentary behavior, active/passive commuting from/to school, and park visits. Body weight status included BMI overweight, and obesity. The 16 studies that provided some quantitative estimates of the influence of built environment on PA and body weight status all reported a statistically significant relationship in the expected direction.

Findings from this review confirmed the documented relationship between neighborhood built environment and PA among children and adults in developed countries. For example, Reed and Phillips50reported increased duration and intensity of PA among American college students residing close to exercise facilities.Jilcott et al.51found that perceived distance to gyms was negatively associated with PA among low-income, middle-aged women in the United States. Hanibuchi et al.52documented that proximity to parks or greenspaces was positively associated with PA frequency among older Japanese adults. Edwards et al.53reported that proximity to parks and beaches was positively associated with use of those environments for PA among Australian adolescents. Findings from this review add a data point to the literature on the importance of designing new physical environments and/or modifying existing environments in a way that promotes an active lifestyle for children and adolescents in China.

Childhood obesity in China has reached an alarmingly high level. It has become critical that interventions be designed to target the root causes of obesity and its risk factors. Physical inactivity is a major modifiable risk factor for childhood obesity.54A majority of children and adolescents in China fail to meet the recommended guidelines for PA levels.55-57Although the traditional perspective mainly blamed individuals themselves for their sedentary lifestyles, increasing attention has now shifted to the environmental determinants of health behavior, in particular, how neighborhood built environments may impact local residents’ engagement with PA.Two systematic reviews have investigated the associations between various built environment attributes and childhood obesity.56,58Neighborhood features such as walkability, bikeability,and proximity to parks and recreational facilities were found to be associated with increased PA and active commuting.Interventions providing safe,walkable neighborhood with access to necessary destinations have been recommended in an effort to combat the epidemic of childhood obesity. A recent study reviewed the built environmental correlates of PA in China and documented a positive association between PA and proximal nonresidential locations, pedestrian infrastructure, aesthetics, and recreational facilities, as well as a negative association between PA and living in an urban setting.26Findings from that review are largely consistent with ours,except that our review focused more on the built environment as it pertained to locations for children’s daily activities(e.g., parks, greenspaces, and recreational facilities), whereas the afore mentioned review, which predominantly examined PA among Chinese adults aged ≥18 years, focused more on work-related built environments PA (e.g., transportation systems and proximity to the workplace).26Our review adds to the scientific evidence by systematically identifying and synthesizing literature regarding the impact of neighborhood environments on child and adolescent populations in China.

Improving the availability of local exercise facilities in communities has become a national policy, as highlighted in China’s 13th Five-Year Plan(2016-2020)for the Sport Industry.59The plan includes building new exercise facilities (e.g.,community fitness centers, parks, roadside open spaces with exercise equipments) and enhances the accessibility of existing facilities (e.g., extending the operating hours of school gyms/playgrounds and opening them to the public). Findings from this review suggest that these policy-level interventions can enhance PA engagement among Chinese children and adolescents.In the meantime,urban designs that incorporate sidewalks,bike lanes,and walking paths,reduce motorized traffic,and encourage lower population densities are likely to promote PA and prevent childhood obesity in China.

A few limitations pertaining to this review and the studies it includes should be noted.All studies adopted an observational study design. The absence of experimental designs prevents a causal interpretation regarding the impact of neighborhood built environment on PA and body weight status; therefore,the estimated relationships are correlational in nature.In addition,only 3 of the 20 studies included in this review adopted a longitudinal study design, which is necessary to examine the change in PA or body weight in response to changes in built environment. No studies included in this review reported quantitative estimates focusing on the same built environments, physical activities, and body weight measures, which presented a situation that precluded meta-analysis. The literature search identified articles written in English only and excluded articles written in Chinese and other languages.This decision was made in part owing to concerns about potential discrepancies in study quality, as well as concerns about the lack of heterogeneity in built environment measures and study designs existing in the literature written in English versus Chinese. Also, this review only included published literature;thus, useful and relevant unpublished studies may have been overlooked. Future reviews could explore the unpublished, or“grey” literature to determine whether it is useful in building on our findings.

5. Conclusion

This study systematically reviewed the scientific evidence regarding the influence of neighborhood built environment on PA and body weight status among Chinese children and adolescents. A total of 20 studies met the selection criteria and were included in the review. Availability, accessibility,and proximity to greenspaces, parks, recreational facilities,and sidewalks were associated with increased PA, reduced sedentary behavior, and/or active commuting. In contrast, the absence of bike lanes and living in higher density residential areas were associated with increased likelihood of childhood obesity. Research on the environmental correlates of PA in China and other countries is rapidly increasing, so future reviews may be able to analyze more studies in this field.Future reviews should also include articles written in Chinese and other languages, as well as unpublished studies in the“grey” literature. Longitudinal or experimental study designs and objective environmental and PA measures should also be part of future research into the influence of the neighborhood built environment on PA and body weight status among children and adolescents in China.

Authors’contributions

RA conceived of the study, conducted the systematic review, and wrote the manuscript; JS and QY contributed to literature review, data synthesis, and manuscript drafting; YY contributed to data synthesis and manuscript drafting. All authors have read and approved the final version of the manuscript,and agree with the order of presentation of the authors.

Competing interests

authors declare that they have no competing interests.