Access to bike lanes and childhood obesity: A systematic review and meta‐analysis

Summary

The lack of bike lane access has been a proven risk factor for childhood obesity due to its role in discouraging healthy lifestyles. However, there has not been a systematic review of this important association in the existing literature. This study aims to fill this gap. A literature search was conducted in the Cochrane Library, PubMed, Embase, and Web of Science for studies published from 1 January 2019 onwards that examined the association between bike lane access and weight‐related behaviours and outcomes among children aged <18 years. A total of 21 studies were included in this systematic review. Among them, most of the studies showed that bike lane access was significantly associated with children and adolescents' physical activity (PA), whereas only two studies showed a negative association. Meta‐analysis also supported these findings and showed that bike lane access was significantly associated with children and adolescents' PA (odds ratio [OR] = 1.57, 95% confidence interval [CI]: 1.37–1.81). Additionally, we reviewed how bike lane characteristics and microenvironment variables such as children and adolescents' choice of bicycle travel mode, the degree of separation of cycle path, cycle path unevenness, and street maintenance were associated with adolescents' preferences and intention to cycle. This systematic review and meta‐analysis strongly suggests that bike lane access is associated with children and adolescents' PA. Nonetheless, it was difficult to draw a conclusion on the association between bike lane access and weight‐related outcomes.

1. INTRODUCTION

Obesity is a leading cause of morbidity and premature mortality worldwide. In recent decades, the global prevalence of adult obesity has increased, from nearly 30% in 1980 to 40% in 2013. ¹ Furthermore, the global prevalence of obesity in children and adolescents increased from 0.7% to 5.6%. ² Obesity‐related comorbidities for adults include cardiovascular disease, hypertension, type 2 diabetes mellitus, and certain cancers, ³ as well as depression, anxiety, and low confidence. ⁴ , ⁵ , ⁶ In children and adolescents, obesity is related with a higher risk of cardiovascular and cardiometabolic metabolic risk factors; pulmonary, endocrine, gastrointestinal, and musculoskeletal complications; lower quality of life and reduced psychological health. ⁷ , ⁸ What is more, children and adolescents with obesity are more likely to be classified as being obese or overweight in adulthood. ⁷ , ⁹ Lobstein and Jackson‐Leach ¹⁰ have estimated that by 2025 some 268 million children and adolescents aged 5–17 years may be overweight, including 91 million obese, assuming no policy interventions have proven effective at changing current trends. We have also estimated the likely numbers of children and adolescents in 2025 with obesity‐related comorbidities: impaired glucose tolerance (12 million), type 2 diabetes (4 million), hypertension (27 million) and hepatic steatosis (38 million). ¹⁰ Therefore, childhood obesity has become a major public health problem because of its health risks and fast‐growing prevalence. As we all know, the rapid growth of the prevalence of childhood obesity is directly related to the poor lifestyle. Among them, lack of physical activity (PA) plays an important role in the occurrence and development of childhood obesity. ¹¹ , ¹² , ¹³

The term ‘obesogenic environment’ has been coined and defined as the environmental factors that may increase children and adolescents' weight status. ¹⁴ Such environments impact body size through enabling or hindering healthy eating and PA. Obesogenic environmental factors can be divided into macro‐scale (e.g. connectivity and land‐use mix) and micro‐scale environmental factors (e.g. evenness of cycle path, presence of speed bumps, vegetation and environmental maintenance). ¹⁵ Micro‐scale environmental factors are easier to modify compared with macro‐scale factors. ¹⁵ , ¹⁶ Bike lanes, as a portion of the roadway, ¹⁷ are one such factor. There are multiple aspects of bike lane design, including the density of bike lanes, proximity to the nearest bike lanes, width of bike lanes and other bike ability indices. ¹⁸ , ¹⁹ Bike lane design and availability can indirectly influence childhood obesity through cycling behaviours, as it is assumed that a bikeable environment should encourage children and adolescents to conduct more PA such as cycling, ²⁰ which would be an important contributor to reducing rates of childhood overweight and obesity, within a broader socio‐ecological context. ²¹

Some studies have revealed an association between childhood obesity and the environmental factors that may support cycling. However, these study findings have been at a general level, often addressing this association using broader concepts of active transport infrastructure or behaviours (including overall PA) and remaining inconclusive in terms of effect direction and size. By way of example, Pont et al²² conducted a systematic review of the associations between environmental factors and active transportation, finding that bike lane access may be associated with higher rates of active transportation among young people aged 5–18 years. In previous studies, bike lane was only discussed as a subgroup variable, but no studies specifically examined cycling behaviours directly in relation to cycling infrastructure. ²²

Furthermore, Europe is developing active transportation policies to explore how environmental factors both positively and negatively influence cycling behaviors. ²³ However, associations of the policies promoting bike lane construction with PA are mixed, because many environmental and injury prevention policies have complex effects that have not been rigorously evaluated. It is necessary to conduct a systematic review of the relevant global research, in order to more robustly establish the relationship between bike lane access and childhood obesity.

This review aims to summarize the associations between bike lane access and childhood obesity. We tested our hypothesis that a lack, or lower level, of bike lane access was associated with lower levels of PA, higher levels of sedentary behaviours, and therefore greater weight among children and adolescents compared with their counterparts with greater bike lane access. ²⁴ Furthermore, we conducted a meta‐analysis to quantify the association of bike lane access with childhood obesity and PA.

2. METHODS

We conducted a systematic review and meta‐analysis based on the Cochrane Handbook 5.1.0, and the results of this systematic review and meta‐analysis were reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISRM). ²⁵

2.1. Eligibility criteria

Studies that met all of the following criteria were included in the review: (1) study designs: experimental studies, cross‐sectional studies and longitudinal studies including prospective and retrospective cohort studies; (2) study participants: children and adolescents aged <18 years; (3) exposures of interest: bike lanes (e.g. bike lanes access, length of bike lanes and availability of bike lanes); (4) study outcomes: weight‐related behaviours (e.g. PA, sedentary behaviours) and/or outcomes (e.g. overweight or obesity measured by body mass index [BMI, kg/m²] or waist circumference); (5) article types: peer‐reviewed original research; (6) time of publication: from the inception of the given electronic bibliographic database to 1 January 2019 and (7) language: written in English.

2.2. Search strategy

A keyword search was performed in four electronic bibliographic databases: Cochrane Library, PubMed, Embase, and Web of Science. The search strategy included all possible combinations of keywords from the three groups related to bike lanes, children and weight‐related behaviours or outcomes. The specific search strategy is provided in Appendix S1.

Titles and abstracts of the articles identified through the keyword search were screened against the study selection criteria. Potentially relevant articles were retrieved for an evaluation of the full text. The reviewers L. Z. and Y. L. independently conducted the title and abstract screening and identified potentially relevant articles for the full‐text review. Discrepancies were compiled by L. Z. and screened by two other reviewers X. P. and P. J. L. Z., Y.L., X. P. and J. P. jointly determined the list of articles for the full‐text review through discussion. Then, L. Z. and Y. L. independently reviewed the full texts of all articles in the list and determined the final pool of articles included in the review. Figure 1 shows the search and filtering process.

FIGURE 1.

Study exclusion and inclusion flowchart

2.3. Data extraction and preparation

A standardized data extraction form was used to collect methodological and outcome variables from each selected study, including authors, year of publication, country, sampling strategy, sample size, age at baseline, follow‐up years, number of repeated measures, sample characteristics, statistical model, attrition rate, measures of the bike lane access, measures of weight‐related behaviours, measures of body‐weight status and key findings on the association between bike lane access and weight‐related behaviours and/or outcomes. L. Z. and Y. L. independently extracted data from each study included in the review, and discrepancies were resolved by X. P. and P. J.

2.4. Meta‐analysis

A meta‐analysis was performed to estimate the pooled association size of bike lane access on each weight‐related behaviour and outcome. Weight‐related outcomes included BMI z‐score, overweight status and obesity status. Overweight status (BMI at or above the 85th percentile) and obesity status (BMI at or above the 95th percentile) were based on the 2000 age‐sex‐specific Centers for Disease Control and Prevention Growth Charts. Weight‐related behaviours included PA (e.g. cycling and active commuting to school), sedentary behaviours and diet. Several studies were excluded from the meta‐analysis due to the following reasons: neither standard error nor confidence interval (CI) was reported; association size was unable to be transformed into a standardized coefficient (i.e. beta coefficient) due to the limited information reported; the unit of the association size was inconsistent with others and less than two studies reported the same outcome variable.

Study heterogeneity was assessed by using the I ² index. The level of heterogeneity represented by I ² was interpreted as modest (I ² ≤ 25%), moderate (25% < I ² ≤ 50%), substantial (50% < I ² ≤ 75%) or considerable (I ² > 75%). ²⁶ A fixed‐association model was estimated when modest‐to‐moderate heterogeneity was present, and a random‐association model was estimated when substantial‐to‐considerable heterogeneity was present. ²⁷ Publication bias was assessed by a visual inspection of the funnel plot and Begg's and Egger's tests to see if more than 10 studies were included. ²⁸ All meta‐analyses were performed by the ‘meta’ packages using R software (Version R i386 3.4.2). ²⁹ All analyses used two‐sided tests, and p < 0.05 were considered statistically significant.

2.5. 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. This assessment tool rates each study based on 14 criteria (Tables S3). For each criterion, a score of one was assigned if ‘yes’ was the response, whereas a score of zero was assigned otherwise (e.g. an answer of ‘no’, ‘not applicable’, ‘not reported’ or ‘cannot determine’). A study‐specific global score ranging from 0 to 14 was calculated by summing up 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.

3. RESULTS

3.1. Study selection

Figure 1 shows the flowchart of study selection. We identified 2312 articles in three databases, and 1361 non‐duplicated articles were included for the title and abstract screening. After excluding 1271 irrelevant records, the full texts of the remaining 90 articles were reviewed against the study selection criteria. A total of 21 studies were included in this systematic review. Some of them were excluded from the meta‐analysis due to the following reasons: neither standard error nor CI was reported (n = 10); the unit of association size (use β instead of odds ratio [OR]) was inconsistent with others (n = 3) and less than two studies reported the same outcome variable (n = 5).

Table 1 summarizes the basic characteristics of the 21 included studies. All the studies were published between 2005 and 2018, with 14 cross‐sectional studies, 2 longitudinal studies and 5 experimental studies (e.g. experimental study using manipulated photographs to find cycling‐friendly environments for children). The sample size in these studies ranged widely from 53 to 1 244 862. The largest number of studies was conducted in Belgium (n = 6), followed by the United States (n = 4), Australia (n = 3) and the Netherlands (n = 3), and with one study each in Greece, Ireland, Spain, Sweden and the United Kingdom. Three of these studies were conducted at the national level, and the rest were conducted at the state (n = 8) and city (n = 10) levels.

TABLE 1.

Basic characteristics of the studies included

First author (year)	Study area, country [scale]	Study design	Sample size (% of boys)	Age at baseline (years + SD) a or range (years)	Sample characteristics (follow‐up status for longitudinal studies)	Statistical model
Boarnet (2005) 30	California, US [S1]	C	862 (NA)	Grades 3–5 in 2002–2003	ECLS‐K survey participants	Two sample t test
Carver (2015) 31	Victoria, Australia [S1]	C	640 (49%)	Aged 11.6 ± 2.0 in 2010	The sample was recruited for the Australian arm of an international comparison study of children's independent mobility. They are 411 primary and 229 secondary school‐age children.	Logistic regression
Chen (2018) 32	Seattle, WA, US [C1]	L	53 (NA)	NA	Students from elementary schools (K‐5) or elementary and middle school (K‐8) students. (followed up from 2005 to 2016 with eight repeated measures).	Generalized linear model
Chomitz (2011) 33	Somerville, MA, US [C1]	C	926 (49%)	Grades 6–8 in 2007	2007 youth risk surveillance survey (YRBS) participants. Middle school students (grades 6–8).	Multivariate logistic regression
Estevan (2018) 34	Valencia, Spain [C1]	C	465 (45%)	Aged 16.5 ± 0.8 in 2013–2015	International Physical Activity and the Environment Network (IPEN) adolescent study participants. Students from nine high schools.	Mixed associations regression models
Ghekiere (2015) 15	Flanders, Belgium [S1]	E	1232 (50%)	Aged 10.5 ± 0.6 in 2014–2015	Students (grades 5–6) from 45 primary school located across Flanders.	Hierarchical Bayes analyses.
Ghekiere (2015) 35	Flanders, Belgium [S1]	C	305 (52%)	Aged 11.3 ± 0.6 in 2014	Students (grades 5–6) from 12 primary school located across Flanders.	Hierarchical Bayes analyses
Ghekiere (2016) 36	Melbourne, Australia [C1]	C	677 (47%)	Aged 11.5 ± 0.6	Children Living in Active Neighborhoods (CLAN) study participants. Students from 19 primary schools.	Multilevel linear regressions
Ghekiere (2018) 37	Flanders, Belgium [S1]	E	1289 (49%)	Aged 10–12 in 2014–2015	Students (grades 5–6) from 45 primary school located across Flanders.	Hierarchical Bayes estimation
Helbich (2016) 38	Amersfoort, Haarlem, Hengelo, Rotterdam, Vlaardingen, Netherlands [C5]	C	97 (40%)	Aged 6–11 in 2008–2009	Part of ‘Spatial Planning and Children's Exercise’ (SPACE) project. Students from six elementary schools located in five neighbourhoods in mid‐to large‐sized Dutch cities.	Generalized linear mixed models and Spearman's correlation
Kamargianni (2015) 39	Greece and Cyprus[N2]	C	9554 (48%)	Aged 15.7 in 2012–2013	Students from public high schools in different types of cities (urban, rural and insular) of two different countries.	Logistic mixture model
Lee (2017) 40	Austin, Texas, US,[C1]	E	165 (50%)	Grades 1–5 in 2011	Elementary students from a new school that opened in late August 2010.	Binomial logistic regression
Mandic (2017) 41	Dunedin, New Zealand [C1]	C	764 (45%)	Aged 13–18 (15.2 ± 1.4) in 2014–2015	Built Environment and Active Transport to School (BEATS) study participants. Students from 12 secondary schools.	Mean (SD) and frequency (%)
Nelson (2010) 42	Ireland [N]	C	2159 (53%)	Aged 15–17 (16.04 ± 0.66)	The take Physical Activity Research for Teenagers (PART) study participants. Individuals who lived within 2.5 miles of their school.	Bivariate logistic regression and multivariate mode
Oliveira (2018) 43	Northern region of Portugal [N]	L	583 (49%)	Aged 12–18 (14.28 ± 1.79) in 2011–2013	Part of the longitudinal analysis of biomarkers and environmental determinants of physical activity (lab med physical activity study). (followed up from 2011 to 2013 with two repeated measures)	Linear regression
Veitch (2017) 44	Melbourne, Australia [C1]	C	92 (42%)	Aged 13–16 (14.7 ± 1.0) in 2014	Students from different socio‐economic status (SES) backgrounds.	Hierarchical Bayes analyses
Verhoeven (2017) 45	Flanders, Belgium [S1]	E	882 (55%)	Aged 12–16 (13.9 ± 1.6) in 2016	Students (grades 1–4) from 12 secondary schools across Flanders.	Hierarchical Bayes estimation and logistic regression analyses
Verhoeven (2018) 46	Flanders, Belgium [S1]	E	882 (55%)	Aged 12–16 (13.9 ± 1.6) in 2016	Students from 12 secondary schools across Flanders.	Hierarchical Bayes estimation
Verhoeven (2018) 47	Ghent, Flanders, Belgium [C1]	C	204 (47%)	Aged 12–16 (14.4 ± 1.2) in 2015	Students (grade1–4) from six secondary schools in and around Ghent.	Univariate multilevel logistic regression analyses
Vries (2007) 48	Netherlands [C6]	C	422 (49%)	Aged 6–11 (8.3 ± 1.4) in 2004–2005	Part of the Spatial Planning and Children's Exercise (SPACE) study. Students from 20 elementary schools.	Univariate and multivariate linear regression
Weimann (2015) 49	Scandinavia, Sweden [S1]	C	205 (50%)	Aged 4–11 (8.5 ± 1.6) in 2009–2010	Part of Identification and Prevention of Dietary‐ and Lifestyle‐Induced Health Associations in Children and Infants (IDEFICS) study. They from 168 families.	Mixed linear regression

Note. Scale: [N] – National; [S] – State (e.g. in the United States) or equivalent unit (e.g. province in China, Canada); [Sn] – n states or equivalent units; [CT] – County or equivalent unit; [CTn] – n counties or equivalent units; [C] – City; [Cn] – n cities.

Abbreviations: C, cross‐sectional; E, experimental; L, longitudinal.

^aNA = Not available.

Table S3 reports criterion‐specific and global ratings from the National Institutes of Health's Quality Assessment Tool for Observational Cohort and Cross‐Sectional Studies. The 21 studies scored between 9 and 14 with an average of 10.3.

3.2. Measures of bike lane access

The bike lane access was measured by Geographic Information Systems (GIS) as the length of walking/cycling tracks within buffer zones centred on individual addresses or schools (n = 6), with varying radii (from 0.25 to 5 km). The most commonly‐used buffer zone was a 0.25‐km road‐network buffer, followed by a 0.8‐km road‐network buffer, a 0.5‐km road‐network and a 5‐km road‐network. In terms of the numbers of studies by methodology, experimental studies using manipulated photographs to investigate which micro‐scale environmental factors determine children's or adolescents' preferences towards cycling for transport (e.g. evenness of cycle path including the categories of ‘very uneven’, ‘moderately uneven’ and ‘even’; type of cycle path including the categories of ‘no cycle path’; ‘cycle path separated from traffic with lines’, ‘not separated from walking path’, ‘cycle path separated from traffic with a curb, not separated from walking path’, ‘cycle path separated from traffic with a hedge, not separated from walking path’) (n = 6); cycling conditions by a checklist adapted from the Neighborhood Environment Walkability Scale (NEWS) (n = 2) and Assessing Levels of Physical Activity environmental (ALPHA) (n = 1); the presence of bike lanes by questionnaire (n = 2); the frequency of walk/bike path usage by questionnaire (n = 2); the presence of bike lanes by self‐reporting (n = 1) and the availability of bike lanes by questionnaire (n = 1; Table S1).

3.3. Association between bike lane access and weight‐related behaviours

Twenty studies examined the association between bike lane access and weight‐related behaviours, including those that only measured PA (n = 19) and mixed findings that measured both PA and sedentary behaviour (n = 1; Table S2). When PA was used as the outcome variable, 10 studies reported the relationship between bike lane access and children's PA, such as active commuting to school, active commuting to home and cycling for leisure purpose. Among them, six studies reported a significant positive correlation between bike lane access and children's sports activities. ³¹ , ³² , ³⁸ , ³⁹ , ⁴² , ⁴⁸ Two studies reported that there was no significant correlation between bicycle tracks and children's PA. ³⁴ , ³⁶ One study reported a significant negative correlation between bike lane access and children's sports activities. ⁴⁰ One study reported that bike path usage was significantly associated with moderate PA but not associated with vigorous PA. ³³ Six studies reported the relationship between bike lane access and children's preferences and intention to cycle. ¹⁵ , ³⁵ , ³⁷ , ⁴⁴ , ⁴⁶ , ⁴⁷ The degree of separation of cycle path, evenness of cycle path and street maintenance were associated with adolescents' preferences and intention to cycle for transport. Three studies found that parents or children reported that bike lane access was an important condition for children's PA. ³⁰ , ⁴¹ , ⁴⁵ One study reported that children with low access to bike paths had more sedentary time than those with medium or high access. ⁴⁹

Figure 2 summarizes the modelling results from the meta‐analysis. It shows the forest plot of change in children's PA in response to bike lane access with OR. A meta‐analysis was conducted to estimate the pooled estimation size of the association between measures of bike lane access and PA outcomes. We observed significant association in the meta‐analyses (OR = 1.57, 95% CI: 1.37–1.81) and with moderate heterogeneity (I ² = 38%).

FIGURE 2.

Forest plot of changes in children's physical activity in response to bike lane access with OR. OR, Odds ratio; Norah M. Nelson (2010 A): Male; Norah M. Nelson (2010 B): Female; Virginia Rall Chomitz (2011 A): Moderate Physical Activity; Virginia Rall Chomitz (2011 B): Vigorous Physical Activity; Virginia Rall Chomitz (2011 C):60+Minutes Physical Activity Appendix A. Search strategy and search results

3.4. Association between bike lane access and weight‐related outcomes

One study reported that the presence of bike lanes was associated with a lower BMI and waist circumference in girls; the availability of bike lanes was associated with a higher BMI and waist circumference in boys. ⁴³ The study also showed that perceptions of distant facilities at baseline were associated with lower fitness at follow‐up in boys. Also, the positive perception of a pleasant environment at baseline was associated with better fitness at follow‐up among boys. Additionally, for girls, higher bike lane availability and positive aesthetic perception at baseline were associated with healthier body composition at follow‐up.

4. DISCUSSION

The aim of this research was to systematically review the association between bike lane access and childhood obesity. We identified and systematically reviewed 21 studies that assessed the association between the bike lane access and weight‐related behaviours and outcomes in children and adolescents. We included 14 cross‐sectional studies, 2 longitudinal studies and 5 experimental studies. The majority of studies measured bike lane access using GIS‐based measures, and PA was the most commonly studied outcome variable. Mixed results were observed for this association across the studies. Although only a few studies reported null associations for weight‐related behaviour/outcomes with increased bike lane access, most of the studies reported positive associations for children's weight‐related behaviour/outcomes with increased bike lane access. Our meta‐analysis also found that the availability of bike lanes was associated with PA among children.

Overall, our results showed that children's increased active transportation and PA were related to bike lane access. This may suggest that as the presence of bike lanes increase, so does the likelihood that children or their parents will choose to cycle, which has considerable potential for increasing health promoting levels of PA. It is widely accepted that the neighbourhood environment may interact with personal characteristics to affect individual weight status and, at times, even outweigh personal factors. ¹⁶ The elements external to the individual that are involved in the development of obesity have become known as the obesogenic environment. ⁵⁰ A systematic map of reviews on social and environmental interventions to reduce childhood obesity identified a need for reviews focusing on interventions or changes to the built environment. ⁵¹ Meanwhile, our result is consistent with other previous reviews. For instance, Pont et al ²² found a possible positive correlation between children's active transportation and recreational facilities, bicycles and/or walking facilities near home. Lorenc et al ⁵² found a positive correlation between the presence of walking and/or bike paths and PA among children.

Interestingly, Lee et al ⁴⁰ found a significant negative correlation between bike lanes and children's sports activities, which may be caused by a small sample size (n = 465). Smith et al ⁵³ found parental licence for independent mobility was only associated with a need for safer places to cycle (positive) and objectively assessed cycling infrastructure (negative) in adjusted models. This finding could be due to parents allowing their children to be independently mobile, but more so for walking rather than cycling. ⁵³ Moreover, PA was measured using questionnaires, which can lead to recall bias, thus affecting the final results. At the same time, Ghekiere et al ³⁶ and Estevan et al ³⁴ reported an insignificant correlation between bike lane access and children's sports activities, which may also be caused by a small sample size (677 and 165, respectively). In addition, the linear regression model used by Ghekiere et al ³⁶ did not adjust for related confounding. We believe that this may also be a negative result caused by study bias, which needs to be further expanded for verification.

On the other hand, a review by Harrison and Jones ⁵⁴ supports our results from another perspective. They found that children attending schools with the best nearby conditions for walking and cycling (e.g. cycle lanes and traffic calming) spent more time in PA during commuting times to and from school compared with those at schools that have the worst provision. Fraser and Lock ¹⁹ found a positive correlation between commuting by bike and bike lane access. This is also consistent with our results, which indicate that parents or children reported bike lane access to be an important condition for children's PA. More directly, our results showed that bike lane access was not only associated with children's activities but also with obesity levels in children. Our results showed that the presence of bike lanes was associated with a lower BMI and waist circumference for girls but a higher BMI and waist circumference for boys. Nonetheless, it was difficult to draw a firm conclusion about the association between bike lane access and BMI/waist circumference.

Logically, after observing the results related to bike lane access and children's activities, it was important to further explore how specific micro‐environment factors related to children's PA. Our results show that bike lanes that were well‐separated and good maintenance are likely to encourage adolescents to cycle. It is also important to note that cycling infrastructure is an important factor that may positively influence children's cycling. For example, Giles‐Corti et al ⁵⁵ found that improving micro‐scale attributes may increase the suitability of a street for children's cycling activities, such as improving well‐separated of bike lane, even of bike lane and good maintenance of bike lane.

To advance the research on the association between the bike lane access and children's weight‐related behaviours and outcomes, future studies should overcome or mitigate several limitations of this study. First, objective measurement should be conducted in a more precise and consistent way, for example, using GIS‐based road‐network distance and a set of radii a priori for better comparability and better reporting of methods. ⁵⁶ , ⁵⁷ More advanced spatial approaches, such as remote sensing and citizen science, ⁵⁸ , ⁵⁹ are alternative methods to obtain such environmental measures where GIS‐based road‐network are not available. ⁶⁰ , ⁶¹ Second, some novel objective measures and subjective measures should be added to measure all dimensions of bike lane access and its affiliated microenvironment, such as evenness of the cycle path and the degree of separation of cycle paths. Third, more pathway‐based analyses need to be conducted to elucidate underlying mechanisms from bike lane access in individual's home and/or school neighbourhoods to child weight‐related behaviours and outcomes. For example, greater bike lane access would also increase the access to food venues in the neighbourhood, including both healthful and unhealthful ones. ²⁴ , ⁶² The actual pathway from bike lane access through PA to weight‐related outcomes may be affected by the neighbourhood food environment and hence may be more complex than our hypothesis. However, these research questions need to be answered in longitudinal study designs that account for multiple levels of influence on body size outcomes across the socio‐ecological model. ⁶³ , ⁶⁴ Lastly, bike lanes may have different degrees of completeness across regions or be occupied by vehicles (or parking) and/or pedestrians, which could all affect the actual utilization of bike lanes and the association between bike lane access and children's PA and weight status through factors such as parental perception of road or neighbourhood safety for cycling. ⁶⁵ Moreover, the emerging bike‐sharing systems (widely popular in many cities and countries already) are another important factor that could make the utilization of bike lanes different from traditional scenarios where bikes are owned exclusively by individuals. ⁶⁶ , ⁶⁷ Therefore, changes in the perceptions and behaviours of cyclers could also be an important research consideration.

5. CONCLUSIONS

Although most studies included in this systematic review and meta‐analysis revealed a positive association between bike lane access and PA of children and adolescents, it was difficult to draw a conclusion on the association between bike lane access and weight‐related outcomes. However, according to many reasonable scientific hypotheses and evidence from some high‐quality research, improving bike infrastructure and the relevant microenvironment may be an effective way to improve the support for children and adolescents to cycle safely outdoors and engage in more PA. Such actions would also help us design more longitudinal studies to further elucidate the association between bike lane access and children's obesity.

CONFLICT OF INTEREST

We declare no conflicts of interest.

Supporting information

Data S1 Supporting information

Table S1. Measures of bike lane access and weight‐related behaviors or body‐weight status in the studies included (see [ref] in the reference list of this document)

Table S2. Estimated associations between bike lane access and weighted‐related behaviors and body‐weight status in the studies included (see [ref] in the reference list of this document)

Table S3. Study quality assessment (see 14 questions at the footnote) (see [ref] in the reference list of this document)

Pan X, Zhao L, Luo J, et al. Access to bike lanes and childhood obesity: A systematic review and meta‐analysis. Obesity Reviews. 2021;22(S1):e13042. 10.1111/obr.13042