Zougheibe et al. Int J Geogr (2021) 20:2 https://doi.org/10.1186/s12942-020-00254-w International Journal of Health Geographics

REVIEW Open Access Children’s outdoor active mobility behaviour and neighbourhood safety: a systematic review in measurement methods and future research directions Roula Zougheibe1* , Jianhong (Cecilia) Xia1, Ashraf Dewan1, Ori Gudes2 and Richard Norman3

Abstract Background: Numerous studies have examined the association between safety and primary school-aged children’s forms of active mobility. However, variations in studies’ measurement methods and the elements addressed have con- tributed to inconsistencies in research outcomes, which may be forming a barrier to advancing researchers’ knowl- edge about this feld. To assess where current research stands, we have synthesised the methodological measures in studies that examined the efects of neighbourhood safety exposure (perceived and measured) on children’s outdoor active mobility behaviour and used this analysis to propose future research directions. Method: A systematic search of the literature in six electronic databases was conducted using pre-defned eligibil- ity criteria and was concluded in July 2020. Two reviewers screened the literature abstracts to determine the studies’ inclusion, and two reviewers independently conducted a methodological quality assessment to rate the included studies. Results: Twenty-fve peer-reviewed studies met the inclusion criteria. Active mobility behaviour and health charac- teristics were measured objectively in 12 out of the 25 studies and were reported in another 13 studies. Twenty-one studies overlooked spatiotemporal dimensions in their analyses and outputs. Delineations of children’s neighbour- hoods varied within 10 studies’ objective measures, and the 15 studies that opted for subjective measures. Safety perceptions obtained in 22 studies were mostly static and primarily collected via parents, and dissimilarities in actual safety measurement methods were present in 6 studies. The identifed schematic constraints in studies’ measurement methods assisted in outlining a three-dimensional relationship between ‘what’ (determinants), ‘where’ (spatial) and ‘when’ (time) within a methodological conceptual framework. Conclusions: The absence of standardised measurement methods among relevant studies may have led to the cur- rent diversity in fndings regarding active mobility, spatial (locality) and temporal (time) characteristics, the neighbour- hood, and the representation of safety. Ignorance of the existing gaps and heterogeneity in measures may impact the reliability of evidence and poses a limitation when synthesising fndings, which could result in serious biases for policymakers. Given the increasing interest in children’s health studies, we suggested alternatives in the design and method of measures that may guide future evidence-based research for policymakers who aim to improve children’s active mobility and safety.

*Correspondence: [email protected] 1 School of Earth and Planetary Sciences, Curtin University, Kent Street, Perth, WA 6102, Australia Full list of author information is available at the end of the article

© The Author(s) 2021. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativeco​ ​ mmons.org/licen​ ses/by/4.0/​ . The Creative Commons Public Domain Dedication waiver (http://creativeco​ mmons​ .org/publi​ cdoma​ in/​ zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 2 of 24

Keywords: Children’s active mobility, Perceived safety, Measured crime, Geographic information system (GIS), Global positioning system (GPS), Activity tracking, Spatiotemporal analysis, Methodological conceptual framework

Introduction More importantly, variations in active behaviour from Extensive evidence supports that high levels of physi- childhood to adolescence [22] suggest that measure- cal activity (PA) have profound long-term health ben- ment methods may vary by age group. Tus, the dearth efts for children [1–5]. Among primary school-aged of review in measures of active behaviour in safety con- children, outdoor active (non-motorised) forms of text and the profound importance of this age group mobility such as free play, organised sport or active [31] on lifelong PA have directed our focus onto pri- travel between destinations are signifcant contributors mary school-aged children. We expand upon the earlier to children’s overall PA [6]. Within the feld of child- approach and examine children outdoor active mobility specifc research, one well-established potential infu- behaviour (COAMB) in terms of free play, scheduled ence on children’s active behaviour is neighbourhood activities and the associated active travel between des- safety in terms of personal safety and road dangers [7, tinations, as opposed to more typical reviews that only 8]. Although, the efects of neighbourhood safety on discuss one type of PA, such as trips to and from school PA may vary according to whether dangers are meas- or free play [23]. ured or perceived [7], neighbourhood safety continues In this review, we aim to synthesis and assess the dif- to threaten children’s PA [9] and afect their activity ferent methodological measures found in studies that space [10]. However, existing literature has produced examined the impact of safety exposures on modify- mixed fndings on the efect of safety on children’s PA ing primary school-aged COAMB. To accomplish this [11, 12]. Inconsistency in methods of measures may goal, this systematic review seeks to answer three ques- largely explain the disagreements between fndings. For tions: First, where does current research stand in terms example, perceived safety has often been assessed via of measurement methods used to examine the efect of diferent questionnaires developed to ft the objectives neighbourhood safety perceived (by parents or children) of individual studies [13]. Moreover, measuring chil- and measured (actual), on primary school-aged children? dren’s forms of active mobility behaviour have varied Secondly, what are the gaps in current practice? Finally, across studies from parent or child questionnaires [14] what are the future directions that could be taken in the to travel diaries [15] to time-latitude–longitude records area to better inform the decision-making process? of child mobility [16]. Additionally, a child’s neighbour- hood—a place that ofers opportunities for the major- ity of children’s daily routine activities (i.e., where they Methods live, go to school, visit a destination or play)—is seen as A systematic literature search was conducted based on playing a crucial role in the outcomes of the examined the methodological guidelines of the Preferred Report- contextual determinants of child health behaviour [10]. ing Items for Systematic Reviews and Meta-Analyses Te emerging tools that measure the intensity of (PRISMA) statement checklist of 27 items [24]. However, movement (e.g., accelerometers) or the geographic some items on the list were unchecked as they do not (spatial) location of movement (e.g., global position- apply to a methodological review (Additional fle 1). No ing systems [GPS]), and geographic technology, have protocol was published for this literature review. proven advantageous in improving our understanding of evidence-based child-related research [17]. How- ever, the pathway to obtaining reliable data for analysis Eligibility criteria remains laden with challenges in terms of measure- To be included in the review, papers had to meet the fol- ment, integration and technological limitations. lowing criteria: be peer-reviewed, be published in Eng- Te methodological reviews of safety exposures in lish, report participants’ characteristics (size, age, etc.) child-related studies are scarce. Available manuscripts examined primary school-aged children, measure forms have either addressed broad age groups combining of children’s outdoor active mobility (i.e., PA, outdoor environmental determinants measurements meth- active playing, , , biking and active travel), ods or were specifc [18–21]. A rigorous synthesis of and address (a) perceived or (b) measured (actual) cor- methodological measures based on an interdisciplinary relations with safety (personal and road). Reports, theses, vision in research examining neighbourhood safety protocols, non-peer-reviewed studies and studies that efects on active mobility behaviour is still lacking. assessed the efect of interventions (e.g., trafc calming) Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 3 of 24

were excluded. Detailed inclusion and exclusion criteria described; if a criteria were not applicable, it was dis- used for this review are also listed in Additional fle 2. counted from the total score. Each study could score a total of 8 points, and this maximum score was used Search strategy to calculate a percentage of study quality [25]. Study Te literature search began in February 2019 based on the quality was rated robust if the study secured a percent- above eligibility criteria. Six electronic databases, Google age of ≥ 66.7%, was rated fair if it scored between ≥ 50 Scholar, PubMed, Scopus, ScienceDirect, SpringerLink and < 66.6% or was rated weak if it scored < 50% [25]. Fur- and Web of Science were searched. Works were retrieved ther details on quality assessment can be found in Addi- using combinations of search terms explicitly developed tional fle 2. to meet the objectives of this review. We used a combi- nation of a minimum of three keywords rotated in turn, Results with each belonging to one of the following groups: (1) Study selection the target population, AND (2) active behaviour, AND (3) Te title scan that was carried out using the defned key the safety neighbourhood’s exposures, OR (4) moderators terms has identifed 231 papers, out of 13,091 title refer- and mediators OR (5) spatiotemporal aspects. Details of ences across six databases, as potentially relevant. After terms used per group are found in Additional fle 2. 128 duplicates were removed, the remaining 103 articles’ titles and abstracts were screened, and 64 studies were Selection process thoroughly reviewed. A manual search of the reference Te primary author (RZ) carried out a comprehensive lists of the individual studies yielded three additional screening of the retrieved studies. Te fnal date of the studies. A total of 25 articles passed a thorough review search was July 2020. Furthermore, RZ also scanned the and met the inclusion criteria (see Fig. 1). reference lists of the individual papers to identify further studies. An independent screening of abstracts for inclu- Study characteristics sion was conducted by the second author (CX). Te third Aside from three longitudinal studies, 22/25 studies author (RN) resolved disputes regarding which studies to (88%) had a cross-sectional design. Sample sizes ranged include. from 35 [28] children to 3,200 children [29]. Forms of COAMB were examined in correlation to perceived Data extraction safety (19/25 studies; 76%), measured safety (3/25 studies; After the study selection, RZ organised and extracted the 12%) or both measured and perceived safety (3/25 stud- relevant data into three main categories: (1) study char- ies; 12%). Multiple studies (10/25 studies; 40%) extracted acteristics, namely the author(s), year of publication, data for their analysis from larger projects (e.g. the Per- study location and participant demographics (e.g. age, sonal and Environmental Association with Child’s Health gender and ethnicity); (2) measures of safety (perceived [PEACH; UK] project), see Table 1. With the exception and/or actual) and children’s outdoor active mobility of two Iranian manuscripts [30, 31], 23/25 studies (92%) behaviour (COAMB); and (3) neighbourhood measures. explored populations in developed countries. Tese 23 We extracted the studies’ tools and methods for measur- studies were geographically distributed to include fve ing safety, COAMB and neighbourhoods, as well as the studies each in the and the United King- signifcant results, study variables and methods used to dom; four in Canada; three in Australia; two in New Zea- examine the relationships between these factors. land; and one each in Portugal, Finland, the and Austria (Additional fle 2: Figure 1). Methodological assessments of individual studies quality A formal assessment of the included studies’ quality was Study methodological quality assessment completed independently and critically by two review- Diferences between the frst reviewer (RZ) and the sec- ers RZ and BJ. Any rating discrepancies were discussed, ond reviewer (BJ), in the output of the methodological and a shared decision was reached in required cases. quality assessment, was resolved via discussion to achieve RZ compiled a 20 priori methodological quality crite- full agreement. Of the twenty-fve studies included, 8/25 ria, of which 13 were adopted from earlier reviews [20, studies were rated robust (32%), 14/25 studies (56%) were 25–27]. Te remaining seven criteria expanded upon fair, and three studies (12%) were weak. Further details measurement methods that were believed to be funda- are in Additional fle 2: Table 2. mental when examining the efect of safety exposures on COAMB (see Additional fle 2: Table 1). Each study was Measuring COAMB or health indices allocated a point if a criterion was present or was allo- Over half of the studies (13/25 studies; 52%) obtained cated no points if the criteria were absent or inadequately forms of COAMB from parent questionnaires and/or Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 4 of 24

Fig. 1 Flow chart of the PRISMA for the systematic methodological review

travel diaries. Te remainder (12/25 studies; 48%) used count from accelerometers was used in three stud- objective measures, including 2/25 studies (8%) used ies [33, 36, 41, 45], while another study used par- (BMI) measures, with remaining used ent questionnaires [44] to calculate children’s active an accelerometer activity tracker in fve studies (two play (both by frequency and duration of participa- Actigraphs, two Acticals, and two GT1M Actigraphs), tion in vigorous-intensity or moderate-intensity an Accusplit pedometer (AH 120M8 KS10), a GPS spa- activities and by reaching 60 min of daily PA based tial location tracker in fve studies (two Garmin Forerun- on the recommended guidelines). Another study ner 220 s, VGPS-900, GSM22, and QStarz BT-Q1000/ depicted intensity by measuring children’s daily BT-Q1000XT), and an ArcGIS to derive index average minutes of outdoor active play [40]. in one study. Some studies used a single tool such as an • Activity space: Two studies used children’s move- accelerometer [33, 36, 45] or GPS [50, 52] while a few ments across space to determine the size and studies combined more than one, such as a GPS with an geometry of movements when visiting places. Te accelerometer and travel diary/activity log [40, 41]. Vari- activity space is depicted by an area feature (closed ation of utilised tools and the discrete output units across polygon) and obtained by collecting multiple longi- studies are illsutrated below, Fig. 2. tude and latitude points of active mobility. Circu- lar multi-bufer rings represent various spaces, and Outcomes each ring is built from the frequency with which Te temporal characteristics of COAMB (weekends children visited the places and the time spent there vs. weekdays and time of day, such as before and after [52]. Alternatively, a minimum convex polygon school) were accounted for in 4/25 studies (16%) [28, 33, connecting at least three points of the visited desti- 36, 41]. Dissimilarities in the tools and methods used to nations is used to depict the activity space [49]. measure forms of COAMB, as displayed in Fig. 2, fol- • Active route (derived path of the active travel lowed by disparity in outcome being measured in difer- between destinations): One study used GPS way- ent units across studies: points embedded in watches to derive a linear fea- ture representation of active child transportation • Activity intensity: Moderate-to-vigorous physi- (i.e., the routes or path that a child uses to connect cal activities (MVPA) was used in fve studies. Step to his/her target destination) [28]. A longitudinal Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 5 of 24 derive objectively environmental measures/ Spatial analysis – Spatial analysis to Spatial analysis to statistical analysis Statistical analysis Statistical analysis Analysis method Analysis Spatial/ statistical/ spatial analysis statistical - - + SLICES) ownership, parents’ parents’ ownership, education) the home istics around and (within 100 m of the shortestbufer school to route Minority ethnic group, Minority ethnic group, Black,Asian, Chinese, Not speci - Other, Mixed, character fed), Family ownership, istics (car in the number of rooms house) Sex/gender IM licence (using Types Land use Sex/gender, age Sex/gender, Sex/gender (cars Sociodemographic character Environmental Sex/gender. age Sex/gender. Sex/gender, race/ethnicity (White, Examined variables Individual Family Neighbourhood home trian network bufer the home around (10 min walk) Bufer 500-m from 500-m from Bufer Local area Local Within 800 m pedes - Within Local area Local Local area Local Neighbourhood Mobility (AIM) to school to walkers) *active route to home to *active route Active IndependentActive Independent mobility *MVPA Walking frequencyWalking (high walkers/low The outcome outcome The Active measured Behaviour/ + individual diary interview/ (focussed group group (focussed discussion)/- questionnaire questionnaire on independent mobility non-school hours and after before school weekdays and weekend tionnaire/ GPS, mobility travel GPS, mobility travel 7 days From children children From Derived from children children Derived from Accelerometer/ days 8 consecutive From children ques - children From past 7 days Active behaviour Active period/ tool/recall - - - - + pre + per + percep + mobility defensive + defensive naire (children and (children naire mobility parents patterns perceived licences, safety) (focussed group group (focussed discussion) tion of trafc safety tion of trafc safety concerns survey- (social/physi and cal environment their rules regarding physical children activity and percep ance behaviour risk) and ceived active transporta - 15 destina - tion to tion tion of the local environment ferred travel method travel ferred naire frequency Parents question - Parents From children children From Parents perception perception Parents Parents questionnaire Parents - (avoid Indices for Children question - Children Self-reported (walking Studies’ measures Studies’ Safety Questionnaire by) (perceived (actual)measured (CLAN) from from (CLAN) a longitudinal study Finland cross-section United KingdomUnited cross-section United KingdomeUnited Longitudinal (SPEEDY)Data from project Australia SectionCross United KingdomUnited SectionCross Country study design Summary of and characteristics methods of to measures the examine between association and active behaviour outdoor children neighbourhood 170 (51% M) 473 (250 M/160 F) = 491 (39%M) 1 (18 M/17F)/ not reported)/ 2007)/ n longitudinal study/ longitudinal = 492 (Sex/gender = 1121(43%M) = = [ 28 ]/ Sep.–Oct. 2009/ 35 children 10–11 years Aged [ 35 ]/- n 9–11-year-old Aged [ 34 ], T1 (April–July n 2008) T2 (April-July 9–10 years Aged ], From 2001–2005 [ 33 ], From n Aged 10–11 years Aged [ 32 ]/- n years old 9–11 years Aged Study characteristics Study Citation (by (by Citation alphabetical order)/ of data year collection/ n: sex/ Sample Size gender (M/F) (or mean Age/grade age) 5 4 3 2 1 Table in the included studies safety # Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 6 of 24 Spatial analysis/ statistical analysis Spatial analysis/ statistical analysis Statistical analysis Analysis method Analysis Spatial/ statistical/ spatial analysis statistical - - - + , - + Par + older + connect and + Neighbourhood parents + parents neigh - sibling (sex ethnicity of (New European, Zealand Island, Maori, Pacifc Indian,Samoan, Asian, Others), study or work outside the home, household (dwelling type, cars availability length of residency) IM bourhood perception of safety cohesion, ents IM licences tion from child and par tion from ents type (land use) the neighbourhood (neighbourhood income), neighbour (roads, hood perception acces - personal safety, sibility of facility) age, Sex/gender, demographics Parents Sex/gender, age Sex/gender, percep Environmental Age, sex/gender, SES of sex/gender, Age, Examined variables Individual Family Neighbourhood around the home around home, the second home, set those found within 1,600 m The immediate street street immediate The AS 400, 800 m of model Moore’s Neighbourhood Neighbourhood (IM) Activity Space time Independent Mobility Neighbourhood (NAS) *Outdoor playing playing *Outdoor MVPA The outcome outcome The Active measured Behaviour/ Travel Diary/ Travel 7 days GPS/ 7 days Accelerometer/ 7 days Active behaviour Active period/ tool/recall - - - - + envi + connec - + mobility parents’ + parents’ + social + safety parental + parental con - + neighbour - (Demograph (CATI) ics hood percep tion cohesion tion IM cerns), children naire (child habitual (child naire neighbourhood activities behaviour percep ronmental tion (child outdoor active outdoor (child play on the perception neighbourhood) Parents questionnaire questionnaire Parents Parents questionnaire Parents question - Children Parents questionnaire questionnaire Parents Studies’ measures Studies’ Safety Questionnaire by) (perceived (actual)measured Cross-section project (KITC) From Canada the (STEAM) From project Canada Cross-section [BEAT] project from Country study design (continued) 254 (100 M/133F) 143 (49 M/94F) 736 (47% M and and 2012/ years aged 8–13 years (mean age of 10.5) and 239 parents May of 2010 and 2011 and 5—8 May2011/ 52%F) included in the analysis/ 5–6 grade = = = [ 38 ]/ Between 2011 n [ 37 ]/During April and n two groups aged 9–11 two groups 12–13 years/grade [ 36 ] /April2010 – n years, 10–12 years, Aged Study characteristics Study Citation (by (by Citation alphabetical order)/ of data year collection/ n: sex/ Sample Size gender (M/F) (or mean Age/grade age) 8 7 6 1 Table # Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 7 of 24 cal analysis Spatial analysis/Statisti - Statistical analysis Statistical analysis Analysis method Analysis Spatial/ statistical/ spatial analysis statistical + school travel perceived + perceived non-white) (single or dual parents’ number household, household of siblings, edu - parental income, value of cation, parents’ and income outdoor trafc speed, volume, trafc calming and pedestrian infrastruc - ture environment (access to (access to environment media in the bedroom, parents’ IM derived from Area questionnaires, Deprivation*,level safety perceived characteristics (father/ mother driving licence, father/ cars, owned mother occupation status) of walkingsafety to school attitude, parental mode, walking school time to Sex/gender Race/ethnicity (white, characteristics: Family (trafc safety Pedestrian Sex/gender, home Sex/gender, Sex/gender, household Sex/gender, Examined variables Individual Family Neighbourhood around participantsaround home deprived areas km bufer zone 1 km zone bufer High and low School to home area School to Neighbourhood per day of active per day play outdoor (BMI) /Self-reported PA school Average of minutes of minutes Average Body Mass Index Perceived Walking Walking to Perceived The outcome outcome The Active measured Behaviour/ - GPS in the watch derived from a derived from questionnaire ceived Walking Time Time ceived Walking school (PWTS)to in min Accelerometer and Accelerometer 7 days And activity log Self-reported PA Self-reported PA Parents reported Per Parents Active behaviour Active period/ tool/recall - Pedestrian safety Pedestrian pedestrian ceived safety NEWS_Y Index to derive ment using PAQ-C PA (mode of transport in the previous week,- demograph school access to ics, service and public transportation, attitude towards waking Objective measures of Objective measures survey per for Parents Parents Survey used Parents - environ perceived self-reported Children Parents questionnaire questionnaire Parents Studies’ measures Studies’ Safety Questionnaire by) (perceived (actual)measured Canada SectionCross United KingdomUnited cross-sectional Iran Cross-Section Country study design (continued) 458 (230 M/228F)/ 2015–2016/ dren (364 M/371F)/ dren aged 7–9 years the survey = = 194 / [ 40 ]/ study between n aged 10–12 [ 39 ]/2014/ n aged 9–10 years - [ 30 ] /-/ of chil - 735 parents in 9 schools returned Study characteristics Study Citation (by (by Citation alphabetical order)/ of data year collection/ n: sex/ Sample Size gender (M/F) (or mean Age/grade age) 11 10 9 1 Table # Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 8 of 24 derive the area of derive the area statistical exposures/ analyses Spatial Analysis/ Statistical Analysis Spatial analysis to Spatial analysis to Analysis method Analysis Spatial/ statistical/ spatial analysis statistical - + neighbour streetscape + streetscape hood exposures hood exposures GIS street (measured distance connectivity, destina - school, to tion accessibility, Ratio of High-speed around roads school audit) ethnicity, (New Zealand (New Zealand ethnicity, Maori, Pacifc European, Indian/Asian/ Island, Other Ethnicity) pick up) assistance) Sex/gender, age, race/ age, Sex/gender, availabilitySES (car for SES (car ownership, public ownership, SES (car race/ethnicity) Examined variables Individual Family Neighbourhood participants home address distance of 3.2 km (using GIS and geocoded students’ home address Bufer 1000-m around 1000-m around Bufer Within walkingWithin Neighbourhood derived from derived from Route National Safe School Survey To *%MVPA Walking to school to Walking The outcome outcome The Active measured Behaviour/ + GPS hours tionnaire that participants within walkingare distance between school) home to participants address Travel + Travel diary/ Accelerometer outside school 7 days From parents’ ques - parents’ From Inclusion criteria were geocode GIS used to Active behaviour Active period/ tool/recall - + measured + measured naire on neighbour naire hood perception hood perception from using items Ranui Action Survey networkroad adapted from sev - from adapted eral surveys includ - ing the National Safe for Centre School to Routes Survey, Parents (UH-PEAK), SPAN, NEWS, and EnVivo) CATI-Parents question - CATI-Parents Parents questionnaire questionnaire Parents safety Personal Safety Trafc Studies’ measures Studies’ Safety Questionnaire by) (perceived (actual)measured dinal project New Zealand Cross-section project (KITC) from United States United from Cross-section [T-COPPE]- longitu Country study design

th 4 (continued) 830 parents of ­ 830 parents 236 (104 M/132) and 2012/ for weekday weekday for and 210 analyses, (91 M/119F) for weekend days mean Age analyses. this study 9.8 for 9 schools, from 5–8 grade grade (412 M/418F) grade = = [ 42 ]/ - n [ 41 ]/ Between 2011 n Study characteristics Study Citation (by (by Citation alphabetical order)/ of data year collection/ n: sex/ Sample Size gender (M/F) (or mean Age/grade age) 13 12 1 Table # Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 9 of 24 Statistical analysis Statistical analysis Analysis method Analysis Spatial/ statistical/ spatial analysis statistical ethnicity (Hispanic/ African Ameri - Latino, can, American Indian/ Alaska Asian Native, American, White) built environment features ethnicity (white, ethnicity (white, but not non-white, in analysis) accounted Social Norms, Safety, Constraints, Nuisance, of minutes accessibility, 3 pm till from daylight of deprivasunset), level - tion (using Index of Multiple Deprivation from (IMD) and derived categories of seven deprivation, Daylight, BMI Pubertal status, Sex/gender, age, race/ age, Sex/gender, of perception Parents SES Family Sex/gender, age, race/ age, Sex/gender, of (Aesthetics, Perception Examined variables Individual Family Neighbourhood 10–15 min Walking distance Walking Not reported Neighbourhood play, exercise and exercise play, sport, active com - muting Active play Active Frequency of outdoor of outdoor Frequency Local-IM -IM Area The outcome outcome The Active measured Behaviour/ questionnaire naire Frequency Frequency naire in participation in active active play, and struc - travel and exercise tured sport From the parents’ the parents’ From From the question - From Active behaviour Active period/ tool/recall - environment) ised) perception of ised) perception the environment nuisance, (aesthetic, including traf - safety cross, fc of places to heavy trafc and social norm, road, constraints) naire (computer naire Parents questionnaire Parents of the (perception Children question - Children Studies’ measures Studies’ Safety Questionnaire by) (perceived (actual)measured - longitu (PEACH) dinal study United States United section cross United KingdomUnited from cross-section Country study design M,72F)/ (continued) (mean age of 9.7 children) and 2008/ (639 M/661F)/ 23 schools = 144(72 = 1307 years aged 7–12 years [ 44 ]/Sep.-Dec.2014/ n [ 43 ]/ Between 2006 n old from 10–11 years Study characteristics Study Citation (by (by Citation alphabetical order)/ of data year collection/ n: sex/ Sample Size gender (M/F) (or mean Age/grade age) 15 14 1 Table # Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 10 of 24 Statistical analysis Spatial analysis statistical analysis Statistical analysis Analysis method Analysis Spatial/ statistical/ spatial analysis statistical - - neighbourhood envi walkabilityronmental collective efcacy, scale, crime victimisationPrior stranger danger survey, and crime perception demography (parents (parents demography - education), paren age, percep parents’ tal PA, tion of neighbourhood (sidewalk,safety street stran - from fear safety, crime and trafc gers, safety) Sociodemographic Sex/gender, age, race, age, Sex/gender, Household income, Sex/gender, age, family age, Sex/gender, Examined variables Individual Family Neighbourhood Home-school Census blocks Census Local destinations Local Neighbourhood (IM) Walking to school to Walking MVPA Independent mobility The outcome outcome The Active measured Behaviour/ survey the Difered home to trips from school and from home school to tionnaire of previous previous of tionnaire activ physical - week ity based on IPAQ From the parents’ the parents’ From Accelerometer/ 7 days Derived from a ques - Derived from Active behaviour Active period/ tool/recall - + - + crime + stran - + prior the perception the perception of environmental factors that prevent walk from children school ing to victimization perception ger danger perception) crimes geocoded near participants home naire (demo naire graphics - physi (parental family cal activity, and demographic, (adapted perception NEWSfrom and studies) previous derive to naire mobility style Police + Police reported Parents survey Parents surveyChildren on

Parents question - Parents Parents questionnaire questionnaire Parents question - Children Studies’ measures Studies’ Safety Questionnaire by) (perceived (actual)measured section study - longitudi [SALTA] nal study Iran Cross-section United States United Cohort cross- (NIK) study From Portugal from cross-section Country study design th, ­ 6 (continued) 354 (156 M) of 145 (71 M/74F) / grade grade 2007–2009/ = = [ 46 ]/ 2010/2011/ n ]/the year 2009/ [ 31 ]/the year Grade 3 – 5/- [ 45 ]/Between n 6–11-years-old Study characteristics Study and their parents Citation (by (by Citation alphabetical order)/ of data year collection/ n: sex/ Sample Size gender (M/F) (or mean Age/grade age) 17 18 16 1 Table # Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 11 of 24 Statistical analysis Statistical analysis Analysis method Analysis Spatial/ statistical/ spatial analysis statistical Statistical analysis Crime Risk Index, + + IM licence Consumer Expendi - Consumer ethnicity (White, Asian, African American, Hispanic), demography (median household income and educa - tion) (working status of per vehicles parents, Parental household, attitude (promoters, - protec pragmatists tors) (language, SES, marital (language, education, cars’ status, a dog own ownerships, pub. safety, (trafc, Trans) (neighbourhood and view of parents) 2 ture Data, the density of ture outlet (using walk food PA and places for score Sex/gender, age, race/ age, Sex/gender, Age, family background family background Age, Sex/gender background Family of parents Perception of children Perception Examined variables Individual Family Neighbourhood Urban Zipcode Neighbourhood Walking distance Walking Neighbourhood (BMI) Mobility (AIM) Body Mass Index Active IndependentActive Frequent Walking Walking and Frequent The outcome outcome The Active measured Behaviour/ Travel + Travel Diary assessed at baseline 2012 and three later years tured question - tured naire (using KONTIV- format) trips from parents’ parents’ trips from questionnaire Height and weight Height and weight From semi-struc - From Walking and cyclingWalking Active behaviour Active period/ tool/recall - + par + mobility for each zip code for actual from crime statistic (demographic and (demographic household ents mobility licences habits (understand IM motivation and licences) questionnaire walking (children’s and cycling and safety) Perceived to compared of safety) Measured CrimeMeasured Risk Index from (CRI) Parents questionnaire questionnaire Parents Child interview Parents & children & children Parents questionnaire parents’ (perception Children Studies’ measures Studies’ Safety Questionnaire by) (perceived (actual)measured United States United study Longitudinal Austria Cross-section Australia cross-sectional Country study design = 919 (continued) = 2108/50.5%F/ 190 (49%F) from 190 (49%F) from 291 (150 M/141F)/ lected in 2012 with follow years three up two public schools/ aged 5–6 and n (424 M/495F) aged 19 10–12 from primary schools = = [ 29 ]/ Baseline col - T1 n aged 5–11 years [ 15 ]/- n aged 6–9(10) years old aged 6–9(10) years [ 47 ]/-/ n Study characteristics Study Citation (by (by Citation alphabetical order)/ of data year collection/ n: sex/ Sample Size gender (M/F) (or mean Age/grade age) 20 19 21 1 Table # Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 12 of 24 cal analysis cal analysis Spatial analysis/statisti - Spatial analysis/statisti - Statistical analysis Analysis method Analysis Spatial/ statistical/ spatial analysis statistical - - ethnicity (White) income and family walkabilseason, profle, - ity index (using streets connectivity from the length of roads, intersection density, block length, average connected node ratio), destina - to proximity tions (walk score, school, distance to population density) and pedestrian from safety trafc neighbourhood (Low, neighbourhood (Low, medium, high), mater IM index nal Education, parents derived from and child question - naires school- tion of safety, specifc walkability digitise (high/low), pedestrian network ownership), weather, weather, ownership), connectivitystreet Sex/gender, age, race/ age, Sex/gender, education, family parental Sex/gender, age, Sex/gender, of the school SES level and Child percep Parents Sex/gender, age, Sex/gender, car Household (income, Examined variables Individual Family Neighbourhood network bufer network bufer distance around participants home defne a neigh - to bourhood m of child’s 1600 m of child’s home Home-school Crime in 1 km road within 800 and Participants were of were Participants 1 kmWithin distance Neighbourhood per day of active per day transportation using children using children questionnaires Average minutes minutes Average Activity space IM index computed Travel mode to school mode to Travel The outcome outcome The Active measured Behaviour/ days Pedometer/7 survey (at school) GPS/7 days steps count using steps Derived from children children Derived from Active behaviour Active period/ tool/recall person & Property measures & children question - & children naire self-administered questions safety perception safety Measured crimeMeasured Report a Against 24 months before For Parents questionnaire questionnaire Parents completed Parents Parents survey for Parents Studies’ measures Studies’ Safety Questionnaire by) (perceived (actual)measured Play Study) Play project Canada from Longitudinal (Active Data from Australia cross-section (TREK)Data from The NetherlandThe Cross-section Country study design M/182F)/ (continued) 926 (463 M,463F) 660 (315 M/341F) 2015 and Dec.2016/ (mean age 11.5) 2007 included in the analysis/ 12 (mean age 9.5) = 387(185 = = [ 50 ]/between Jan. n aged 10–13 years [ 49 ]/July – December n aged 10–12 years ]/ Fall of 2018/ [ 48 ]/ Fall n and their parents 5–8 of age 7 – grade Study characteristics Study Citation (by (by Citation alphabetical order)/ of data year collection/ n: sex/ Sample Size gender (M/F) (or mean Age/grade age) 24 23 22 1 Table # Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 13 of 24 analysis Spatial analysis/ spatial statistical Analysis method Analysis Spatial/ statistical/ spatial analysis statistical Non-Hispanic, White) ity index, trafc danger, Neighbourhood-level walkability school) Race/ethnicity (Hispanic, Poverty walkabilDerived—street - walkers (to *Potential Examined variables Individual Family Neighbourhood areas School attendance School attendance Neighbourhood - ability (identify level walkers)potential Neighbourhood Walk The outcome outcome The Active measured Behaviour/ - Questionnaire for Older Children, NIK Neighbourhood Impact Older Children, Activity for CRI Crime on Kids, Risk Questionnaire crime Index using Physical (measured bourhood walk from ability Level potential (estimate pedestrianwalkers, residential facilities, land use density, con - mix, street nectivity) GIS derived Neigh - GIS derived Active behaviour Active period/ tool/recall - Childhood Prevention Policy Education project, IMD Index Education Policy Prevention Texas survey T-COPPE KIC Kinds in the City, Interview, Telephone Computer-aided Environmental Environmental SALTA of neighbourhood design, perceptions assess parental used to Youth Scale for Walkability Neighbourhood Environment Trafc + Trafc dan - ger (crash rate)) to to rate)) ger (crash Neighbour indicate Level hood Safety rate (8 major crimesrate index against the person Measured CrimeMeasured Use geocoded Crime Studies’ measures Studies’ Safety Questionnaire by) (perceived (actual)measured United States United Cross-section Country study design Study measures and analysis accounted for day type (weekend/weekdays and outside school i.e. before and after school hours) before and outside school i.e. type day (weekend/weekdays for accounted and analysis measures Study = Neighbourhood Environment Walkability Scale, NEWS-Y Scale, Walkability Neighbourhood Environment (continued) elementary school/- [ 51 ]/- 73 from Children Study characteristics Study Citation (by (by Citation alphabetical order)/ of data year collection/ n: sex/ Sample Size gender (M/F) (or mean Age/grade age) 25 International Physical Activity Questionnaire, Activity Questionnaire, International Physical questionnaire IPAQ diary systems, information of travel survey non-home activity GIS geographic format for patterns, KONTIV Transport, and Active Support Leisure for IndexWalkability SWI School sports and shopping centre, bus stop, cinema, arcade, library, of swimming pool, Area-IM destinations local shops and park or playground, school, house, of best friend’s destinations Local-IM land use mix), connectivity, street density, residential pedestrian facilities, walkers, of potential walkability an (estimate Neighbourhood-level from network connectivity derived index from volume, derived and trafc and Kids Environment Travel TREK in a year), danger and the crime rate (trafc Neighbourhood-level safety from derived with * denoted Studies Travel Environment and Kids Project, CATI and Kids Project, Environment Travel actual TREK crime statistics), service), to access PEACH housing and geographical skills education and training, health and disability, employment, (income, of deprivation categories based on seven score of Multiple is a composite Deprivation TV reduction intervention and Activity En Vivo Kids, Among of Environment Activity and Nutrition, UH-PEAK Urban Hispanic School Physical Perceptions Health, SPAN Child’s With Association and Environmental Personal NEWS study, 1 Table # MVPA SES Socioeconomic status, Neighbourhoods, CLAN Living in Active NAS Neighbourhood Activity Children Space, System, BMI Body not reported, Mass (-) data Index, GPS Global Positioning F Female, M males, System, GPS Global Positioning Transport, and Active Built Environment BEAT People, Young in Determinants activity Sport, Environmental and Easting Behaviour Physical SPEEDY Activity, Physical vigorous Medium-to PAQ-C kids KITC in the City, and Activity Monitoring, Exposure STEAM Spatio-Temporal Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 14 of 24

study by Vonderwalde also used GPS to identify children’s health research. Te geography of environ- active transportation trips made by children [50]. mental exposures must be captured to examine the asso- • Walkability indices: Objectively derived as the ciation between variables within that specifc area unit index of neighbourhood-level walkability estimates (called a neighbourhood) to explore the efect of expo- for potential walkers by combining land use mix, sures in the analysis. However, a concern that has been street network connectivity, residential density and articulated in earlier research [54, 55] was refected pedestrian facilities [51]. across reviewed studies in the diversity of measurement • Health indices: Body mass index “the ratio of an indi- approaches and outcome units. vidual’s weight in kilograms divided by the height in 2 meters squared (BMI = kg/m )” to measure of body The outcomes fat percentage and obesity [53]. In this review, one Te measurement method adapted from earlier work study measured BMI [29] and another used parents’ [55], has been expanded to delineate how studies’ meas- reported children’s weight and height [39]. urement techniques fall into one of fve recognised types of measures. Te output categories show the geographic In the remaining 13 studies, COAMB was assessed context of the children assessed or measured by studies using a variety of subjective measurement methods: to derive safety exposures (Additional fle 2: Figure 2). Under the most conventional approach within the scope • Walkability indices: Subjectively derived walking fre- of the studies (15/25 studies; 60%), the child’s neigh- quency from parent questionnaires [32]. bourhood was most often defned as the local area, the • Home-school active travel: Studies used parent ques- area within a 10-min walk, the street immediately near a tionnaires and/or travel diaries to assess the percep- child’s home, the walk between school and home, or the tion of walking to school [30, 31], frequency of walk- area of the school [15, 30, 32, 33, 35, 36, 38, 39, 42–44, ing and cycling to school [47], frequency of walking 46, 47, 51]. Two studies (2/25; 8%) defned the neighbour- to school [42] and satisfaction with active travel to hood as the administrative boundary of the census block school or mode of travel to school [48]. or zip code, which falls under residence-based approach • Type of outdoor active behaviour: One study used [29, 45]. Both the above categories are static and subjec- the frequency of outdoor play, exercise, sport and tive representations. Te activity bufer-based neighbour- active commuting [43]. Active independent mobility hood (3/25 studies; 12%) is a GIS-based unit of analysis (AIM) was primarily derived by asking children for a obtained via bufers around a point feature (represent- list of destinations they are allowed to go to on their ing the home or school of each participant). However, own or with a friend without an adult [15, 34, 35, 38, this unit of analysis is depicted by diferent bufer radii 46]. However, one study by Page et al. distinguished (using visual inspection of the GPS waypoint and apply- between two types of IM, namely Local IM and Area ing equal weight to the delineated bufer area) of 800– IM [43]. 1000 m around the school [41]; or 400, 800, 1200 and 1600 [37], and 500 and 1000 m around the child’s home Additionally, a range of various technical data captur- [28]. Te activity space-based measure (1/25 studies; 4%) ing specifcations were depicted in the measures. For delineates the geometry of a child’s activity space and dif- example, epochs, which is the interval of time for cap- fers from the earlier group in that it is an irregular con- turing rapid transitions of an activity, varied among vex polygon composed of joining points (a minimum of and within the tools used. Using global positioning sys- three) representing the destinations visited by children tems, studies used one-second [52], fve-second [28] [49]. A pre-defned bufer radius (using a distance based and 10-s epochs resampled to 30-s [45], and 15-s [40] on earlier research fndings such as 1 km around the par- epochs, while other manuscripts that used an accelerom- ticipant’s home) can also depict the bufer-based neigh- eter adopted 30-s epochs [41, 45]. Additionally, reviewed bourhood (4/25 studies; 16%) [16, 34, 40, 42]. studies used a range of varied inclusion criteria (i.e. the minimum hours or days of measures to include for analy- Measuring safety exposures sis; Table 2). Examined safety exposures in the neighbourhood refected personal safety—either perceived (19/25 Measuring the area of exposures to assess Impact studies; 76%) or actual (3/25 studies; 12%)—or jointly on COAMB addressed perceived and measured safety (3/25 studies; Delineating the shared spaces that ofer opportunities 12%). Road safety concerns were addressed in 15 out of for the majority of daily routine activities is essential in 25 studies (60%), which in all but two studies were per- ceived rather than objective measures. Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 15 of 24

Fig. 2 Means of measures and the outcome indices of children’s active Mobility behaviour. AIM active independent mobility, BMI body = = mass index, FWAC frequent walking and cycling, MVPA moderate-to-vigorous physical activity, NAS neighbourhood activity space, = = = PWTS perceived walking to school, GPS global positioning system, GIS geographic information system = = =

The outcome or careful drivers who pay attention to pedestrians and Perceived personal safety was captured via question- cyclists [42, 48], heavy trafc [32, 40, 46–48], safe places naires mailed to parents, administered through a com- to cross the road [42, 44, 46, 49], road pollution [43] or puter-assisted telephone interview (parents or children) trafc danger [35]. or conducted on school premises (children). A pre- Measured personal safety was examined by using crime designed set of questions that assess parents’ perceptions types (against persons and/or property) in two stud- of environmental characteristics used: the Neighbour- ies. However, these were separated in the analysis in hood Walkability Index for Youth (NEWS-Y) instrument one study [50] but combined in another [45]. Remaining [39], the NEWS [46], Ranui Actions Survey [41], used a studies used indices of; a crime risk index (computing the survey that was adapted from several earlier surveys [42], likelihood of a crime occurring in a neighbourhood based adapted questionnaires from a previous study [32], while on actual crime statistics) [29], a neighbourhood safety the rest of reviewed papers have used questionnaires level index (combining crime data with trafc volumes, to ft the individuality of their research objectives. Te the percentage of high-speed streets and accident rates) respondents were asked about their own fear of being a [51]. Objective measurements of road safety employed victim of a crime in their neighbourhood [44], or to rate GIS to measure trafc speeds around schools, and they their agreement with statements about perceived poten- adopted road hierarchy as a proxy to derive the ratio of tial crime/safety [45] and perceived neighbourhood risk high-speed roads around schools [41] in combination [37]. Studies aimed at capturing general feelings of safety with streetscape audit (via Google Street View). Another varied in their use of open-ended questions [44], close- study generated indices of trafc volume, trafc speed ended 5-point Likert scales ranging from very unsafe and trafc calming as well as pedestrian infrastructure to very safe [30], or 5-point Likert scales ranging from [40]. strongly agree to strongly disagree [41]. Of the 7/25 stud- Controlled confounders of the same questionnaires in ies (28%) that assessed children’s perceptions, one exam- terms of individual characteristics were mainly child sex/ ined how children perceive their parents’ view of safety gender (20/25 studies; 85%), race/ethnicity (9/25 stud- and its efect on encouraging or discouraging active ies; 36%) and family characteristics as a proxy for the mobility [47]. socio-economic status (18/25 studies; 72%). However, Studies that assessed road safety perception (13/25 ethnic/racial categories were inconsistently classifed studies 52%) captured parents’ perceived road safety in across studies, although some studies originated from the 10/13 studies (78%) and children’s perceptions in 5/13 same country [29, 40, 44, 45, 51], Table 1. Indices used to studies (38%). Studies used various indices as a proxy for address socio-economic status varied across studies, with road safety, that can be grouped under the perception car ownership followed by family income being the most of signals on a busy road [44], the availability of a side- discussed. walk [40, 42, 44], driving style in terms of fast drivers [36] Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 16 of 24 days of 10 h 4 days < 10 h GPS data and participants with < 3,581 counts per minute > h of accelerometer or GPS wear time or less than 15-s or GPS wear 10 h of accelerometer < 7 h of data (weekend) MVPA analysis, GPS data (adopted from accelerometer standard measures) standard accelerometer from GPS data (adopted weekend day epochs days of accepted measures of accepted counts and at least 3 days zero for any day if fewer than 10,000 and over 20 million steps counted to to counted 20 million steps than 10,000 and over if fewer day any for MVPA calculate h or more of data (weekdays) and of data (weekdays) 3 h or more outside school hours, Accelerometer GPS inclusion of data was not reported Data with 1,000–30,000 steps daily included Data with 1,000–30,000 steps of data least 4 days At the analysis from removed of measures and last day First h wearing time for 3 weekdays and 1 weekend day for inclusion in the for and 1 weekend 3 weekdays day time for 10 h wearing Data excluded with Data excluded No report of details 3 h of out-of-school time GPS (two wearing and 4 h of one weekdays) inclusion for accelerometer Guided by Remove min of consecutive than 20 min of consecutive inclusion with no more time for 10 h wear Inclusion/exclusion data GPS COAMB of collected 4 weekdays and at least 1 weekend for inclusion; data were excluded excluded inclusion; data were and at least 1 weekend4 weekdays for - GPS 10-s epochs (cleaned and resampled to to and resampled GPS 10-s epochs (cleaned + duration + duration in survey travel + travel diary mobility diary 7 days for accelerometer for 7 days for accelerometer + + 30-s school hours 8:00–8:45 am, after school 2:30–7:00 pm and week ends) school [6 am] to school frst bell) and evening (6–9 pm) school frst bell) and evening school [6 am] to Compared outside school hours for weekdays and weekends (before and weekends weekdays (before outside school hours for Compared Accelerometer for 7 days for Accelerometer Every 749/30-s epochs), Pedometer for 9 days for Pedometer and mapping destination Epoch not reported – Accelerometer for 7 days for Accelerometer 5-secs epochs and weekends weekdays outside school hours for Compared GPS in a wearable watch for 7 days for watch GPS in a wearable – 5-s epochs – GPS GPS for 7 days GPS for each second Recorded and weekends weekdays outside school hours for Compared 15-secsepochs GPS Accelerometer for 7 days for Accelerometer 30-s epochs – Data registryData (epoch) or time weekends vs weekdays (i.e. (if present) measures Temporal school) before and after of the day segments Tools used Tools Accelerometer for 8 consecutive days 8 consecutive for Accelerometer Epoch not reported during non-school hours and weekends weekdays (before outside school hours for Compared [ 41 ] [ 49 ] [ 36 ] [ 16 ] [ 28 ] [ 52 ] [ 40 ] [ 45 ] Citation [ 33 ] Summary of in variety by organised characteristics: objective measures the a citation, output units, brief of the and tools inclusion and exclusions 2 Activity space Active transportationActive Proximity Neighbourhood activity space Active outdoor play (daily average) (daily play outdoor Active Table criteria COAMB output unit of measured The Intensity MVPA Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 17 of 24

Methodological considerations in current measures behaviour. Daily travel diaries collected a child’s daily Overall, the review in fndings revealed the existence destinations visited and the mode of transportation. of signifcant heterogeneity across studies in terms of Both techniques rely on respondents to recall a particu- what we are measuring (active behaviour, exposures and lar event from a specifc time, introducing human recall cofounders) and method of measure (data). Tus, to pro- error or missing data that may impact analysis of the pre- pose alternatives to guide future research, we delve into dictors of PA [19]. For example, a review by Kelly et al. current methodological measurements issues. We illus- 2013, on studies comparing self-reported and GPS-meas- trate schematically current methodological practices in ured journey duration, concluded that participants were layers based on the merit of two components: the deter- consistently over-reporting the duration of the journey. minants (in the dotted line); the methods of measure Tis suggests that when studies use self-reported journey (in dashed line). Accumulated inconsistencies in what behaviour, the journey duration should be treated as an we measure and the uncertainty in how we measure overestimate [59] that agree with the fndings of Klous increases the error of correlates and positional behaviour. et al. [60], 2017 for the rural population. Te comparison Ignorance of the infuences of reliability and the accuracy between questionnaires to accelerometer by Määttä et al. of data (levels 1–5, Fig. 3) poses a serious limitation as it [61], in specifc to children, has as well suggested that may result in the misevaluation or misplacement of expo- accelerometer is a better measure than questionnaires sures and bias in evidence for policymakers. In general, in when addressing PA duration among 11 years old chil- the majority of reviewed studies, a cross-sectional design dren out of school PA. was used, meaning the studies represented a short period Another signifcance observation found in most stud- of data collection (registry) that was not a real represen- ies reviewed is that they lacked understanding of the tation of the spatiotemporal behaviour of child mobility temporality characteristics in the exposures as well as [43]. in the outcome of active behaviour. Such a character- istic of behaviour is confrmed in the review by Brooke Consistency of measures et al. [62], on school-aged children PA. Tus, ignorance In level 1 of Fig. 3, we found an inconsistent account- of the temporal component may lead to over- or under- ing for sex/gender and age and race/ethnicity variation estimated impact of exposures [55]. across studies. Additionally, we found that research cap- turing parents and or children’s views of safety are faced Accuracy of measures an inconsistent subgrouping of race/ethnicity or family Despite the proven advantages in the emerged tracking socio-economic indicators preventing comparison across tools such as accelerometers and spatial technology of studies. GPS ofering the location and dimension of active mobil- In the measure of safety exposures, evidence indicates ity, studies often recognised some drawbacks that are variations between parents’ and children’s perceptions [7, depicted in level 3, Fig. 3: 56, 57]. Despite that, the majority of studies relied on one side view of perceived safety (the parents). Tus, suggest- Inclusion criteria Majority of studies seem to operate ing that an understanding of both parents’ and children’s within the timeframe of 7 to 9 consecutive days of a sur- perceptions of fear of safety may be incomplete. Addi- vey. However, varied inclusion criteria were adopted in tionally, perceived safety, as measured in self-reported each study, Table 2. Research using GPS was faced with questionnaires, are static mostly and lack more specifc the absence of a standard operating protocol for GPS understating such as the intensity and frequency of feel- device usage and thus adapted previous practices and ing unsafe[13], or the geographic linkages to fear. In the established protocols for the accelerometer. Te nature actual personal safety measures, variations in means of of the GPS data difers from accelerometers, and further using real crime data were depicted across studies. Ear- guidelines on GPS protocol are warranted. lier research agrees that diferent police-reported crime types produce diferent efects on various travel options Registration Variations in the epochs (frequency in sec- [45, 58]. Additionally, except two studies [40, 45], an onds of capturing location) used in data registry across understanding of the variation between perceived and studies, as shown in Table 2, refect the individuality of actual safety is lacking. measures. In addition, the increased use of GPS to obtain high-resolution spatiotemporal data was faced with sig- Reliability of measures nals that are weakened or cut of when indoors [37], in In level 2 of Fig. 3, included studies used questionnaires urban settings, inside buildings, near trees and on cloudy for either parents or their children to recall general days, thus resulting in errors as well as the battery life lim- feelings for safety perception as well as the child active itations. Although a new generation of GPS technology Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 18 of 24

Fig. 3 Schematic representation summarising the existing measurement methods in the studies included in the review Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 19 of 24

may facilitate the indoor GPS option, future research that fundamental to research aimed at planning health inter- intends to use such tools must give careful attention to vention strategies. However, our review showed that study area structure to involve other supporting methods. researchers in this feld currently lack agreement on not only what to measure but also primarily how to measure. Completeness Missing data in studies that used GPS or In the past, several conceptual frameworks were devel- accelerometer devices were generally due to recruited oped to guide evidence-based research, yet the repre- study participants not wearing the device (forgetting to sentation of infuencing factors was either stationary and wear, not charged or turned of) during some parts of the specifc to one type of active behaviour (e.g. school active day or had incomplete travel diary data [28]. Missing data travel [7, 68]) or addressed a broad range of children’s means the 7 days measures include a signifcantly shorter ages (5–18 years old) [8]. Te time element introduced period of data in the analysis and that visited destinations by Pont et al. in the simplifed framework [69] lacked or trips are missed [63]. spatial dimensions. Additionally, exposures that modify children’s active behaviour may very well be too com- Confdentiality Tis was not explicitly reported in plex to be depicted in one single framework [70]. Tus, a reviewed papers, possibly due to the fact that the major- framework does not exist that expands beyond the iden- ity of studies that have used GPS to track children’s active tifcations of variables to agree on the representation of mobility were being undertaken in developing countries. exposures that modify primary school-aged children’s Nevertheless, existing evidence suggests that the per- active mobility behaviour in this context. Based on the ceived acceptability of such data collection method was four identifed levels of concern in Fig. 3, it is now crucial lower in older low socio-economic population [65], as well to conceptualise alternative measurement methods and as in specifc ethnicity though more acceptance within the technology to guide future studies. younger population [64] or in developing countries [66] To guide future research, we proposed a triadic con- where health technologies become less acceptable due to ceptual framework (see Fig. 4). In the framework, we perceived privacy threats. distinguish three pillars (represented in yellow) that were derived from the theoretical representation of objects vis- Uncertainty in representing the area of exposures ualised by Peuquet, which is advantageous when examin- (neighbourhood) ing human mobility in research [71]. We portrayed frst Te current review portrayed the persistence of an the interplay of ‘what’ (determinants), which is arranged important issue when using geographical data which is according to the socio-ecological domains of individual, the articulated uncertain geographic context problem family and neighbourhood infuences in the context of (UGCoP) [67]. As level 4 in Fig. 3 illustrates a variation safety. ‘Where’ addresses the spatial variable or dimen- in measurement methods and outcome across reviewed sion of behaviour infuenced by safety. Finally, temporal studies was found. Defning the geometry of the child characteristics (e.g. exposures and COAMB) answer the neighbourhood to assess infuences of environmental question of ‘when’ (i.e. the time of occurrences, such as exposures ranged from subjective, arbitrary representa- weekdays vs. weekends, and the time segments of the day, tions to an objective delineation. Bufers (circular) are a frequency and regularity). better representation of an individual’s mobility in space than subjective measures, yet they remain static in time. Moving forward Additionally, circular proxies do not generally coincide Addressing public health concerns in the feld of children with the area that children access. Te minimum convex and youth active mobility is expanding rapidly in terms of polygon, which measures activity space, is one of the pre- research conduct. However, despite technology advances, vailing spatial methods used to represent the geographi- the challenge will likely, in the absence of standardisation, cal context in spatial epidemiology [54]. A recent study remain primarily in the data measures that may have an by Zhao et al. [54] argued the infuence of the various implication on the evidence. Results of the methodologi- methods of measurement in terms of multiple bufer radii cal quality appraisal showed that the majority of studies or activity spaces (e.g., road network bufer, minimum (88%) fall under robust or fair quality. Nevertheless, the convex polygon, weighted standard deviation eclipse) on heterogeneity in applied measurements and outcome analytical results related to health outcomes. extends across all studies. Tus, it is maybe crucial to draw upon the conceptual framework (Fig. 4) to achieve A methodological conceptual framework to guide future an outlook on alternative measurement methods that research may reduce constraints portrayed in Fig. 3. Relevant data that we use in studies to assess the impact on children’s active health behaviour and outcomes are Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 20 of 24

Study design Another important aspect is a consensus in future Future studies should aim to consistently account for studies for representing the spatial and temporal ele- gender, age and race/ethnicity diferences in their meas- ments of exposures (safety in this context), as well as ures as well as in analyses. A review of the evidence the frequency and intensity of feelings. Tis is funda- revealed that those variables have consistently predicted mental and may help researchers to answer a wider COAMB. In addition, a more comparable understanding array of substantive research questions. Studies that of the efects of ethnicity and socio-economic status can intend to use questionnaires to capture safety percep- be gained with standardisation of indices used and sub- tions may beneft from the location-based approach grouping. At the neighbourhood level, more research on [76], which is likely to ofer more comprehensive safety understanding safety from children’s perspectives is war- strategies. To overcome the human errors of recalling ranted [56]. feeling safe, web-based applications have become more Additionally, understanding the population’s fears popular and can be installed on mobile devices for peri- concerning types of crime may ofer a more precise odic access and instant recall of perceptions. For exam- picture of the state of children’ safety [72]. Tus, future ple, the use of a fear of crime mobile phone application studies would beneft from examining the impact of that ofers instant recording of people fear as welll as actual safety measures to consistently distinguish geolocating partcipants perception of concern [73, 74]. between types of crimes and to assess the most relevant In measuring COAMB, the frst triangle (Fig. 4) sug- crimes (e.g. crimes against persons and crimes against gests that future studies consider the spatial (e.g. geo- properties) to modify COAMB for appropriate inter- graphic location) and temporal (e.g. weekends and ventions. Furthermore, despite fndings in the literature weekdays) elements in both reported and objective confrming parents’ infuences when perceiving low measures of COAMB. For studies that will be conducted safety barriers, it is crucial to emphasise the dynamic in areas where there is perceived threat of confdential- nature of parents’ infuence boundaries. Although gen- ity, potential GPS signal loss (dense urbanity) or limited der-based, older primary school-aged children’s conf- budget, makes wearable technologies unattainable [66]. dence in their neighbourhood (represented in dashed Tus, an alternative map-based questionnaire method line, Fig. 4) results in children expanding the space has proven comparable to GPS in providing activity space they negotiate, and the impact of their parents’ percep- information for exposure assessment [63]. To bypass tion (represented in solid line, Fig. 4) is seen to lessen. errors caused by human recalling activities or daily dia- Finally, exploring the relationship between active behav- ries, future studies may consider taking advantage of the iour and health indices, such as BMI, in the safety con- emerging web-based applications that can be installed on text is currently lacking. Tus, consistent accounting mobile devices or accessed periodically for instant recall for BMI can help uncover this relationship in future of activities, such as the Finnish SoftGIS [77]. Other research. examples include the Public Participation GIS used for elderly active travel [78] or the Ecological Momentary Methodology of measure Assessment mobile application used for health behaviour In the increased interest in evidence-based research assessment [79]. about children’s active health to provide an appropri- In objective measures of COAMB at the foundation ate baseline for targeting interventions, we suggest that level, activity tracking and spatial technology ofer a future studies move away from subjective measures range of tools to collect, visualise and analyse the spatial towards alternatives that ofer a better output in research and temporal outcomes of mobility behaviour. Moving aiming for children health equity. forward, we suggest researchers take advantage of these Regarding perceived safety measaures, we suggest tools and ensure that the temporality of active behaviour that future studies replace the subjective and static (in the survey and the analysis) over several days and at understanding of safety perceptions to obtain more diferent times of the day is considered. However, stand- measurable feelings. Te emergence of redesigned ardisation in the protocols of measures using tools such questionnaires about fear of crime in the last decade as GPS and accelerometers may eliminate some of the may help studies to (i) include specifc geographical studies’ mixed fndings when using objective, rather than and temporal references for participants’ feelings [73, reported, measures of neighbourhood exposures and 74], (ii) capture the frequency and intensity of fear [13], active behaviour. and (iii) assign a timeframe for asking about feelings of Regarding the measure of children’s neighbourhoods, worry (e.g. in the last year or the previous month) and researchers are recommended to avoid assessing area perceived risk (e.g. in the next year, how likely…) [75]. of contextual exposures arbitrarily or measuring it in static space and time where possible. At the same time, Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 21 of 24

Fig. 4 A methodological conceptual framework of children’s outdoor active mobility behaviour in a safety context. Exposures (on the left) are organised as per the socio-ecological model and by the output of measures (as we move right) to minimise the efect of the existing UGCoP (level 4, has provided powerful insights that can help researchers Fig. 3) in this conceptual framework, we suggest alter- to expand the trajectory of this feld. For example, Pap- natives (see the second and third triangle in Fig. 4). Te palardo and Simini recently presented a framework for growing body of evidence confrmes the variation of an algorithm to reproduce real human spatiotemporal safety impact on children’s activity space(s) according patterns from mobile data [82]. Furthermore, expanding to gender and age [49], in the distance covered [28], and the analytical capabilities of space–time (space and time trips made [15, 33, 47, 52, 80]. Tus, we argue the need triangle, Fig. 4) when capturing and analysing data may in future studies for innovative geo-approach and tool uncover hidden spatiotemporal patterns of safety (actual to enhance the accuracy when measuring environmen- or perceived) and children’s active behaviour in a space– tal exposures’ efect. A socio-geo-neighbourhood, for time approach. For example, the space–time budget was example, that accounts for children’s gender and age and used to understand the interaction of the social environ- predict their likelihood of mobility within a space may be ment with people’s crime propensity [83]. Tus, integra- deemed fundamental. Older children require a diferent tion of GPS, GIS with space–time may aid researchers type of environment to be active than younger children; to explore more challenging areas such as to predict the thus, the area and nature of active space is dynamic and spatial heterogeneity in children’s populations in environ- fuctuates by gender, age and perceived safety barriers. In mental context and may help policymakers to better plan considering this direction, it is critical to move away from the safety and mobility needs of active children. circular bufers as they do not coincide with the human mobility pattern. Integration of road network with chil- Strengths and limitations dren socio-geo background could derive a more reliable Tis review focused on specifc topics and omitted some area when deriving exposures. Recent advances in deep closely related issues that are worthy of further investi- learning [81] proven advantageous in detecting roads gation. For example, studies that account for policies, network using high-resolution satellite images. and seasonal/weather variation were not considered. In the era of big data, derived human activities are Similarly, analysis of correlates, such as spatial or sta- embedded in the social aspects of our daily lives. Tis tistical methods used (level 5 in Fig. 3), was beyond the Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 22 of 24

scope of this review, but correlates are likely to lead to Until this problem is resolved, signifcant evidence may diferent inferences due to diferences in the underly- be buried by these measurement and analysis methods. ing assumptions of analysis techniques. Assessment of methods to (a) clean the GPS data collected on COAMB Supplementary Information (e.g. PALMS, Google Fusion Table software [40, 41, The online version contains supplementary material available at https​://doi. org/10.1186/s1294​2-020-00254​-w. 50]) or (b) audit features of the neighbourhood, such as streetscape (e.g. NZSPACE [41]), were not included in the remit of this review. In addition, this review used Additional fle 1. PRISMA checklist. comprehensive combinations of keywords to capture all Additional fle 2. Details of included and excluded criteria. Details of used search key terms. The methodological quality assessment criteria used to studies, but some papers may have been omitted due to appraise each study, including table 1 showing a list of criteria that each the inconsistent terminology used among studies; we study was assessed against. Table 2 shows details of each total study score, tried to mitigate this risk by examining bibliographies, the percentage accumulated, and the assigned level of quality (Robust, fair and weak). but the risk of omission remained. Additionally, reports Additional fle 3. Excel sheet with the total number of studies that were and thesis studies were excluded. Despite these limita- excluded and the reason for exclusion. tions, this study is, to the best of our knowledge, the frst to comprehensively synthesise the methods of measures Abbreviations of primary school-aged children’s COAMB in a neigh- COAMB: Children’s outdoor active mobility behaviour; SAT: School active bourhood safety context. Te paper is pioneer in con- transport; GIS: Geographic information system; GPS: Global positioning structing schematic representation that highlight layers system; MVPA: Medium-to-vigorous physical activity; AS: Activity space; NAS: Neighbourhood activity space; FWAC​: Frequent walking and cycling; AIM: in current methodlogical practices and to put forward a Active independent mobility; BMI: Body mass index; NEWS-Y: Neighbourhood three-dimensional framwork for future measures. Tis walkability index for youth; PALMS: Personal Activity and location measure- review is also the frst to assess the quality of the included ment system; PPGIS: Public participation GIS; EMA: The Ecological momentary assessment; CATI: Computer-aided telephone interview. studies methodologically, to synthesise crucial existing methodological gaps and to outline a framework to guide Acknowledgements future research. Another strength included studies cov- The authors thank Bev Jebson (BJ) for her assistance in the rating of the meth- odological quality. Additionally, the authors would like to thank the thoughtful ered a wide range of countries origin. comments of the three anonymous reviewers of the Journal of Health Geo- graphics that have helped to improve the review considerably.

Conclusion Authors’ contributions RZ carried screening of the literature for inclusion, the rating of the methodo- Tis review is the frst to comprehensively and system- logical quality, conducted the review, synthesised the fndings, drafted, and atically synthesis measurement methodologies of safety fnalised the manuscript. CX screened abstracts for inclusion independently and contributed together with RN and OG and AD to the concept of the exposures that impact COAMB. After reviewing 25 stud- paper. All authors read and approved the fnal manuscript. ies, we identifed mixed methodological designs and an absence of standardisation in measures that may have Funding The research received no specifc grant from any funding agency in public, led to the current diversity in studies’ outcomes. Such commercial or not-for-proft sectors. disparity in research outputs may be reducing the signif- cance of synthesised evidence. Te methodological qual- Availability of data and materials All data generated for synthesising methods during the systematic review are ity assessment that this review undertaken showed that included in the published article, and its supplementary information. most studies were of moderate or weak quality regarding their measurement methods. We also schematically out- Ethics approval and consent for publication Not applicable. lined accumulated layers of heterogeneity in the studies’ method of measures that may afect data reliability. We Consent for publication argued that our constructed three-dimensional con- Not applicable. ceptual framework is vital to guide future research aim- Competing interests ing to assess neighbouirhood exposures impact, such The authors declare that they have no competing interests. as safety, on COAMB. Moreover, we suggested poten- Author details tial alternative methodological measures, tools and solu- 1 School of Earth and Planetary Sciences, Curtin University, Kent Street, Perth, tions for studies that aim to provide children with equal WA 6102, Australia. 2 School of Public Health and Community Medicine, UNSW 3 active health opportunities. Despite advances in spatial Medicine, Sydney, NSW, Australia. School of Public Health, Curtin University, Perth, WA, Australia. technology, ignoring the uncertainty produced by the heterogeneity in current measurement practices may Received: 11 August 2020 Accepted: 17 December 2020 result in misevaluation or misplacement of exposures. Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 23 of 24

References 28. Fagerholm N, Broberg A. Mapping and characterising children’s 1. WHO. Global strategy on diet, physical activity and health. Geneva: World daily mobility in urban residential areas in Turku Finland. Fennia. Health Organization; 2020. 2011;189(2):31–46. 2. WHO. Global strategy on diet, physical activity and health: childhood 29. Suminski RR, et al. Actual neighborhood-level crime predicts body mass obesity and overweight. Geneva: World Health Organization; 2016. index z-score changes in a multi-racial/ethnic sample of children. Prevent 3. WHO, Population-based approaches to childhood obesity prevention. In: Med Rep. 2018;12:164–9. WHO library cataloguing-in-publication data; 2012. World Health Organi- 30. Mehdizadeh M, Mamdoohi A, Nordfaern T. Walking time to school, zation: WHO Document Production Services, Geneva, Switzerland children’s active school travel and their related factors. J Transp Health. 4. WHO. Report of the commission on ending childhood obesity. Geneva: 2017;6:313–26. World Health Organization: WHO Document Production Services; 2016. p. 31. Shokoohi R, Hanif NR, Dali MM. Children walking to and from school in 68. Tehran: associations with neighbourhood safety, parental concerns and 5. WHO. Report of the commission on ending childhood obesity. Imple- children’s perceptions. ProcediaSocBehavSci. 2012;38:315–23. mentation plan: executive summary. Geneva: World Health Organization; 32. Alton D, et al. Relationship between walking levels and perceptions of 2017. the local neighbourhood environment. BMJ. 2007;92(1):29–33. 6. Borghese MM, Janssen I. Development of a measurement approach to 33. Carver A, et al. Are children and adolescents less active if parents restrict assess time children participate in organized sport, active travel, outdoor their physical activity and active transport due to perceived risk? SocSci active play, and curriculum-based physical activity. BMC Public Health. Med. 2010;70(11):1799–805. 2018;18(1):396. 34. Carver A, et al. Independent mobility on the journey to school: A joint 7. McMillan TE. Urban form and a child’s trip to school: the current literature cross-sectional and prospective exploration of social and physical envi- and a framework for future research. J Plan Literature. 2005;19(4):440–56. ronmental infuences. J Transp Health. 2014;1(1):25–32. 8. Panter JR, Jones AP, van Sluijs EMF. Environmental determinants of active 35. Davis A, Jones L. Environmental constraints on health: listening to chil- travel in youth: a review and framework for future research. Int J Behav- dren’s views. Health Educ J. 1996;55(4):363–74. NutrPhys Act. 2008;5(1):34. 36. Faulkner G, et al. Children’s outdoor playtime, physical activity, and 9. Wolfe MK, Noreen C. McDonald, Association between neighborhood parental perceptions of the neighbourhood environment. Int J Play. social environment and children’s independent mobility. J Phys Act 2015;4(1):84–97. Health. 2016;13:970–9. 37. Loebach JE, Gilliland JA. Free range kids? Using GPS-derived activity 10. Jack G. Place matters: the signifcance of place attachments for children’s spaces to examine children’s neighborhood activity and mobility. Environ well-being. Br J Soc Work. 2010;40(3):755–71. Behav. 2016;48(3):421. 11. Davison K, Lawson C. Do attributes in the physical environment infuence 38. Lin E-Y, et al. Social and built-environment factors related to children’s children’s physical activity? A review of the literature. Int J BehavNutrPhys independent mobility: the importance of neighbourhood cohesion and Act. 2006;3(1):19. connectedness. Health Place. 2017;46:107–13. 12. Robinson AI, Carnesb F, Oreskovica NM. Spatial analysis of crime inci- 39. Noonan R, et al. Cross-sectional associations between high-deprivation dence and adolescent physical activity. Prev Med. 2016;85:74–7. home and neighbourhood environments, and health-related variables 13. Gray E, Jackson J, Farrall S. Reassessing the fear of crime. Eur J Criminol. among Liverpool children. BMJ Open. 2016;6:e008693. 2008;5(3):363–80. 40. Nguyen A, Borghese MM, Janssen I. Pedestrian trafc safety and outdoor 14. Brown BB, et al. Physical activity mediates the relationship between active play among 10–13 year olds living in a mid-sized city. Prevent Med perceived crime safety and obesity. Prev Med. 2014;2014(66):140–4. Rep. 2018;10:304–9. 15. Stark J, Frühwirth J, Aschauer F. Exploring independent and active mobil- 41. Oliver M, et al. Associations between the neighbourhood builteeviron- ity in primary school children in Vienna. J TranspGeogr. 2018;68:31–41. ment and out of school physical activity and active travel: an examination 16. Vonderwalde M, et al. Objectively measured crime and active transporta- from the kids in the city study. Health Place. 2015;2015(36):57–64. tion among 10–13 year olds. Prevent Med Rep. 2019;13:48–51. 42. Oluyomi AO, et al. Parental safety concerns and active school commute: 17. Lee NC, et al. Does activity space size infuence physical activity levels of correlates across multiple domains in the home-to-school journey. Int J adolescents?—a GPS study of an urban environment. Prevent Med Rep. BehavNutrPhys Act. 2014;11(1):32. 2016;3:75–8. 43. Page AS, et al. Independent mobility, perceptions of the built environ- 18. Wong BY-M, Faulkner G, Buliung R. GIS measured environmental cor- ment and children’s participation in play, active travel and structured relates of active school transport: a systematic review of 14 studies. Int J exercise and sport: the PEACH Project. Int J BehavNutrPhys Act. BehavNutrPhys Act. 2011;8(1):39. 2010;7(1):17. 19. Smith M, et al. Childre’s geographies for activity and play: an overview of 44. Roberts JD, et al. Parental perceived built environment measures and measurements approach. Res Gate. 2015;9:1–20. active play in Washington DC metropolitan children. Prevent Med Rep. 20. Bates B, Stone MR. Measures of outdoor play and independent mobil- 2016;3:373–8. ity in children and youth: a methodological review. J Sci Med Sport. 45. Kneeshaw-Price SH, et al. Neighborhood crime-related safety and its rela- 2015;18(5):545–52. tion to children’s physical activity. J Urban Health. 2015;92(3):472–89. 21. Jankowska MM, Schipperijn J, Kerr J. A framework for using GPS data 46. Santos MP, et al. Parental physical activity, safety perceptions and chil- in physical activity and sedentary behavior studies. Exerc Sport Sci Rev. dren’s independent mobility. BMC Public Health. 2013;13:584–584. 2015;43(1):48–56. 47. Timperio A, et al. Perceptions about the local neighborhood and walking 22. ABS. Australian Health Survey: Physical Activity, 2011–12 Australian and cycling among children. Prev Med. 2004;38(1):39–47. Bureau of Statistics, Editor. 2013, Australian Bureau of Statistics: Canberra. 48. van den Berg P, et al. Factors afecting parental safety perception, satis- 23. Leung KYK, Loo BPY. Association of children’s mobility and wellbeing: a faction with school travel and mood in primary school children in the case study in Hong Kong. Travel BehavSoc. 2017;9:95–104. Netherlands. J Transp Health. 2020;16:100837. 24. Moher D, et al. Preferred reporting items for systematic reviews and 49. Villanueva K, et al. How far do children travel from their homes? Explor- meta-analyses: the PRISMA statement. BMJ. 2009;339:b2535. ing children’s activity spaces in their neighborhood. Health Place. 25. Schoeppe S, et al. Associations of children’s independent mobility and 2012;18(2):263–73. active travel with physical activity, sedentary behaviour and weight 50. Vonderwalde M, et al. Objectively measured crime and active transporta- status: a systematic review. J Sci Med Sport. 2013;16(4):312–9. tion among 10–13year olds. Prevent Med Rep. 2019;13:48–51. 26. Marzi I, Reimers A. Children’s independent mobility: current knowledge. 51. Zhu X, Lee C. Walkability and safety around elementary schools eco- Fut Direct Public Health Implicat. 2018;2018:2441. nomic and ethnic disparities. Am J Prevent Med. 2008;34(4):282–90. 27. Lubans DR, et al. The relationship between active travel to school and 52. Loebach JE, Gilliland JA. Free Range Kids? Using GPS-derived activity health-related ftness in children and adolescents: a systematic review. Int spaces to examine children’s neighborhood activity and mobility. Environ J BehavNutrPhys Act. 2011;8(5):1–12. Behav. 2016;48(3):421–53. 53. King TC. 3—tissue homeostasis, damage, and repair. In: King TC, editor. Elsevier’s integrated pathology. Philadelphia: Mosby; 2007. p. 59–88. Zougheibe et al. Int J Health Geogr (2021) 20:2 Page 24 of 24

54. Zhao P, Kwan M-P, Zhou S. The uncertain geographic context problem in 71. Peuquet DJ. It’s about time: a conceptual framework for the representa- the analysis of the relationships between obesity and the built environ- tion of temporal dynamics in geographic information systems. Ann Assoc ment in Guangzhou. Int J Environ Res Public Health. 2018;15(308):1–20. Am Geogr. 1994;84(3):441–61. 55. Yi L, et al. Methodologies for assessing contextual exposure to the built 72. Lorenc T, et al. Crime, fear of crime and mental health: synthesis of theory environment in physical activity studies: a systematic review. Health & and systematic reviews of interventions and qualitative evidence, vol. 2. Place. 2019;60:102226. Liverpool: Public Health Research; 2014. 56. Nayak A. ‘Through children’s eyes’: childhood, place and the fear of crime. 73. Solymosi R, Bowers K, Fujiyama T. Mapping fear of crime as a context- Geoforum. 2003;34(3):303–15. dependent everyday experience that varies in space and time. Legal 57. Carver A, Timperio A, Crawford D. Playing it safe: the infuence of neigh- CriminolPsychol. 2015;20(2):193. bourhood safety on children’s physical activity—a review. Health Place. 74. Solymosi R. Exploring spatial and temporal variation in perception of 2008;14(2):217–27. crime and place using crowdsourced data. In: Department of Security 58. Appleyard BS, Ferrell CE. The Infuence of crime on active & sustainable and Crime Science 2017, University College London: University College travel: new geo-statistical methods and theories for understanding crime London. and mode choice. J Transp Health. 2017;6:516–29. 75. Jackson J, Farrall S, Gray E. Feelings and functions in the fear of crime: 59. Kelly P, et al. Quantifying the diference between self-reported and global applying a new approach to victimisation insecurity. Br J Criminol. positioning systems-measured journey durations: a systematic review. 2010;51(1):75–94. Transp Rev. 2013;33(4):443–59. 76. Kyttä M, et al. Perceived safety of the retroft neighborhood: a location- 60. Klous G, et al. Mobility assessment of a rural population in the Nether- based approach. Urban Design Int. 2013;4:1–18. lands using GPS measurements. Int J Health Geogr. 2017;16(1):30. 77. Kyttä AM, Broberg AK, Kahila MH. Urban environment and children’s 61. Määttä S, et al. Validity of self-reported out-of-school physical activity active lifestyle: SoftGIS revealing children’s behavioral patterns and mean- among Finnish 11-year-old children. Arch Public Health. 2016;74(1):11. ingful places. Am J Health Promot. 2012;26(5):e137–48. 62. Brooke HL, et al. A systematic literature review with meta-analyses of 78. Laatikainen TE, Haybatollahi M, Kyttä M. Environmental, individual and within- and between-day diferences in objectively measured physical personal goal infuences on older adults’ Walking in the Helsinki Metro- activity in school-aged children. Sports Med. 2014;44(10):1427–38. politan Area. Int J Environ Res Public Health. 2019;16:58. 63. Kestens Y, et al. Integrating activity spaces in health research: Comparing 79. Mitchell JT, et al. Combined ecological momentary assessment and the VERITAS activity space questionnaire with 7-day GPS tracking and global positioning system tracking to assess smoking behavior: a proof of prompted recall. Spatial Spatio-temporal Epidemiol. 2018;25:1–9. concept study. J Dual Diagn. 2014;10(1):19–29. 64. Zenk SN, et al. Feasibility of using global positioning systems (GPS) with 80. Oliver M, et al. Neighbourhoods for Active Kids: study protocol for a cross- diverse urban adults: before and after data on perceived acceptability, sectional examination of neighbourhood features and children’s physical barriers, and ease of use. J Phys Act Health. 2012;9(7):924–34. activity, active travel, independent mobility and body size. BMJ Open. 65. Schmidt T, et al. Challenges in using wearable GPS devices in low-income 2016;6:e013377. older adults: can map-based interviews help with assessments of mobil- 81. Xu Y, et al. Road extraction from high-resolution remote sensing imagery ity? TranslBehav Med. 2018;9(1):99–109. using deep learning. Remote Sens. 2018;10:1461. 66. Ahlan AR, Ahmad BIE. User acceptance of health information technol- 82. Pappalardo L, Simini F. Data-driven generation of spatio-temporal rou- ogy (HIT) in developing countries: a conceptual model. ProcediaTechnol. tines in human mobility. Data Min Knowl Disc. 2018;32(3):787–829. 2014;16:1287–96. 83. Wikström POH, et al. Activity felds and the dynamics of crime. J Quant 67. Kwan M-P. The uncertain geographic context problem. Ann Assoc Am Criminol. 2010;26(1):55–87. Geogr. 2012;102(5):958–68. 68. Mitra R. Independent mobility and mode choice for school trans- portation: a review and framework for future research. Transp Rev. Publisher’s Note 2013;33(1):21–43. Springer Nature remains neutral with regard to jurisdictional claims in pub- 69. Pont K, et al. The Model of Children’s Active Travel (M-CAT): A conceptual lished maps and institutional afliations. framework for examining factors infuencing children’s active travel. AustOccupTher J. 2011;58(3):138–44. 70. Götschi T, et al. Towards a comprehensive conceptual framework of active travel behavior: a review and synthesis of published frameworks. Curr Environ Health Rep. 2017;4(3):286–95.

Ready to submit your research ? Choose BMC and benefit from:

• fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations • maximum visibility for your research: over 100M website views per year

At BMC, research is always in progress.

Learn more biomedcentral.com/submissions