Home and Neighbourhood Built Environment Latent Class Profiles and Young Children’s Developmental Vulnerability: Findings from the Australian Early Development Census

Abstract

Introduction
Children's developmental outcomes in their early years have lifelong consequences for later health and wellbeing. Despite persistent geographical inequities in early development, the specific role of the neighbourhood built environment remains underexplored. Understanding how different neighbourhood typologies influence early development can inform more equitable urban planning and policy for children.

Objectives
To identify typologies of neighbourhood built environments and assess their association with early childhood developmental vulnerability, and whether these associations differ by area-level disadvantage.

Methods
Australian Early Development Census (AEDC) data were linked with geospatial measures of the neighbourhood built environment for 5,066 children in Perth, Western Australia. Latent profile analysis identified neighbourhood built environment profiles based on features such as access to green space, movement network, and child-relevant destinations. Logistic regression models tested associations between built environment profiles and developmental vulnerability, stratified by area-level disadvantage.

Results
Overall, 21.8% of children were developmentally vulnerable on at least one AEDC domain. Five built environment profiles emerged: 'Average neighbourhood', 'Established neighbourhood', 'Low destinations and transport mix', 'Connected housing', and 'Disconnected semi-rural'. Children in the 'Established neighbourhood' profile had significantly lower odds of developmental vulnerability (18.3%) compared with 'Average neighbourhood', 'Connected housing', and 'Disconnected semi-rural' profiles. In more disadvantaged areas, associations between built environment profiles and developmental vulnerability were less consistent.

Conclusions
Neighbourhoods with better access to child-relevant destinations, attractive green spaces, and public transport were associated with lower developmental vulnerability. Conversely, neighbourhoods with a substandard movement network, less destination access, and higher traffic exposure were linked to worse early development outcomes. Further studies should focus on unpacking the mechanisms through which individual and combined effects of features of the neighbourhood environment influence early development.

Introduction

Early childhood (0-8 years) marks an important phase when one’s environment critically influences how the brain develops, translating to health and wellbeing outcomes across the life course [1]. Access to enriched environments early in life supports optimal childhood development. For example, young children who live with highly educated, high-income neighbours show better achievement at school [2]. In contrast, children who are exposed to challenging environments, such as poverty and frequent relocation, are at increased risk of developmental problems [3].

To map, monitor and evaluate the impact of policy and service provision on early childhood development, some countries have employed population-wide census data collection approaches. For example, the Australian Early Development Census (AEDC) collects data from at least 96% of Australian children in their first year of full-time school (approximately 5 years old), providing a robust representation of the population [4]. Data from the AEDC are used to determine if children are ‘developmentally on track’, ‘developmentally at risk’ or ‘developmentally vulnerable’ across five domains: Physical health and wellbeing, Social competence, Emotional maturity, Language and cognitive skills (school-based), and Communication skills and general knowledge. Children are classified as developmentally vulnerable if they fall below approximately the lowest 10% of scores on each domain. The 2024 AEDC report indicates that 23.5% of Australian children are developmentally vulnerable in at least one developmental domain [4], highlighting the need for research to help develop effective interventions.

The socioecological framework [5] can be used to unpack how environmental factors potentially impact children’s early development. This framework emphasises that a child’s development is influenced by the interplay across nested systems, from the child and family to the broader impact of the neighbourhood and society. Children are exposed to and interact with environments such as their home, and the immediate neighbourhood built environment surrounding their home. The neighbourhood built environment consists of the design and construction of infrastructure created or modified by people, including features such as housing, street design, traffic, parks and other community spaces [6, 7]. The built environment can play a crucial role in early development through exposure and access (or lack thereof) to essential services and resources, providing opportunity for positive and/or negative daily experiences across various socioecological systems [8]. For instance, access to green space (e.g., parks, home yards) has been associated with enhanced cognitive, socio-emotional, and physical development in children through increased opportunities for social interactions, physical activity, and outdoor play, as well as improved parental wellbeing [9, 10]. Access to public transport supports more physical activity [11] as well as accessing services such as education and leisure centres [12], which is important for supporting early childhood development. Importantly, investigating the impact of modifiable built environment attributes on early child development provides an opportunity to guide the optimisation of these environments for young children [13].

Nevertheless, current research investigating the impact of the built environment on early childhood development is limited in its exploration of the underlying pathways. One potential barrier is that many built environment features are interdependent, interact, or are correlated through complex mechanisms. For example, high density areas could have shorter distances to beneficial destinations, but increased traffic exposure [14]. Additionally, built environment features could have a combined, rather than standalone, effect. For example, green space (e.g., parks, nature), blue space (i.e., oceans, rivers and lakes), and home yards may all collectively and accumulatively influence early childhood development by promoting outdoor play [15]. A deeper understanding of how different built environment features interact and influence early child development is needed to disentangle the impacts of their combined effects.

Here, latent profile analysis offers an innovative method of identifying underlying and undefined clusters of built environment features, revealing how multiple features co-occur within neighbourhoods that could inform pathways to child development outcomes [16]. Unlike approaches that examine built environment features in isolation, latent profile analysis captures qualitatively distinct profiles of various environmental attributes [17], offering a more meaningful representation of the built environment as experienced by children in their everyday lives. This method has been used in studies exploring the relationship between built environments and physical activity in children, adolescents, and adults [18, 19]. However, to our knowledge, no study has used latent profile analysis to identify built environment profiles and their relationship with early childhood developmental vulnerability.

Moreover, increasing social inequity has amplified concerns about the negative effects of sociodemographic disadvantage on early childhood development [20], which tends to accumulate in geographical areas [21]. According to the 2024 AEDC report, children in the most disadvantaged locations had more than twice the rate of developmental vulnerability compared to those in the least disadvantaged areas, with this gap widening over time [4]. Area-level disadvantage is associated with limited access to important child-relevant services and play spaces, contributing to poor developmental outcomes [22]. A more nuanced understanding of how sociodemographic disadvantage interacts with the built environment is thus needed.

Therefore, the present study sought to: (1) identify distinct built environment typologies of young children’s neighbourhoods using latent profile analysis; (2) examine associations between these neighbourhood built environment typologies and early childhood development vulnerability; and (3) examine whether these associations differ by area-level disadvantage.

Methods

Study Design and Population

This cross-sectional study included a random sample of 5,356 children in the Perth/Peel region in Western Australia who participated in the AEDC (2012 wave), a nationwide data collection occurring every three years [23, 24]. Children were assessed in their first year of full-time school (around 5 years old), with a questionnaire completed by their teachers if they had known the child for at least a month. Children were sampled proportionally to suburb size from suburbs with at least 25 children with AEDC data and residential addresses that could be assigned a latitude and longitude point [13]. Children were included in the analysis if they had complete neighbourhood environment data and at least one score on any of the five AEDC domains (n = 5,066). The 2012 AEDC had a national participation rate of 96.5%, and the final analytic sample represents 16.5% of the 2012 AEDC population in Western Australia (N = 30,631) [25].

Data Linkage

Data linkage of built environment measures to children’s early development outcomes was performed by the Western Australian Government Department of Health’s Data Linkage Branch, an accredited linkage agency. AEDC data were de-identified for linkage, involving the separation of identifying information (e.g., name, residential address) from AEDC scores and replacing it with an encrypted identifier (‘linkage key’). Children’s residential addresses (as recorded in the AEDC data set) were geocoded by the linkage agency and provided to a Geographic Information Systems specialist research team member (who was not involved in subsequent data analysis) with the corresponding linkage key to create the neighbourhood built environment variables. These variables were then returned to the data linkage agency for removal of the geocoded residential address information and provided to a different research team member for merging with AEDC data using the linkage keys.

Child Development

Child development data were obtained from the AEDC (2012 wave). The AEDC uses the Australian version of the Early Development Instrument, which was adapted from the Canadian Early Development Instrument [26]. The instrument has evidenced reliability and validity and has been shown to predict later academic and wellbeing outcomes [27]. Teachers completed the AEDC survey based on their observations of the children in their class. The AEDC measures five domains of early child development: Physical health and wellbeing, Social competence, Emotional maturity, Language and cognitive skills (school-based), and Communication skills and general knowledge [23]. For each domain, children receive a score between 0 (low) and 10 (high) on a developmental spectrum. On the basis of national benchmarks established in the 2009 AEDC data collection cycle, children are classified as developmentally vulnerable if they fall below a cut point (approximately the lowest 10% of scores) on each domain. Domain cut-offs also account for age variations (i.e. a different cutoff depending on the child’s age in years). A binary outcome variable was created to indicate whether a child was developmentally vulnerable on one or more of the five domains (AEDC DV1 indicator; yes/no). Analyses for DV1 included children with valid scores on all five domains (n = 5,027). Vulnerability on each domain separately was also considered. Children with special needs were not included within domain indicators because of the already identified substantial developmental needs of this group.

Neighbourhood Built Environment

Geographic Information Systems measures of the neighbourhood built environment were based on a 1600 m street network service area around each child’s geocoded residential address, capturing neighbourhood features exposed to each child specifically. This 1600 m service area (i.e., distance along walkable roads) was chosen to represent the child’s ‘neighbourhood’. Although this service area is commonly used for adults, young children are typically accompanied by an adult when navigating their neighbourhoods. The service area for some child-relevant destinations (i.e., early childhood education and care, playgroups, child health centres, schools) was increased to 3 km or 10 km to account for their reduced presence. Details of the 17 neighbourhood built environment measures, as well as data sources, have been described in detail previously [13, 28, 29] and are outlined in brief below.

Outdoor Home-Yard Environment

Outdoor home-yard space was measured as the percentage of residential land within the service area built around each child’s geocoded residential address that was not part of the total building footprint for the 1600 m service area. It was not possible to calculate this variable for the child’s home address due to the need to prioritise data security (i.e., potential participant identification) over data quality (i.e., individual residential level measures of the home-yard space).

Density and Movement Network

Residential density was measured as the number of residential dwellings per square kilometre of residential land. Measures of the movement network included street connectivity (the number of three-way - or more - intersections per square kilometre) and number of one-way nodes (cul-de-sacs, which are low traffic streets near home). Road traffic exposure (the percentage of minor roads carrying fewer than 3000 vehicles/day to total length of roads) was calculated as well as the number of transport stops (school bus, public bus, train) and presence/absence of a railway station.

Green Space

Green space was measured as the presence of different sizes and types of green space, including: medium-large park (>0.5 to ≤5 ha); district-regional park (>5 ha); natural public open space; school grounds; club-pay facilities (e.g., sport club facilities for hiring); and an above average attractiveness park. Park attractiveness was derived from objectively measured park attributes (e.g., presence of trees, sporting facilities, amenities, water features, lighting) [30, 31]. The highest scoring park within the neighbourhood was categorised as above or below the average attractiveness score (67.58, SD = 7.41) to indicate for each child whether an above average attractiveness park was present in their neighbourhood.

Child-Relevant Destinations

Child-relevant destinations included four measures: the number of playgroups, kindergartens and early childhood education and care centres within 3km of the child’s home, and the number of child health centres (i.e. community-based health service centres) within 10km of the child’s home.

Sociodemographic Variables

Child sex (male/female), Aboriginal/Torres Strait Islander status (yes/no), and speaking English as a second language (yes/no) were included from the AEDC dataset as covariates in analyses. Children’s age group (in three-month categories) was provided in the AEDC dataset. Sociodemographic disadvantage was determined by the Socio Economic Index For Areas-Index of Relative Socio Economic Disadvantage (SEIFA) for the child’s suburb of residence. SEIFA is measured with factors reflecting area-level disadvantage, such as low levels of educational attainment, low income, and high unemployment, as derived from the 2011 Australian census [32]. SEIFA deciles based on national distribution were assigned to each child according to their home address and classified into five equal intervals for the present sample, with the lowest interval representing areas of greatest disadvantage.

Data Analysis

Continuous or count neighbourhood built environment measures were standardised using z-scores. Latent profiles of the neighbourhood built environment and the association with child developmental vulnerability were determined using the three-step latent profile analysis approach with a distal outcome [33]. In the first step, the 17 neighbourhood built environment variables were entered into the latent profile analysis. Models with 1 through 10 profiles were examined to determine the optimal number of profiles. Model selection was based on the log-likelihood (LL); the Akaike Information Criterion (AIC), Consistent Akaike Information Criterion (CAIC), Bayesian Information Criterion (BIC), and sample-size adjusted Bayesian Information Criterion (SABIC) information criteria; entropy, interpretability, posterior probabilities, and profile size. Lower information criterion (LL, AIC, CAIC, BIC, SABIC) indicate better fit, while higher entropy (>0.8) and higher average latent posterior probabilities (>0.9) indicate better fit [34]. Characteristics of the emerging profiles were considered to determine whether profiles were qualitatively and quantitatively distinguishable and made conceptual sense within the Perth, Western Australia geographical context. Given the latent profiles represented shared neighbourhood characteristics, and children were drawn from 156 suburbs with most suburbs contributing fewer than 1% of the sample, additional adjustment for neighbourhood-level clustering was not included to avoid over-adjustment.

Second, children were assigned the most appropriate profile using proportional assignment adjusted for uncertainty when modelling profiles using the maximum likelihood correction method [33, 35].

In the third step, latent profiles were entered into logistic regression models to examine the association between profiles and AEDC outcomes, adjusting for child sex, Aboriginal or Torres Strait Islander status, and speaking English as a second language. Age was not included as a covariate given that AEDC developmental vulnerability scores are age-adjusted. Effect modification by socioeconomic status was tested by repeating each model, including an interaction term for SEIFA interval-by-profile. Where the interaction was significantly associated with the outcome, logistic regression models stratified by SEIFA intervals were performed. Proportions of children developmentally vulnerable on one or more AEDC domains (DV1) and each domain within each latent profile are presented alongside model results (odds ratios, OR, and 95% CI). Analysis samples varied by outcome and are presented in the results tables (range n = 5,027 to n = 5,060). The three-step latent profile analysis with distal outcome was undertaken in LatentGOLD 6.0 using the cluster and step3 procedures [36].

Results

Children ranged in age from 52 to 79 months, with 99.5% of children around 5 years of age (4 years 10 months to 6 years 1 month; Table 1). Half were male (50.4%), 12% spoke English as a second language (15.6% across Western Australia [37]), and 3% were Aboriginal or Torres Strait Islander (6.9% statewide [37]). About one-in-five children were developmentally vulnerable on at least one AEDC domain. Characteristics of the neighbourhood built environment features are presented in Table 2.

Table 1: Sample Characteristics (n = 5,066). *Children’s age group in three-month categories was included in the AEDC dataset, each interval defined as inclusive of the lower bound and exclusive of the upper bound. AEDC confidentiality rules require suppression of cells that contain less than four children to ensure individuals cannot be identified.

	n (%)
Age groups*
4 years 4 months–4 years 7 months	≤ 3 (0.0)
4y 7m–4y 10m	12 (0.2)
4y 10m–5y 1m	875 (17.3)
5y 1m–5y 4m	1224 (24.2)
5y 4m–5y 7m	1261 (24.9)
5y 7m–5y 10m	1297 (25.6)
5y 10m–6y 1m	382 (7.5)
6y 1m–6y 4m	7 (0.1)
6y 4m–6y 7m	4 (0.1)
6y 7m–6y 10m	≤ 3 (0.0)
Sex (female)	2514 (49.6)
Aboriginal or Torres Strait Islander	174 (3.4)
Speaks English as a second language	620 (12.2)
SEIFA Socioeconomic disadvantage invertal
1, most disadvantaged	358 (7.1)
2	601 (11.9)
3	1215 (24.0)
4	1008 (19.9)
5, least disadvantaged	1883 (37.2)
Developmentally vulnerable on one or more domains (DV1)	1095 (21.8)
Developmentally vulnerable on Physical health and wellbeing	442 (8.7)
Developmentally vulnerable on Social competence	402 (8.0)
Developmentally vulnerable on Emotional maturity	415 (8.2)
Developmentally vulnerable on Language and cognitive skills (school-based)	358 (7.1)
Developmentally vulnerable on Communication skills and general knowledge	444 (8.8)
DV1 per SEIFA socioeconomic disadvantage interval
1, most disadvantaged	108 (30.2)
2	173 (28.8)
3	278 (22.9)
4	222 (22.0)
5, least disadvantaged	314 (16.7)

Table 2: Characteristics of the Built Environment Feature.

	Mean (SD)	Range	n (%)
Outdoor home environment
Neighbourhood home-yard space (m2)	64.6 (7.8)	41.6-98.1
Neighbourhood built environment
Dwelling density	16.4 (5.0)	1.1–66.7
Street connectivity	62.6 (21.1)	8.6–166.7
One-way node count	50.9 (26.7)	2.0–150.0
Road traffic exposure	26.1 (10.9)	0.1–77.5
Transport stop count	41.0 (21.4)	0.0–121.0
Presence of railway station			635 (12.5)
Green space
Presence of medium/large park (>0.5 to ≤5 ha)			4920 (97.1)
Presence of district/regional park (>5 ha)			3964 (78.3)
Presence of natural public open space			4402 (86.9)
Presence of school grounds			4701 (92.8)
Presence of club-pay facilities			3161 (62.4)
Presence of above average attractive park			2790 (55.1)
Child-relevant destinations
Playgroup count within a 3 km radius	3.8 (2.9)	0.0–15.0
Early childhood education and care centre count within a 3 km radius	8.8 (5.0)	0.0–30.0
Kindergarten/pre-primary count within a 3 km radius	6.1 (3.5)	0.0–21.0
Child health centre count within a 10 km radius	9.9 (5.8)	0.0–26.0

Latent Profiles of Neighbourhood Built Environments

Model Selection and Number of Profiles

Information criterion improved with an increasing number of latent profiles, however improvements to fit were minimal beyond four profiles (Supplementary Table 1). Entropy declined to two profiles and thereafter increased, though generally remained high at around 0.9. Models with seven to 10 profiles contained at least one profile with fewer than 5% of the sample and/or profiles that were not conceptually distinct. Models with four to six profiles were examined further for interpretability. The model with five profiles was selected as the best fitting model. The posterior probabilities indicated good classification accuracy (Supplementary Table 2). Profiles are interpreted below with respect to their key points of differentiation. Each of the five profiles was labelled based on its distinguishing built environment feature/s relative to the other profiles.

Characteristics of Profiles

The first profile, ‘Average neighbourhood’, represented the largest proportion of children in the sample at 37.4%. This profile was characterised by average scores on most features, including child-relevant destinations, transport accessibility, and movement network. Given its relatively balanced mix of neighbourhood features and high prevalence, this profile provides a baseline comparison for other profiles.

The second profile, ‘Established neighbourhood’, represented 22.5% of children. This profile was featured by the highest residential density, the greatest number of public transport stops, and the greatest access to child-relevant destinations compared to the other profiles. This profile also had the highest presence of green spaces, with the greatest presence of an above average attractive park (85%), and a moderate level of traffic exposure, street connectivity, and one-way nodes.

The third profile, termed ‘Low destinations and transport mix’ (16.0%), had lower counts of child-relevant destinations, lower percentage of public open spaces (particularly district/regional park and club-pay facilities), as well as fewer transport stops and the lowest traffic exposure (similar to the next profile, ‘Connected housing’). This profile had a moderate level of density, street connectivity, and one-way nodes.

The fourth profile, ‘Connected housing’ (14.3%), had the lowest home-yard space, highest street connectivity and one-way nodes, as well as the lowest traffic exposure. This profile had a relatively moderate level of child-relevant destinations and public open spaces.

The final profile, ‘Disconnected semi-rural’ (9.9%), had the lowest residential density, equally lowest public transport stops and child-relevant destinations (similar to the ‘Low destinations and transport mix’ profile). This profile had the lowest presence of built green spaces, particularly above average attractive parks (5%). This profile also had the highest traffic exposure, the largest home-yard space, the lowest street connectivity, and one-way node count.

Overall, there were minimal differences between profiles for the presence of different park types and the presence of parks was generally high. Characteristics of each profile can be seen in Figure 1 and Table 3.

Figure 1: Characteristics of the Five Latent Profiles of Children’s Neighbourhood Built Environments.

Table 3: Characteristics of the Five Built Environment Profiles. Bold text indicates highest scores on each built environment feature; italicised text indicates lowest scores.

	Average neighbourhood (37.4%)	Established neighbourhood (22.5%)	Low destinations and transport mix (16.0%)	Connected housing (14.3%)	Disconnected semi-rural (9.9%)
Outdoor home-yard environment
Neighbourhood home-yard space (m2), mean (SD)	65.4 (4.8)	63.1 (3.9)	61.1 (6.8)	58.6 (6.1)	79.6 (8.6)
Neighbourhood built environment
Dwelling density, mean (SD)	15.0 (1.6)	22.6 (5.6)	15.4 (2.0)	16.7 (2.0)	9.0 (3.9)
Street connectivity, mean (SD)	55.0 (7.3)	61.1 (17.0)	72.4 (16.4)	91.2 (25.4)	38.1 (13.4)
One-way node count, mean (SD)	63.1 (27.2)	35.1 (16.6)	43.2 (18.1)	68.4 (25.0)	27.5 (13.2)
Road traffic exposure, mean (SD)	31.0 (9.2)	26.4 (7.6)	17.5 (8.1)	17.1 (8.5)	33.1 (13.1)
Transport stop count, mean (SD)	41.4 (13.1)	68.8 (17.4)	21.5 (7.5)	32.8 (9.4)	20.0 (13.5)
Presence of railway station, n (%)	237 (12.4)	316 (27.8)	30 (3.7)	15 (2.2)	37 (7.5)
Green space
Presence of medium/large park, n (%)	1883 (98.3)	1134 (99.8)	788 (95.8)	698 (99.9)	417 (84.6)
Presence of district/regional park, n (%)	1718 (89.7)	1040 (91.6)	365 (44.4)	614 (87.8)	227 (46.0)
Presence of natural public open space, n (%)	1676 (87.5)	776 (68.3)	798 (97.0)	673 (96.3)	479 (97.2)
Presence of school grounds, n (%)	1869 (97.6)	1134 (99.8)	664 (80.7)	691 (98.9)	343 (69.6)
Presence of club/pay grounds, n (%)	1151 (60.1)	967 (85.1)	367 (44.6)	410 (58.7)	266 (54.0)
Presence of above average attractive park, n (%)	1264 (66.0)	960 (84.5)	240 (29.2)	300 (42.9)	26 (5.3)
Child-relevant destinations
Playgroup count within 3 km radius, mean (SD)	8.5 (3.3)	14.4 (4.7)	4.1 (1.6)	9.9 (3.2)	3.3 (2.3)
Early childhood education and care centre count within 3km radius, mean (SD)	6.0 (2.2)	10.4 (2.8)	2.7 (1.3)	6.1 (1.8)	2.6 (1.6)
Kindergarten/pre-primary count within 3 km radius, mean(SD)	3.1 (1.7)	6.5 (3.2)	1.4 (1.1)	5.4 (3.1)	1.9 (2.0)
Child health centre count within 10 km radius, mean (SD)	11.5 (4.6)	15.2 (4.6)	4.9 (3.7)	6.2 (3.7)	5.0 (3.5)

Associations between built Environment profiles and early child developmental vulnerability

Built environment latent profiles were associated with being developmentally vulnerable on at least one domain (Wald p = 0.007). The odds of being developmentally vulnerable on at least one domain were significantly lower for the ‘Established neighbourhood’ profile (20% lower odds) compared with the ‘Average neighbourhood’. Follow up pair-wise comparisons found that the ‘Established neighbourhood’ profile had significantly lower developmental vulnerability than the ‘Average neighbourhood’, ‘Connected housing’, and ‘Disconnected semi-rural’ profiles (paired Wald tests p < 0.05, Table 4). Overall, 18.3% of children in the ‘Established neighbourhood’ profile were developmentally vulnerable on one or more domains, compared with 22.5% of children in the ‘Average neighbourhood’ profile, 22.9 % of children in the ‘Connected housing’ profile, and 25.8% of children in the ‘Disconnected semi-rural’ profile. The proportion of children being developmentally vulnerable on at least one domain did not differ for all other paired profile comparisons (all p > 0.050).

Table 4: Associations Between Built Environment Profiles and Developmental Vulnerabilities. Logistic regression models adjusted for child sex, Aboriginal or Torres Strait Islander status, and speaking English as a second language. OR = odds ratio. CI = Confidence interval. DV1 = Vulnerable on at least one AEDC domain. DV1: ‘Established neighbourhood’ significantly different (lower) to ‘Average neighbourhood’ (p = 0.006), ‘Connected housing’ (p = 0.018), ‘Disconnected semi-rural’ (p < 0.001). No other significant between-group differences (all paired comparisons p > 0.05). Physical health and wellbeing: ‘Established neighbourhood’ significantly different (lower) from ‘Average neighbourhood’ (p = 0.016), and ‘Disconnected semi-rural’ (p = 0.003). No other significant between-group differences.

	DV1 (n = 5027)		Physical health and wellbeing (n = 5059)		Social competence (n = 5055)		Emotional maturity (n = 5036)		Language and cognitive skills (school-based) (n = 5060)		Communication skills and general knowledge (n = 5058)
	%	OR (95% CI)	%	OR (95% CI)	%	OR (95% CI)	%	OR (95% CI)	%	OR (95% CI)	%	OR (95% CI)
Average neighbourhood (37.4%)	22.5	Ref	9.3	Ref	7.9	Ref	8.4	Ref	7.6	Ref	8.8	Ref
Established neighbourhood (22.5%)	18.3	0.8 (0.6, 0.9)	6.7	0.7 (0.5, 0.9)	7.3	0.9 (0.7, 1.2)	8.0	0.9 (0.7, 1.3)	5.7	0.7 (0.5, 1.0)	7.3	0.8 (0.6, 1.1)
Low destinations and transport mix (16.0%)	21.6	0.9 (0.8, 1.2)	9.1	1.0 (0.7, 1.3)	7.9	1.0 (0.7, 1.4)	7.2	0.8 (0.6, 1.2)	7.0	0.9 (0.6, 1.3)	8.8	1.0 (0.7, 1.4)
Connected housing (14.3%)	22.9	1.0 (0.8, 1.3)	8.3	0.9 (0.6, 1.2)	7.4	0.9 (0.6, 1.3)	8.9	1.1 (0.8, 1.5)	7.8	1.0 (0.7, 1.5)	10.8	1.3 (0.9, 1.8)
Disconnected semi-rural (9.9%)	25.8	1.2 (0.9, 1.5)	11.2	1.2 (0.9, 1.7)	10.5	1.4 (1.0, 1.9)	9.0	1.1 (0.7, 1.6)	7.3	1.0 (0.6, 1.5)	9.5	1.1 (0.8, 1.6)
p-value		0.007		0.036		0.270		0.760		0.310		0.150

Latent profiles were also associated with being developmentally vulnerable on the Physical health and wellbeing domain (Wald p = 0.036), with fewer children in the ‘Established neighbourhood’ profile being vulnerable (6.7%) compared to the ‘Average neighbourhood’ (9.3%) and ‘Disconnected semi-rural’ (11.2%) profiles. Built environment latent profiles were not associated with the Social competence, Emotional maturity, Language and cognitive skills (school-based), and Communication skills and general knowledge domains.

Socio Economic Effect Modification of the Association Between Built Environment Profiles and Early Child Developmental Vulnerability

Latent profiles did not significantly differ by SEIFA intervals for most outcomes (interaction p > 0.05), except for being developmentally vulnerable on at least one or more domains (interaction p = 0.018). In the third (middle) SEIFA interval, children in the ‘Established neighbourhood’ profile had lower odds of being developmentally vulnerable on at least one domain compared with the ‘Connected housing’ profile, and children in the ‘Disconnected semi-rural’ profile had lower odds of being developmentally vulnerable compared with the ‘Average neighbourhood’, ‘Low destinations and transport mix’, and ‘Connected housing’ profiles (Table 5). In the least disadvantaged interval, children in the ‘Disconnected semi-rural’ profile had higher odds of being developmentally vulnerable on at least one domain compared with all other profiles. There were no differences by built environment profile and developmental vulnerability on one or more domains for children in the most and second most disadvantaged intervals, and the second least disadvantaged interval.

Table 5: Associations Between Built Environment Profiles and Developmental Vulnerabilities on at Least one AEDC Domain, Stratified by SEIFA Intervals. Estimates for ‘Connected housing’ within interval 1 (most disadvantaged) are not available due to too few children in these categories. Interval 3: ‘Established neighbourhood’ significantly different (lower) to ‘Connected housing’ (p = 0.034); ‘Disconnected semi-rural’ significantly different (lower) to ‘Average neighbourhood’ (p = 0.029), ‘Low destinations and transport mix’ (p = 0.023), ‘Connected housing’ (p = 0.009). No other significant between-group differences (all paired comparisons p > 0.05). Interval 5: ‘Disconnected semi-rural’ significantly different (higher) than ‘Average neighbourhood’ (p < 0.001), ‘Established neighbourhood’ (p < 0.001), ‘Low destinations and transport mix’ (p < 0.001), and ‘Connected housing’ (p = 0.007). No other significant between-group differences (all paired comparisons p > 0.05).

	Average neighbourhood (37.4%)	Established neighbourhood (22.5%)	Low destinations and transport mix (16.0%)	Connected housing (14.3%)	Disconnected semi-rural (9.9%)
Interval 1, most disadvantaged (n = 357)
% children in profile	62.6	11.2	15.6	11.0	9.5
% vulnerable	28.9	39.0	25.1	–	31.2
OR (95% CI)	Ref	1.6 (0.7, 3.5)	0.8 (0.4, 1.8)	Not estimable	1.1 (0.5, 2.8)
p-value	0.660
Interval 2 (n = 601)
% children in profile	54.9	18.8	3.2	4.7	15.4
% vulnerable	27.8	25.5	34.9	49.2	27.4
OR (95% CI)	Ref	0.9 (0.5, 1.5)	1.4 (0.6, 3.6)	2.7 (0.9, 7.7)	1.0 (0.5, 1.8)
p-value	0.260
Interval 3 (n = 1,198)
% children in profile	38.4	22.1	14.0	17.6	8.0
% vulnerable	24.2	19.3	26.6	27.6	13.3
OR (95% CI)	Ref	0.7 (0.5, 1.1)	1.1 (0.7, 1.8)	1.2 (0.8, 1.9)	0.5 (0.2, 0.9)
p-value	0.035
Interval 4 (n = 1,003)
% children in profile	27.0	26.4	18.4	15.1	13.0
% vulnerable	22.1	18.5	24.6	20.8	27.8
OR (95% CI)	Ref	0.8 (0.5, 1.2)	1.2 (0.7, 1.9)	0.9 (0.5, 1.6)	1.4 (0.8, 2.3)
p-value	0.260
Interval 5, least disadvantaged (n = 1,867)
% children in profile	32.0	24.0	19.0	17.3	7.7
% vulnerable	16.5	13.9	15.0	17.9	29.0
OR (95% CI)	Ref	0.8 (0.6, 1.2)	0.9 (0.6, 1.3)	1.1 (0.8, 1.6)	2.2 (1.4, 3.4)
p-value	< 0.001

Discussion

This study aimed to examine the associations between early childhood development and the built environment by using latent profile analysis to determine whether the resulting neighbourhood profiles were associated with early developmental vulnerability, while also considering the role of area-level socioeconomic disadvantage. Five distinct profiles emerged, including ‘Average neighbourhood’, ‘Established neighbourhood’, ‘Low destinations and transport mix’, ‘Connected housing’, and ‘Disconnected semi-rural’. Differences in children’s developmental vulnerability were observed across these profiles. Consistent with the socioecological framework, this suggests that the classification captures environmental influences relevant to early development at the neighbourhood-level, identifying meaningful combinations of built environment features as they co-occur in real world settings.

Children who lived in an ‘Established neighbourhood’ had the lowest rate of developmental vulnerability in any domain (18.3%), with particularly low vulnerability in the physical health and wellbeing domain (6.7%). One feature of this profile is the greatest number of child-relevant destinations, which is consistent with previous literature showing that better access to key learning, care, and health destinations was associated with improved early development outcomes [8]. For example, availability of daycare and kindergartens was associated with better physical development [13] and cognitive development [38]. While all profiles had a high presence of parks, the ‘Established neighbourhood’ profile had the highest presence of attractive parks. This aligns with past studies showing that while access to public green space generally has positive influences on early development [8], the quality or attractiveness of the space has a stronger impact [13]. Qualitative evidence similarly highlights the crucial role of park quality in influencing the actual usage of the space [39]. This profile also featured the greatest accessibility of public transport, as well as the highest dwelling density, which is consistent with previous studies linking these features to better physical development [13]. Accessible public transport is seen by parents as important for easy access to destinations and services, facilitating social networks, and local community participation, which are beneficial to early development [6].

Interestingly, the ‘Established neighbourhood’ profile had a moderate level of connectivity and one-way node counts, along with moderate to high traffic exposure. This could suggest that moderately connected and walkable streets might be sufficient to support positive early development outcomes when public transport is accessible, with slightly higher traffic exposure an acceptable trade-off in dense, active areas where resources and destinations are more concentrated. With past studies showing null associations between street connectivity, walkability, and early development [13, 28], these findings suggest that movement network features, like street connectivity, one-way nodes, public transport, traffic exposure, and density, may yield greater insights when investigated in combination rather than in isolation.

Children whose neighbourhood featured limited child-relevant destinations, few quality green spaces, limited public transport, but highest traffic exposure (‘Disconnected semi-rural’) had the highest rate of developmental vulnerability (25.7%) in any domain, and especially in the physical health and wellbeing domain (11.2%). Prior research has shown that low traffic exposure is associated with better independent mobility and physical activity [40], while high traffic exposure is associated with poorer development outcomes [13, 28]. The evidence suggests this profile reflects the most disadvantageous environment for early child development. Nevertheless, the ‘Low destinations and transport mix’ profile also had limited destinations, green spaces, and public transport, yet did not show poorer development outcomes. This suggests the buffering role of a well-connected and safe neighbourhood movement network when access to transport and destinations is more limited.

The ‘Disconnected semi-rural’ profile also had the largest home yards. While access to larger private yards and gardens has been frequently associated with better early social-emotional [28, 41] and physical development [42], the present analysis suggests that the benefits may be attenuated by other disadvantageous factors, such as limited local child-relevant destinations and public transport options. Nevertheless, differences in yard sizes across profiles were relatively small. Further research is needed to clarify the role of home yards, particularly their interaction with neighbourhood built environment features and the impact on early child development.

The built environment profiles showed the strongest associations with early physical health and wellbeing. This may be because physical development is more directly influenced by the availability of the neighbourhood features measured in this study. For instance, access to parks and playgrounds [43], high density [10], and accessible public transportation [13], are associated with higher levels of children’s physical activity. Child-relevant destinations such as childcare and playgroups also provide ample opportunities for outdoor play and activity [44], which directly supports physical development. In contrast, other developmental domains (e.g., social-emotional, cognitive, language) may be influenced by a wider set of factors that extend beyond just access to destinations and services. For example, while proximity can determine whether families access childcare at all, the quality of childcare, which reflects the learning experience children receive, has been shown to more strongly predict cognitive [45] and language development [46] than proximity to care. Studies also suggest pathways linking the built environment to these domains may be more indirect, through activity enjoyment, social interaction, and parenting [47]. Future research should continue to examine the pathways through which the built environment influences early development, particularly beyond physical health.

In general, more children were developmentally vulnerable if they lived in more disadvantaged suburbs, aligning with previous research highlighting inequity in early development [28]. Past studies have generally observed access to beneficial built environments is lower in more disadvantaged areas [6]. In this study, built environment profiles also varied by area-level socioeconomic disadvantage. For children living in the least disadvantaged suburbs, associations between the profiles and early development were consistent with the overall findings. While high socioeconomic status is expected to support better outcomes, this finding suggests that the negative impact of low-mobility, resource-poor environments (‘Disconnected semi-rural’) may persist in these areas. In moderately disadvantaged areas, the ‘Disconnected semi-rural’ and the ‘Established neighbourhood’ profiles appeared to be associated with better development. Similar patterns appeared in the second most disadvantaged areas, although not statistically significant, possibly due to relatively smaller sample sizes. Future studies should explore potential protective factors that may buffer the effects of limited access to destinations and services in these lower socioeconomic status areas.

This inconsistency across socioeconomic status levels aligns with previous findings. For instance, Collyer and colleagues [48] observed that in high socioeconomic status areas, associations between the built environment and early development were in the expected direction, whereas unexpected associations appeared in low socioeconomic status areas. This discrepancy suggests that certain built environment features may function differently across socioeconomic contexts. For example, high density housing in high socioeconomic status areas may indicate access to more amenities and facilities, but is more likely to be associated with noise, crime/incivilities, and overcrowding in low socioeconomic status areas. Overall, more research is needed to understand how area-level disadvantage shapes the impact of the built environment on early development.

Strengths, Limitations and Future Directions

This study has a number of strengths. First, is the application of linking population-level child development census (AEDC) with geospatial neighbourhood built environment data, which allows for an integrated understanding of early childhood development by examining exposure to immediate environmental factors. This study also provides a broad coverage representative of Australia’s urban-dwelling preschool children from a large metropolitan city. Second, examining the combined influence of built environment features on early child development moves beyond considering individual built environment effects, and is a promising start toward creating a child liveability index that further integrates built environment indicators. Third, while the current profile analysis considered the availability and accessibility of 17 built environment features, future research could include other relevant built environment attributes, such as perceived or objective quality of key built environment features, to further unpack the mechanisms underlying profile differences.

Limitations include the generalisability of the findings to children from more disadvantaged areas, those living in rural and remote communities, and non-Australian urban contexts. In this Perth-based sample, most children had relatively good access to public green spaces, private yards, and lived in less disadvantaged suburbs. This may have reduced variability in built environment exposure and limited the detection of certain effects, for example, the extent to which green space differentiates neighbourhood environments and the impact this has on early child development. Future research should expand the analysis to include more diverse cultural, geographic, and socioeconomic contexts. While the present study was interested in exploring area-level socioeconomic disadvantage for capturing the geographic concentration of social inequities and their interaction with each child’s neighbourhood built environment, area-level measures may not fully reflect disadvantage at the child or household level and should be interpreted as contextual.

The present study focused on children who were at school entry age (around 5 years) in 2011, which may limit the applicability of the findings to other developmental stages or contemporary urban contexts. Although longitudinal studies examining the long-term influence of neighbourhood built environments across developmental stages are limited, growing evidence indicates there could be positive or negative effects on child development from the cumulative exposure to different neighbourhood environment features. For example, access to high quality green space has been associated with improved socio-emotional development from 4 to 13 years, with associations strengthening as children age [49]. As children grow older and gain greater independent mobility, their interactions with neighbourhood environments increase [50], potentially strengthening the influence of local destinations on developmental outcomes. While neighbourhood infrastructure in Perth (Western Australia) has evolved since 2011, evidence from a large Perth-based cohort commencing in 2015 suggests that changes in key built environment attributes, such as residential density and vegetation cover, have been relatively modest over the past decade and have had limited impacts on children’s behaviours [51]. As such, the neighbourhood built environment profiles identified in this study remain relevant, though they should be interpreted in the urban context at the time.

Conclusion

This study utilises linked datasets and classifies and identifies important neighbourhood built environment patterns associated with early childhood development vulnerability, highlighting how specific built environment profiles, particularly those characterised by greater access to child-relevant destinations, attractive green spaces, and public transport, are associated with lower developmental vulnerability, especially in the physical health and wellbeing domain. Conversely, profiles with poor movement network, low destination access, and higher traffic exposure were linked to worse developmental outcomes. These findings underscore the importance of considering the combined influence of built environment features, rather than isolated elements, in shaping early child development. Further studies should continue to build on the derived profiles and explore the quality of built environment features, caregiver perceptions, and broader family and community-level factors to better understand the mechanisms linking neighbourhood environments to child development.

Acknowledgements

This research was supported by the Australian Government through the Australian Research Council’s Centre of Excellence for Children and Families over the Life Course (#CE200100025). HC is supported by The Kids Research Institute Australia Ascend Senior Research Fellowship. EKA is supported by an Australian Government Research Training Program Fees Offset and Australian Government Research Training Program Stipend, a UWA and Graduate Women (WA) Research Scholarship, a Stan and Jean Perron Top Up Scholarship, and a Stan and Jean Perron Award for Excellence. The authors gratefully acknowledge prior contributions to components of this research program from Prof Billie Giles-Corti, Dr Tess Gregory, Bridget Beesley and Alex Saunders. This paper used data from the Australian Early Development Census (AEDC). The AEDC is funded by the Australian Government Department of Education. The findings and views reported are those of the authors and should not be attributed to the Department or the Australian Government.

Statement of Conflicts of Interest

The authors declare no potential conflicts of interest with respect to the research, authorship or publication of this research.

Ethics Statement

Ethics approval was granted by The University of Western Australia Human Research Ethics Committee (2021/ET000722).

Data Availability Statement

Australian Early Development Census data can be obtained via the Social Research Centre Australia (https://www.aedc.gov.au). All other data access enquiries should be directed to the Corresponding author.

References

1. Hertzman, C., Making early childhood development a priority. Lessons from Vancouver. Canadian Centre for Policy Alternatives, 2004, https://policyalternatives.ca/sites/default/files/uploads/publications/BC_Office_Pubs/early_childhood.pdf.

2. Dupere, V., et al., Understanding the positive role of neighborhood socioeconomic advantage in achievement: the contribution of the home, child care, and school environments. Dev Psychol, 2010. 46(5): p. 1227-44. 10.1037/a0020211

   10.1037/a0020211

3. Ziol-Guest, K.M. and C.C. McKenna, Early childhood housing instability and school readiness. Child Dev, 2014. 85(1): p. 103-13. 10.1111/cdev.12105

   10.1111/cdev.12105

4. Department of Education Skills and Employment Australian Government, Australian Early Development Census National Report 2024. 2025: Canberra, Australia. Available from: https://www.aedc.gov.au/resources/detail/2024-aedc-national-report.

5. Bronfenbrenner, U., The Ecology of Human Development: Experiments by Nature and Design. 1979: Harvard University Press.

6. Villanueva, K., et al., The built environment and early childhood development: qualitative evidence from disadvantaged Australian communities. Children’s Geographies, 2023. 21(2): p. 330-346. 10.1080/14733285.2022.2059651

   10.1080/14733285.2022.2059651

7. Saelens, B.E. and S.L. Handy, Built environment correlates of walking: a review. Med Sci Sports Exerc, 2008. 40(7 Suppl): p. S550-66. 10.1249/MSS.0b013e31817c67a4

   10.1249/MSS.0b013e31817c67a4

8. Christian, H., et al., The influence of the neighborhood physical environment on early child health and development: A review and call for research. Health Place, 2015. 33: p. 25-36. 10.1016/j.healthplace.2015.01.005

   10.1016/j.healthplace.2015.01.005

9. Flouri, E., E. Midouhas, and H. Joshi, The role of urban neighbourhood green space in children’s emotional and behavioural resilience. Journal of Environmental Psychology, 2014. 40: p. 179-186. 10.1016/j.jenvp.2014.06.007

   10.1016/j.jenvp.2014.06.007

10. Roemmich, J.N., et al., Association of access to parks and recreational facilities with the physical activity of young children. Prev Med, 2006. 43(6): p. 437-41. 10.1016/j.ypmed.2006.07.007

    10.1016/j.ypmed.2006.07.007

11. Durand, C.P., et al., Transit Use by Children and Adolescents: An Overlooked Source of and Opportunity for Physical Activity? J Phys Act Health, 2016. 13(8): p. 861-6. 10.1123/jpah.2015-0444

    10.1123/jpah.2015-0444

12. Berg, J. and J. Ihlström, The Importance of Public Transport for Mobility and Everyday Activities among Rural Residents. Social Sciences, 2019. 8(2): p. 58, https://www.mdpi.com/2076-0760/8/2/58.

13. Bell, M.F., et al., Children’s neighbourhood physical environment and early development: an individual child level linked data study. J Epidemiol Community Health, 2020. 74(4): p. 321-329. 10.1136/jech-2019-212686

    10.1136/jech-2019-212686

14. Binter, A.C., et al., Urban environment and cognitive and motor function in children from four European birth cohorts. Environ Int, 2022. 158: p. 106933. 10.1016/j.envint.2021.106933

    10.1016/j.envint.2021.106933

15. Chawla, L., Benefits of nature contact for children. Journal of planning literature, 2015. 30(4): p. 433-452. 10.1177/0885412215595441

    10.1177/0885412215595441

16. Collins, L.M. and S.T. Lanza, Latent class and latent transition analysis: With applications in the social, behavioral, and health sciences. 2013: John Wiley & Sons.

17. Spurk, D., et al., Latent profile analysis: A review and “how to” guide of its application within vocational behavior research. Journal of Vocational Behavior, 2020. 120: p. 103445. 10.1016/j.jvb.2020.103445

    10.1016/j.jvb.2020.103445

18. Bai, P., et al., Preschool children’s physical activity in the home, childcare and neighbourhood environment: A latent profile analysis using device-based measures. J Sports Sci, 2024. 42(13): p. 1173-1183. 10.1080/02640414.2024.2376392

    10.1080/02640414.2024.2376392

19. Caetano, I.T., et al., Adolescent’s movement behaviors and built environment: a latent class analysis. BMC Public Health, 2021. 21(1): p. 1937. 10.1186/s12889-021-11974-4

    10.1186/s12889-021-11974-4

20. Poulain, T., M. Vogel, and W. Kiess, Review on the role of socioeconomic status in child health and development. Curr Opin Pediatr, 2020. 32(2): p. 308-314. 10.1097/MOP.0000000000000876

    10.1097/MOP.0000000000000876

21. Pearson, A.L., P. Apparicio, and M. Riva, Cumulative disadvantage? Exploring relationships between neighbourhood deprivation trends (1991 to 2006) and mortality in New Zealand. Int J Health Geogr, 2013. 12: p. 38. 10.1186/1476-072X-12-38

    10.1186/1476-072X-12-38

22. Leventhal, T. and J. Brooks-Gunn, The neighborhoods they live in: the effects of neighborhood residence on child and adolescent outcomes. Psychol Bull, 2000. 126(2): p. 309-37. 10.1037/0033-2909.126.2.309

    10.1037/0033-2909.126.2.309

23. Australian Early Development Census, Australian Early Development Census (AEDC) Data Guidelines. 2023, Commonwealth of Australia. Available from: https://www.aedc.gov.au/resources/detail/aedc-data-guidelines.

24. Brinkman, S.A., et al., Data resource profile: the Australian early development index (AEDI). Int J Epidemiol, 2014. 43(4): p. 1089-96. 10.1093/ije/dyu085

    10.1093/ije/dyu085

25. Department of Education, Australian Early Development Census: 2012 Summary Report – November 2013. 2013, Department of Education: Canberra, ACT 2601, Australia. Available from: https://www.aedc.gov.au/docs/default-source/default-document-library/aedc_2012_summary_report_-1501-083-5-2.pdf?sfvrsn=88a6de98_1.

26. Janus, M., et al., The Early Development Instrument: A population-based measure for communities. A handbook on development, properties, and use. 2007, Ontario, Canada: Offord Centre for Child Studies, McMaster University.

27. Gregory, T., et al., Associations between School Readiness and Student Wellbeing: A Six-Year Follow Up Study. Child Indicators Research, 2021. 14(1): p. 369-390. 10.1007/s12187-020-09760-6

    10.1007/s12187-020-09760-6

28. Christian, H., et al., Relationship between the neighbourhood built environment and early child development. Health Place, 2017. 48: p. 90-101. 10.1016/j.healthplace.2017.08.010

    10.1016/j.healthplace.2017.08.010

29. Villanueva, K., et al., The impact of the built environment on health across the life course: design of a cross-sectional data linkage study. BMJ Open, 2013. 3(1). 10.1136/bmjopen-2012-002482

    10.1136/bmjopen-2012-002482

30. Edwards, N., et al., Development of a Public Open Space Desktop Auditing Tool (POSDAT): A remote sensing approach. Applied Geography, 2013. 38: p. 22-30. 10.1016/j.apgeog.2012.11.010

    10.1016/j.apgeog.2012.11.010

31. Giles-Corti, B., et al., Increasing walking: how important is distance to, attractiveness, and size of public open space? Am J Prev Med, 2005. 28(2 Suppl 2): p. 169-76. 10.1016/j.amepre.2004.10.018

    10.1016/j.amepre.2004.10.018

32. Australian Bureau of Statistics. Census of Population and Housing: Socio-Economic Indexes for Areas (SEIFA), Australia, 2011. 2013 [cited 2026 29 January]; Available from: https://www.abs.gov.au/ausstats/abs@.nsf/Lookup/2033.0.55.001main+features100052011.

33. Bakk, Z., F.B. Tekle, and J.K. Vermunt, Estimating the Association between Latent Class Membership and External Variables Using Bias-adjusted Three-step Approaches. Sociological Methodology, 2013. 43(1): p. 272-311. 10.1177/0081175012470644

    10.1177/0081175012470644

34. Sinha, P., C.S. Calfee, and K.L. Delucchi, Practitioner’s guide to latent class analysis: methodological considerations and common pitfalls. Critical care medicine, 2021. 49(1): p. e63-e79. 10.1097/CCM.0000000000004710

    10.1097/CCM.0000000000004710

35. Vermunt, J.K. and J. Magidson Technical Guide for Latent GOLD 5.1: Basic, Advanced, and Syntax. 2016. https://www.statisticalinnovations.com/wp-content/uploads/LGtechnical.pdf.

36. Powerful and User-Friendly Latent Class Analysis. Available from: https://www.statisticalinnovations.com/latent-gold-6-1/.

37. Australian Government, A Snapshot of Early Childhood Development in Australia 2012 – AEDI National Report Re-issue. 2013, Australian Government: Canberra, Australia. Available from: https://www.aedc.gov.au/docs/default-source/default-document-library/021213-14b-aedi_national_report_2012.pdf?sfvrsn=14feda94_1.

38. Prado-Galbarro, F.J., et al., Early childhood development and urban environment in Mexico. PLoS One, 2021. 16(11): p. e0259946. 10.1371/journal.pone.0259946

    10.1371/journal.pone.0259946

39. McCormack, G.R., et al., Characteristics of urban parks associated with park use and physical activity: a review of qualitative research. Health Place, 2010. 16(4): p. 712-26. 10.1016/j.healthplace.2010.03.003

    10.1016/j.healthplace.2010.03.003

40. Riazi, N.A., et al., Correlates of Children’s Independent Mobility in Canada: A Multi-Site Study. Int J Environ Res Public Health, 2019. 16(16). 10.3390/ijerph16162862

    10.3390/ijerph16162862

41. Richardson, E.A., et al., The role of public and private natural space in children’s social, emotional and behavioural development in Scotland: A longitudinal study. Environ Res, 2017. 158: p. 729-736. 10.1016/j.envres.2017.07.038

    10.1016/j.envres.2017.07.038

42. Van Kann, D.H.H., et al., Applying an ecosystem approach to explore modifiable factors related to the risk for low motor competence in young children. J Sci Med Sport, 2022. 25(11): p. 890-895. 10.1016/j.envres.2017.07.038

    10.1016/j.envres.2017.07.038

43. Zare Sakhvidi, M.J., et al., Greenspace and health, wellbeing, physical activity, and development in children and adolescents: An overview of the systematic reviews. Current Opinion in Environmental Science & Health, 2023. 32: p. 100445. 10.1016/j.coesh.2023.100445

    10.1016/j.coesh.2023.100445

44. Davison, K.K. and C.T. Lawson, Do attributes in the physical environment influence children’s physical activity? A review of the literature. International Journal of Behavioral Nutrition and Physical Activity, 2006. 3(1): p. 19. 10.1186/1479-5868-3-19

    10.1186/1479-5868-3-19

45. de Souza Morais, R.L., et al., Quality of the home, daycare and neighborhood environment and the cognitive development of economically disadvantaged children in early childhood: A mediation analysis. Infant Behavior and Development, 2021. 64: p. 101619. 10.1016/j.infbeh.2021.101619

    10.1016/j.infbeh.2021.101619

46. De Marco, A. and L. Vernon-Feagans, Rural Neighborhood Context, Child Care Quality, and Relationship to Early Language Development. Early Educ Dev, 2013. 24(6): p. 792-812. 10.1080/10409289.2013.736036

    10.1080/10409289.2013.736036

47. Putra, I.G.N.E., et al., Do physical activity, social interaction, and mental health mediate the association between green space quality and child prosocial behaviour? Urban Forestry & Urban Greening, 2021. 64: p. 127264. 10.1016/j.ufug.2021.127264

    10.1016/j.ufug.2021.127264

48. Collyer, C., M.F. Bell, and H.E. Christian, Associations between the built environment and emotional, social and physical indicators of early child development across high and low socioeconomic neighbourhoods. Int J Hyg Environ Health, 2022. 243: p. 113974. 10.1016/j.ijheh.2022.113974

    10.1016/j.ijheh.2022.113974

49. Feng, X. and T. Astell-Burt, Residential Green Space Quantity and Quality and Child Well-being: A Longitudinal Study. Am J Prev Med, 2017. 53(5): p. 616-624. 10.1016/j.amepre.2017.06.035

    10.1016/j.amepre.2017.06.035

50. Villanueva, K., et al., Where Do Children Travel to and What Local Opportunities Are Available? The Relationship Between Neighborhood Destinations and Children’s Independent Mobility. Environment and Behavior, 2013. 45(6): p. 679-705. 10.1177/0013916512440705

    10.1177/0013916512440705

51. Robinson, T., et al., Understanding variations in the built environment over time to inform longitudinal studies of young children’s physical activity behaviour - The BEACHES project. Health Place, 2024. 90: p. 103345. 10.1016/j.healthplace.2024.103345

    10.1016/j.healthplace.2024.103345