Table 1 Study characteristics

Boone-Heinonen et al. 2010, USA

Adolescents and young adults

N = 12,701 (50.9%)

6 years

Pay and public PA facilities availability (count per 10,000 population)

Street connectivity (Alpha index)

Landscape diversity (Simpson’s diversity index)

Leisure time MVPA (S)

Fixed effects Poisson regression

In the whole sample landscape diversity, public facility availability, and alpha index were unrelated to MVPA.

Leisure-time MVPA bouts associated with increased public facility availability among female movers (0.053, 95% CI: 0.008, 0.097) and with pay facility availability among men (0.024, 95% CI: 0.006, 0.042, p < 0.05)

No significant gender specific effects of landscape diversity and street connectivity on leisure-time MVPA bouts.

Buck et al 2019, Germany, Italy, and Sweden

Children and adolescents 3–15 years old

N = 2488 (48.1%)

6 years

Movability index (land use mix, street connectivity, availability of public transport and public open spaces)

LPA and MVPA (O)

Linear mixed model

Results presented separately for genders.

In girls, the movability index showed a consistent significantly positive effect on MVPA (β = 2.14, 95% CI: (0.11; 4.16)) for all ages, and in boys, on LPA with age for each year (β = 2.68, 95% CI: (0.46; 4.90)). Availability of public open spaces was more relevant for MVPA in girls (β = 2.38, 95% CI: (0.43; 4.34)) and LPA in boys (β = 10.6, 95% CI: (4.78; 16.3)) during childhood, whereas in adolescence, intersection density (β = 3.36, 95% CI: (1.14; 5.57)) became more important for boys LPA.

0.86

Carver et al. 2010, Australia

Children 8–9 years old

N = 170 (49%)

Adolescents 13–15 years old

N = 276 (57%)

2 years

Road environment; GIS

MVPA (O)

Active transport (walking cycling) (S)

Linear regression

Results presented separately for genders.

Children

Boys: Change in MVPA was positively associated with the number of slow points (chicanes) in the neighborhood (β = 1.55, 95% CI = 0.25 to 2.86) (before school), the total length of locals roads (B = 3.81, 95% CI = 0.95 to 6.67) (weekends) and intersection density (B = 0.49, 95% CI = 0.14 to 0.84) (weekends).

Girls: Change in MVPA was negatively associated with the intersection density (B = − 0.05, 95% CI = − 0.09 to − 0.003) and the number of traffic/pedestrian lights (B = − 0.88, 95% CI = − 1.41 to − 0.35) (evenings).

Adolescents.

Boys: The number of speed humps in the neighborhood was positively associated with change in MVPA after school (p < 0.05), and the local road index was negatively associated with change in MVPA on weekend days (p < 0.05).

Girls: Total length of local roads (p < 0.01), intersection density (p < 0.01), and the number of traffic/pedestrian lights (p < 0.05) were each positively associated with change in MVPA before school. The local road index was negatively associated with change in MVPA after school (p < 0.01). The numbers of speed humps (p < 0.05) and gates/barriers (p < 0.01) were positively associated with change in MVPA during evenings. The total length of local roads (p < 0.01), intersection density (p < 0.05), and the number of speed humps (p < 0.05) were positively associated with change in MVPA during non-school hours, while the local road index was negatively associated with MVPA (p < 0.01). The number of speed humps was positively associated with change in MVPA on weekend days (p < 0.05).

0.82

Coogan et al. 2009, USA

Adult women

N = 20,354 (100%)

6 years

Housing density, land use, street connectivity, traffic, public transit availability, presence of sidewalks, distance to parks; GIS

Utilitarian and exercise walking (S)

Multinomial logistic regression generalized estimating equation model

(Only females)

Increases in utilitarian walking were associated with increased housing density (OR = 2.72, 95% CI 2.22, 3.31) and bus availability (OR = 1.44, 95% CI: 1.21, 1.72).

Increased housing density led to increased exercise walking (OR = 1.28, 95% CI: 1.07, 1.52)

Land use, street connectivity, traffic, presence of sidewalks, and distance to parks were not associated with utilitarian or exercise walking.

0.91

Coombes et al. 2014, UK

Children 10–11 years old

N = 518 (55.8%)

1 year

Availability of greener environments and destinations, density of the road network, school-home distance;

GIS

Overall PA (O).

Travel mode (S)

Multiple regression models

No significant associations between change in school commute environment or home neighborhood supportiveness and overall PA.

No gender specific effects.

0.82

Crawford et al. 2010, Australia

Children

N = 301 (57.5%)

5 years

Destinations (PA related and school), road connectivity, traffic exposure; GIS.

MVPA (O)

Generalized estimating equation

Results presented separately for genders.

No significant associations between BE and MVPA, only the presence of dead-end roads was positively associated with MVPA in boys (b = 0.007, 95% CI: 0.01, 0.13; p < 0.05).

0.86

Dowda et al. 2020, USA

Children

N = 555 (56%)

3 years

Outdoor PA equipment

Number of PA days and PA location (S)

Longitudinal Poisson regression

Outside PA equipment were positive significant predictors of street PA in total sample (p < 0.05).

No gender specific effects.

0.82

Evenson et al. 2018, USA

Adolescents girls

N = 265

1 year

Parks availability; GIS

Park visits;

MVPA in parks (O)

Wilcoxon sign rank test for two dependent samples, Pearson correlation coefficients

(Only females)

Parks were an under-used resource for adolescent girls, particularly for MVPA.

Only one-fifth of the sample (20% at baseline, 19% at follow-up) visited a park at least once in six days of observation. The average duration of park visits was higher at baseline (63.9 min) compared to follow-up (38.4 min). On days when a park was visited, MVPA was higher than on days when a park was not visited. However, only 1.9% (baseline) and 2.8% (follow-up) of MVPA occurred in parks.

0.77

Hou et al. 2010, USA

Adults

N = 5115 (54.5%)

15 years

Street connectivity (intersection density, link-node ratio) , characteristic of local roads (density, proportion local relative to total road); StreetMap data, TIGER/line road classification

Overall walking, cycling, and jogging (S)

Two-part marginal effect model (probit model and an ordinary least squares regression model)

Results presented separately for genders.

Intersection density was positively associated with walking, bicycling and jogging frequencies in low urbanicity areas for both genders (men: β = 1.0; 95% CI: 0.04, 1.9, p = 0.04; women: β = 1.3; 95% CI: 0.6, 2.0, p = 0.001). Density of local roads was positively associated only among men in low urbanicity areas (β = 1.0; 95% CI: 0.1, 2.0, p = 0.03).

In high urbanicity areas walking, cycling and jogging frequencies in women were negatively associated with local road density (β = − 1.3; 95% CI: − 2.2, − 0.3, p = 0.007) and proportion of local roads (β = − 1.4; 95% CI: − 2.3, − 0.6, p = 0.001).

No significant associations between link-note ratio and PA for both genders.

0.86

Michael et al. 2010, USA

Adult men (> 65 years old)

N = 513

Average 3.6 years

Availability of proximate PA resources: parks, trails, and recreational facilities; GIS

Walking (S)

Log-binomial regression

(Only males)

Proximity to recreational facilities was not associated with walking.

Distances to a park and a trail were positively associated with maintaining or increasing walking between baseline and follow-up, but was not significant for the whole sample. Proximity to parks and proximity to trails, respectively, were associated with a 22% (95% CI: 1.01, 1.47) and 34% (95% CI: 1.16, 1.55) higher likelihood of maintaining or increasing walking time in high-SES neighborhoods, but there was no association in low-SES neighborhoods.

0.77

Sanders et al. 2015, Australia

Children

N = 4983 (49.1%)

8 years

Availability of green areas

Overall PA (S)

Multilevel linear regression and multilevel logistic models

Results presented separately for genders.

Boys living in areas with 10 % more neighbourhood green space had a 7 % (95% CI = 1.02, 1.13) greater odd of choosing physically active pastimes; and 7% (95% CI = 1.02; 1.12) and 9% (95% CI = 1.03; 1.15) greater odds of meeting PA guidelines on weekdays and weekends, respectively. A 10% difference in green space was associated with a mean of 1.9 min greater time spent physically active on a weekday (β = 1.88, 95% CI = 0.22, 3.53; p = 0.026), and 3.0 min more weekend PA (β = 3.01, 95% CI = 0.37, 5.66; p = 0.026) after adjusting for confounders, but only at younger ages.

No statistically significant results were observed for girls.

0.91

Schipperijn et al. 2015, Denmark

Young adults,

N = 177 (57%)

6 years

Movability index, recreational facilities, density of daily destinations, street connectivity; GIS

Overall PA (O)

Multivariable analysis of variance

No significant associations between changes in movability index, availability in recreational facilities, density of daily destinations, street connectivity and PA for the whole sample.

Increases in mean daily total PA associated with increases in movability index (β = 10.15, 95% CI: 2.08, 18.21, p = 0.014) and daily destinations (β = 31.24, 95% CI: 10.64, 51.84, p = 0.003) in females.

Increased intersection density (street connectivity) was negatively associated with mean daily total PA in males (β = − 35.47, 95% CI: − 67.10, − 3.83, p = 0.03)

0.91

Andersen et al. 2017, Denmark

Adolescents

N_baseline = 354 (53%)

N_post-renewal = 319 (59.6%)

1 year

New urban green spaces and playgrounds were created

PA within the renovated area (O).

Linear mixed model

Post-intervention sample spent 7.8 min per day in LPA (p = 0.012) and 4.5 min per day in MPVA (p < 0.001).

No gender specific effects.

0.77

Brown and Werner 2007, USA

Adults

N = 51 (47%)

1 year

New rail stop

Moderate PA bouts (O).

Ordinary least squares regression

Rail ridership was positively associated with moderate activity bouts, β = 0.39 (SE = 0.01), p = 0.01.

No gender specific effects.

0.77

Burbidge and Goulias 2009, USA

Children and adults from 5 years old

N = 82 (54.9%)

1 and 5 months

New trail

Overall PA, walking and biking trips (S)

A fixed-effects panel analysis regression

The new trail was associated with significant decrease in total PA (-0.245, t value = − 2.13, R-square = 0.045, p = 0.036) and in the number of walking trips (− 0.265, t value = − 2.71, R-square = 0.070, p = 0.008).

No gender specific effects.

0.77

Chang et al. 2017, Mexico

Adults

N_{pre-intervention} = 1067 (52%)

N_{post-intervention} = 1420 (51%)

3 years

Bus rapid transit and streetscape redesign (widened sidewalks, road diets, recovery of green and public space)

Walking for transport, walking for recreation, and cycling for transport (S)

Propensity score matching, cluster analysis

The average treatment effect of living in post-intervention neighbourhood on walking for transport was 24.37 min per week, on walking for transport and recreation—31.72 min, on cycling for transport—4.81 min.

Cluster analyses showed that females with low education experienced the greatest increases in PA. All of the female clusters experienced significant growth in recreational walking and transport walking except female homemakers with high education.

Male clusters experienced either minimal improvements in recreational walking or decreases, but significant improvement in walking for transport (the greatest increase in male students with mid-level education).

0.82

Cohen et al. 2015, USA

N_baseline = 922

N_post-renewal = 1043

3 years

Park improvements

Park use and PA (O)

Mixed effect model and logit models

Use of the two renovated parks and PA level of users increased compared with baseline. %-change in total park use (β = 233.1, SE = 55.9, p = 0.001); %-change in MET-hours expended in park (β = 254.8, SE = 70.1, p = 0.001).

The total park use and MET-hours expended in unrenovated parks significantly decreased.

No gender specific effects.

0.86

Cohen et al 2014, USA

N/A

2 years

Creation of pocket parks

Park use and PA (O)

Self-reported park use

Generalized estimating equation

The new pocket parks had significantly more users than comparison park playgrounds (β = − 1.21, SE = 0.28, p = 0.001).

More females were observed at the pocket parks during follow-up than at comparison park playgrounds (63% vs. 56%, p = 0.0068).

Females were somewhat less active than males in the pocket parks, with 22% engaged in MVPA vs. 29% of males (p = 0.08).

0.68

Cranney et al. 2016, Australia

Children and adults

N_baseline = 8560 (46%)

N_{post-installation} = 7097 (46%)

N_follow-up = 8248 (46%)

1 year

Park improvement (outdoor gym)

Park use and PA (O)

Two sample z-test

The proportion of all park users engaged in MVPA increased significantly (p < 0.001) from baseline (9.4%) to post-installation (12.8), then decreased toward the baseline proportion at follow-up.

The proportion of outdoor area users from all park users increased from 2.4% to 6% (p < 0.001) and decreased back to 3.3% at follow-up, which is still a significant increase in comparison with baseline (p < 0.001).

The proportion of male park users engaged in MVPA during follow-up measurement increased on 1.9% in comparison with baseline (p < 0.01), for female users increase was not significant.

The proportion of male outdoor area users from all park users during follow-up measurement increased on 1.1% in comparison with baseline (p < 0.01), for female users increase was not significant.

0.77

Dill et al. 2014, USA

Adults

N = 353

2 years

New bicycle boulevard

MVPA (O).

Walking and cycling trips (S)

Binomial logit regression, negative binomial regression and linear regression models

No significant associations between installation of bicycle boulevards and increases in PA and active transportation.

Women engaged in less MVPA (ß = − 4.46, p = 0.02) and minutes of bicycling (ß = − 0.311, p = 0.01), were less likely to cycle more than 10 min (ß = − 0.475, p = 0.04) and make bike trips (ß = − 0.58, p = 0.01), but were more likely to walk more than 20 min (ß = 0.616, p < 0.001).

0.85

Goodman et al. 2013, UK

Adults

N = 1510 (57%)

2 years

New local walking and cycling routes

Walking and cycling at new routs (S)

Longitudinal Poisson regression

After one year 32% of sample started to use new routes (29% walking, 13% cycling), after two years the proportion of users increased to 38% (35% walking, 16% cycling).

Men were more likely to use Connect2 (rate ratio 1.14 for men vs. women, p < 0.05). However, results were significant only in one of three cities.

0.91

Heien et al. 2015, UK

3 years

New transport infrastructure (busway with path for walking and cycling)

Commute mode share (active travel) (S)

Multivariable multinomial logistic regression models

Commuters living 4 km from the busway were almost twice as likely to report a substantial increase (> 30%) in their active travel mode share (relative risk ratio [RRR] 1.80, 95% confidence interval [95% CI] 1.27 to 2.55), and half as likely to report a small decrease RRR 0.47, 95 % CI 0.28 to 0.81), than those living 9 km away.

Proximity to the busway also predicted a large decrease in the share of trips made entirely by car (RRR 2.09, 95 % CI 1.35, 3.21).

No gender specific effects.

0.86

Heinen et al. 2018, Australia

Adults

N = 4637 (57.7%)

4 years

New public bicycle-sharing scheme

Time spent cycling (S)

Multinomial logistic regression

On average, the respondents decreased the total time spent cycling by 1.98 minutes a week. Time spent cycling for transport decreased by 2.34 min per week, whereas the average time spent cycling or recreation increased by 0.35 min.

No significant associations between proximity to a bicycle-sharing station and changes in time spent cycling.

Women, when compared to men, were less likely to increase or decrease the time spent cycling.

0.82

King et al. 2015, USA

Children and adults

N_baseline = 2888 (42.2%)

N_{post-intervention} = 4525 (46.4%)

2 years

New park

Park use and PA (O)

T-test

The total number of people observed using the park post-construction significantly increased (p = 0.004). Three-fold increase in energy expended within the park after construction (p = 0.002). Increase in the proportion of park users who were engaged in moderate (p = 0.007) or vigorous activity (p = 0.04).

The number of visitors increased for both genders.

Proportion of visitors engaged only in sedentary activities decreased both in females (from 59 to 42%) and in males (from 44 to 26%).

Proportion of females observed engaging in vigorous PA inside of the park boundaries increased from 0 to 20% (mostly children).

Adolescent females were very under-represented within the park. Of the adolescent females counted, few were engaged in vigorous PA. On the other hand, there was a significant increase in the proportion of adolescent males observed engaging in vigorous PA (P < 0.001), especially in playfields.

0.77

Ng et al. 2020, Australia

6–12 months

Upgrade od childcare outdoor spaces (installation of outdoor PA equipment)

Overall PA, MVPA (O)

Multivariable linear regression

Intervention preschoolers were more active than control at follow-up (58.09 vs. 42.13 min/day increase in total PA; 30.46 vs. 19.16 min/day increase in MVPA (all p < 0.001)).

Boys were significantly more active than girls (p < 0.01).

0.86

Panter et al. 2017, UK

Adults

N = 1257

2 years

New walking and cycling infrastructure

Walking for transport and recreation (S)

Latent class analysis and multinomial regression

Short-lived and sustained increase as well as uptake in walking for transport and recreation were associated with use of new walking and cycling infrastructure.

Proximity to the intervention was associated with both uptake of and short-lived increases in walking for transport.

Increase and uptake in walking for transport or recreation were not associated with gender.

0.82

Parker et al. 2011, USA

Adults

N/A

1 year

New bike lane

Number of cyclists (O)

Negative binomial regression

The mean number of cyclists observed per day increased by 57% (p < 0.001).

The increase among adult female riders (133%, p < 0.001) was greater than among adult male riders (44%, p < 0.001).

0.6

Parker et al. 2013, USA

Youth and adults

N/A

1 year

New bike lane

Number of cyclists (O)

Negative binomial regression

There was an increase in cyclists on all three streets after the installation of the bike lanes [62.5 (± 28.8) vs. 110 (± 109); Z = 8.97, p < 0.000], with the largest increase on the street with the new lane [pre 79.2 (± 30.5), post 257.1 (± 50.9); Z = 10.79, p < 0.000].

The increase in cyclists was greater among females (4.69) than males (3.12).

0.75

Rissel et al. 2015, Australia

Adults

N = 512 (62.5%)

1 year

New cycling infrastructure

Number of cyclist (O).

Cycling behavior (S)

Mixed-effects logistic regression models

Bike counts at two sites on the new bicycle path reported an increase of 23 % and 97 % respectively at 12 months. Weekly frequency of cycling remained higher in the intervention (29.2–25.8% at follow-up) area than the comparison area (22.4–23.2% at follow-up) (p = 0.04).

Among the participants in the cohort, there was no change in the self-reported weekly frequency of cycling. Only 15 % participants reported using the new bicycle path, with most users (76 %) living in the intervention area.

No gender specific effects.

0.91

Schultz et al. 2017, USA

Children and adults

N_baseline = 2080 (46%)

N_{1 year} = 2275 (45%)

N_{2 years} = 2276 (46%)

2 years

Improved access to the park

Park use and PA (S)

One-way ANCOVA model and Sidak post-hoc comparisons

Total park use increased from 2012 (n = 2080) to 2013 (n = 2275) and remained constant in 2014 (n = 2276). However, despite increases in safe access and overall park use, there was a significant decrease in total energy expenditure following the installation of the crosswalk that was sustained in 2014.

Male park use increased from 2012 to 2013 (6.95 to 10.49) but significantly decreased from 2013 to 2014 (10.49 to 7.82); however, there was still a significant increase from 2012 to 2014 (6.95 to 7.82). Females also showed a significant increase of park use from 2012 to 2013 (7.45 to 9.78); however, unlike males, the increased use was maintained in 2014 (9.63).

The total energy expenditure both for males and females in 2014 was significantly lower than in 2012.

0.91

Smith et al. 2019, Australia

Adults

N = 389 (58.4%)

1 year

New recreational infrastructure Peninsula Aquatic and Recreation Centre (PARC)

Use of PARC, MVPA (S)

Multivariable logistic regression

After 12 months 17,5% of sample reported occasional use of PARC and 8.7% used it on regular basis.

PARC users were not significantly more likely than non-users to show improvements in their level of leisure-time PA over 12 months.

Females used PARC more often (odds ratio 2.30, 95% CI: 1.37–3.87) than males.

0.82

Sun et al. 2014, China

Young adults,

N = 169 (56%)

10 months

(3 after intervention)

Increase in land use, street connectivity, and bus accessibility

Walking behavior (S)

Multivariable linear regression model

Intervention had positive effect on walking distance and walking ratio.

An increase in pedestrian network connectivity (road intersections) positively predicted walking distance (p < 0.001) and walking ratio (p < 0.001). An increased exposure to life area buildings and an increase in population density were associated with longer walks (p < 0.001). At the lower campus level, an increase in the number of work area buildings was associated with a decrease in subjects’ altitude ranges (p < 0.001), while the increased bus service resulted in more people frequenting higher elevation levels (p = 0.01), and a change in middle-class bus service was inversely associated with subjects’ predicted movement across hilly terrain (p = 0.02).

No gender specific effects.

0.82

Tannis et al. 2019, USA

Adults

N = 88 (78.4%)

1 year

Move into houses with active design (more attractive stairwells, outdoor community garden area, outdoor fitness area, community gym)

PA (S)

Steps per day (O)

T-tests and Mann–Whitney U test

The greater daily steps increase had AD residents who moved from an elevator building (p = 0.051).

Difference in MVPA between AD and non-AD residents was not significant.

AD building women reported more work-related MVPA overall (p = 0.01).

AD men engaged in more moderate recreational PA (p = 0.044).

0.91

Tester and Baker 2009, USA

Children and adults

N/A

1 year

Park improvements (playfields)

Park use and PA (O)

T test

Results presented separately for genders.

There were significant increases of the amount of playfield users among children and adults of both genders at the intervention parks (p < 0.05), but not in the control park.

There were statistically significant increases among males and females who were observed at each respective PA level (sedentary, moderate, vigorous) in the intervention parks (p<0.05).

On the control playfield, only moderately active males increased.

0.75

Wells and Yang 2008, USA

Adult women

N = 32 (100%)

4 years

Moving to neo-traditional neighbourhood

Overall walking (O)

Mixed modeling

(only females)

Women who moved to places with fewer cul-de-sac, on average, walked more (5303 more steps per week, or 757 more steps per day, p = 0.025).

Increases in land-use mix were associated with less walking (31,820 fewer steps per week, or 4545 fewer steps per day), p = 0.013.

0.86

West and Shores 2015, US

Adults

N = 273 (41.1%)

1 year

New greenway

Walking, MVPA (S)

Ordinary least squares regressions

No significant differences between the experimental and control groups in days of walking, moderate activity, or vigorous activity before and after the greenway was constructed.

No gender specific effects.

0.88