Table 4. Relationships between total PA and BE attributes across included studies.

BE attributes	PA-BE relationships			% studies supporting the predicted association ab	Summary codes c
	Significant (+)	Significant (-)	Non-significant
Residential density	N/A	N/A	[12]	0/1 = 0%	0
Leisure amenities availability	N/A	N/A	[12]	0/1 = 0%	0
Recreational facilities proximity	[53]	N/A	N/A	1/1 = 100%	+
Transit stops proximity	[12]	[53]	[54]	1/3 = 33%	00
Services proximity	N/A	[54]	[12,33]	1/3 = 33%	00
Aesthetics	N/A	N/A	[12,53]	0/2 = 0%	0
Infrastructure for cycling/walking	[55] d	N/A	N/A	1/1 = 100%	+ d
Street connectivity	N/A	[12] e	N/A	1/1 = 100%	− e
Sidewalks	[33]	N/A	[12,53]	1/3 = 33%	00
Terrain slope	[33]	N/A		1/1 = 100%	+
Bike path availability	[33]	N/A	[12]	1/2 = 50%	?
Crime safety during day	[33,53]	[52]	[12]	2/4 = 50%	?
Crime safety at night	[33,52,53]	N/A	[12]	3/4 = 75%	++
Traffic safety	N/A	[12]	[33,53]	1/3 = 33%	00
Urbanicity	[56–58]d	N/A	N/A	3/3 = 100%	++ d

^a Number of articles supporting the predicted association divided by the entire number of studies investigated each BE variable.

^b Studies investigating variables signed ^d or ^e was not considered in the coding unless all studies solely investigated the same variable.

^c The principles of evidence summary coding were adapted from Sallis et al [32]; +/- = positive or negative association (60–100% of articles supporting the predicted association); 0 = No relationship (0–33% of articles supporting the predicted association); ? = inconsistent relationship (34–59% of articles supporting the predicted association). Single signed codes (+, - or 0) were given for BE variables that were investigated only by 1–2 studies with respect to certain PA domains; When relationships were investigated in (3–4) or > 4 studies, double (++, - or 00) and triple (+++, - or 000) signed summary coding was applied, respectively.

^d association with physical inactivity.

^e reduced/lack of access.