Cross-sectional and longitudinal associations between the built environment and walking: effect modification by socioeconomic status

Table 4 Longitudinal associations between walkability and walking, Alberta’s Tomorrow Project, Canada (n = 703), 2008–2015

	Overall sample	Overall sample			Highest education level attained		Household income^a
	Crude (unadjusted) (n = 703)	Adjusted (n = 703)	P-value for interaction term		High school or less (n = 149)	> High school (n = 554)	Low-income (n = 204)	Not low-income (n = 499)
	b (95% CI)	b (95% CI)	education income		b (95% CI)	b (95% CI)	b (95% CI)	b (95% CI)
Change in built environment characteristics (baseline to follow up)
Walkability (per 10-unit increase)	0.44 (− 20.21, 21.08)	0.41 (− 20.02, 20.84)	0.076	0.502	− 31.08 (− 71.78, 9.63)	11.52 (− 11.87, 34.90)
% Walkability	0.003* (0.00, 0.01)	0.003 (0.00, 0.01)	0.030	0.487	*0.01 (0.00, 0.01)**	0.00 (− 0.00, 0.00)
Walkability-change tertiles (REF = minimal change)
Decreased walkability	−9.89 (− 79.06, 59.29)	−9.03 (− 77.48, 59.41)	0.785	0.772
Increased walkability	13.11 (− 56.13, 82. 36)	12.60 (− 56.12, 81.31)	0.666	0.865

Note. Each model included only one of the built environment variables. Overall adjusted models included relationship status, children at home, season of survey completion, follow-up survey type, and years between surveys, as time-varying coefficients. Models with stratum-specific estimates were adjusted for the same covariates and were included in the table wherever the P value for the interaction term was less than 0.10 (model estimates from models with statistically significant interaction terms (P < 0.05) were bolded). Models testing for effect modification by baseline education and household low-income status were each run separately
^aHousehold low-income status was determined using the 2006 median income level for Alberta, estimated from the 2006 Statistics Canada Census
*P < 0.05

ISSN: 1471-2458