. 2020 Aug 25;128(8):087005. doi: 10.1289/EHP6161

Table 4.

Sensitivity analysis for associations between urban green spaces within $250 m$ of study participants’ postal code residence and cardiovascular incidence, readmission, and mortality in four population-based cohorts of urban residents in Ontario, Canada, 2000–2014.

Outcome	Number of events	Further adjusted for smoking and education^c	Further adjusted for access to primary care^d
Outcome	Number of events	Hazard ratio (95% CI)	Hazard ratio (95% CI)
Full cohort^a
Mortality
CVD	114,208	0.91 (0.89, 0.94)	0.89 (0.87, 0.91)
Nonaccidental	330,560	0.91 (0.89, 0.93)	0.89 (0.88, 0.90)
Incidence cohort^a
Mortality
CVD	88,263	0.92 (0.89, 0.95)	0.90 (0.88, 0.92)
Nonaccidental	277,174	0.91 (0.89, 0.93)	0.89 (0.88, 0.90)
Incidence
AMI	58,553	0.95 (0.92, 0.98)	0.94 (0.92, 0.96)
HF	134,655	0.96 (0.93, 0.98)	0.95 (0.93, 0.96)
AMI cohort^b
Readmission: CVD	4,419	—	1.03 (0.98, 1.07)
Mortality
CVD	2,788	—	0.99 (0.93, 1.04)
Nonaccidental	5,463	—	0.99 (0.95, 1.03)
HF cohort^b
Readmission: CVD	5,482	—	1.00 (0.95, 1.05)
Mortality
CVD	4,981	—	0.98 (0.93, 1.03)
Nonaccidental	9,151	—	0.98 (0.94, 1.02)

Note: —, no data; AMI, acute myocardial infarction; CI, confidence interval; CVD: cardiovascular disease; HF, heart failure.

^{^a}

Two-level nested, spatial random-effects Cox proportional hazards model (level one: census division, level two: census tract). Hazard ratios were scaled to an interquartile increase in NDVI (full and incidence cohorts: $IQR = 0.17$ ; AMI cohort: $IQR = 0.12$ ; HF cohort: $IQR = 0.13$ ). The fully adjusted model included age, sex, region (lived or not in the Greater Toronto Area), area-level unemployment, percent less than high school education, percent recent immigrants, and household income (quintiles), and population density.

^{^b}

For the AMI and HF cohorts, the models further adjusted for clinical severity, in-hospital care, medications at discharge, smoking, and individual-level SES variables (AMI cohort: $IQR = 0.12$ ; HF cohort: $IQR = 0.13$ ).

^{^c}

Indirect adjustment of smoking was conducted using data from the 2000/2001, 2003, and 2005 cycles of Canadian Community Health Survey. This was not applicable to AMI and HF cohort.

^{^d}

Access to primary care was derived using the density of family physicians. This sensitivity analysis was based on the main analysis presented in Figure 1.