Objective  To examine the relationship between exposure to residential green space and the incidence of cardiovascular disease (CVD) among rural residents in Xinjiang Production and Construction Corps (Xinjiang).

  Methods  Totally 12 813 rural residents aged ≥ 18 years and living in local regions at least one year were recruited with typical sampling in Xinjiang for a face-to-face baseline questionnaire survey during September 2016 and 9 659 of them without CVD and with complete residence information at the baseline survey were followed up four times in April 2019, June 2020, July 2021, and June 2022. Generalized linear model (GLMM) and Cox proportional hazards regression model were used to analyze the relationship between exposure to residential green space (measured with normalized difference vegetation index NDVI for the region between a participant′s residential location and the center of the residential area) and CVD incidence among the participants. Subgroup analysis was performed to explore potential impact of each stratification factor on the relationship between exposure to residential green space and CVD incidence risk.

  Results  Of the 9 659 cohort participants, 9 004 were followed-up and 655 (6.78%) were lost to follow-up by the end of the study (June 2022). During the follow-up of 49 565.76 person-years (PYs) totally and 5.50 ± 1.09 PYs averagely among the 9 004 participants finally included in the analysis, overall 762 CVD cases were identified; the CVD incidence density was 153.73/10 000 PYs and the CVD cumulative incidence was 8.46%. The mean values of NDVI_{250 m}, NDVI_{500 m}, and NDVI_{1 000 m} were 0.229 ± 0.068, 0.265 ± 0.067, and 0.298 ± 0.060 for the followed-up participants with CVD incidence and 0.242 ± 0.063, 0.278 ± 0.060, and 0.308 ± 0.053 for those without CVD, respectively, with significant differences in the NDVI values between the two groups (all P < 0.001). After adjusting for gender, age, education, marital status, physical activity and particulate matter less than 2.5 microns in aerodynamic diameter (PM_2.5), the results of GLMM model analysis revealed the β coefficients of – 0.200, – 0.169, and – 0.200 for CVD incidence associated with each quartile increase in the values of NDVI_{250 m}, NDVI_{500 m}, and NDVI_{1 000 m}. The results of Cox proportional hazards regression model analysis also showed that each quartile increase in the values of NDVI_{250 m}, NDVI_{500 m}, and NDVI_{1 000 m} was associated with 14.4% (hazard risk HR = 0.856, 95% confidence interval 95%CI: 0.798 – 0.919), 14.3% (HR = 0.857, 95%CI: 0.800 – 0.918), and 16.7% (HR = 0.833, 95%CI: 0.755 – 0.896) reduction in CVD incidence risk. After adjusting for gender, age, education, marital status, physical activity and PM_2.5 (except for when used as a stratifica-tion variable), subgroup analysis results further demonstrated that each quartile increase in the values of NDVI_{250 m}, NDVI_{500 m}, and NDVI_{1 000 m} was associated with much higher reduction of 20.7% (HR = 0.793, 95%CI: 0.720 – 0.874), 20.1% (HR = 0.797, 95%CI: 0.726 – 0.875) and 24.1% (HR = 0.759, 95%CI: 0.688 – 0.837) in CVD incidence risk for the followed participants aged 50 years and younger; and for the followed participants with higher intensity of physical activity, each quartile increase in the values of NDVI_{250 m} was associated with a 28.9% (HR = 0.711, 95%CI: 0.613 – 0.825) reduction in CVD incidence risk.

  Conclusion  Higher exposure to residential green space could be associated with reduced CVD incidence risk among rural residents in Xinjiang.