Evaluation Of Causal Relationship Between Vegetation Cover And Stroke Incidence Based On The Difference-in-differences Approach

Objective:To explore the causal relationship between vegetation cover and the incidence of stroke in residents.Method:Relying on the disease surveillance system of Shandong Province to collect stroke incidence surveillance data from 20 counties and districts（267 township-level units）from 2013 to 2020.Vegetation coverage（Normalized Difference Vegetation Index（NDVI）,Soil-adjusted Vegetation Index（SAVI）,Enhanced Vegetation Index（EVI））of patients was obtained by satellite remote sensing inversion technique using the usual address of stroke patients as exposure indices.The association between vegetation cover and stroke incidence was first assessed using a generalized linear mixed model（GLMM）,followed by a difference in differences model（DID）to assess the causal association between vegetation cover and stroke incidence.In addition,the study used stratified analysis to examine the effect of vegetation cover on the risk of stroke in different age groups and by gender.Results:A total of 539662 stroke patients were selected in the database,and then their temporal and spatial descriptive analysis was carried out.Vegetation cover is based on Enhanced Vegetation Index（EVI）,Soil-Adjusted Vegetation Index（SAVI）and Normalized Difference Vegetation Index（NDVI）.The rationality of the results is further verified by comparing the effect values of the three vegetation indices.The GLMM results showed a negative correlation between each 0.1 unit increment of NDVI,EVI,and SAVI and stroke incidence,with RR values of 0.882（0.785-0.991）,0.917（0.830-0.995）and 0.856（0.724-0.986）respectively.DID results showed a protective causal effect of both EVI and SAVI on stroke incidence in the total population,with RR values of 0.530（0.470-0.598）and 0.545（0.496-0.600）for stroke incidence per 0.1 unit increment of EVI and SAVI,respectively.NDVI demonstrated a significant protective effect only for the female group and the≥65 years old group due to the influence of soil and atmospheric aerosols.Further analysis of stroke incidence typing revealed that all three indicators of vegetation cover had a protective effect on the incidence of ischaemic stroke,with RR values of 0.529（0.460-0.610）,0.489（0.439-0.545）and 0.786（0.708-0.872）per 0.1 unit increment of EVI,SAVI and NDVI on the incidence of ischaemic stroke,respectively.This study is limited to the basic assumption that there are no other confounding factors that show different spatiotemporal changes related to exposure except seasonal temperature and PM_2.5,so the results need to be interpreted carefully.Conclusion:This study verifies that vegetation cover has a protective causal effect on the incidence of stroke,which is very important for scientifically understanding the impact of vegetation cover on the incidence of stroke and providing scientific information for policymakers.