Spatial analysis of 10-year predicted risk and incident cardiovascular atherosclerotic diseases in a swiss urban population

Abstract Funding Acknowledgements Type of funding sources: Private grant(s) and/or Sponsorship. Main funding source(s): Swiss National Science Foundation GlaxoSmithKline. Introduction Whether cardiovascular risk score geographically aggregates and informs on spatial development of atherosclerotic cardiovascular disease (ASCVD) remains unknown. Aims We first determined the spatial distribution of cardiovascular risk predicted in an urban population. Second, we compared risk maps with clustering of incident ASCVD over a 10-year follow-up. Sensitivity analyses were conducted to identify explanatory risk factors for both clusters of predicted risk and observed ASCVD. Methods Using data from a population-based cohort study, we computed the Systematic Coronary Risk Evaluation Score 2 (SCORE2 and SCORE2-OP) in participants free from ASCVD at baseline (2003-2006). Geographical distributions of predicted risk and incident ASCVD were determined using global and local spatial autocorrelation statistics. Different models were tested to identify explanatory variables (alcohol consumption, body mass index, socio-economic status, anxiety disorder, major depressive disorder, Mediterranean diet, sport activity, polygenic risk score, mean noise, concentration of PM2.5, normalized difference vegetation index, land surface temperature). Results 6203 individuals (56% women, mean age 52.5 ± SD 10.7) with a median follow-up of 10 years (IQR, 6-10) were included in the analyses. Over the period, there was persistent geographical clustering of predicted risk and ASCVD (Figures 1 and 2). Predicted risk and incident ASCVD marginally overlapped spatially. Body-mass index (BMI) and alcohol consumption explained most of the spatial distribution of predicted risk, reducing high-risk clusters from 286 to 5 and low-risk clusters from 391 to 44. For ASCVD, high clusters either persisted or were reinforced depending upon locations, after adjustment for potential risk factors, with an increase from 100 to 118. Low clusters reduced from 293 to 76 after the adjustment. Incidence rate of ASCVD was 2.5% (IC95%, 3.7-2.4%]) in high-risk clusters compared to the rest of the population. Conclusions Using a population-based cohort, we showed that in an urban area there exists clusters of high and low incidence of ASCVD. Geographical distribution of clusters of high clinical risk score was not congruent with that of incident ASCVD, limiting the use of clinical risk scores in identifying areas where ASCVD may develop.