Skip to main content
. 2024 Feb 27;14:4752. doi: 10.1038/s41598-024-54900-5

Figure 3.

Figure 3

Clusters of predicted risk, with age fixed, unadjusted (A), adjusted for alcohol (B), body mass index (C) and multivariable adjusted at follow-up 1 (2009–2012). Getis-Ord Gi* statistic for the individuals geo-referenced at their postal address. White dots correspond to individual with values that do not exhibit spatial dependence. Red/Blue dots indicate individuals with a positive/negative Z score considering a p value < 0.05 (α = 0.05). This means that high or low values clusters, respectively, within a spatial lag of 600 m, are found closer together than expected if the underlying spatial process was random. The multivariable model (MGWR) included body mass index, alcohol consumption, living in couple, low education status, Townsend index, Swiss origin, anxiety disorder and major depressive disorder, a polygenic risk score and at follow-up 1 (2009–2012) a Mediterranean diet score, and daily moderate intensity physical activity.