Predicting the geographical distribution of drug use disorder in Sweden from the geographical variation in social deprivation, genetic risk and urbanization

Drug use disorder (DUD) is distributed unevenly over geographical areas of Sweden. We explore, using geographically weighted regression, the proportion of this variation associated with community levels of genetic risk, social deprivation and urbanicity. In the 5983 Demographic Statistical areas (DeSOs) within Sweden, we predict DeSO prevalence from mean genetic risk for DUD, measured by the family genetic risk score (FGRS DUD ), social deprivation (SD) and degree of urbanization. We first modeled results for genetic risk and SD only finding, in all of Sweden, the proportions of predicted variance (±95% CIs) in DeSO levels of DUD associated with genetic risk and SD were, respectively, 0.58 (0.58–0.59) and 0.42 (0.41–0.42). Similar results were obtained in the major cities: Stockholm, Gothenburg, and Malmö. Genetic effects were more predictive in males while levels of SD were more predictive in females. In maps of FGRS DUD and SD prediction of DUD levels in DeSOs, we found results consistent with prior geographical patterns known in Sweden. Adding urbanicity to our model, only modestly predicted DeSO levels of DUD. To explore the potential origins of geographic variation in FGRS DUD , we showed that young adults with high FGRS DUD preferentially migrated to areas in Sweden with high levels of DUD. Our findings provide an initial tentative step toward clarifying the role of genetic and social-demographic factors in the geographical distribution of DUD within Sweden.