Fig. 1: Mean Treatment Usage Frequencies and Regional Excess Suicide Deaths across 2016–2020 [Both sexes].

The Y-axis depicts the regional mean of min-max normalized year-wise differences in suicide deaths per 100,000 inhabitants between adolescence and young adulthood. The X-axis depicts regional mean values of min-max normalized clozapine, ECT and lithium treatment usage frequencies across 2016–2020 values. The slope and confidence intervals (CI:s) of the robust linear regression model contrasting these two variables are depicted as a blue line (slope) with grey shading (CI:s). Region population in relation to the national population is illustrated by the circle diameter, and counties affiliated with medical universities are highlighted in blue. These include Skåne region with Lund University, Stockholm region with Karolinska Institutet, Uppsala region with Uppsala University, Västerbotten region with Umeå University, Västra Götaland region with Gothenburg University/Sahlgrenska Academy, and Östergötland region with Linköping University. The figure demonstrates that regional mean of min-max normalized clozapine, ECT and lithium usage frequencies across 2016–2020 are inversely correlated with regional excess adolescent suicide deaths in the combined sexes group (β = –0.613, p-value = 0.0003, multiple R-squared: 0.123, adjusted R-squared: 0.077, 95% CI: –0.338, –0.889). Abbreviations: 95% CI, 95% confidence interval; ECT, electroconvulsive therapy; U.Affil, medical university affiliation (regions affiliated to medical universities are coloured in blue and regions unaffiliated to medical universities are coloured in black); W, weights (regional population size expressed as a percentage of the total national population).