Fig. 2. Comparisons between AD/LB groups and the independent effects of LB, Aβ and tau pathologies on clinical outcomes.

a–l, Analyses were performed using linear regression models with AD/LB groups (a–c,g–i) or all three binarized pathologies (d–f,j–l) as independent variables in the same model, adjusted for age, sex, education (for cognitive outcomes) and cognitive stage (MCI/dementia). In g,h,j,k, logistic regression models with the same covariates were used because the outcomes were binary. Outcomes were z scored (according to the distribution in Aβ-negative controls) cognitive tests (a–f) and motor questionnaires (i,l) or binary assessment of correct visuospatial task (g,j) or presence of hallucinations (h,k). Boxes (a–c,g–i) show interquartile range, the horizontal lines are medians and the whiskers were plotted using the Tukey method. In d–f,l, the dot/center shows the estimate of the pathology and the error bars show the 95% CI, where negative values equal worse performance. In j,k, the dot/center represents ORs, where values <1 equal a decrease, and error bars show the 95% CI of the ORs. Worse performance is marked in red. AD positivity was defined as the presence of both Aβ and T. LB positivity was defined as the presence of an abnormal α-syn SAA result. The effect of LB pathology on clinical outcome with/without adjusting for Aβ and T is shown in Extended Data Table 2. Overall, 302 participants were AD−/LB−, 106 were AD−/LB+, 377 were AD+/LB− and 98 were AD+/LB+ and 204 were LB+, 607 were Aβ+ and 489 were tau+. The statistical analyses with corrections for multiple comparisons are shown in Supplementary Fig. 1. Missing data are shown in Supplementary Table 1. *P < 0.05; **P < 0.01; ***P < 0.001 (two-sided).