Fig. 2.

Structural equation models assessing the indirect effects of cognitive measures.

Models reveal the magnitude and strength of the indirect effect of cognitive measures on the relationship between UPSIT and the diagnostic categories, accounting for age and sex as covariates. Since the indirect effects between GENUN and HVLT in each model were non-significant, the associations between GENUN and HVLT were fixed to zero to give the models one degree of freedom. Significant and nonsignificant associations between study variables are depicted by connecting lines as described in the legend. The three sets of values above the lines between diagnostic categories and UPSIT convey the magnitude (and strength, as p-value), in order, of their direct association, the indirect association of the HVLT along the arrowed lines from the diagnostic category to HVLT to UPSIT, and, the indirect association of the second cognitive measure along the arrowed lines from the diagnostic category to that measure to UPSIT. Fit statistics and the variance (R²) in scores explained by the models are indicated. (A) Model assessing the indirect effects of the HVLT and MoCA. (B) Model assessing the indirect effects of the HVLT and the visuospatial-executive-functioning subscore of the MoCA. (C) Model assessing the indirect effects of the HVLT and the delayed-memory-recall subscore of the MoCA.

Model statistics: χ² = model chi-square, CFI = comparative fit index, RMSEA = root mean square error of approximation, SRMR = standard root mean square residual.

PPMI-defined diagnostic categories: HC = healthy controls; GENUN = individuals with asymptomatic genetic (LRRK2, SNCA, or GBA) PD; GENPD = individuals with symptomatic genetic PD; SPD = individuals with sporadic PD at baseline; PROD (possible prodromal PD) = individuals diagnosed with hyposmia and/or RBD.