Higher median MD is related to higher risk of dementia while accounting for clinical markers
Cohort | Baseline marker predicting dementia conversion | ||||
β (SE) | P value | HR (95% CI) | Ng R2 | AUC | |
SCANS | 0.717 (0.181) | 7.1e-05 | 2.048 (1.438 to 2.918) | 0.206 | 0.794 |
RUN DMC | 0.310 (0.136) | 0.016 | 1.364 (1.060 to 1.755) | 0.165 | 0.825 |
HARMONISATION | 0.579 (0.253) | 0.023 | 1.784 (1.085 to 2.935) | 0.096 | 0.761 |
Change in DTI predicting dementia conversion | |||||
Cohort | β (SE) | P value | HR/OR (95% CI) | Ng R 2 / R 2 L | AUC |
SCANS | 0.951 (0.227) | 2.49e-05 | 2.588 (1.663 to 4.027) | 0.202 | 0.785 |
RUN DMC | −0.068 (0.305) | 0.825 | 0.935 (0.498 to 1.667) | 0.248 | 0.891 |
HARMONISATION | 0.453 (0.237) | 0.056 | 1.573 (0.998 to 2.597) | 0.109 | 0.738 |
Values show standardised regression coefficients β (SE) for the predictor variables in the Cox regression or logistic regression models of dementia conversion. P-value < 0.05 are marked in bold. The HR or OR together with the CI are shown. The model parameters Nagelkerke’s R 2 (Ng R2) or Hosmer and Lemeshow’s R2 (R2 L) give an estimated amount of variation in the dependent variable explained by the model. The AUC evaluates how well the model classifies dementia conversion vs. no-dementia conversion at all possible cutoffs respectively. All regression models control for the effects of age, sex and NART-IQ or education.
Statistical significance p<0.05.
AUC, area under the curve; DTI, diffusion tensor imaging; MD, mean diffusivity; Ng R2, Nagelkerke R2; R2 L, Hosmer and Lemeshow’s R2.