Table 2.

Results of logistic regression analysis for prediction of abnormal CSF biomarkers across all three centers

Model/predictors	OR	(95% CI)		p value	R 2	ΔR2	OR	(95% CI)		p value	R 2	ΔR2
Model variables	Abnormal CSF-Aβ42 (< 813 pg/ml)						Abnormal CSF-tau (> 375 pg/ml)
Step 1: covariates
Age (per 1-year increase)	1.09	1.03–1.15		0.003	0.080	0.080	1.18	1.09–1.28		< 0.001	0.187	0.187
Positive APOE ε4 status	2.34	1.10–5.00		0.028	0.126	0.046	6.20	2371–16.2		< 0.001	0.329	0.142
Step 2: center (ADC = reference)				0.001	0.271	0.145				0.562	0.339	0.010
DELCODE	6.54	2.14–20.0		0.001			0.841	0.241–2.94		0.786
IDIBAPS	5.78	2.14–15.6		0.001			1.53	0.476–4.89		0.477
Model variables	Abnormal CSF-ptau181 (> 53 pg/ml)						Abnormal Aβ42 and abnormal ptau181
Step 1: covariates
Age (per 1-year increase)	Not selected						1.15	1.05–1.26		0.002	0.117	0.117
Positive APOE ε4 status	2.66	1.28–5.54		0.009	0.069	0.069	5.23	1.80–15.2		0.002	0.234	0.117
Step 2: center (ADC = reference)				0.728	0.075	0.006				0.393	0.255	0.021
DELCODE	0.983	0.401–2.41		0.970			1.68	0.374–7.54		0.499
IDIBAPS	1352	0.574–3.19		0.490			2.60	0.643–10.5		0.180

R² and ΔR² give an estimate of explained variance of the full model and each predictor’s contribution of explained variance within the model, respectively

No values are reported for nonsignificant predictor variables in step 1 of the models, as we used a forward selection process at this step

Aβ amyloid beta, ADC Amsterdam Dementia Cohort, APOE apolipoprotein E, CI confidence interval, CSF cerebrospinal fluid, DELCODE German Center for Neurodegenerative Diseases (DZNE), IDIBAPS l’Institut d’Investigacions Biomèdiques August Pi i Sunyer Hospital Clinic Barcelona, OR odds ratio, ptau181 tau phosphorylated at position 181