Skip to main content

Table 4 Summary of logistic regression models and AUROC analysis

From: Correlations between plasma and PET beta-amyloid levels in individuals with subjective cognitive decline: the Fundació ACE Healthy Brain Initiative (FACEHBI)

Characteristic

Model 1

Model 2a

Model 2b

Model 3

Model 4

OR (95% CI)

p

OR (95% CI)

p

OR (95% CI)

p

OR (95% CI)

p

OR (95% CI)

p

Age

1.091 (1.016–1.172)

0.017

1.114 (1.033–1.202)

0.005

1.097 (1.017–1.183)

0.017

1.113 (1.029–1.205)

0.008

APOE

7.319 (2.503–21.401)

2.8E-04

7.208 (2.431–21.373)

3.7E-4

Log TP42/40

0.131 (0.021–0.819)

0.030

0.126 (0.017–0.944)

0.044

0.133 (0.021–0.829)

0.031

AUROC

0.702

0.806

0.754

0.818

0.681

Youden’s index

0.42

0.54

0.49

0.55

0.42

Cut-off

0.101

0.147

0.115

0.092

0.081

Specificity

69.6

87.3

76.8

77.3

59.1

Sensitivity

72.2

66.7

72.2

77.8

83.3

PPV

19.1

34.3

23.6

25.5

16.7

NPV

96.2

96.3

96.5

97.2

97.2

  1. Logistic regression models were used to assess predictors of FBB-PET SUVR positivity (cut-off > 1.45) after adjustment by selected covariates in 199 participants
  2. Model 1 included only age as a predictor; model 2 requires a blood extraction (APOE ε4 carrier status (0–1) or log TP42/40, model 2a and 2b respectively); model 3 included age, APOE, and log TP42/40, and model 4 only included the target plasma biomarker log TP42/40
  3. The criterion for choosing the operating point along the ROC curve was Youden’s index maximum
  4. AUROC area under the receiver operating characteristic curve, CI confidence interval, NPV negative predictive value, OR odds ratio, PPV positive predictive value, TP total plasma