Observed Verified Data | Imputed Data | |||||||
---|---|---|---|---|---|---|---|---|
Cutoff TICS Score | Sensitivity, % | Specificity, % | PPV, % | NPV, % | Mean Sensitivity, % | Mean Specificity, % | Mean PPV, % | Mean NPV, % |
≤18 | 7.7 | 100 | 100 | 96.3 | 5.0 | 100 | 100 | 97.6 |
≤19 | 15.4 | 100 | 100 | 96.6 | 10.0 | 100 | 100 | 97.7 |
≤20 | 15.4 | 100 | 100 | 96.6 | 10.0 | 100 | 100 | 97.7 |
≤21 | 15.4 | 99.4 | 50.0 | 96.5 | 10.0 | 99.7 | 50.0 | 97.7 |
≤22 | 23.1 | 98.4 | 37.5 | 96.8 | 15.1 | 99.4 | 37.5 | 97.9 |
≤23 | 30.8 | 98.1 | 40.0 | 97.1 | 20.1 | 99.2 | 40.0 | 98.0 |
≤24 | 38.5 | 96.4 | 31.3 | 97.4 | 25.1 | 98.6 | 31.3 | 98.1 |
≤25 | 46.2 | 94.5 | 26.1 | 97.7 | 30.1 | 97.8 | 26.1 | 98.2 |
≤26 | 46.2 | 92.6 | 20.7 | 97.6 | 30.1 | 97.1 | 20.7 | 98.2 |
≤27 | 69.2 | 89.6 | 22.0 | 98.6 | 45.3 | 95.8 | 21.5 | 98.5 |
≤28 | 76.9 | 81.2 | 14.7 | 98.8 | 50.3 | 92.5 | 14.5 | 98.6 |
≤29 | 100 | 69.3 | 12.0 | 100 | 65.4 | 87.8 | 11.9 | 99.0 |
≤30 | 100 | 54.4 | 8.4 | 100 | 65.4 | 82.0 | 8.4 | 98.9 |
≤31 | 100 | 47.6 | 7.4 | 100 | 68.3 | 74.4 | 6.3 | 98.9 |
≤32 | 100 | 42.7 | 6.8 | 100 | 73.3 | 64.0 | 4.9 | 98.9 |
≤33 | 100 | 35.0 | 6.1 | 100 | 77.0 | 53.6 | 4.0 | 98.9 |
≤34 | 100 | 25.9 | 5.4 | 100 | 82.8 | 39.7 | 3.4 | 98.9 |
≤35 | 100 | 18.1 | 4.9 | 100 | 87.6 | 28.4 | 3.0 | 98.9 |
≤36 | 100 | 10.0 | 4.5 | 100 | 93.5 | 15.1 | 2.7 | 98.9 |
≤37 | 100 | 5.2 | 4.2 | 100 | 97.6 | 6.1 | 2.6 | 99.0 |
≤38 | 100 | 3.2 | 4.2 | 100 | 98.7 | 3.5 | 2.6 | 99.0 |
≤39 | 100 | 1.9 | 4.1 | 100 | 99.6 | 1.5 | 2.5 | 99.3 |
≤40 | 100 | 1.0 | 4.1 | 100 | 100 | 0.5 | 2.5 | 100 |
≤41 | 100 | 0 | 4.0 | 100 | 100 | 0 | 2.5 | 100 |
NPV = negative predictive value; PPV = positive predictive value; TICS = Telephone Interview for Cognitive Status.
Note: Sensitivity, specificity, PPV, and NPV presented at cutoff scores between 18-41 of the observed verified data are listed in the left columns, and the total population with additional imputed data for individuals with missing dementia status are listed in the right columns. In the study, dementia status was only verified in a subsample (n = 168) of all the participants with a TICS score >30 (n = 655), for feasibility reasons. Data imputation was performed as a sensitivity analysis, wherein we imputed dementia status for the 487 individuals with a TICS score >30 without verified dementia status.