Table 3.
Diagnostic Power Comparisons Between Different Cutoffs in MCI and Dementia Subjects
Sensitivity (95%CI) | Specificity (95%CI) | AUC (95%CI) | p-value | |
---|---|---|---|---|
MCI vs normal | 0.000*,a | |||
Single cutoff (21/22) | 0.861 (0.786, 0.917) | 0.723 (0.598, 0.827) | 0.792 (0.719, 0.865) | |
7th percentile | 0.287 (0.209, 0.376) | 0.969 (0.893, 0.996) | 0.628 (0.549, 0.707) | |
16th percentile | 0.410 (0.322, 0.503) | 0.908 (0.810, 0.965) | 0.659 (0.581, 0.737) | |
Dementia vs normal | <0.000* | |||
Single cutoff (18/19) | 0.922 (0.861, 0.962) | 0.923 (0.830, 0.975) | 0.922 (0.876, 0.969) | |
2nd percentile | 0.359 (0.276, 0.442) | 0.985 (0.917, 1.000) | 0.672 (0.598, 0.746) | |
Cognitive Impairmentb vs normal | 0.000*,a | |||
Single cutoff (21/22) | 0.932 (0.893, 0.960) | 0.723 (0.598, 0.827) | 0.828 (0.760, 0.895) | |
7th percentile | 0.496 (0.432, 0.560) | 0.969 (0.893, 0.996) | 0.733 (0.675, 0.790) | |
16th percentile | 0.572 (0.508, 0.634) | 0.908 (0.810, 0.965) | 0.740 (0.680, 0.800) |
Notes: *Significant at 0.05. aSingle cutoff is significantly higher than both 7th and 16th percentiles. bCognitive impairment (MCI + dementia).
Abbreviations: AUC, area under curve; MCI, mild cognitive impairment.