Skip to main content
. 2017 Apr 4;12(4):e0174451. doi: 10.1371/journal.pone.0174451

Table 9. DOC-Cog results.

Diagnostic Characteristics Single cut-point (raw scores) Single cut-point (regression) Two cut-points (raw scores) Two cut-points (regression)
AUC 0.776 0.814 0.776 0.814
High-risk Raw score 0–8 POI e > .122 Raw score 0–5 POI e > .26
% of population 180 (47%) 180 (47%) 28 (7%) 58 (15%)
Impaired 44 (24%) 47 (26%) 12 (41%) 25(42%)
Not Impaired 136 (76%) 133 (74%) 16 (59%) 33 (58%)
Specificity 59% 61% 95% 91%
+LR a 2 2.2 4.7 4.8
PPV b 24% 26% 43% 43%
Intermediate-risk N/A f N/A f Raw score 6–9 POI e .08-.26
% of population N/A f N/A f 254 (66%) 178 (46%)
Impaired N/A f N/A f 41 (16%) 26 (15%)
Not Impaired N/A f N/A f 213 (84%) 152 (85%)
Specificity/Sensitivity N/A f N/A f N/A g N/A g
+/-LR ac N/A f N/A f N/A g N/A g
PPV/NPV bd N/A f N/A f N/A g N/A g
Low-risk Raw score 9–10 POI e < .122 Raw score 10 POI e < .08
% of population 207 (53%) 207 (53%) 105 (27%) 151 (39%)
Impaired 9 (4%) 6 (3%) 0 (0%) 2 (1%)
Not Impaired 198 (96%) 201 (97%) 105 (100%) 149 (99%)
Sensitivity 83% 88% 100% 96%
-LR c 0.3 0.2 1.5 1.7
NPV d 96% 97% 100% 99%

a +LR = positive likelihood ratio,

b PPV = positive predictive value,

c -LR = negative likelihood ratio,

d NPV = negative predictive value,

e POI = probability of impairment,

f N/A, no intermediate group using single cut-point method,

g cannot calculate statistics for intermediate group