Task | Cut-off score | Sensitivity to delirium (% of cases) | Delirium versus dementia | Delirium versus cognitively normal controls | ||||||
Specificity (% of cases) | + Likelihood ratio | − Likelihood ratio | AUC (95% CI) | Specificity: % of cases | + Likelihood ratio | − Likelihood ratio | AUC (95% CI) | |||
1 | 5‡ | 95.0 | 66.7 | 2.85 | 0.07 | 0.92** | 80.0 | 4.75 | 0.06 | 0.91** |
4† | 85.0 | 88.9 | 7.66 | 0.17 | (0.83 to 1.01) | 90.0 | 8.50 | 0.17 | (0.81 to 1.01) | |
3 | 55.0 | 94.4 | 9.82 | 0.48 | 90.0 | 5.50 | 0.50 | |||
2 | 5 | 100 | 55.6 | 2.25 | – | 0.94** | 61.1 | 2.57 | - | 1.00** |
4†‡ | 100 | 83.3 | 5.99 | – | (0.86 to 1.03) | 94.4 | 17.86 | - | (0.98 to 1.01) | |
3 | 83.3 | 88.9 | 7.50 | 0.19 | 100 | - | 0.17 | |||
3 | 5 | 94.1 | 66.7 | 2.83 | 0.09 | 0.93** | 78.9 | 4.46 | 0.07 | 0.95** |
4†‡ | 82.4 | 88.9 | 7.42 | 0.20 | (0.83 to 1.02) | 94.7 | 15.55 | 0.19 | (0.87 to 1.03) | |
3 | 76.5 | 94.4 | 13.66 | 0.25 | 100 | - | 0.24 | |||
4 | 5‡ | 93.3 | 41.2 | 1.59 | 0.16 | 0.80* | 83.3 | 5.59 | 0.08 | 0.92** |
4† | 80.0 | 70.6 | 2.72 | 0.28 | (0.65 to 0.96) | 88.9 | 7.21 | 0.22 | (0.83 to 1.02) | |
3 | 66.7 | 82.4 | 3.79 | 0.40 | 94.4 | 11.91 | 0.35 | |||
5 | 5 | 85.7 | 41.2 | 1.46 | 0.35 | 0.84* | 42.1 | 1.48 | 0.34 | 0.89** |
4‡ | 85.7 | 82.4 | 4.87 | 0.17 | (0.67 to 1.00) | 100 | - | 0.14 | (0.74 to 1.04) | |
3† | 78.6 | 94.1 | 13.32 | 0.23 | 100 | - | 0.21 | |||
6 | 5† | 100 | 56.3 | 2.29 | – | 0.88* | 61.1 | 2.57 | - | 0.94** |
4 | 75.0 | 75 | 3.00 | 0.33 | (0.76 to 1.01) | 88.9 | 6.76 | 0.28 | (0.86 to 1.02) | |
3‡ | 75.0 | 81.3 | 4.01 | 0.31 | 100 | - | 0.25 | |||
7 | 5 | 90.9 | 62.5 | 2.42 | 0.15 | 0.88* | 77.8 | 4.09 | 0.12 | 0.91** |
4† | 81.8 | 93.8 | 13.19 | 0.19 | (0.73-to 1.03) | 88.9 | 7.37 | 0.20 | (0.79 to 1.03) | |
3‡ | 72.7 | 93.8 | 11.73 | 0.29 | 94.4 | 12.98 | 0.29 | |||
8 | 5 | 100 | 66.7 | 3.00 | – | 0.94** | 66.7 | 3.00 | – | 0.98** |
4†‡ | 100 | 86.7 | 7.52 | – | (0.84 to 1.05) | 83.3 | 5.99 | – | (0.93 to 1.02) | |
3 | 81.8 | 93.3 | 12.21 | 0.20 | 94.4 | 14.61 | 0.19 |
Specificity values are shown separately for discriminating delirium from dementia and from cognitively normal patients. Optimal cut-off points (determined by the highest combined sensitivity and specificity value, with greater weight placed on sensitivity values in the case of tied scores) are also indicated for when discriminating delirium from dementia (†) or from cognitively normal patients (‡). Note that positive and negative likelihood ratios are not given for cut-off scores with a sensitivity or specificity value of 100%.
* p<0.01; **p<0.001.