Performance of the three models and self-reported diagnosis of gout
| Sensitivity | Specificity | Positive LR | Negative LR | Correctly classified | AUC | |
|---|---|---|---|---|---|---|
| First logistic regression model | 88.0% (79.6% to 93.4%) | 93.0% (87.0% to 96.4%) | 12.5 (6.8 to 22.8) | 0.13 (0.08 to 0.22) | 90.0% (86.4% to 94.1%) | 0.97 (0.95 to 0.99) |
| Second logistic regression model | 87.5% (78.8% to 93.1%) | 89.8% (82.9% to 94.3%) | 8.6 (5.1 to 14.5) | 0.14 (0.08 to 0.24) | 88.8% (83.8% to 92.5%) | 0.95 (0.92 to 0.98) |
| CART | 81.3% (72.2% to 88.1%) | 93.7% (88.0% to 96.9%) | 12.8 (6.8 to 24.3) | 0.20 (0.13 to 0.30) | 88.5% (83.7% to 92.1%) | 0.90 (0.86 to 0.94) |
| Self-reported gout | 92.9% (85.5% to 96.9%) | 81.7% (74.1% to 87.5%) | 5.1 (3.6 to 7.2) | 0.09 (0.04 to 0.18) | 86.3% (81.2% to 90.3%) | 0.87 (0.83 to 0.91) |
CIs are in brackets.
AUC, area under the receiver-operating characteristic curve; CART, classification and regression tree; LR, likelihood ratio.