Table 2.
Screening accuracy of the 5–2-1 screening criteria in identifying patients with advanced Parkinson’s disease
| Advanced PD | Adjusted Model a | ||||
|---|---|---|---|---|---|
| Indicator | No | Yes | OR (95% CI) | Correct Classification (%) b | AUC c |
| 5–2-1 screening criteria | |||||
| Negative | 3018 | 150 | 1 | – | – |
| Positive | 994 | 552 | 6.84 (5.50, 8.51) | 88.1 | 0.89 |
| Individual clinical indicators | |||||
| ≥2 h off-time/day | |||||
| Negative | 3299 | 208 | 1 | – | – |
| Positive | 713 | 494 | 7.07 (5.76, 8.68) | 88.4 | 0.89 |
| ≥1 h troublesome dyskinesia/day | |||||
| Negative | 3944 | 591 | 1 | – | – |
| Positive | 68 | 111 | 5.56 (3.85, 8.02) | 87.1 | 0.85 |
| ≥5 doses of oral levodopa/day | |||||
| Negative | 3501 | 413 | 1 | – | – |
| Positive | 511 | 289 | 3.03 (2.45, 3.73) | 87.1 | 0.86 |
aRegressions adjusted for age, sex, time since diagnosis, Charlson Comorbidity Index, and country. bCorrect classification is the percentage of patients correctly classified per 5–2-1 criteria (sum of true positive and true negatives divided by total number of patients). C AUC is a screening accuracy measure that balances sensitivity and specificity. AUC is interpreted as follows: AUC = 0.5, non-informative; AUC = 0.5 to ≤0.7, less accurate; AUC = 0.7 to ≤0.9, moderately accurate, AUC 0.9 to < 1, highly accurate; AUC = 1, perfect test [27–29]
AUC Area under the curve, CI Confidence interval, OR Odds ratio