Table 1 |
||||
| Definitions of test and performance characteristics | ||||
| Gold standard | ||||
| Yes | No | |||
| Yes | X | Y | ||
| No | Z | K | ||
| Test/Performance Characteristic | Definition | Explanation | ||
| Patient Classification, % | ||||
| True-Positive | X / (No. of Patients) × 100% | |||
| False-Positive | Y / (No. of Patients) × 100% | |||
| True-Negative | K / (No. of Patients) × 100% | |||
| False-Negative | Z / (No. of Patients) × 100% | |||
| Sensitivity, % | X / (X+Z) × 100% | % of FN hospitalizations correctly identified by claims-based definition | ||
| Specificity, % | K / (Y+K) × 100% | % of non-FN hospitalizations correctly identified by claims-based definition | ||
| Positive Predictive Value (PPV), % | X / (X+Y) × 100% | % of hosp. identified as FN hosp. by claims-based definition that were true-positives | ||
| Negative Predictive Value (NPV), % | K / (Z+K) × 100% | % of hosp. identified as non-FN hosp. by claims-based definition that were true-negatives | ||
| Accuracy, % | (X+K) / (No. of Patients) × 100% | % of all hospitalizations that were true-positives and true-negatives | ||
| Likelihood ratio positive (LR+) | [X/(X+Z)] / [1-K/(Y+K)] | Odds of hosp. being an FN hosp. when claims-based algorithm is positive | ||
| Likelihood ratio negative (LR-) | [1-X/(X+Z)] / [K/(Y+K)] | Odds of hosp. being a non-FN hosp. when claims-based algorithm is negative | ||
FN: febrile neutropenia.
Weycker et al. BMC Health Services Research 2013 13:60 doi:10.1186/1472-6963-13-60
