Table 2.
Score cutoff | Sensitivity | Specificity | Likelihood ratiob | Prevalence of 1% | Prevalence of 10% | Prevalence of 20% | |||
|
|
|
|
NPVc, % | Yieldd, % | NPV, % | Yield, % | NPV, % | Yield, % |
1 | 95.9 | 41.7 | 0.099 | 99.9 | 41.3 | 98.9 | 38.0 | 97.6 | 34.2 |
2 | 92.6 | 60.0 | 0.124 | 99.9 | 59.4 | 98.6 | 54.7 | 97.0 | 49.5 |
5 | 85.5 | 78.5 | 0.185 | 99.8 | 77.8 | 98.0 | 72.1 | 95.6 | 65.7 |
10 | 79.4 | 87.6 | 0.235 | 99.8 | 86.9 | 97.4 | 80.9 | 94.4 | 74.2 |
aThe maximum score was 100; a higher score indicates higher model prediction of COVID-19 positivity.
bThe likelihood ratio uses the equation for negative tests.
cNPV: negative predictive value.
dYield refers to diagnostic yield, which is the percentage of patients that can be ruled out (ie, those with a score below the cutoff).