Table 3 Analyses of data in an example by Sackett et al [8 (Table 4–10, pp. 95–98)].
From: New patient-oriented summary measure of net total gain in certainty for dichotomous diagnostic tests
| Â | Patient A: High prior suspicion of coronary disease | Patient B: Low prior suspicion of coronary disease | Patient C: Intermediate prior suspicion of coronary disease |
|---|---|---|---|
Prevalence | 90% | 5% | 50% |
A | 540 | 30 | 300 |
B | 9 | 86 | 45 |
C | 360 | 20 | 200 |
D | 91 | 864 | 455 |
PPV | 540/549 = 98.36% | 30/116 = 25.86% | 300/345 = 87% |
NPV | 91/451 = 20.18% | 864/884 = 97.74% | 455/655 = 69.47% |
FPR | 9/549 = 1.64% | 86/116 = 74.14% | 45/345 = 13.04% |
FNR | 360/451 = 79.82% | 20/884 = 2.26% | 200/655 = 30.53 |
PLR | 6.67 | 6.67 | 6.67 |
J = sensitivity+specificity-1 | 51.41% | 51.41% | 51.41% |
NND = 1/J | 1.95 | 1.95 | 1.95 |
PSI = Ψ = PPV+NPV-1 = = [PPV-Prevalence]+ [NPV-(1-Prevalence)] | 18.54% | 23.60% | 56.47% |
NNP = 1/PSI = 1/Ψ | 5.4 | 4.2 | 1.8 |