From: Compbdt: an R program to compare two binary diagnostic tests subject to a paired design
Confidence intervals for the parameters of each diagnostic test (95% confidence) | ||
---|---|---|
Test 1 | Test 2 | |
Sensitivity | 79.550%; 85.582% | 88.857%; 93.380% |
Specificity | 68.853%; 79.436% | 69.665%; 80.145% |
Positive LR | 2.594; 3.931 | 2.942; 4.481 |
Negative LR | 0.195; 0.284 | 0.091; 0.154 |
Positive PV | 85.409%; 90.731% | 86.927%; 91.783% |
Negative PV | 59.388%; 70.180% | 73.403%; 83.570% |
Comparison of the parameters of the two diagnostic tests (α = 5%) | ||
Sensitivities | ||
McNemar test statisics: test statistic = 24.582, p ‐ value = 0 | ||
Exact test: p ‐ value = 0 | ||
95% Tango confidence interval for Se2 − Se1: 5.278%; 11.966 | ||
Specificities | ||
McNemar test: test statistic = 0.044, p ‐ value = 0.833 | ||
Exact test: two ‐ sided p ‐ value = 0.916 | ||
Likelihood ratios (Method of Leisenring et al. [21] and Pepe [1]) | ||
Positive LRs: test statistic = − 0.898, p ‐ value = 0.369 | ||
Negative LRs: test statistic = 4.663, p ‐ value = 0 95% confidence interval for NLR1/NLR2: 1.487; 2.644 | ||
Predictive values (Method of Leisenring et al. [13]) | ||
Positive PVs: test statistic = 0.802, p ‐ value = 0.371 | ||
Negative PVs: test statistic = 23.579, p ‐ value = 0 | ||
Predictive values (Method of Kosinski [14]) | ||
Positive PVs: test statistic = 0.807, p ‐ value = 0.369 | ||
Negative PVs: test statistic = 22.502, p ‐ value = 0 | ||
Relative predictive values (Method of Moskowitz and Pepe [22]) | ||
Positive PVs: test statistic = − 0.895, p ‐ value = 0.371 | ||
Negative PVs: test statistic = − 4.737, p ‐ value = 0 95% confidence interval for NPV1/NPV2: 0.762; 0.894 |