Skip to main content
. Author manuscript; available in PMC: 2022 Sep 27.
Published in final edited form as: Biostat Epidemiol. 2021 Sep 27;5(2):267–286. doi: 10.1080/24709360.2021.1975255

Table 2.

Power and medical costs using AWA, Accuracy or AUC in comparing two diagnostic tests in two independent samples. The biomarker values from the two diagnostic tests in the infected population are distributed as N(μ1=0.6,σ1=1) and N(μ2=0.5,σ2=0.5), respectively.

Method Sample Size (N1, N2=500) Relative Importance (r) Prevalence Range (p) Prevalence rate Threshold (Dx1) Sensitivity (Dx1) Specificity (Dx1) AWA, Accuracy or AUC (Dx1) SD (Dx1) Threshold (Dx2) Sensitivity (Dx2) Specificity (Dx2) AWA, Accuracy or AUC (Dx2) SD (Dx2) SD (Dx1-Dx2) Power Medical Cost (Dx1) Medical Cost (Dx2)
AWA 500 0.25 5%−10% 0.075 2.189 0.056 0.986 0.759 0.01 0.817 0.263 0.949 0.782 0.0192 0.0216 0.274 37.01875 33.534375
500 0.25 10%−30% 0.200 0.366 0.592 0.643 0.618 0.027 0.27 0.677 0.705 0.692 0.0257 0.0373 0.628 76.5 61.8
500 0.1 5%−10% 0.075 0.677 0.469 0.751 0.626 0.0378 0.363 0.608 0.766 0.696 0.037 0.053 0.374 31.42875 25.5225
500 0.1 10%−30% 0.200 −1.11 0.956 0.134 0.709 0.0152 −0.173 0.911 0.365 0.747 0.0212 0.0261 0.419 39.04 34.3
1000 0.25 5%−10% 0.075 2.189 0.056 0.986 0.759 0.0071 0.817 0.263 0.949 0.782 0.0192 0.0204 0.294 74.0375 67.06875
1000 0.25 10%−30% 0.200 0.366 0.592 0.643 0.618 0.0191 0.27 0.677 0.705 0.692 0.0257 0.032 0.742 153 123.6
1000 0.1 5%−10% 0.075 0.677 0.469 0.751 0.626 0.0268 0.363 0.608 0.766 0.696 0.037 0.0457 0.455 62.8575 51.045
1000 0.1 10%−30% 0.200 −1.11 0.956 0.134 0.709 0.0107 −0.173 0.911 0.365 0.747 0.0212 0.0238 0.474 78.08 68.6
Accuracy 500 0.25 5%−10% 0.075 3 0.008 0.999 0.924 0.0019 1.506 0.022 0.999 0.925 0.0024 0.003 0.0989 37.315625 36.790625
500 0.25 10%−30% 0.200 2.61 0.022 0.995 0.801 0.004 0.943 0.188 0.97 0.814 0.0103 0.011 0.319 98.3 84.2
500 0.1 5%−10% 0.075 3 0.008 0.999 0.924 0.0019 1.506 0.022 0.999 0.925 0.0024 0.003 0.0989 37.24625 36.72125
500 0.1 10%−30% 0.200 2.61 0.022 0.995 0.801 0.004 0.943 0.188 0.97 0.814 0.0103 0.011 0.319 98 82.4
1000 0.25 5%−10% 0.075 3 0.008 0.999 0.924 0.0014 1.506 0.022 0.999 0.925 0.0024 0.003 0.106 74.63125 73.58125
1000 0.25 10%−30% 0.200 2.61 0.022 0.995 0.801 0.0028 0.943 0.188 0.97 0.814 0.0103 0.011 0.334 196.6 168.4
1000 0.1 5%−10% 0.075 3 0.008 0.999 0.924 0.0014 1.506 0.022 0.999 0.925 0.0024 0.003 0.106 74.4925 73.4425
1000 0.1 10%−30% 0.200 2.61 0.022 0.995 0.801 0.0028 0.943 0.188 0.97 0.814 0.0103 0.011 0.334 196 164.8
AUC 500 0.25 5%−10% 0.075 0.638 0.76 0.0608 0.474 153.125 153.125
500 0.25 10%−30% 0.200 0.664 0.76 0.04 0.774 200 200
500 0.1 5%−10% 0.075 0.664 0.76 0.0608 0.474 83.75 83.75
500 0.1 10%−30% 0.200 0.664 0.76 0.04 0.774 140 140
1000 0.25 5%−10% 0.075 0.664 0.76 0.0515 0.586 306.25 306.25
1000 0.25 10%−30% 0.200 0.664 0.76 0.0339 0.882 400 400
1000 0.1 5%−10% 0.075 0.664 0.76 0.0515 0.586 167.5 167.5
1000 0.1 10%−30% 0.200 0.664 0.76 0.0339 0.882 280 280
*

The cost of a false negative is defined as 1, and the cost of a false positive is assumed to be r.