Table 2.
Results at various prediction threshold cut-offs that can be used to select the threshold used by any future epidemiology study when selecting an end of observations due to death
Finding people who are dead … | |||||
---|---|---|---|---|---|
Prediction threshold | Sensitivity of death | Specificity of death | Positive predictive value of death | Proportion of target population | |
If you choose by prediction threshold to be greater than … | |||||
0.9 | 26.170 | 99.901 | 86.898 | 0.007 | |
0.5 | 61.895 | 99.474 | 74.754 | 0.020 | |
0.1 | 90.282 | 98.149 | 55.095 | 0.040 | |
If you choose by sensitivity … | |||||
0.666 | 50 | 99.676 | 79.549 | 0.015 | |
0.1046 | 90 | 98.184 | 55.489 | 0.040 | |
0.0019 | 99 | 69.321 | 7.510 | 0.324 | |
If you choose by specificity… | |||||
0.253 | 78.608 | 99 | 66.515 | 0.029 | |
0.905 | 25.599 | 99.9 | 87.174 | 0.007 | |
0.990 | 7.207 | 99.99 | 95.054 | 0.002 |