Table 1.
Physician | Mild or worse ROP
|
Type 2 or worse ROP
|
Treatment-requiring ROP
|
||||||
---|---|---|---|---|---|---|---|---|---|
Sensitivity | Specificity | AUC | Sensitivity | Specificity | AUC | Sensitivity | Specificity | AUC | |
Fellow 1 | 0.841 | 0.925 | 0.871 | 0.709 | 0.777 | 0.743 | 0.556 | 0.805 | 0.680 |
Fellow 2 | 0.944 | 0.915 | 0.933 | 0.545 | 0.990 | 0.762 | 0.764 | 0.996 | 0.880 |
Fellow 3 | 0.670 | 0.957 | 0.812 | 0.356 | 1.000 | 0.678 | 0.324 | 0.972 | 0.648 |
Fellow 4 | 0.962 | 0.832 | 0.933 | 0.465 | 0.985 | 0.725 | 0.267 | 1.000 | 0.633 |
Fellow 5 | 0.833 | 0.964 | 0.900 | 0.560 | 0.939 | 0.751 | 0.667 | 0.974 | 0.810 |
| |||||||||
Mean | 0.850 | 0.919 | 0.890 | 0.527 | 0.938 | 0.732 | 0.515 | 0.949 | 0.730 |
Results are calculated for diagnosis of mild or worse ROP, type-2 or worse ROP, and treatment-requiring ROP compared with the expert reference standard of diagnosis. Results are displayed as proportions, and unknown diagnoses by graders are considered incorrect responses. Unknown diagnoses were provided for 12 (5%) by fellow 1, 2 (1%) by fellow 2, 21 (8%) by fellow 3, 0 (0%) by fellow 4, and 7 (3%) by fellow 5.