Table 3.
Clinically significant thresholds | Image-based classification (expert 1 versus expert 2) | Image-based classification (expert 1) vs. BIO | Image-based classification (expert 2) vs. BIO | |||
---|---|---|---|---|---|---|
Weighted Kappa (95% CI) | Agreement (95% CI) | Weighted Kappa (95% CI) | Agreement (95% CI) | Weighted Kappa (95% CI) | Agreement (95% CI) | |
Zone I disease | 0.6 (0.5–0.7) | 93 (91–94) | 0.6 (0.5–0.7) | 93 (92–94) | 0.6 (0.5–0.6) | 93 (92–94) |
≥ Stage 3 | 0.6 (0.6–0.7) | 95 (94–96) | 0.6 (0.5–0.7) | 95 (94–96) | 0.6 (0.5–0.7) | 94* (93–95) |
Plus disease | 0.7 (0.6–0.8) | 98 (97–99) | 0.5 (0.4–0.7) | 97* (96–97) | 0.6 (0.4–0.7) | 97 (97–98) |
Treatment-Requiring (Type I ROP) | 0.7 (0.6–0.8) | 97 (97–98) | 0.6 (0.5–0.7) | 96* (95–97) | 0.6 (0.5–0.7) | 96 (95–97) |
N = 1553 study eye examinations. BIO = binocular indirect ophthalmoscopy.
P<0.05 compared to the image-based inter-expert discrepancy frequency (column 1).