Diagnostic performance of StrPerimeter and NoLongStr in the training and validation cohorts
| Training cohort (n=206) | Validation cohort (n=138) | |||||
| Variables | StrPerimeter | NoLongStr | Combined | StrPerimeter | NoLongStr | Combined |
| Cut-off for F2–F4 | ≥3.01 | ≥0.0046 | ≥/< both cut-offs (92.7%) | ≥3.01 | ≥0.0046 | ≥/< both cut-offs (97.1%) |
| F0–F1, n | 15 | 14 | 10/99 | 17 | 17 | 16/60 |
| F2–F4, n | 82 | 80 | 78/4 | 55 | 53 | 53/5 |
| Sensitivity (%) | 93.2 (85.2–97.2) | 90.9 (82.4–95.7) | 95.1 (87.3–98.4) | 91.7 (80.9–96.9) | 88.3 (76.8–94.8) | 91.4 (80.3–96.8) |
| Specificity (%) | 87.3 (79.6–92.5) | 88.1 (80.6–93.1) | 90.8 (83.4–95.3) | 78.2 (67.1–86.4) | 78.2 (67.1–86.4) | 78.9 (67.8–87.1) |
| Positive predictive value (%) | 84.5 (75.5–90.8) | 85.1 (75.9–91.3) | 88.6 (79.7–94.1) | 76.4 (64.7–85.3) | 75.7 (63.7–84.8) | 76.8 (64.8–85.8) |
| Negative predictive value (%) | 94.5 (87.9–97.7) | 92.9 (86.0–96.6) | 96.1 (89.8–98.7) | 92.4 (82.5–97.2) | 89.7 (79.3–95.4) | 92.3 (82.2–97.1) |
| Positive likelihood ratio | 7.33 (4.55–11.80) | 7.66 (4.67–12.58) | 10.37 (5.73–18.76) | 4.21 (2.74–6.45) | 4.05 (2.64–6.23) | 4.34 (2.79–6.76) |
| Negative likelihood ratio | 0.08 (0.04–0.17) | 0.10 (0.05–0.20) | 0.05 (0.02–0.14) | 0.11 (0.05–0.25) | 0.15 (0.07–0.30) | 0.11 (0.05–0.25) |
| Cut-off for F3–F4 | ≥3.77 | ≥0.0056 | ≥/< both cut-offs (96.6%) | ≥3.77 | ≥0.0056 | ≥/< both cut-offs (93.5%) |
| F0–F2, n | 18 | 17 | 15/124 | 17 | 21 | 15/73 |
| F3–F4, n | 54 | 54 | 53/7 | 40 | 39 | 39/2 |
| Sensitivity (%) | 87.1 (75.6–93.9) | 87.1 (75.6–93.9) | 88.3 (76.8–94.8) | 95.2 (82.6–99.2) | 92.9 (79.4–98.1) | 95.1 (82.2–99.2) |
| Specificity (%) | 87.5 (80.7–92.2) | 88.2 (81.5–92.8) | 89.2 (82.5–93.6) | 82.3 (72.9–89.1) | 78.1 (68.3–85.7) | 83.0 (73.1–89.8) |
| Positive predictive value (%) | 75.0 (63.2–84.1) | 76.1 (64.2–85.1) | 77.9 (65.9–86.7) | 70.2 (56.4–81.2) | 65.0 (51.5–76.6) | 72.2 (58.1–83.1) |
| Negative predictive value (%) | 94.0 (88.2–97.2) | 94.1 (88.3–97.2) | 94.7 (88.9–97.6) | 97.5 (90.5–99.6) | 96.2 (88.4–99.0) | 97.3 (89.8–99.5) |
| Positive likelihood ratio | 6.97 (4.48–10.85) | 7.38 (4.67–11.65) | 8.19 (5.03–13.32) | 5.38 (3.48–8.32) | 4.24 (2.88–6.25) | 5.58 (3.50–8.89) |
| Negative likelihood ratio | 0.15 (0.08–0.28) | 0.15 (0.08–0.28) | 0.13 (0.07–0.26) | 0.06 (0.01–0.22) | 0.09 (0.03–0.27) | 0.06 (0.02–0.23) |
| Cut-off for F4 | ≥5.14 | ≥0.0079 | ≥/< both cut-offs (99.5%) | ≥5.14 | ≥0.0079 | ≥/< both cut-offs (97.1%) |
| F0–F3, n | 16 | 15 | 15/157 | 10 | 11 | 9/100 |
| F4, n | 31 | 31 | 31/2 | 25 | 24 | 24/1 |
| Sensitivity (%) | 93.9 (78.4–98.9) | 93.9 (78.4–98.9) | 93.9 (78.4–98.9) | 96.2 (78.4–99.8) | 92.3 (73.4–98.7) | 96.0 (77.7–99.8) |
| Specificity (%) | 90.8 (85.2–94.5) | 91.3 (85.8–94.9) | 91.3 (85.8–94.9) | 91.1 (83.8–95.4) | 90.2 (82.7–94.8) | 91.7 (84.5–95.9) |
| Positive predictive value (%) | 66.0 (50.6–78.7) | 67.4 (51.9–80.0) | 67.4 (51.9–80.0) | 71.4 (53.5–84.8) | 68.6 (50.6–82.6) | 72.7 (54.2–86.1) |
| Negative predictive value (%) | 98.7 (95.1–99.8) | 98.8 (95.1–99.8) | 98.7 (95.1–99.8) | 99.0 (93.9–99.9) | 98.1 (92.5–99.7) | 99.0 (93.8–99.9) |
| Positive likelihood ratio | 10.16 (6.32–16.33) | 10.83 (6.63–17.71) | 10.77 (6.59–17.60) | 10.77 (5.93–19.55) | 9.40 (5.30–16.65) | 11.63 (6.19–21.85) |
| Negative likelihood ratio | 0.07 (0.02–0.26) | 0.07 (0.02–0.25) | 0.07 (0.02–0.25) | 0.04 (0.01–0.29) | 0.09 (0.02–0.32) | 0.04 (0.01–0.30) |