Table 3.
Validation of the PMcardio app for the primary and key secondary outcomes in the overall sample (n = 290)
| Major ECG abnormalities | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Ref + | Ref - | Sensitivity | Specificity | LR+ | LR- | PPV | NPV | Kappa† | ||
| PMcardio | + | 61 | 18 | 85.9% (75.6–93.0) | 91.7% (87.3–95.0) | 10.4 (6.6–16.4) | 0.15 (0.09–0.27) | 77.2% (68.3–84.2) | 95.2% (91.8–97.3) | 0.95 (0.91–0.99) |
| - | 10 | 200 | ||||||||
| Atrial fibrillation or flutter | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Ref + | Ref - | Sensitivity | Specificity | LR+ | LR- | PPV | NPV | Kappa† | ||
| PMcardio | + | 34 | 2 | 97.1% (85.1–99.9) | 99.2% (97.2–99.9) | 123.4 (31.0–491.2) | 0.03 (0.00–0.20) | 94.4% (81.0–98.5) | 99.6% (97.3–99.9) | 1.00 (1.00–1.00) |
| - | 1 | 252 | ||||||||
| Indication of (past) ischemia | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Ref + | Ref - | Sensitivity | Specificity | LR+ | LR- | PPV | NPV | Kappa† | ||
| PMcardio | + | 14 | 10 | 53.6% (33.4–73.4) | 96.2% (93.1–98.2) | 14.2 (7.0–28.6) | 0.48 (0.32–0.73) | 58.3% (40.9–73.9) | 95.5% (93.3–97.0) | 0.91 (0.81–1.00) |
| - | 12 | 253 | ||||||||
| Clinically relevant impulse or conduction abnormality | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Ref + | Ref - | Sensitivity | Specificity | LR+ | LR- | PPV | NPV | Kappa† | ||
| PMcardio | + | 24 | 20 | 88.9% (70.8–97.7) | 92.4% (88.5–95.3) | 11.6 (7.5–18.1) | 0.12 (0.04–0.35) | 54.6% (43.6–65.1) | 98.8% (96.5–99.6) | 0.91 (0.84–0.98) |
| - | 3 | 242 | ||||||||
Data are point estimate (95% confidence interval). Reference in each analysis is expert panel consensus on presence of the outcome of interest.
ECG = electrocardiogram; LR+ = positive likelihood ratio; LR- = negative likelihood ratio; NPV = negative predictive value; PPV = positive predictive value.
†
Kappa for interobserver agreement between Android and iPhone.