Table 2 Agreement statistics for coding algorithms ascertaining acute and prevalent heart failure according to coding position
Coding algorithms according to event type and code position | Algorithms (N) | Sensitivity (95% CI) | I2 for sensitivity (95% CI) | Specificity (95% CI) | I2 for specificity (95% CI) |
|---|---|---|---|---|---|
Acute HF | |||||
All | 17 | 63.5% (51.3–74.1) | 99.3 (99.0–99.2) | 96.2% (91.5–98.3) | 99.7 (99.6–99.7) |
All studies with > 200 GS events | 14 | 59.8% (48.1–70.5) | 99.3 (99.1–99.4) | 96.2% (92.1–98.2) | 99.6 (99.6–99.7) |
Studies at low risk of bias | 9 | 55.5% (45.1–65.4) | 98.9 (98.7–99.2) | 97.2% (89.7–99.3) | 99.7 (99.7–99.8) |
Any diagnostic position | 13 | 62.3% (47.7–75.0) | 99.5 (99.4–99.6) | 94.2% (84.0–98.1) | 99.7 (99.7–99.8) |
1ry diagnostic position | 7 | 71.0% (49.4–86.0) | 99.8 (99.7–99.8) | 97.8% (93.4–99.3) | 99.7 (99.7–99.8) |
Prevalent HF | |||||
All | 21 | 63.7% (55.3–71.3) | 98.6 (98.3–98.8) | 98.1% (97.0–98.8) | 98.7 (98.5–98.9) |
All studies with > 200 GS events | 10 | 60.8% (50.9–70.6) | 99.4 (99.3–99.5) | 98.1% (96.4–99.0) | 99.2 (99.0–99.4) |
Studies at low risk of bias | 8 | 64.3% (54.0–73.4) | 98.9 (98.6–99.2) | 97.7% (96.2–98.6) | 97.9 (97.2–98.6) |
Any diagnostic position | 17 | 63.0% (53.9–71.3) | 98.7 (98.4–98.9) | 98.2% (96.9–99.0) | 99.0 (98.8–99.2) |
2ry diagnostic position | 4 | 66.4% (45.8–82.2) | 99.2 (98.9–99.5) | 97.1% (96.0–98.0) | 88.9 (79.6–98.3) |