Subgroup analysis
| Subgroup | Number of studies | Heterogeneity (I²) | Diagnostic error rate and 95% CI |
| Overall pooled estimate | 22 | 95% | 0.7% (0.5 to 1.1) |
| Studies from North America | 4 | 73% | 0.4% (0.2 to 0.6) |
| Studies from Europe | 8 | 62% | 0.5% (0.3 to 0.8) |
| Studies from rest of world | 10 | 94% | 1.1% (0.7 to 1.6) |
| Studies by decade, 1990s | 1 | 0% | 1.8% (1.6 to 2.0) |
| 2000–2009 | 10 | 90% | 0.7% (0.4 to 1.0) |
| 2010–2019 | 11 | 94% | 0.7% (0.3 to 1.3) |
| Single hospital studies | 6 | 89% | 0.7% (0.1 to 1.6) |
| Multiple hospital | 16 | 96% | 0.8% (0.5 to 1.2) |
| Studies with exclusions | 14 | 94% | 0.7% (0.4 to 1.1) |
| Studies with no exclusions | 8 | 96% | 0.9% (0.3 to 1.7) |
| Studies that excluded obstetrics | 5 | 55% | 0.5% (0.3 to 0.8) |
| Studies that included obstetrics | 17 | 96% | 0.8% (0.5 to 1.2) |
| Studies that excluded paediatrics | 8 | 61% | 0.6% (0.4 to 0.9) |
| Studies that included paediatrics | 14 | 97% | 0.8% (0.4 to 1.3) |
| Studies that used HMPS screening questions | 19 | 95% | 0.7% (0.4 to 1.1) |
| Used 18 HMPS criteria | 12 | 96% | 0.8% (0.5 to 1.3) |
| Used 19 HMPS criteria | 5 | 91% | 0.6% (0.1 to 1.4) |
| Studies that did not use HMPS | 2 | 0% | 0.5% (0.3 to 0.8) |
| Studies that used HMPS AE definition | 7 | 81% | 0.5% (0.3 to 0.7) |
| Studies that used other AE definitions | 13 | 93% | 1.0% (0.6 to 1.4) |
| Studies that used Likert scale for causation | 17 | 95% | 0.8% (0.5 to 1.2) |
| Studies that used yes/no | 5 | 91% | 0.6% (0.1 to 1.4) |
| Studies that defined diagnostic error | 10 | 98% | 0.6% (0.2 to 1.2) |
| Studies that did not define diagnostic error | 12 | 85% | 0.8% (0.5 to 1.2) |
| Studies with ROB=8 | 4 | 99% | 0.7% (0.1 to 1.7) |
| Studies with ROB=7 | 9 | 91% | 0.8% (0.4 to 1.2) |
| Studies with ROB <7 | 9 | 90% | 0.8% (0.3 to 1.5) |
AE, adverse event; HMPS, Harvard Medical Practice Study; ROB, risk of bias.