Table V.
The performance of (i) M4 triage, (ii) triage based on hCG ratio cut-offs (low risk if hCG ratio <0.87 or >1.66) and (iii) triage based on a single progesterone cut-off (low risk if progesterone ≤10 nmol/l).
| PUL classified as low risk | NPV | Sensitivity | FPR | |
|---|---|---|---|---|
| Using all data (n = 835) | ||||
| M4 decision | 70% (67 to 73) | 97% (95 to 98) | 82% (73 to 88) | 23% (20 to 27) |
| hCG ratio cut-offs | 81% (78 to 83) | 96% (94 to 97) | 68% (58 to 77) | 13% (11 to 15) |
| Difference | −11% (−13 to −9) | nc | 14% (7 to 21) | 10% (8 to 13) |
| Using QCCH and HH patients with progesterone levels (n = 560) | ||||
| M4 decision | 68% (64 to 72) | 97% (94 to 98) | 82% (72 to 89) | 25% (21 to 29) |
| Progesterone cut-off | 47% (43 to 51) | 91% (87 to 94) | 68% (57 to 78) | 51% (46 to 55) |
| Difference | 21% (16 to 26) | nc | 14% (−1 to 28) | −26% (−31 to −21) |
95% confidence intervals are shown in parentheses.
NPV, negative predictive value; FPR, false positive rate; QCCH, Queen Charlottes' & Chelsea Hospital; HH, Hillingdon Hospital; nc, not computed.