Table 4.
Performance of RAMs applied postpartum to predict VTE
| Risk assessment models | Threshold or cut-off | Endpoint | Data source | Performance measures | |||||
| TP | FP | FN | TN | Sensitivity (95% CI) | Specificity (95% CI) | ||||
| Predicting postpartum VTE following vaginal and caesarean delivery | |||||||||
| ACCP (one study) | NR | VTE | Gassmann et al30 | 0 | 34 | 0 | 310 | unable to estimate – no VTE | 0.90 (0.86 to 0.93) |
| ACOG (one study) | NR | VTE | Gassmann et al30 | 0 | 30 | 0 | 314 | unable to estimate – no VTE | 0.91 (0.88 to 0.94) |
| ASH (one study) | NR | VTE | Gassmann et al30 | 0 | 0 | 0 | 344 | unable to estimate – no VTE | 1.00 (0.99 to 1.00) |
| Caprini (one study) | Risk score ≥2 | VTE | Tran et al37 | 3 | 5780 | 0 | 311 | 1.00 (0.44 to 1.00) | 0.05 (0.05 to 0.06) |
| Caprini | Risk score ≥3 | VTE | Tran et al37 | 1 | 3066 | 2 | 3025 | 0.33 (0.06 to 0.79) | 0.50 (0.48 to 0.51) |
| Caprini | Risk score ≥4 | VTE | Tran et al37 | 0 | 1257 | 3 | 4834 | 0.00 (0.00 to 0.56) | 0.79 (0.78 to 0.80) |
| Padua (one study) | Risk score ≥4 | VTE | Tran et al37 | 0 | 50 | 3 | 6041 | 0.00 (0.00 to 0.56) | 0.99 (0.99 to 0.99) |
| RCOG (three studies) | NR | VTE | Gassmann et al30 | 0 | 138 | 0 | 206 | unable to estimate – no VTE | 0.60 (0.55 to 0.65) |
| RCOG | Risk score ≥2 | VTE | Tran et al37 | 1 | 3837 | 2 | 2254 | 0.33 (0.06 to 0.79) | 0.37 (0.36 to 0.38) |
| RCOG | ≥2 low risk factors or 1 high risk factor | VTE | Sultan et al35 | 197 | 149 205 | 115 | 283 836 | 0.63 (0.58 to 0.68) | 0.66 (0.65 to 0.66) |
| SFOG (two studies) | Risk score ≥2 | VTE | Lindqvist et al32 | 18 | 111 | 19 | 2273 | 0.49 (0.33 to 0.64) | 0.95 (0.94 to 0.96) |
| SFOG | ≥2 risk factors | VTE | Sultan et al35 | 109 | 41 145 | 412 | 620 721 | 0.21 (0.18 to 0.25) | 0.94 (0.94 to 0.94) |
| Chau, 2019 (one study*) | Risk score ≥3 (2012 data set) | VTE | Chau et al26 | 0 | 101 | 1 | 456 | 0.00 (0.00 to 0.79) | 0.82 (0.78 to 0.85) |
| Chau, 2019 | Risk score ≥3 (2015 data set) | VTE | Chau et al26 | 0 | 113 | 1 | 393 | 0.00 (0.00 to 0.79) | 0.78 (0.74 to 0.81) |
| Ellis-Kahana, 2020 (full model) (one study†) | Risk score >3 (high risk) | VTE | Ellis-Kahana et al39 | 68 | 7942 | 41 | 75 449 | 0.62 (0.53 to 0.71) | 0.90 (0.90 to 0.91) |
| Ellis-Kahana, 2020 (without antepartum thromboembolic disorder) | Risk score >3 (high risk) | VTE | Ellis-Kahana et al39 | 63 | 9926 | 46 | 73 465 | 0.58 (0.48 to 0.67) | 0.88 (0.88 to 0.88) |
| Sultan, 2016 (one study‡) | ≥2 risk factors: top 35% (threshold: 7.2 per 10 000 deliveries) | VTE | Sultan et al35 | 355 | 231 480 | 166 | 430 386 | 0.68 (0.64 to 0.72) | 0.65 (0.65 to 0.65) |
| Sultan, 2016 | ≥2 risk factors: top 25% (threshold: 8.7 per 10 000 deliveries) | VTE | Sultan et al35 | 310 | 164 976 | 211 | 496 890 | 0.60 (0.55 to 0.64) | 0.75 (0.75 to 0.75) |
| Sultan, 2016 | ≥2 risk factors: top 20% (threshold: 9.8 per 10 000 deliveries) | VTE | Sultan et al35 | 278 | 131 921 | 243 | 529 945 | 0.53 (0.49 to 0.58) | 0.80 (0.80 to 0.80) |
| Sultan, 2016 | ≥2 risk factors: top 10% (threshold: 14 per 10 000 deliveries) | VTE | Sultan et al35 | 185 | 66 053 | 336 | 595 813 | 0.36 (0.32 to 0.40) | 0.90 (0.90 to 0.90) |
| Sultan, 2016 | ≥2 risk factors: top 6% (threshold: 18 per 10 000 deliveries) | VTE | Sultan et al35 | 158 | 41 096 | 363 | 620 770 | 0.30 (0.27 to 0.34) | 0.94 (0.94 to 0.94) |
| Sultan, 2016 | ≥2 risk factors: top 5% (threshold: 19.7 per 10 000 deliveries) | VTE | Sultan et al35 | 139 | 32 980 | 382 | 628 886 | 0.27 (0.23 to 0.31) | 0.95 (0.95 to 0.95) |
| Sultan, 2016 | ≥2 risk factors: top 1% (threshold: 41.2 per 10 000 deliveries) | VTE | Sultan et al35 | 47 | 6576 | 474 | 655 290 | 0.09 (0.07 to 0.12) | 0.99 (0.99 to 0.99) |
| Predicting postpartum VTE following caesarean delivery only | |||||||||
| ACOG (one study) | Risk score ≥3 | VTE | Lok et al33 | 0 | 0 | 0 | 859 | unable to estimate – no VTE | 1.00 (1.00 to 1.00) |
| RCOG (two studies) | NR | VTE | Binstock and Larkin (abstract)24 | 11 | 2692 | 0 | 172 | 1.00 (0.74 to 1.00) | 0.06 (0.05 to 0.07) |
| RCOG | Risk score ≥3 | VTE | Lok et al33 | 0 | 649 | 0 | 210 | unable to estimate – no VTE | 0.24 (0.22 to 0.27) |
| Binstock, 2019 (one study) | NR | VTE | Binstock and Larkin (abstract)24 | 11 | 2635 | 0 | 229 | 1.00 (0.74 to 1.00) | 0.08 (0.07 to 0.09) |
| Cavazza, 2012 (one study) | Moderate/high/very high | VTE | Cavazza et al25 | 0 | 268 | 1 | 232 | 0.00 (0.00 to 0.79) | 0.46 (0.42 to 0.51) |
| Lok, 2019 (one study) | Risk score ≥3 | VTE | Lok et al33 | 0 | 28 | 0 | 831 | unable to estimate – no VTE | 0.97 (0.95 to 0.98) |
*Data discrepancy in paper—text states analysis included 1069 women: 557 in the 2012 time frame and 512 in the 2015 time frame; however, data in tables suggest 558 women included in the 2012 time frame and 507 in the 2015 time frame.
†Internal validation study. Full risk prediction model: C-statistic, 0.817 (95% CI: 0.768 to 0.865) with Hosmer-Lemeshow p value=0.297; model without antepartum thromboembolic disorder: C-statistic, 0.778 (95% CI: 0.729 to 0.826) with Hosmer-Lemeshow p value=0.114.
‡Sultan et al,35 final risk prediction model in external Swedish cohort: C-statistic, 0.73 (95% CI: 0.71 to 0.75) and calibration slope, 1.11 (95% CI: 1.01 to 1.20).
ACCP, American College of Chest Physicians; ACOG, American College of Obstetricians and Gynecologists; ASH, American Society of Hematology; FN, false negative; FP, false positive; NR, not reported; RAMs, risk assessment models; RCOG, Royal College of Obstetricians and Gynaecologists; SFOG, Swedish Society of Obstetrics and Gynecology; TN, true negative; TP, true positive; VTE, venous thromboembolism.