Introduction
California Maternal Quality Care Collaborative (CMQCC), American College of Obstetricians and Gynecologists (ACOG), and Association of Women’s Health, Obstetrics and Neonatal Nurses (AWHONN), have tools to classify patients into low, medium, and high-risk to predict postpartum hemorrhage. However, these tools are often imprecise with limited predictive power, posing a threat to patient morbidity, mortality, and health equity.
Methods
From an observational cohort study, we calculated the hemorrhage risk for n=525 subjects according to the three tools and compared these data to actual events of recorded blood volume loss, use of uterotonics beyond prophylactic oxytocin and against two volume thresholds.
Results
15.65% had ≥1000ml blood loss. CMQCC and ACOG’s risk ratings are concordant in 82.3% of cases. The AWHONN tool is less correlated with CMQCC and ACOG. In the high-risk category to predict the use of post-prophylactic uterotonics, CMQCC, ACOG and AWHONN predicted the need 23.0%, 23.7% and 43.9% of the time, respectively. Those medicated had higher average BMI (p=.045), Cesarean births (p=.013), and were primipara (<.001). When high or medium risk is used to predict blood loss for ≥ 1000 ml, AWHONN had the best sensitivity (92.7%) and worst specificity (27.4%), the negative predictive value (NPV) was 95.3% and the false positive rate was high, with only 19.1% of the high or medium risk cases experiencing blood loss ≥ 1000ml. The tools often produced false positives rather than correctly identifying negatives. ACOG and CMQCC had a sensitivity of 52.4% and 63.4%, respectively, for the high and medium risk categories, and NPVs of 88.9% and 90.8%. For either measure of PPH, the areas under the curve for the tools are poor, ranging from 0.58 to 0.67.
Conclusion
A significant proportion of hemorrhages requiring medications are missed by the tools, necessitating a more accurate solution to prepare birthing teams, curtail blood loss, reduce patient mortality, and support maternal health equity.