Predicting intrauterine vacuum‐induced postpartum hemorrhage‐control device treatment success

Abstract Objective To identify predictors of postpartum hemorrhage (PPH) treatment success with the JADA intrauterine vacuum‐induced hemorrhage‐control device in post‐market settings. Methods We performed a retrospective cohort study of PPH managed using the JADA system across 16 centers in the United States (NCT04995887). Treatment success was defined as no treatment escalation or PPH recurrence after device placement. Multivariable logistic regression was used to determine significant predictors of PPH treatment success with a stepwise model selection technique. Selection of the best performing model was determined using area under the receiver operating characteristic (AUC) curves, Akaike information criterion (AIC), and data completeness. In exploratory analyses, we assessed differences in treatment success by cause of bleeding and difference in time to device insertion by self‐reported race, including an interaction term between race and insertion time. Results From October 2020 to March 2022, 800 individuals had PPH managed with the JADA device. Most individuals delivered vaginally (n = 530, 66.3%), self‐identified as White (n = 463, 58.7%), and had a body mass index at delivery ≥35 kg/m2 (n = 322, 40.2%). Cumulative blood loss (CBL) ≥1000 mL before device insertion occurred in 59.5% of individuals, and 65.9% experienced isolated uterine atony. Overall, the treatment success rate was 89.5%, varying across the top predictors. Device success ranged from 90.0% to 94.3% when CBL was <2000 mL prior to insertion. The best stepwise logistic regression model had an AUC of 76.5% and an AIC of 462.15. The most important predictors of treatment success were CBL at time of device insertion, bleeding cause, delivery mode, and time from placental‐delivery‐to‐insertion (p value range: 0.00001–0.05924). Multiple causes of bleeding (atony and non‐atony) were associated with lower treatment success compared to isolated atony (81.9% vs. 92.6%, p < 0.001), but not with only non‐atony bleeding causes (91.3%, p = 0.12). Individuals who identified as White had shorter time to insertion compared to non‐White (38.5 vs. 54.0 min, p < 0.01); however, the interaction between race and insertion time was not significant (p = 0.52). Conclusion The two most important predictors of treatment success with JADA device were CBL prior to device insertion and the etiology of the bleeding. Understanding these factors may guide PPH treatment algorithms.