A Pragmatic Randomized Controlled Trial to Predict Postpartum Hemorrhage

Logistic Regression Prediction Model vs. Standard of Care for Prediction of Postpartum Hemorrhage - A Pragmatic Randomized Controlled Trial

This research project aims to enhance the safety of childbirth by using advanced computer models to predict the risk of postpartum hemorrhage (PPH).

PPH is a significant concern for mothers during and after delivery. Current risk assessment tools are basic and do not adapt to changing conditions. This study will investigate whether a new and recently validated model for predicting PPH, combined with a provider-facing Best Practice Advisory (BPA) regarding currently recommended strategies triggered by an increased predicted risk, can improve perinatal outcomes.

This study will compare the current category based risk assessment tool with a new, enhanced prediction model which calculates risk based on 21 factors, automatically updates as new information becomes available during labor and, if elevated, provides a provider-facing Best Practice Advisory (BPA) recommending consideration of strategies that are institutionally agreed to represent high-quality practice.

Investigators hypothesize that the enhanced care approach will result in improved perinatal outcomes.

The goal of the study is to improve the wellbeing of mothers during childbirth by harnessing the power of modern technology and data analysis.

Postpartum hemorrhage (PPH) is a common complication following vaginal or cesarean delivery and contributes significantly to maternal morbidity and mortality in the United States. There are numerous clinical factors which contribute to a patient's risk of developing PPH. Utilization of an evidence-based tool for PPH risk prediction is recommended by national societies and required by the Joint Commission.

Most currently used tools are category based and assign a low, medium, or high risk of hemorrhage. These tools fail to take advantage of the vast amounts of data and computing power available via modern electronic medical records. Predictive modeling and informatics-based solutions could help to modernize PPH risk prediction and improve patient outcomes.

This study proposes to continue standard of care risk assessment for all patients, including those randomized to the intervention arm (ARM B). Those patients in the intervention arm (ARM B) will have an additional risk prediction displayed, which will show the quantitative output from the logistic regression PPH risk prediction model, (validated in a previous study). In addition to this display, patients above a preset threshold of 3% risk will have a Best Practice Advisory (BPA) deployed to clinicians with recommended actions. These recommended actions, including the prophylactic use of tranexamic acid and second-line uterotonics, are supported by best evidence in those patients deemed to be at elevated a priori risk of PPH. These prophylactic treatments are accepted standard of care for those patients deemed high risk, and may be administered, at the discretion of the covering clinician, to patients rated high risk by the current risk assessment tool in the comparator arm (Arm A) of the study. The recommendations within the best practice advisory serve as a reminder of best practices as defined by the department and providers are not forced to follow the recommendations of the best practice advisory.

Inclusion Criteria:

• All vaginal and cesarean deliveries occurring at Vanderbilt University Medical Center

Exclusion Criteria:

• All patients will be randomized at the time of admission to the obstetric service. Patients who are discharged prior to delivery will be excluded from subsequent analysis. Any patients with a pre-delivery planned hysterectomy (for placenta increta or percreta) will be excluded from the treatment algorithm and primary analysis.