Risk factor stratification for urgent and nonurgent transfusion in patients giving birth

Background

A common approach to attempt to reduce maternal morbidity from hemorrhage is to recognize patients at increased risk, and to make advance preparations for possible blood transfusion in these patients. Preparation may consist of a hold clot, type, and screen, or crossmatch. Most hospitals, including ours, have pathways or guidelines that lay out which of these preparations should be made at the time a patient is admitted to labor and delivery. These are often based on risk factors for hemorrhage but do not take into account the probability that transfusion will be needed. The cost-effectiveness of performing a type and screen or routine crossmatch on patients admitted for delivery has been questioned. Several studies have shown that the chance of transfusions in individuals giving birth is very low. In terms of the need for routine blood preparation, the need for urgent transfusion is most relevant. This has not been included in studies of transfusion rates.

Objective

The purpose of this study was to quantify the relative importance of risk factors present on admission for needing a blood transfusion and to develop a formula to define each individual's risk. This could then be used to decide an appropriate level of initial blood preparation for patients at different risk levels.

Study Design

Risk factors for hemorrhage and the level of transfusion preparation were extracted from the medical records of a cohort of 89,881 patients delivering in an 18-hospital healthcare system over 40 months. We tabulated the number who required at least one RBC transfusion and the number needing an urgent transfusion-defined as receiving blood during labor or within 4 hours after delivery. Odds ratios for requiring a transfusion were calculated for each risk factor. We then calculated the probability of needing a transfusion for each patient based on their risk factor profile.

Results

A total of 643 patients had any transfusion during their hospitalization (0.72% of deliveries), and 311 had an urgent transfusion (0.35% of deliveries). The calculated probability of needing a transfusion was less than 1% in 87.8% of patients and was greater than 5% in 1.2% of patients. The chance of needing a transfusion was highest for placenta accreta spectrum, admission Hgb <8.0, and placenta previa. A second tier of risk factors included abruption, bleeding with no specific diagnosis, and Hgb between 8.0 and 10.0.

Conclusion

In our cohort, very few patients received a transfusion. Applying a formula derived from patient-specific risk factors, we found that almost all patients have a very low probability of needing a transfusion, especially an urgent transfusion. Based on these results, we suggest that a hold clot be used except for the highest-risk patients or in settings with barriers to procuring blood in the rare case of urgent transfusion need. Making this change would greatly reduce hospital blood bank charges.