Patient Blood Management in Obstetrics

Overview

Obstetric anemia and hemorrhage are major causes of maternal morbidity and increased healthcare utilization. Although red blood cell (RBC) transfusion is commonly used, it is associated with higher rates of postpartum complications, including pneumonia, renal failure, and cardiac events (Kloka et al.). Patient Blood Management (PBM) is an evidence-based approach that aims to optimize a patient's own blood and reduce avoidable transfusion through three pillars: treating anemia and iron deficiency, minimizing blood loss, and avoiding unnecessary transfusion. International guidelines support PBM in obstetrics, but data on comprehensive program implementation remain limited due to barriers such as resource constraints and the need for multidisciplinary coordination. Aims , objectives and Hypotheses Primary Aim To assess the association between PBM implementation and blood product utilisation among all obstetric deliveries at Corniche Hospital between 2018 and 2025. Primary Objective To estimate the change over time in the mean number of blood product units transfused per delivery (combined and by product type). Primary Hypothesis Increasing PBM maturity over time, particularly following comprehensive PBM implementation in 2022, is associated with a significant reduction in mean blood product units transfused per delivery, after adjusting for changes in case-mix. Secondary Objectives Secondary objectives include evaluating changes in transfusion-related practice and anemia outcomes, including proportion of deliveries receiving any transfusion; pretransfusion hemoglobin thresholds in non-actively bleeding patients (where definable); predelivery anemia prevalence (and iron therapy utilization where captured). Maternal outcomes will be evaluated by reporting composite morbidity; postpartum hysterectomy; hospital length of stay; High Dependency unit (HDU)/ICU admission and length of stay; 28-day all-cause emergency readmissions; in-hospital mortality. Neonatal outcomes will not be included. At Corniche Hospital, prior evaluation in cases of major obstetric hemorrhage showed that PBM reduced blood product use and improved hemoglobin recovery without increasing morbidity (Ansari et al.). This study extends the assessment to all deliveries from 2018-2025 to evaluate hospital-wide changes in transfusion practice and maternal outcomes as PBM implementation matured.

Aims / Objectives / Hypotheses Primary Aim To assess the association between PBM implementation and blood product utilization among all obstetric deliveries at Corniche Hospital between 2018 and 2025. Primary Objective To estimate the change over time in the mean number of blood product units transfused per delivery (combined and by product type). Primary Hypothesis Increasing PBM maturity over time, particularly following comprehensive PBM implementation in 2022, is associated with a significant reduction in mean blood product units transfused per delivery, after adjusting for changes in case-mix. Secondary Objectives Secondary objectives include evaluating changes in transfusion-related practice and anemia outcomes, including proportion of deliveries receiving any transfusion; pretransfusion hemoglobin thresholds in non-actively bleeding patients (where definable); predelivery anemia prevalence (and iron therapy utilization where captured). Maternal outcomes will be evaluated by reporting composite morbidity; postpartum hysterectomy; hospital length of stay; HDU/ICU admission and length of stay; 28-day all-cause emergency readmissions; in-hospital mortality. Neonatal outcomes will not be included. Study Type Retrospective observational cohort study. Setting Corniche Hospital is a 180-bed tertiary obstetric referral center in the Emirate of Abu Dhabi, United Arab Emirates, with a 64-cot level III neonatal intensive care unit. Over the past 8 years, it has averaged approximately 5500 deliveries annually, with an average cesarean delivery rate of 35%. PBM strategies were implemented progressively, with a comprehensive PBM program established in 2022, supported by governance structures (PBM committee), information systems (dashboards/decision support), education, and updated policies. Study Period 1 January 2018 to 31 December 2025. Participants Inclusion Criteria All women who delivered at Corniche Hospital during the study period (vaginal, operative vaginal, caesarean; livebirth or stillbirth depending on local definition). Exclusion Criteria : None PBM Program Description (Exposure) The main exposure of interest is the implementation of a comprehensive PBM program in 2022. While some PBM strategies were in place in 2018, the comprehensive PBM program at Corniche Hospital was established in 2022 with the formation of a PBM committee led by a consultant anesthetist serving as the program coordinator (Principal Investigator T.A.), and representatives from most clinical and nonclinical departments. PBM strategies at Corniche Hospital comprise clinical and enabling interventions across three pillars. The program is implemented at scale hospital-wide, with key strategies described in the prior major haemorrhage evaluation and the earlier protocol framework. Data Collection For each study participant we have the following data: patient age, body mass index (BMI), parity, gestational age at delivery, mode of delivery, comorbidities (eg. placenta previa / placenta accreta spectrum, hypertensive disorders including preeclampsia, diabetes), units of RBC, FFP, and platelet transfused, mortality, composite morbidity (transfusion reaction, acute lung injury, thrombosis, sepsis), postpartum hysterectomy, hospital length of stay, high-dependency unit length of stay, and all-cause emergency readmissions within 28 days. Data Sources Outcome Measures Primary Outcome Mean number of blood product units transfused per delivery, measured combined units of RBC, FFP, and platelet units transfused per delivery. Secondary Outcomes Secondary outcomes are broken up into two categories: PBM Program Outcomes and Maternal Outcomes. PBM Program Outcomes * Proportion of deliveries receiving any blood product (RBC/Fresh frozen plasma (FFP)/platelets) * Mean pretransfusion Hb in non-actively bleeding patients (where definable) * Predelivery anemia prevalence (define threshold consistent with local policy; e.g., Hb \<11 g/dL) * Utilization of IV iron pre- and post-delivery (if available and reliable) * Optional: Hb level at follow-up timepoint (e.g., 3 weeks post discharge) where routinely captured (expected missingness; report availability) Maternal Outcomes * Composite morbidity (e.g., transfusion reaction, acute lung injury/TRALI, thrombosis, sepsis, postpartum hysterectomy-final components to match data feasibility) * Postpartum hysterectomy (standalone) * In-hospital mortality * Hospital length of stay (LOS) * HDU/ICU LOS (if available) * 28-day all-cause emergency readmissions Statistical Analysis Plan Overview The evaluation will use a year-by-year comparisons from 2018 to 2025 with 2018 as the baseline comparison year. We will investigate the feasibility of applying an interrupted time series (ITS) with a prespecified intervention point at 2022 (comprehensive PBM program established), to estimate changes in level and trend in outcomes. The descriptive analysis will include a summary of deliveries and baseline characteristics by year as well as unadjusted trends in primary and secondary outcomes. Primary outcome modelling (individual-level) Because units transfused per delivery is count data with many zeros and potential overdispersion, we will use negative binomial regression, with robust standard errors as appropriate. The outcome will be a composite of total units (RBC, FFP, and platelets) transfused per delivery. The statistical model will adjust for patient age, body mass index (BMI), parity, gestational age at delivery, mode of delivery, and comorbidities. Secondary outcome modelling For our secondary outcomes we will model binary variables (e.g., % transfused, composite morbidity, readmission) with logistic regression. For length of stay outcomes we will apply negative binomial regression models. Missing data We will quantify missingness by variable and year. We will use complete-case analysis for the primary analysis if covariate missingness is low/moderate. Multiple imputation will be considered for key covariates if missingness is meaningful and plausibly missing at random. Outcomes likely missing not at random (e.g., 3-week Hb) will be reported with availability and interpreted cautiously. Software R will be used for the statistical analysis.