Prediction of escape red blood cell transfusion in expectantly managed women with acute anaemia after postpartum haemorrhage.

ABSTRACT

OBJECTIVE: To determine clinical predictors of escape red blood cell (RBC) transfusion in postpartum anaemic women, initially managed expectantly, and the additional predictive value of health-related quality of life (HRQoL) measures. DESIGN: Secondary analysis of women after postpartum haemorrhage, either randomly allocated to, or opting for expectant management. SETTING: Thirty-seven hospitals in the Netherlands. POPULATION A total of 261 randomised and 362 nonrandomised women. METHODS: We developed prediction models to assess the need for RBC transfusion one using clinical variables (model 1), and one extended with scores on the HRQoL-measures Multidimensional Fatigue Inventory (MFI) and EuroQol-5D (model 2). Model performance was assessed by discrimination and calibration. Models were internally validated with bootstrapping techniques to correct for overfitting. MAIN OUTCOME MEASURES: Escape RBC transfusion. RESULTS: Seventy-five women (12%) received escape RBC transfusion. Independent predictors of escape RBC transfusion (model 1) were primiparity, multiple pregnancy, total blood loss during delivery and haemoglobin concentration postpartum. Maternal age, body mass index, ethnicity, education, medical indication of pregnancy, mode of delivery, preterm delivery, placental removal, perineal laceration, Apgar score and breastfeeding intention had no predictive value. Addition of HRQoL-scores (model 2), significantly improved the model's discriminative ability c-statistics of model 1 and 2 were 0.65 (95% CI 0.58-0.72) and 0.72 (95% CI 0.65-0.79), respectively. The calibration of both models was good. CONCLUSIONS: In postpartum anaemic women, several clinical variables predict the need for escape RBC transfusion. Adding HRQoL-scores improves model performance. After external validation, the extended model may be an important tool for counselling and decision making in clinical practice.