A predictive score model based on complete blood count for a single-unit red blood cell transfusion

Red blood cell (RBC) transfusions are life-saving options for those who are unable to carry oxygen. Although administering a single RBC unit usually elicits a sufficient response, identifying the optimal transfusion trigger can be challenging because this variable remains unknown.

This study aimed to develop a one-unit RBC Transfusion Requirement Scoring Model (1RTRSM), a predictive scoring system through studying the value of each CBC parameter and basic demographics of the patients.

A five-year, retrospective study of patients requesting a single unit of RBC transfusion was carried out to determine the variables influencing transfusion, develop a scoring tool, and pilot test it one year prospectively. Age, sex, ABO grouping and Rh(D) typing department and CBC parameters were included. Based on transfusion status, the results were divided in cases and controls. The score was determined, and the cut-off was assigned using the receiver operating characteristic curve.

Totally, 10,995 individuals - 4635 controls and 6320 cases - were included. They were selected to construct the score considering they possessed positively predicted one-unit RBC transfusion: male, age ≥ 43 years, Hb < 8.9, g/dL, RDW ≥ 15.0 %, PLT < 229.9 × 10 ³ /μL, and MPV < 9.9 fL. The total score value of 5 was chosen as the cut-off with strengths of 87 % accuracy, 89 % sensitivity and 85 % specificity. Within a pilot study, the score's accuracy over all departments was 85 %.

We established a novel 1RTRSM model that predicts patients' transfusion needs across a variety of departments using a simple, valid and reliable tool demonstrating high accuracy.