RESUMO
BACKGROUND: Despite the prophylactic use of platelet transfusion, hemorrhagic complications still represent an important cause of morbidity and mortality in patients with hematologic malignancies. Patient-related factors and characteristics of the transfused product can affect transfusion efficacy. STUDY DESIGN AND METHODS: The aim of this study was to develop and validate the Platelet Efficacy Score (PEscore), based on patient and product characteristics, to predict the likelihood of a satisfactory platelet transfusion (absolute increment ≥10.5 × 109 /L). This study utilized data relative to 16,265 platelet transfusions performed in 1592 oncohematologic patients. The whole sample was divided into two random samples: a training set, in which different patient-related and transfusion-related characteristics were included in a predictive model to develop the PEscore; and a validation set, in which the predictive properties of the PEscore were confirmed. In the training set, multilevel logistic regression analysis was performed in which the likelihood of attaining a satisfactory transfusion was modeled. RESULTS: The predictive score ranged between 0 and 30. Predictive properties of the PEscore were confirmed by the observed rates of satisfactory transfusions in the validation sample; the probability of a satisfactory transfusion was less than 10% for a score less than 12 and exceeded 50% if the score was 22 or higher. The likelihood of a satisfactory transfusion increased by 29% for a 1-unit increase in the PEscore (odds ratio, 1.29; 95% confidence interval, 1.27-1.31). CONCLUSION: The availability of a prediction score can increase transfusion efficacy, help the transfusion medicine specialist in the choice of the best product for the individual patient, and avoid waste of resources.
