Platelet function after cardiac surgery and its association with severe postoperative bleeding: the PLATFORM study.

Platelet dysfunction after cardiac surgery is a determinant of postoperative bleeding. The existing guidelines suggest the use of desmopressin and/or platelet concentrate transfusions in case of platelet dysfunction in bleeding patients, but no cut-off values for platelet activity exist in the literature. The Platelet Function in the Operating Room (PLATFORM) study aims to identify the relationship between platelet function after cardiopulmonary bypass and severe bleeding, finding adequate predictive values of platelet function for severe bleeding. The PLATFORM is a prospective cohort study on 490 adult patients receiving cardiac surgery with cardiopulmonary bypass. Patients received platelet function tests (multiple electrode aggregometry ADPtest and TRAPtest) before surgery and after cardiopulmonary bypass, and routine coagulation tests before surgery and at the arrival in the intensive care unit. The post-cardiopulmonary bypass ADPtest and TRAPtest were significantly (P = 0.001) associated with severe bleeding, as well as the post-cardiopulmonary bypass activated partial thromboplastin time, the international normalized ratio, and the fibrinogen concentration. At a multivariable analysis, the ADPtest (odds ratio 0.962, 95% confidence interval 0.936-0.989, P = 0.005) and the activated partial thromboplastin time (odds ratio 1.097, 95% confidence interval 1.016-1.185, P = 0.017) remained independently associated with severe bleeding. The post-cardiopulmonary bypass ADPtest had the best discrimination, with an area under the curve of 0.712. The best positive predictive value (42%) was found at a cut-off ≤8 U. In conclusion, platelet function tests after cardiopulmonary bypass are significantly associated with postoperative bleeding. However, postoperative bleeding has a multifactorial nature, and the measure of platelet function alone does not provide a high positive predictive value for severe bleeding.