Abstract 17197: The Value of Routine Near-patient Haemostasis Testing for Predicting Bleeding After Cardiac Surgery

Abstract

Introduction: Near-patient haemostasis tests assist diagnosis and treatment of blood clotting abnormalities in bleeding patients after cardiac surgery. Aim: To quantify the predictive value of routine near-patient haemostasis tests for coagulopathic bleeding in a large clinical trial. Methods: We performed a prospective cohort study of the patient characteristics and the results of 23 near-patient platelet and viscoelastometry tests in adults undergoing a range of cardiac surgery procedures. The primary outcome was clinical concern about bleeding (CCB), defined as a composite of high blood loss or pro-haemostatic treatment or re-operation for bleeding. A multivariable predictive model for CCB was developed based on the pre-operative characteristics of 1833 patients at a single cardiac centre recruited between March 2010 and August 2012. This model was validated using data for 1611 patients recruited at 16 other UK cardiac surgery centres between July 2009 and February 2013. The additional predictive value of near-patient test results was estimated in the development cohort. Results: CCB occurred in 449 (24.5%) patients in the development cohort. Male sex, absence of diabetes, complex procedure, anti-platelet treatment, short interval since withdrawal of anti-platelet treatment, reduced BMI, reduced pre-operative haematocrit and reduced pre-operative platelet count were independently associated with increased risk of CCB (area under RoC curve (AUR)=0.72 (0.69 to 0.75)). At 75% sensitivity, the specificity of the model was 57%. In the validation cohort, the AUR of the patient characteristics model was 0.63 (0.60 to 0.66). Adding the results of the most predictive near-patient tests performed immediately after surgery improved prediction of CCB in the development cohort (AUR 0.74; 0.72 to 0.77), but at 75% sensitivity this increased the specificity by only 4%. Adding the results of near-patient platelet tests performed before surgery did not further improve prediction of CCB (AUR 0.75; 0.72 to 0.77). Conclusions: Routine near-patient haemostasis tests offer little improvement in the prediction of CCB after cardiac surgery compared to pre-operative patient characteristics alone.