Sensitivity and specificity of strategies to identify patients with hemostasis abnormalities leading to an increased risk of bleeding before scheduled intervention: the Hemorisk study

Abstract

BackgroundPreoperative identification of patients with hemostasis abnormalities leading to an increased bleeding risk is based on routine hemostasis tests: prothrombin time (PT), activated partial thromboplastin time (APTT), and platelet count. Because of their low predictive performance, guidelines recommend replacing them with structured bleeding risk questionnaires, but none is validated in this population. ObjectivesTo assess the diagnostic accuracy of 3 strategies, performed at the preanesthesia visit before scheduled interventions, and to identify patients with hemostasis abnormalities leading to an increased bleeding risk MethodsA multicenter study was performed in 7 French academic hospitals, involving patients scheduled for surgical intervention, without antiplatelet/anticoagulant treatment. The 3 strategies consisted of 1-a structured screening questionnaire; 2-PT, APTT, and platelet count ordered in selected patients; and 3-systematic PT, APTT, and platelet count. The reference standard comprised von Willebrand factor activity/antigen, factor (F)VIII, FIX, FXI, platelet function analyzer, and, when required, FII, FV, FX, and FVII and hemostasis consultation. ResultsEighteen (1.2%) of 1484 patients had a hemostasis abnormality leading to an increased bleeding risk according to reference standard. In the overall cohort, sensitivity of the questionnaire-based strategy was 50% (95% CI, 26%-74%; specificity, 87% [95% CI, 85%-88%]); sensitivity was 0% (95% CI, 0%-41%) in men vs 82% (95% CI, 48%-98%) in women. For selective routine tests, sensitivity was 33% (95% CI, 13%-59%) and specificity 97% (95% CI, 96%-98%). Corresponding values for systematic routine tests were 44% (95% CI, 22%-69%) and 93% (95% CI, 91%-94%). ConclusionSensitivity was low for all 3 strategies investigated. The structured screening questionnaire had clinically acceptable diagnostic accuracy only in women.