The bleeding severity index: Validation and comparison to other methods for classifying bleeding complications of medical therapy

Abstract Reports of bleeding complications of medical therapy should be based on valid methods of classification, but the reproducibility of existing methods has not been tested. Therefore, we prospectively studied three methods to classify the severity of bleeding: a purely subjective implicit method, a previously published explicit method using brief criteria, and the bleeding severity index, which is a new explicit method using detailed criteria about the amount, rate, and consequences of bleeding. Three physicians independently reviewed abstracts of 168 patients treated with anticoagulants. The proportion of cases classified as major bleeding varied widely when the implicit method was used (2, 14 and 39%), less when the old explicit method was used (28, 40 and 47%), and not at all when the new bleeding severity index was used (20, 20 and 20%). Intraobserver agreement was excellent for both explicit methods (κ ⩾ 0.95). However, interobserver agreement was better for the bleeding severity index (κ = 0.87) than for the old explicit method κ = 0.69) or the implicit method (κ = 0.39). We conclude that the classification of bleeding complications of medical therapy depends on the method used. In comparison to older methods, the bleeding severity index is highly reproducible and should be tested more widely to determine whether it can be applied to the burgeoning clinical research in anticoagulation and thrombolysis.