Predictors of von Willebrand disease diagnosis in individuals with borderline von Willebrand factor plasma levels.

Background: In individuals with borderline von Willebrand factor (VWF) plasma levels, second-level tests are required to confirm or exclude von Willebrand disease (VWD). These tests are time-consuming and expensive.

Objective: To assess which parameters can predict VWD diagnosis in individuals with borderline VWF levels (30-60 IU dL(-1) ).

Methods: Nine hundred and fifty individuals with bleeding episodes or abnormal coagulation test results were investigated with first-level tests (blood count, prothrombin time, activated partial thromboplastin time, blood clotting factor VIII, VWF ristocetin cofactor activity [VWF:RCo], and VWF antigen), and 93 (62 females and 31 males; median age, 28 years; interquartile range 15-44) had borderline VWF:RCo levels. All underwent second-level investigations to confirm or exclude VWD. A multivariable logistic regression model was fitted with sex, age, bleeding score, family history, VWF:RCo and ABO blood group as predictors, and used to predict VWD diagnosis.

Results: Forty-five of the 93 individuals (48%) had VWD (84% type 1). A negative linear relationship between VWF:RCo levels and risk of VWD diagnosis was present, and was particularly evident with blood group non-O [adjusted odds ratio 7.00 (95% confidence interval [CI] 1.48-33.11) for every 5 IU dL(-1) decrease in VWF:RCo]. The other variable clearly associated with VWD diagnosis was female sex (adjusted odds ratio 5.76 [95% CI 1.47-22.53]). The area under the receiver operating characteristic curve of the full logistic model was 0.89 (95% CI 0.82-0.95).

Conclusions: In individuals with borderline VWF, the two strongest predictors of VWD diagnosis are low VWF:RCo levels (particularly in those with blood group non-O) and female sex. This predictive model has a promising discriminative ability to identify patients with borderline VWF levels who are likely to have VWD.