Background: Diagnosis of von Willebrand disease (VWD) in patients with a mild to moderate bleeding tendency (mild bleeding disorders, MBD) bears special challenges, as von Willebrand factor (VWF) levels are influenced by several factors, such as age, acute phase reaction and blood group, and might be unstable. Unstable VWF levels might lead to under- or over-diagnosis of VWD in MBD patients, and repeated measurements to identify VWD patients might be required. Aims: We aimed to identify clinical and epidemiological factors associated with unstable VWF levels, leading to change of the diagnosis based on the recommended diagnostic cut-off levels. Furthermore, we wanted to investigate cut-off of baseline VWF levels with a low probability to decrease below the diagnostic cut-off and, thus, not require subsequent VWF measurements. Methods: All patients enrolled into the Vienna bleeding biobank (VIBB, EC no. 603/2009) before October 2021 with at least two separate assessments of VWF levels, in particular VWF antigen (VWF:Ag) and activity (VWF:GPIbM and/or ristocetin-cofactor activity [VWF:RCo]), were included in this analysis. Detailed inclusion and exclusion criteria of the VIBB, a single-center cohort study on patients with MBD, have been published previously (Gebhart et al. Haemophilia.2018). For this analysis, pathological VWF levels were defined as any measurement of VWF:Ag, VWF:GPIbM or VWF:RCo ≤50 IU/dL, according to current guidelines (James et al. Blood Adv. 2021). We defined 3 subgroups of patients according to lowest VWF (VWF:Ag or VWF:GPIbM or VWF:RCo) at baseline and at each measurement: patients with steadily pathological measurements ≤50 IU/dL ("steady pathological"), patients with repeatedly normal measurements >50 IU/dL ("steady normal") and patients crossing the 50 IU/dL threshold between study visits ("changing"). Measurements were excluded from the analysis if an evident VWF influence such as pregnancy, infection, surgery or acute phase reaction was identified. Results: In total, 281 of 691 patients (40.7 %) from the VIBB study who had records of at least two VWF measurements of VWF antigen and activity levels were included in this analysis. Patients had a median number of 2 VWF measurements (Q1-Q3, 2-2) over a median follow-up time of 781 days (Q1-Q3, 105-1679) for a total of 660 measurements. 212 patients had 2 measurement of VWF antigen and activity, followed by 3 (n=47), 4 (n=11), 5 (n=5) and 6 (n=1) measurements. Of all analyzed patients, 217 patients (77.2 %) had VWF levels consistently above the cut-off of 50 IU/dL ( "steady normal"), 28 patients (10.0%) had repeatedly values ≤50 IU/dL ("steady pathological"), while 36 patients (12.8%) had changing results with values both below and above the cut-off of 50 IU/dL ( "changing"; Figure 1). The clinical and laboratory characteristics of each subgroup are summarized in Table 1. Patients in the "steady pathological” and "changing” groups compared to the "steady normal” group were younger and more commonly had blood group O. The values of VWF activity (VWF:RCo or VWF:GPIbM) at inclusion appeared to be more unstable than the VWF:Ag values (Figure 1). Seventeen (47.2%) patients in the changing group had pathological VWF activity values but VWF:Ag values above the cut-off of 50 IU/dL at study inclusion. VWF antigen and activity values at baseline within the changing group ranged from 36 IU/dL to 101 IU/dL and 28 IU/dL to 86 IU/dL, respectively. In a logistic regression model, modeling changing status as a binary outcome and lowest VWF at baseline using linear splines with a knot set at 50 IU/dL, the probability of changing was 37.2% (95%CI, 25.7-50.3), 18.6% (95%CI, 12.2-27.4) and 1.1% (95%CI, 0.2-4.8) at 40, 60 and 80 IU/dL, respectively (Figure 2). The added value of the clinical predictors age, sex, the Vicenza score and blood type O beyond lowest VWF at baseline, however, was only weak (χ2 5.0, p = 0.290). Conclusion: Based on our data, the reproducibility of VWF measurements is poor in patients with MBD. Re-evaluation of suspected patients for VWD who do not clearly fulfill the diagnostic criteria should be considered, especially if VWF levels are close to the cut-off level, as targeted treatment before surgical interventions might be warranted. Further studies are needed to better understand the reproducibility of VWF measurements and improve the diagnosis of VWD in patients with MBD. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal