There were 2776 children ages 2 to 12 years with a VWD diagnosis in UDC. Children with VWD diagnosed with an additional bleeding disorder were excluded from this study (n = 64). The remaining 2712 children formed the study population.

Month and year of birth data were converted to a birthdate by assuming the 15th of the month and the resulting date was subtracted from the registration date to calculate age at registration. We then categorized age of registration into ages 2 to 5 years and 6 to 12 years to correspond with preschool and elementary school ages. Age of diagnosis was categorized into three age groups (<2 years, 2–5 years, 6–12 years) to differentiate children diagnosed before 2 years of age. Race/ethnicity was categorized as non-Hispanic white, non-Hispanic black, Hispanic, Asian/Pacific Islander (PI) or American Indian/ Alaskan Native (AI/AN), and other.

Diagnosis of VWD was at the discretion of the physician at the HTCs. The UDC did not specify laboratory definitions nor diagnostic criteria for subtypes. The VWD type 2 subtypes were collapsed into a type 2 VWD category. Additionally, the text from an “other specify” option was reclassified into type 1, type 2, type 3, or unknown VWD by coauthor Sarah O’Brien. For subjects with a history of bleeding, the site of the first bleed was categorized on the form as head (intracranial/extracranial), circumcision, intramuscular injection, oral mucosa, joint, unknown, and other. The HTC staff provided text to specify the location in ‘other’. These text responses were reclassified by Sarah O’Brien. Based on the International Society on Thrombosis and Haemostasis (ISTH)-BAT.⁷ At registration, the presence of a bleeding history was based on the response to the yes/no question “Has the patient ever had a bleed?” Information on the type of health insurance was categorized as private (commercial insurance), public (Medicare, Medicaid, state plan, TRICARE), other insurance or uninsured. Measurements of height and weight were used to calculate body mass index (BMI) as weight (kg)/[height (m)],² and the corresponding BMI-for-age percentile, based on the CDC growth charts for children ages 2 to 19 years was used to categorize BMI ranges as underweight, normal, overweight or obese.¹⁹

Treatment products used by subjects in practice, reported at any annual visit, were grouped into one of three product categories: non-plasma and topical products (i.e., intravenous or intranasal desmopressin (DDAVP), antifibrinolytics, fibrin glue, or other); blood bank products (ie, cryoprecipitate, fresh frozen plasma, platelets, packed red blood cells or whole blood); or factor concentrates (ie, human FVIII containing VWF).

We analyzed frequency of HTC utilization at registration categorized as frequent (one visit/year), infrequent (one visit/ 2–3 years), or rare (one visit/4+ years) or first visit to an HTC. To evaluate well-being status, the use of assistive devices for ambulation or mobility including a cane, crutches, walker or a wheelchair at any annual visit was assessed and categorized as any or none. The use of a central venous access device (CVAD) for treatment administration on any annual visit was also assessed. The number of days missed from school or work due to joint problems was calculated by combining the reported number of missed days for both upper and lower extremities for all annual visits.

Data on the use of orthopedic appliances (cast, splint, orthosis, brace), and invasive procedures (arthrodesis, joint replacement, arthroscopic synovectomy, open synovectomy, radiosynovectomy, other invasive procedure) at any annual visit were assessed. Level of activity at registration was assessed by a check box on the following statements: (a) Unrestricted school/work and recreational activities; (b) Full school/work with limited recreational activity levels due to pain, loss of motion, weakness; (c) Limited school/work and recreational activity levels due to pain, loss of motion, weakness; (d) Limited school/work, recreational activity levels, and self-care activity levels due to pain, loss of motion, weakness; (e) Requires assistance from another person for school/work/self-care, and unable to participate in recreation due to pain, loss of motion, weakness. We categorized levels 3 to 5 into an “limited activity” category.

For analysis, data from any annual visit prior to age 13 were assessed. Clinical characteristics were assessed by sex and by type of VWD using chi-square analysis and Wilcoxon rank sum tests, as appropriate. To assess associations between patient characteristics and the prevalence of ever bleeding among children with type 1 VWD, we used Cox proportional hazard (PH) regression with discrete ties, and estimated adjusted hazard ratios (HRS) and Wald 95% confidence limits (Cl). We used time to first bleed in years as the survival time/failure event and censored the children without any bleeding episodes at the time of registration into UDC. All variables used in the model (sex, health insurance, race/ethnicity, family history, and BMI) were reported at the time of registration. All analyses were performed using SAS version 9.4 and P-values were considered statistically significant when P ≤ .05.

Differences in bleeding characteristics according to VWD type and sex are shown in Table 2. The mean age of first bleed was lower by approximately a year among children with type 3 VWD compared to other VWD types and was lower among boys than girls among all types of VWD, with a statistically significant difference among children with type 1 VWD [3.3 year (boys) vs 4.3 year (girls); P < .001]. The mean age of diagnosis was also much lower among children with type 3 VWD compared to the other types and was significantly lower among the boys compared to girls with type 1 (4.2 year vs 5.0 year; P < .001) or type 3 VWD (0.8 year vs 1.5 year, P < .01). Among children with type 1 VWD, boys were younger at their first HTC visit and at registration than girls. At the time of registration, a higher proportion of boys than girls reported ever having had a bleeding episode among children with type 1 (77.5% vs 72.8%; P = .01) and type 2 (87.7% vs 78.7%; P = .02). When accounting for bleeds reported at registration and any annual visit prior to age 13, boys reported significantly more bleeding episodes compared to girls with type 1, but not with type 2. At registration, a higher percentage of treatment product use was reported among children with type 3 compared to type 1 and type 2 VWD and boys reported receiving treatment product significantly more often than girls among type 1 VWD (19.7% vs 11.9%, P < .01). When using the registration and annual visit forms, a significantly higher proportion of treatment product use was seen among boys than girls with type 1 and type 2.

The results of analyses of five life quality or well-being status questions by VWD type are shown in Table 3. Children with type 3 VWD utilized HTCs more frequently than those with type 1 or type 2. About 2% of children with VWD had a CVAD, with children with type 3 VWD having a higher proportion of usage (16%) than children with type 1 or type 2. Greater than 95% of children with type 1 (n = 2006; 96.9%) and type 2 (n = 327; 96.5%) reported unrestricted physical activity, however, the proportion was lower among those with type 3 VWD (n = 109; 91.6%). No child reported any days missed from school due to joint problems and there was only one reported invasive joint procedure performed in a child with type 3 VWD (data not shown).

In contrast, a far higher proportion of children with type 3 reported any use of a cane, crutches, walker, or wheelchair for mobility assistance compared to those with type I or type 2 VWD. Less than 1% of children among all types of VWD had any use of an orthopedic appliance reported at any annual visit; among children with type 3 VWD, usage was less than 5%.

The results of analyses of the site of the first bleed among the children who experienced a bleed based on registration form (n = 1965) are shown in Table 4. The most common sites of first bleed were epistaxis and oral cavity bleeding among all children. However, boys experienced a higher proportion of epistaxis and girls experienced a higher proportion of oral cavity bleeding as the site of their first bleed. Children with type 3 had lower rates of epistaxis compared to type 1 or type 2 VWD. Boys with type 3 VWD showed equivalent frequencies of reporting both circumcision (21%) and epistaxis (21%) as site of first bleed. Surgical sites were also reported among children with type 1 as a site of first bleed more frequently than type 2 or type 3 VWD.

We also evaluated treatment product usage among children who received treatment (n = 1893) and did not see any significant differences by sex. This population was coming out of the HIV hepatitis epidemic and none of the plasma-derived products that contained VWF (and currently used for VWD) were FDA approved for this indication. From our data, 2% of the clinicians prescribed FVIII concentrate to the VWD population. Usage of human FVIII containing VWF among children was 13.8% with type 1, 48.4% with type 2, and 85.7% with type 3 VWD. For non-plasma and topical products, usage was 65.1% (30.9% DDAVP) with type 1, 58.7% (19.2% DDAVP) with type 2, and 59.7% (1.7% DDAVP) with type 3 VWD. For blood bank products, usage was 0.2% with type 1, 0.9% with type 2, and 4.2% with type 3 (Figure S1). Sixteen percent of children used more than one treatment product.

Table S5 shows the proportion of children with type 1 VWD who experienced a bleeding episode and the adjusted HRS for associations with demographic characteristics. Children with type 1 VWD who had experienced a bleed were more likely to be male (HR 1.4; 95% Cl 1.2–1.5), be Asian/Pl/Al/AN race (HR 1.4; 95% a 1.1–1.8) or Hispanic ethnicity (HR 1.3; 95% Cl 1.2–1.6). We did not have a large enough sample to perform a Cox proportional hazard regression among children with type 2 or type 3 VWD.

It is not clear why boys compared to girls would show different bleeding characteristics resulting, for example, in almost a year difference in first bleed episode among children with type 1 VWD. Boys may be experiencing both hemostatic challenges (eg, circumcision) and bleeding episodes much earlier compared to girls. There may be a higher index of suspicion for a bleeding disorder among boys, such as hemophilia, resulting in earlier testing and diagnosis. Boys have been reported to be more physically active than girls, even as infants ^20,21 which could result in more frequent bleeding episodes. Only one other study, the Willebrand in the Netherlands (WiN) study, evaluated children with moderate to severe VWD by sex.¹¹ The WiN study (n = 113, 0–16 year olds) included girls who experienced menstruation, which may have precluded any observable bleeding score differences by sex. Only 5% (n = 35) of girls reported heavy menstrual bleeding as a site of first bleed in our study. Other studies have only evaluated bleeding characteristics by sex among undiagnosed children in need of further hematologic evalution.^3,9,12 Our study underscores the importance of using a large dataset to gain a comprehensive understanding of the bleeding history and symptoms among children with diagnosed VWD by age, stages of development, and sex.

Consistent with other clinical studies, our surveillance data also showed a more severe bleeding phenotype among children with type 3 VWD, compared to children with type 1 or type 2 VWD. ^1,11 Sanders et al¹¹ reported ISTH-BAT pediatric bleeding symptoms and scores to be the highest among children with type 3, then type 2, and then type 1, as well as an earlier mean age of diagnosis among children with type 3 VWD. The study also reported cutaneous bleeding (32%) and epistaxis (31%) as the top sites of first bleed among children with VWD, but did not analyze by sex. ¹¹ We observed that among the two most frequently reported sites of first bleed, epistaxis and oral cavity bleeding, boys frequently reported more epistaxis and girls reported more oral cavity bleeding among all types of VWD.

As with other bleeding disorders such as hemophilia, severe bleeding symptoms in people with VWD can also result in a decrease of quality of life.^22,23 Van Galen et al²² reported that 23% of patients with VWD reported joint bleeds, 47% among patients with type 3 VWD. Individuals who reported joint bleeds also self-reported joint damage and lower health-related quality of life (HR-QoL) measures compared to patients with VWD without joint bleeds. The study also found that the majority of joint bleeds occurred before 16 years of age.²² A decrease in the joint range of motion was also noted in a 2005 UDC report among VWD participants, as early as two years of age.^23.24 We did not evaluate individual joint bleeding or range of motion in our study, but assessed missed days due to joint problems (data not shown), of which, there were no reported cases. Although we did not implement standardized HR-QoL measures, the surveillance data did assess other well-being type indicators. Our findings further emphasize the early age onset of these bleeding episodes and its impact on mobility and physical activity.

Of clinical importance is the identification of potential higher risk populations for bleeding among children with VWD in order to conduct more patient and provider awareness and education for prevention and treatment. We found that children with type 1 VWD who were Hispanic or Asian/Pl/Al/AN were significantly more likely to have experienced a bleed, compared to non-Hispanic white children at the time of registration. Further attention is needed to address potential reasons for this health disparity. The HR was not statistically significant for an association between obesity and ever bleeding. But there were proportional increases in the population ever experiencing a bleed by increasing weight category, the lowest among the underweight (71.1%) to the highest among the obese (80%). Studies in the hemophilia population using UDC data have found an association between obesity and decreases in joint range of motion over time.²⁵ The VWD population should be advised of the potential outcomes with obesity, especially among children with type 3 VWD.

This study has many strengths. It is the largest cohort of pre-adolescent children with VWD reported to date, allowing for more robust analysis of bleeding characteristics, treatment and well-being status by type and sex. We were able to evaluate risk factors for bleeding among children with type 1 VWD. However, there are several limitations. The UDC is a surveillance system, designed to capture minimal data elements. Therefore, bleeding scores were not collected as part of UDC, nor was information on VWF:antigen levels or VWF ristocetin cofactor activity, blood type, and presence of anemia. Therefore, the number of low von Willebrand levels included as type 1 VWD is not known. Diagnosis of VWD by a physician should not have differed by sex. Also, data on bleeding episodes were based on recall of the patient, but again the recall bias should not have differed by sex. There also was no data collection regarding age of menarche, long-term prophylaxis, or any standardized HR-QoL questionnaires. So our well-being status questions were inferred based on other survey questions, and diagnosis for VWD may not have been uniformly interpreted across HTCs.

This study contributes to the understanding of the pre-adolescent history of bleeding among children with diagnosed VWD in the U.S. Boys showed higher proportions of bleeding episodes and treatment product use than girls with type 1 VWD. Minority children (Hispanic or Asian/Pl/Al/AN) in the U.S. are at higher risk of ever bleeding, compared to children who are non-Hispanic white. Children with type 3 VWD also show the early stages of lower well-being status compared to children with type 1 VWD and type 2 VWD. Awareness of these bleeding differences and continued vigilance of care may improve outcomes for children with VWD.