Von Willebrand disease screening in women undergoing hysterectomy for heavy menstrual bleeding.

In this retrospective cohort study, we used the Truven Health MarketScan Research Databases which include the medical prescription claims of over 109 million total covered lives across the US as well as Medicaid data on 8.6 million patients from 2011 to 2013. While all patient identifiers are removed to protect privacy, all health information during periods of eligibility is linked via a unique member identification number, allowing for patient-level longitudinal analysis. The MarketScan Databases contain patient demographics and physician and facility claims from over 100 payers and are representative of all 50 states of the US commercially insured population. Additionally, the databases include the same comprehensive claims for a sample of Medicaid patients (from 14 states representing all regions of the country). As all data are de-identified and collected for administrative purposes, this study did not qualify as Human Subject research and was considered exempt by the Nationwide Children’s Hospital Institutional Review Board. Procedure codes were used to identify females age <40 years undergoing hysterectomy or HA ( Table 1 ). HA’s included endometrial ablation, uterine artery embolization, or hysteroscopy with dilatation and curettage. Subjects were required to have 12 months of continuous enrolment prior to surgery date. We only included females with a diagnosis of excessive bleeding or anaemia prior to surgery. In an effort to focus on patients undergoing surgery for excessive bleeding primarily due to menstrual bleeding, females with fibroids and genital tract malignancy were excluded. We also excluded females with previously diagnosed bleeding disorders, defined as the presence of an ICD-9 code for VWD, platelet function defect, thrombocytopenia, or other haemorrhagic or coagulation defect in women without evidence of haemostatic screening tests in the year preceding hysterectomy/HA. We defined VWD screening as the presence of a laboratory claim for either VWF:Antigen and/or VWF:RCo plasma levels within the 12 months preceding hysterectomy/HA. To determine if patient and facility level characteristics impacted screening for VWD, we collected the following information: (a) known bleeding disorder diagnosis and/or endometriosis prior to surgery; (b) age; (c) type of insurance; (d) whether patient was living in metropolitan statistical area (MSA; used as a proxy for urban vs rural inhabitance); (e) number of miles and approximate travel time to the nearest haemophilia treatment centre (HTC). In the US, a MSA is a geographical region with a relatively high population density and close economic ties throughout the area. These regions are not incorporated as cities or towns nor do they have legal administrative divisions like counties or separate entities like states. We used ArcMAP ® software (Environmental Systems Research Institute, Redlands, CA, USA) to calculate distance between the MSA centroid and nearest HTC. The MarketScan database only provides MSA data for commercially insured patients. Frequencies were calculated to describe the sample. Bivariate analyses using the chi-squared test were performed initially to examine relationships between the independent factors and VWD screening. Multivariable logistic regression including the a priori identified variables listed above was performed to assess the factors related to the occurrence of VWD screening. Outcomes from the multivariable logistic regression models are expressed as adjusted odds ratios.

We identified 9104 females ages 10-39 years who underwent hysterectomy or HA for HMB. We excluded 106 with known bleeding disorders, most commonly thrombocytopenia (n = 50), other coagulation defects (n = 34), and VWD (n = 11). Of the remaining 8998 women without a reported diagnosis of a bleeding disorder, 57 (0.6%) were screened for VWD within 12 months preceding surgery. There were 1276 women (14%) who underwent other coagulation tests prior to surgery that did not include VWD screening, most commonly prothrombin time and partial thromboplastin time. We had MSA data on 7325 commercially insured females, of whom 70.2% lived within a MSA. Among those living in a MSA, 77.2% lived <100 miles from a HTC and 52.6% lived within 1 hour of a HTC ( Table 1 ). Females living in a MSA were significantly more likely to be screened than those outside of a MSA ( P = 0.047). For those living within a MSA, the odds of being screened for VWD was lower in females with endometriosis (OR = 0.41, 95% CI 0.19, 0.86; P = 0.019) and those living >100 miles from the nearest HTC (OR = 0.28, 95% CI 0.09, 0.90; P = 0.032) after adjusting for age. Nearly half of the study population (42.2%) had a diagnosis of endometriosis. When evaluating the entire commercially insured population (those living within and outside of a MSA), we found that a diagnosis of endometriosis was still a significant predictor of under-screening (OR = 0.51, 95% CI 0.27, 0.96; P = 0.038) after adjustment for age and rural inhabitance ( Table 2 ). However, endometriosis did not predict screening when females with Medicaid were included in the analysis.

This study demonstrated that the frequency of VWD screening in the 12 months prior to hysterectomy/HA in a nationally representative population of publically and commercially insured women undergoing hysterectomy/HA for HMB was very low. We found that greater distance to a HTC was strongly predictive of decreased likelihood of undergoing VWD screening. Additionally, we found that a diagnosis of endometriosis was a more modest predictor of under-screening but was only statistically significant in females with commercial insurance. There are important limitations of our study to consider. Administrative data are dependent on billing codes and not complete medical records. Though our database included a large and diverse population representative of the US population, Medicaid patients were underrepresented compared to commercially insured patients. We were unable to account for ethnicity, family history of bleeding disorders, past medical history, or medications since these data were not included in our database. Therefore, important elements of the medical history that may have guided decision-making for VWD screening were not identifiable in the retrospective claims-based cohort. Importantly, our data only included VWD screening that occurred up to 12 months prior to hysterectomy/HA, meaning that females screened for VWD in the more distant past who did not ultimately have a diagnosis of a bleeding disorder would have been considered as “not-screened” in our study. Despite this major limitation, we believe our results (screening occurred in <1% of patients <40 years requiring surgical intervention for HMB) highlight under-consideration of bleeding disorders in this patient population. One essential reason to identify females with VWD prior to hysterectomy or HA is to avoid peri- and postoperative complications. In a large database study that included 545 women with VWD who underwent hysterectomy, those with VWD were significantly more likely to experience bleeding and require transfusion than women without VWD. 8 Identification of bleeding disorders may also provide new treatment considerations (factor replacement, etc) that could delay or avoid need for major surgery. Our finding that women with endometriosis were even less likely to be screened highlights the need to increase awareness that a bleeding disorder may be a contributing factor to HMB, even in the presence of gynaecologic disease. 5 This study brings to light the need for the haematology community to improve education and awareness among women’s health providers in order to identify females with bleeding disorders and allow for optimal medical management of HMB prior to surgical consideration, inform decision-making when considering surgery, and to optimize haemostasis for those who require surgery. 9

Bleeding disorders in women are under-recognized and undertreated. Women are equally as likely as men to have bleeding disorders other than haemophilia and are disproportionately affected by these diseases due to the bleeding challenges of menstruation and childbirth. A common disease identified in women with excessive bleeding is von Willebrand disease (VWD), an autosomal dominant disorder caused by a quantitative or qualitative deficiency of von Willebrand factor (VWF). Heavy menstrual bleeding (HMB) is the most common symptom in women with VWD, occurring in up to 93% of patients. 1 , 2 Among women with HMB, the reported prevalence of VWD ranges from 5% to 20%, with an overall prevalence of 13% in a systematic review. 3 Women with VWD are also more likely to be diagnosed with haemorrhagic ovarian cysts due to ovulation-associated bleeding and endometriosis due to increased retrograde menstruation. As a result, women with bleeding disorders are more likely to undergo hysterectomy and also undergo hysterectomy at an earlier age than women without bleeding disorders. 4 Consideration of an underlying bleeding disorder is recommended for women with HMB and the presence of any additional risk factors such as HMB since menarche, family history of abnormal bleeding and/or personal history of other bleeding symptoms. 5 In a 2001 Committee Opinion, the American College of Obstetricians and Gynaecologists (ACOG) recommended VWD screening prior to hysterectomy in women with HMB. 6 In an updated 2019 Committee Opinion, ACOG refined the statement to recommend an evaluation for bleeding disorders in patients with HMB and a positive screening test for an underlying disorder of haemostasis. 7 However, the actual frequency of VWD screening prior to hysterectomy in clinical practice is unknown. In the present study, we used a longitudinal health care insurance claims database to assess patterns of VWD screening in a nationally representative sample of females age 15 years of expert recommendations.