Correspondence: Miss R. A. Kadir, Obstetrics and Gynaecology, Royal Free Hospital, Pond Street, London NW3 2QG, UK.
Objective To determine the prevalence of von Willebrand disease in women presenting with menorrhagia.
Design Systematic review of studies evaluating the prevalence of von Willebrand disease in women with menorrhagia.
Setting Hospital outpatient clinics (mainly gynaecological) and population surveys.
Population Women presenting with menorrhagia.
Methods Relevant studies were extracted from MEDLINE search, bibliographies of identified articles and published proceedings of meetings and conferences.
Main outcome measures Number of women with von Willebrand disease.
Results Eleven studies were included, totalling 988 women with menorrhagia. One hundred and thirty-one women were diagnosed to have von Willebrand disease with prevalences in individual studies ranging from 5% to 24%. The overall prevalence was 13% (95% CI 11–15.6%). The prevalence was higher in the European studies—18% (95% CI 15–23%) compared with that in North American studies—10% (95% CI 7.5–13%). This difference (P= 0.007) is likely to be the result of differences in the studies, which include method of recruitment of study population, method of assessing menstrual blood loss ethnic composition of study population, criteria for diagnosis and use of race- and ABO blood group-specific values for von Willebrand factor.
Conclusions The prevalence of von Willebrand disease is increased in women with menorrhagia and is the underlying cause in a small but significant group of women with menorrhagia across the world. Testing for this disorder should be considered when investigating women with menorrhagia, especially those of Caucasian origin, those with no obvious pelvic pathology or with additional bleeding symptoms.
Over the last six years, there have been international efforts to assess the prevalence of von Willebrand disease in women suffering from menorrhagia. Menorrhagia is a common complaint in women of reproductive ages with 5% of women aged between 30 and 49 suffering from this condition1; menorrhagia comprises 12% of all gynaecology referrals.2 The contributory aetiology may be local or systemic but a specific cause is identified in less than 50% of affected women.3 Although menorrhagia is a very common symptom of women with von Willebrand disease4–6 and other inherited disorders of coagulation, its value as a predictor of mild bleeding disorders has only been studied seriously since the late 1990s. Type 1 von Willebrand disease, which accounts for 70% of cases, is usually mild and affected women may not have problems until faced with a haemostatic challenge, such as menstruation or childbirth. Thus, screening of women with menorrhagia would be an ideal way of identifying women with this disorder. However, establishing a diagnosis of von Willebrand disease, especially in its mild form, is difficult and complex7 and, in the absence of genetic diagnosis, we rely on a combination of clinical features and laboratory assessment of von Willebrand factor antigen (VWF:Ag), von Willebrand factor functional activity (VWF:Ac) or FVIII assay. The level of these coagulation factors in plasma is influenced by many factors such as age, race and blood group. Levels also fluctuate within the same woman at different phases of the menstrual cycle.8 While the prevalence of von Willebrand disease in the general population varies between 0.8% and 1.3%,9,10 the prevalence in women with menorrhagia has been reported to be much higher. This paper presents a systematic review of all the studies that have assessed the prevalence of von Willebrand disease in women with menorrhagia. It is hoped that the review will help to establish whether testing for this disorder should be part of routine investigation for women presenting with menorrhagia.
Information on studies was extracted from MEDLINE, bibliographies of published articles, abstracts from meetings of the American Society of Haematology, the International Society of Thrombosis and Haemostasis and the World Federation of Haemophilia. The key words for the MEDLINE search were von Willebrand, menorrhagia, women, prevalence and inherited bleeding disorders. Bibliographies of identified articles were scanned for any unidentified articles. Published proceedings of relevant conferences were hand searched. The authors of publications were contacted, where appropriate, to obtain additional information.
Studies were included if they assessed the prevalence of von Willebrand disease in women with menorrhagia. Studies looking exclusively at adolescent menorrhagia were excluded. All other studies were included.
Twelve studies were identified.11–22 Five14–16,20,22 of these were published as abstracts from Congress proceedings and we were able to contact the authors of one abstract.15 The abstract included results from preliminary data and after contacting the authors we were given data from the completed studies. The designs of the studies were compared. Although there were some similarities, the studies varied in the method of recruitment of the study population, demographic characteristics (particularly ethnic composition), criteria for diagnosis (the diagnostic test and its cutoff values), use of race- and ABO blood group-specific values and the method of assessing menstrual blood loss. One study21 was excluded because testing for von Willebrand disease was selective (criteria not clear) and not done in all participants. Tables 1 and 2 present details of the studies reviewed. Four studies11–13,15 tested two or more samples before diagnosis was made, while four studies11,12,15,19 tested only once in the first week of the menstrual cycle.
Table 1. Prevalence rates, age, race and blood group distribution of the studies reviewed.
A total of 988 women were involved in the 11 studies and 131 were diagnosed as having von Willebrand disease, with prevalence in the individual studies ranging from 5% to 24%.11–22 The prevalence rates with confidence intervals are given in Fig. 1. All studies reported prevalences that were higher than in the general population. Only three15,17,18 studies used controls for comparison while all other studies used general population prevalence quoted in the literature.9,10
In the various studies, the study population was recruited from either (1) the general population where women with a self-diagnosis of menorrhagia were invited to participate, (2) women with diagnosis of menorrhagia on the medical database or (3) from gynaecological clinics. Diagnosis of menorrhagia was made using objective methods (alkaline haematin or PBAC) in most European studies11–13 but in only one study15 from North America. Some uniformity is noted in the actual diagnosis of von Willebrand disease as all studies with one exception have used VWF activity as a means of diagnosing von Willebrand disease. The North American studies used blood group-specific levels and race-specific levels unlike European studies, which used standard common values. The threshold levels for diagnosis were also different in the European and North American studies. European studies used a value below the lower limit of normal range as diagnostic and this level was 50 IU/L. In contrast, most North American studies used two standard deviations below the control mean or the 2.5th centile as the threshold for diagnosis. In six studies a detailed bleeding history was taken and evaluated, the deatils are given in Table 3.
Table 3. Evaluation of additional bleeding symptoms.
Additional bleeding symptom in women with von Willebrand disease
Women without von Willebrand disease and additional bleeding symptoms (P)
All studies have shown that the prevalence of von Willebrand disease is higher in women with menorrhagia than in the general population although the prevalence rates vary. The differences in the studies could explain the varying prevalence rates. Studies that recruited women from the general population16,17 reported the lowest prevalence. It is possible that in population-based studies where a self-diagnosis of menorrhagia is used, some women may not have menorrhagia on objective testing accounting for the lower prevalence generally seen in these studies. Recruitment based on a database diagnosis of menorrhagia, undertaken in one American study,18 is fraught with the problem of not knowing how a diagnosis of menorrhagia was made in the first instance and, as such, may again include women who do not have true menorrhagia after objective testing. This again falsely increases the size of population under study and dilutes the prevalence rates.
As perception of menorrhagia is influenced by cultural, social and demographic factors, the way menorrhagia is diagnosed is also important. Subjective diagnosis is often inaccurate and the need for objective measurements of menstrual blood loss is well documented.23–26 Objective assessment of menorrhagia using alkaline haematin method26 or PBAC27 could avoid this problem. Although the alkaline haematin method27 is the most accurate estimate of menstrual blood loss, it is cumbersome for routine clinical practice and the alternative of PBAC28 is a simple tool that can be used in day-to-day clinical practice.
Another important factor that could affect prevalence levels is the ethnic mix of the population tested. The populations from North America are mixed with a significant proportion of black women. It has been shown that black women have higher levels of VWF antigen, activity and Factor VIII than white women,8,29,30 and more recently, studies have shown that ethnic variations in VWF levels can influence the diagnosis of von Willebrand disease.31 Studies in the review that have assessed the different prevalence in white and black women have found a lower prevalence in black women.18 This could affect the overall prevalence in studies with a significant mix of black women. Indeed, North American studies18,19 have shown prevalence rates in white women to be similar to those reported in European studies.11–13 The difference in prevalence between white and black women persists even after using race-specific levels for African women.18,19 It is possible, as suggested by one of the studies,18 that perception of menorrhagia is different between white and black women or that the prevalence of an alternate cause (e.g. fibroids in black women) may be higher and this could explain the lower prevalence of von Willebrand disease in black women (by increasing the number of women with menorrhagia and hence the denominator). It can also be hypothesised that the genetic defects that cause von Willebrand disease are truly less common in women of African descent than those of European ancestry. In studies like that of Kadir et al.,12 which included a significant number of black women (15.4%) but did not use race-specific cutoffs, there is a possibility that the prevalence could theoretically have been under-estimated. The lower prevalence of von Willebrand disease in black women needs further confirmation in studies using both objective methods of assessing menorrhagia and race-specific threshold levels for diagnosis.
Blood group O has been shown to be associated with lower VWF:Ag and VWF:Ac levels.9,32 The European studies11–14 did not use blood group specific levels and it can be argued that this could potentially inflate the prevalence of von Willebrand disease in these populations. Similarly, as all North American studies have used blood group-specific levels, one can speculate that their lower prevalence may be due to the use of ABO-specific levels. However, using ABO-specific levels for the diagnosis of von Willebrand disease may not be as relevant as previously thought. In a study by Nitu-Whalley et al.,33 bleeding symptoms were similar in blood groups O and non-O type I von Willebrand disease patients with VWF:Ac levels of 35–50 IU/mL.
The diagnosis of von Willebrand disease is complex and the diagnosis of mild forms, in particular, can be difficult.7 The laboratory diagnosis and classification of von Willebrand disease involves screening, diagnostic and discriminating tests. The diagnostic tests commonly used are plasma factor VIII level, VWF antigen and assessment of functional ability of VWF to bind platelets as determined by the ristocetin cofactor agglutination activity assay [VWF:RCo or RiCoF (older terminology)]. Measurement of VWF:Ac is the single most sensitive test for most forms of von Willebrand disease34,35 and the gold standard method to measure VWF:Ac is by VWF:RCo assay. Due to the poor reproducibility and considerable inter-assay and interlaboratory variability of the VWF:RiCo assay, there has been renewed interest in the VWF collagen-binding assay (VWF:CBA) as a measure of VWF function. Some authors believe VWF:CBA should not be considered as a replacement for VWF:RCo assay36 and using both VWF:RCo and VWF:CB can improve the ability to detect variants of von Willebrand disease. Different thresholds, either absolute values or lower centiles of a control population, are used for diagnosis and this could contribute to an apparent difference in prevalence rates.
Published data are conflicting with regard to variations in VWF levels during the menstrual cycle. Intra-individual variations in VWF:Ag and FVIII, especially in patients with von Willebrand disease, have been described by Abildgaard et al.37 and this may in part be related to the menstrual cycle as suggested by Mandalaki et al.38 Kadir et al.,8 using longitudinal data, have reported a strong cyclic variation related to the menstrual cycle in the levels of VWF:Ag and VWF:Ac. More recently, Miller et al.39 reported lower levels of VWF during menstruation compared with the follicular phase; however, Onundarson40 reported no change in the VWF level during the cycle. As most female hormones are at their lowest levels in the follicular phase, some authors recommend41,42 that blood sampling for clotting factors is done during the early follicular phase. Edlund et al.11 reported a very low variation when sampling was restricted to days five to seven of the menstrual cycle. In view of the variable levels, diagnosis should be based on more than one sample especially when the mild form of the disease is considered.37,43 As the day of testing and the number of samples taken varied in the studies reviewed, it is open to debate whether this was contributory to the different prevalence rates observed in the studies from North America and Europe. Oral contraceptives have been associated with increased VWF and this could lead to a false normal test result, however, the effect is dose dependent and appreciable above an oestrogen dose of 50 μg of ethinylestradiol.44 Studies looking at the effect of low dose oral pills have shown no difference in VWF levels between users and non-users.8,45,46 Kadir et al.8 noted that cyclic variations were dampened with the use of combined oral contraceptive pills.
Finally, it is difficult to comment on the significance of the prevalence of bleeding symptoms as a discriminatory factor. In studies where the frequency of bleeding symptoms in women without von Willebrand disease was available,12,14,18 the difference was statistically significant in only one study (Table 3). Some symptoms, like bleeding after tooth extraction, post-operative bleeding and postpartum bleeding, may be more valuable, as bleeding after trauma is the event most predictive of an inherited bleeding disorder.47 It is possible that statistical significance was not attained due to the small numbers involved and because many of the women presenting with menorrhagia have not been challenged by trauma or surgery. As menorrhagia is a common problem, it is not feasible to perform extensive coagulation testing in all women in view of the cost implications. The tests referred to in this review are secondary confirmatory and diagnostic tests. It would be desirable to have a screening test and some centres have recommended the use of PFA-100 (platelet function analyser) or bleeding time.48 The value of eliciting history of additional bleeding symptoms like postpartum haemorrhage, post-operative bleeding and bleeding after dental extraction lies in using it as screening tool in women with menorrhagia to select a population that would merit from coagulation testing.
It is reasonable to conclude that the prevalence of von Willebrand disease is increased in women with menorrhagia. Whether the regional differences noted are true remains to be evaluated. A multicentre study with similar recruitment criteria including objective assessment of menorrhagia and uniform diagnostic tests, threshold levels for diagnosis, number of samples and timing of samples may answer this question but its value and a justification of the costs involved are debatable. The results of this systematic review suggest it is reasonable to believe that von Willebrand disease is the underlying cause for menorrhagia in a small but significant proportion of women. Testing for this bleeding disorder should be considered when investigating women with menorrhagia, especially those of Caucasian origin, those with no obvious pelvic pathology (normal pelvic examination and normal pelvic scan) or additional bleeding symptoms and for all women prior to opting for surgical intervention. It now remains to make family doctors and gynaecologists more aware of this problem so that timely referral may be made and diagnosis established. However, this should ideally be preceded by the development of guidelines for screening and subsequent management of von Willebrand disease. Close collaboration between gynaecologists and haematologists should be established in the management of women with menorrhagia. Awareness of von Willebrand disease as a cause for menorrhagia is important for the following reasons—health implications like future surgery and childbirth; effective medical treatment of menorrhagia with desmopressin nasal spray49 in addition to oral contraceptive pill and antifibrinolytic agents, used singly or in combination; avoidance of major surgical intervention. Furthermore, von Willebrand disease is an inherited disorder with autosomal dominant inheritance and identification of a case would lead to screening and early diagnosis of von Willebrand disease in relatives of the index case.