The risk of venous thromboembolism associated with the use of tranexamic acid and other drugs used to treat menorrhagia: a case–control study using the General Practice Research Database
Mr A Sundström, Centre for Pharmacoepidemiology, Clinical Epidemiology Unit, Karolinska University Hospital, Solna, Building M9:01, SE-171 76 Stockholm, Sweden. Email firstname.lastname@example.org, email@example.com
Objective To assess whether use of tranexamic acid is associated with an increased risk of venous thromboembolism (VTE).
Design Nested case–control study.
Setting Database study using the General Practice Research Database for the years 1992–1998.
Population Women aged 15–49 years with a diagnosis of menorrhagia.
Methods Multivariate conditional logistic regression was used to estimate the risk for VTE associated with different drug treatments for menorrhagia, adjusting for confounders.
Main outcome measures Adjusted odds ratios with 95% CI.
Results A total of 134 cases of VTE and 552 matched controls were identified. Recent use of tranexamic acid was scarce, yielding an adjusted odds ratio for VTE of 3.20 (95% CI 0.65–15.78). The use of mefenamic acid (ORadj 5.54 [95% CI 2.13–14.40]) or norethisterone (ORadj 2.41 [95% CI 1.00–5.78]) was associated with an increased risk of VTE, as was a recent—in relation to menorrhagia—diagnosis of anaemia or a haemoglobin value <11.5 g/dl (ORadj 2.23 [95% CI 1.02–4.86]).
Conclusions We found that tranexamic acid was associated with an increased risk of VTE, although the risk estimate did not reach statistical significance. Increased risks of VTE associated with other treatments for menorrhagia were observed. The increased risk of VTE observed with a diagnosis of anaemia—a proxy for more severe menorrhagia—suggests that menorrhagia could be a prothrombotic condition. The observed association between VTE, tranexamic acid and other treatments for menorrhagia may thus partly be explained by confounding by indication. The possibility that menorrhagia is itself a risk factor for VTE merits further investigation.
Tranexamic acid has been used since 1966 for the treatment of menorrhagia.
Menorrhagia is defined as excessive menstrual bleeding occurring over several consecutive cycles. Objectively, it is a total menstrual blood loss of at least 80 ml per menstruation. Approximately 30% of women complain of menorrhagia, and the condition accounts for almost 25% of gynaecological operations.1,2 Menorrhagia may be the result of uterine, systemic or iatrogenic causes. Uterine causes include conditions such as fibroma, adenomyosis or dysfunctional uterine bleeding. Systemic causes could be either haematological or endocrinological disorders such as hypothyroidism. Of the iatrogenic causes, intrauterine devices and use of anticoagulants are the most common causes.
Tranexamic acid prevents the decomposition of fibrin in clotted blood by blocking the activation of plasminogen. This leads to a cessation of further bleeding. Since its introduction, there has been a suspicion that tranexamic acid, as with any agent that stops bleeding and inhibits the dissolution of thrombi, might indirectly predispose to thrombosis.
In the British National Formulary, thromboembolic events are listed as possible adverse effects to use of tranexamic acid.3 In Martindale, however, it is stated that reported thrombotic complications are usually a consequence of its inappropriate use:4 tranexamic acid is contraindicated in women with a predisposition to thrombosis and in women with active intravascular clotting.
Some reports have suggested an association between tranexamic acid and thrombosis,5–9 whereas others could not confirm this suspicion.10–14 A Cochrane review has highlighted the need for population-based studies on the association between plasminogen activator inhibitors and venous thromboembolism (VTE).15 To our knowledge, the only observational, population-based study on the association between use of tranexamic acid for menorrhagia and VTE showed a negative association between tranexamic acid and VTE, with an odds ratio of 0.55 (95% CI 0.31–0.97).16
Tranexamic acid is used in primary care mainly to treat menorrhagia—other indications for its use are to prevent bleeding in patients who have undergone surgery or for other bleeding disorders such as epistaxis.
The primary objective of this study was to investigate whether the use of tranexamic acid, when used to treat menorrhagia, is associated with an increased risk of VTE (deep vein thrombosis [DVT] or pulmonary embolism) and to put that risk into context by investigating the risk of VTE associated with use of other drugs approved for treatment of menorrhagia.
We used data from the General Practice Research Database (GPRD) for our investigation. The GPRD is an anonymised database containing general practice patient records for approximately 4% of the population in the UK; data on diseases and symptoms, prescribed drugs and lifestyle factors are linked to the individual patient’s records. The database has been used extensively in studies of morbidity, drug use and adverse effects, and its validity has been described in a large number of previous studies.17–19 Since 1992, most data are considered to have reached research standard (up to standard).
The use of the GPRD for studies of this kind has multicentre research ethics committee approval (06/MREC 04/72). Furthermore, this study was approved by the Scientific and Ethical Advisory Group for the GPRD.
Women were eligible to be included in the study population after 6 months of registration with a practice contributing up-to-standard data. Women with a diagnosis of menorrhagia who were aged between 15 and 49 years were selected in the study population and remained ‘at risk’ of VTE until they left the practice, reached the age of 50 years, died or had a first episode of VTE, whichever occurred first, during the period 1992–1998.
Identification of cases
A case was defined by a diagnosis of pulmonary embolism or DVT, confirmed by evidence of anticoagulant therapy or death from a cause consistent with VTE during the study period. This method for identifying cases of VTE in the GPRD has been validated in a previous study.17 The first date of recording a diagnosis indicating symptoms of the VTE (e.g. haemoptysis, shortness of breath, chest pain and swelling or redness of a limb) was established and recorded as the ‘event’ (index) date. Women with known risk factors for VTE, for example history of VTE, recent trauma or surgery within 42 days before the event, a diagnosis of cancer recorded within 90 days before and after the event, a diagnosis of systemic lupus erythematosus, multiple sclerosis or Crohn’s disease, were excluded. Women with a record indicating pregnancy (e.g. antenatal care, termination of pregnancy, miscarriage or birth) within 42 days before the VTE event date were also excluded because pregnancy is a well-known risk factor for VTE and also because they were not at risk of menorrhagia during the time preceding the event. Women who had a prescription for a combined oral contraceptive (OC) that covered the event date were also excluded.20
Selection of controls
For each case, up to eight controls were selected at random from the study base and matched by practice, year of birth and index date. The exclusion criteria applied to cases were applied similarly to controls. After exclusions, there were on average four controls matched to each case. The matching on index date meant that the control had to qualify for inclusion in the study at that date, that is to have a recorded diagnosis of menorrhagia before the index date.
Definitions of exposure and confounding factors
Cases and controls were deemed to be exposed to tranexamic acid or to other medications used to treat menorrhagia, that is mefenamic acid, norethisterone or ethamsylate, if they had had a prescription during 90 days before the index date. Treatments for menorrhagia are symptomatic treatments, so defining exposure status on a particular day is difficult. We calculated the prescription intervals for menorrhagic treatments among women with menorrhagia, and these were used for classifying individual subject’s exposure status.
A diagnosis of fibroma within 90 days before or after a diagnosis of menorrhagia or ever in the study period was identified. Anaemia was identified by records of that diagnosis within 14 days before or after a diagnosis of menorrhagia or by a haemoglobin (Hb) measurement <11.5 g/dl in the same time window.21
Smoking habit and body mass index (BMI) closest to the index date were ascertained: records on smoking habits were extracted, as were records on height and weight, from which measurements BMI was calculated. A number of chronic conditions were identified: treated asthma first recorded at least 42 days before the index date, treated diabetes mellitus or thyroid disease before the index date and renal disease at any time in the study period.
In the analyses, the risk of VTE among women exposed to different potential risk factors was compared with unexposed subjects by calculating the odds ratio with 95% CI by means of conditional logistic regression analysis.
For the multivariate regression analysis, risk factors were included if, when adding a single variable to a model containing the treatment variables indicating use of tranexamic acid, mefenamic acid or norethisterone, that risk factor changed the odds ratio of any of the treatment variables by more than 10%. All variables listed in Table 1 were tested for inclusion in the multivariate model in this way.
Table 1. Characteristics of cases and controls with associated crude odds ratios with 95% CI derived by univariate conditional logistic regression analysis
|Obesity (BMI ≥ 30 kg/m2)*||36 (26.9)||99 (17.9)||1.65||(1.04–2.63)|
|BMI unknown*||26 (19.4)||113 (20.5)||0.92||(0.53–1.58)|
|Smoker (yes/no)**||49 (36.6)||146 (26.4)||1.75||(1.14–2.69)|
|Smoking status unknown**||24 (17.9)||80 (14.5)||1.49||(0.81–2.73)|
|Diagnoses in relation to menorrhagic diagnosis|
|Diagnosis of anaemia or Hb <11.5 g/l within 14 days before or after diagnosis of menorrhagia||12 (9.0)||27 (4.9)||1.93||(0.94–3.96)|
|Diagnosis of fibroma within 90 days before or after a diagnosis of menorrhagia||3 (2.2)||6 (1.1)||2.72||(0.68–10.94)|
|Diagnosis of fibroma (at any time)||6 (4.5)||38 (6.9)||0.71||(0.29–1.72)|
|Concomitant diseases in relation to index date|
|Varicose veins (within 1 year before index)||5 (3.7)||7 (1.3)||3.61||(1.07–12.15)|
|Treated asthma (1st diagnosis at least 6 weeks before index)||15 (11.2)||40 (7.2)||1.45||(0.76–2.77)|
|Treated diabetes mellitus (before index)||6 (4.5)||3 (0.5)||7.43||(1.82–30.34)|
|Treated thyroid disease (before index)||8 (6.0)||20 (3.6)||1.75||(0.76–4.05)|
|Renal disease (at any time)||8 (6.0)||12 (2.2)||2.87||(1.10–7.45)|
Subgroup analyses were performed among (a) women with more severe or a more definite diagnosis of menorrhagia defined as a diagnosis of anaemia or a measurement of Hb <11.5 g/dl, within 14 days either before or after a diagnosis of menorrhagia, (b) women with a diagnosis of fibroma, either within 90 days before or after a diagnosis of menorrhagia or ever during the study period. Finally, subgroup analyses were also performed among women without these diagnoses.
All statistical analyses were performed in Stata for Windows, version 8.0 (Stata, College Station, TX, USA).
Among the cohort of women with menorrhagia (122 237 women), we identified 134 cases and 552 matched controls after applying the exclusion criteria.
Fifty-two cases (39%) with a mean age of 41.0 years suffered a pulmonary embolism [of which 4 (7.7%) with fatal outcome] and 82 (61%) with a mean age of 41.7 years had a DVT (of which none with a fatal outcome). The characteristics of cases and controls are presented in Table 1, including lifestyle factors and comorbidity. The odds ratios presented in Table 1 estimate the unadjusted odds ratio of VTE associated with each variable listed.
The use of ethamsylate was scarce: only one case and one control. In the subsequent analyses, use of ethamsylate medication was not considered. Twenty-three (17%) of the 134 cases and 33 (6%) of the 552 controls had been exposed to tranexamic acid, mefenamic acid or norethisterone. No concurrent use of these drugs was present, with the exception of one case who used both norethisterone and mefenamic acid. When excluding that single case, the observed odds ratios changed marginally (data not shown).
We present crude and adjusted odds ratios with 95% CI for VTE with exposure to these three drugs in Table 2. All three treatments appeared to be associated with an increased risk of VTE, but the risk associated with tranexamic acid was not statistically significant (ORadj 3.20 [95% CI 0.65–15.78]). The strongest association with VTE was seen with the use of mefenamic acid (ORadj 5.54 [95% CI 2.13–14.40]).
Table 2. Crude and adjusted odds ratios for VTE associated with different factors as derived by univariate and multivariate conditional logistic regression analysis
|Tranexamic acid||3||4||2.97 (0.65, 13.65)||3.20 (0.65–15.78)|
|Mefenamic acid||10||12||4.14 (1.70, 10.09)||5.54 (2.13–14.40)|
|Norethisterone||10||17||2.67 (1.15, 6.19)||2.41 (1.00–5.78)|
A diagnosis of anaemia or an Hb level <11.5 g/dl during 14 days before or after a record of menorrhagia was found to be associated with VTE (ORadj 2.23 [95% CI 1.02–4.86]). In addition, obesity, that is a BMI ≥30 kg/m2, was associated with an increased risk of VTE (ORadj 1.96 [95% CI 1.20–3.21]), as was smoking (ORadj 2.18 [95% CI 1.36–3.49]) and having an unknown smoking status (ORadj 2.81 [95% CI 1.31–6.04]).
The importance of the timing of a diagnosis of menorrhagia was assessed by performing subgroup analyses of cases and controls who had had their most recent diagnosis of menorrhagia recorded within 1 year before or after the index date (56 cases and 150 controls). The unadjusted odds ratio for the association between VTE and tranexamic acid (three cases and four controls) in that subgroup was 1.79 (95% CI 0.38–8.53), for mefenamic acid (nine cases and four controls) unadjusted odds ratio was 6.70 (95% CI 1.96–22.89) and for norethisterone (eight cases and ten controls) unadjusted odds ratio was 2.31 (95% CI 0.86–6.21).
A diagnosis of anaemia or an Hb level <11.5 g/dl within 14 days of a diagnosis of menorrhagia was found for 9% of the cases compared with 5% of the controls. Use of drugs to treat menorrhagia among women with anaemia was very rare: no case or control used tranexamic acid, two cases and two controls used mefenamic acid, no case and two controls used norethisterone and no case or control used ethamsylate. A recent diagnosis of fibroma was also scarce (only 2.2% of cases and 1.1% of controls). A diagnosis of fibroma at any time during the study period was less common among cases than controls (4.5 versus 6.9%). In women with any record of a fibroma, exposure to drugs for menorrhagia was extremely scarce: no cases or controls were using tranexamic acid, one case and one control used mefenamic acid, one case and six controls used norethisterone and one control used ethamsylate. In these subgroups, use of drugs to treat menorrhagia was thus too infrequent to allow for estimations of odds ratios for VTE.
Among women without anaemia or fibroma, the estimated odds ratios for VTE associated with the three treatments for menorrhagia were very similar to those for the whole population (data not shown).
The use of tranexamic acid for the treatment of menorrhagia was scarce, and no statistically significant association with risk of VTE was observed, although the point estimate of the odds ratio was 3.2. In fact, the study only had a 45% power to detect an odds ratio of 3 (exposure prevalence 0.007; four controls per case; z-alpha 1.96); it had an 80% power to detect an odds ratio of ≈5. Use of mefenamic acid and norethisterone was associated with significantly increased risks of VTE, with observed odds ratios of 5.5 and 2.4, respectively. Even though the primary objective of the study was to assess the relationship between VTE and use of tranexamic acid, the results regarding the two latter associations in this observational study may be regarded as suggestive evidence for these associations. The number of users of ethamsylate was too low to allow for any conclusions regarding the risk for VTE.
This study was based on data from the GPRD—a database that has been used extensively for epidemiological studies. The validity of the method by which cases of VTE were identified from the database for this study has been examined and found to be reliable.17 A number of limitations to the study are important to consider however.
First is the potential for misclassification of disease and/or outcomes. With regard to the use of drugs prescribed to treat menorrhagia, the database holds information on prescriptions but not on actual use. This might constitute less problem regarding the use of chronic treatments requiring repeated prescriptions (e.g. OCs, antihypertensives, lipid-lowering drugs etc.), but drugs used to treat menorrhagia are often used intermittently making the exposure classification uncertain regarding timing and duration of use. However, if a misclassification has occurred, there is no reason to assume that it should be different between cases and controls. A nondifferential misclassification would only mean that the associations between these treatments and VTE may have been diluted, leading to an underestimation of the odds ratios associated with treatment.
Since menorrhagia is a symptom with many possible underlying causes, all with different latency and induction times, the use of the first date of diagnosis as the entry into ‘time at risk’ is the most reasonable, even if prescriptions for the condition may have been issued previously, but without the entry of a corresponding diagnosis. The inclusion of unexposed (to menorrhagia) time as ‘exposed’ to the study base may cause biased results regarding the associations between the treatments and VTE.
Misclassification of concomitant medical conditions is probably not an issue regarding chronic diseases; individuals with such diseases can be assumed to attend medical care regularly, and their concomitant diseases are therefore often well recorded. For menorrhagia, however, it is probably only women with pronounced problems with this condition who seek medical advice. Thus, some underdiagnosis may have occurred, making the size of the cohort of women with menorrhagia smaller than it potentially could have been. At the same time, it is possible that the women with a diagnosis of menorrhagia suffer from the condition with varying degrees of severity.
Even if fibroma is a common cause of menorrhagia, fibromas were too scarce in this study population to allow for any certain conclusions regarding their role in the aetiology of VTE. A diagnosis of fibroma was neither a risk factor for VTE nor a confounder in the association between treatments for menorrhagia and VTE in this analysis.
In our cohort of 686 women, we identified 39 (12 cases and 27 controls) women with a record of anaemia or an Hb measurement <11.5 g/dl within 14 days before or after a diagnosis of menorrhagia in an attempt to identify women with more severe menorrhagia. The low number of women with anaemia did not allow for estimations of odds ratios for exposure to the drugs for menorrhagia in this specific subgroup. However, in the absence of anaemia, the observed odds ratios for menorrhagic drugs differed very little from those observed in the analysis of all cases and controls.
However, anaemia was associated with an increased risk of VTE, also after adjustment for the drugs used to treat menorrhagia, with an adjusted odds ratio of 2.23 (95% CI 1.02–4.86); this could probably indicate that a more severe or definite menorrhagia is associated with the outcome. This raises the possibility that menorrhagia is a condition that in itself entails an increased risk for VTE and that the observed association between VTE and treatments for menorrhagia may partly be explained by the underlying disease.
Menorrhagia without apparent causes has in some studies been attributed—to varying extents—to different bleeding disorders, for example platelet defects such as von Willebrand’s disease.22–25 The association between menorrhagia and haemophilic conditions may make a relation between menorrhagia and thromboembolism counterintuitive.
However, disturbed haemostasis can change rapidly between increased coagulation and fibrinolysis. External factors, such as medications, may influence the delicate balance in haemostasis and direct it towards coagulation, resulting in thromboses.
Development of pulmonary embolism in a woman with severe acquired haemophilia and treated with tranexamic acid has been described previously.8
The observed relationship between norethisterone and risk of VTE could be causal—according to the British National Formulary,3 progestogens should be used with caution in those susceptible to thromboembolism. VTEs associated with the use of high doses of therapeutic progestogens—albeit not norethisterone—have been reported,26 and a mechanistic study showed that norethisterone had a very small effect on the procoagulant activity of blood vessels compared with controls and with other progestins.27
Use of mefenamic acid was strongly associated with VTE in the multivariate model, with an odds ratio of 5.54 (95% CI 2.13–14.40). An association between mefenamic acid and VTE has not been suspected previously. Mefenamic acid (an anthranilic acid derivative) is a nonsteroidal anti-inflammatory drug, a drug group associated with serious cardiovascular thrombotic events, stroke and myocardial infarction.28,29 However, to our knowledge, a causal link between the use of NSAIDs and venous thrombotic events has not been observed or suggested.
We observed an increased risk of VTE with tranexamic acid, albeit not statistically significant, which is interesting since, to our knowledge, the only observational, population-based study on the use of tranexamic acid for menorrhagia and its relation with VTE performed by Berntorp et al.16 in Sweden showed a negative association between venous thrombosis and use of the drug. However, the use of tranexamic acid in Sweden is much more common—4% of the controls in the Berntorp study had used tranexamic acid during 1 month before the event—than in the UK, where only 0.7% of the controls in this study had had a prescription for the drug during 90 days before the index. According to sales data from IMS Health, the number of defined daily doses (DDD; 1 DDD of tranexamic acid = 2 g30) sold per 1000 inhabitants per day in Sweden was 0.53 in 1997, compared with 0.14 in the UK, that is sales were more than three times as high as in Sweden. The underlying risk profile of women receiving tranexamic acid in the UK may thus be assumed to be different from that of women in Sweden. Although more importantly, the Swedish study presented only unadjusted odds ratios, and it included users of both oral and injected hormonal contraceptives in the analyses. In our opinion, the results of the Berntorp study and the present one are not comparable because the study methods and analyses differed.
As described by Rosendaal,31 VTE is a disease caused by several interacting factors. An individual can carry or be exposed to various factors at different ages, and those factors can have a different impact on the total risk for VTE depending on age. One single, and in itself not very impressive, factor can be added to the individual’s other combination of risk factors, and thus ‘exceed the thrombosis threshold’, as formulated by Rosendaal, which leads to a thromboembolic episode. Some women with menorrhagia could carry with them genetic and/or acquired risk factors that make them more susceptible to VTE, and again for some of those women, untreated menorrhagia or medications used for the condition, or indeed a combination of both disease and its therapy, may be a more or less important part of the multifactorial causal pathway to a VTE.
In summary, the use of mefenamic acid or norethisterone among women with a diagnosis of menorrhagia was associated with an increased risk of VTE. An increased risk of VTE was found for tranexamic acid too, but the number of exposed women was too few to reach a definite conclusion. The finding that a diagnosis of anaemia or a low Hb, which we considered a proxy for more severe menorrhagia, was associated with an increased risk of VTE points to the possibility that menorrhagia could be a prothrombotic condition in itself. We thus cannot rule out the possibility that the observed associations between treatments for menorrhagia and risk of VTE are partly due to confounding by indication, that is partly caused by the underlying condition. The possibility of menorrhagia being a risk factor of VTE merits further investigation, as does possibly the observed association between mefenamic acid and thromboembolism.
Disclosure of interests
A.S., H.S., H.K. and L.A.: no financial, personal, political, intellectual or religious interests to state in relation to the subject matter of this article.
Contribution to authorship
A.S. participated in the design of the study and in some of the data extraction. He was responsible for the majority of the statistical analyses and was responsible for writing the first drafts and for finalising the manuscript. H.S. participated in the design of the study, in the extraction of data and in writing and revising the manuscript. H.K. initiated additional steps in the analyses and took part in the analyses and in the revisions of the manuscript. L.A. was the principal investigator in this study and he was responsible for the design and overlooked the analyses and the revisions of the manuscript.
Details of ethical approval
This study was approved by the Scientific and Ethical Advisory Group (SEAG) for the GPRD on 12 April 2006, protocol 757.
The use of the GPRD for studies of this kind has multicentre research ethics committee approval (06/MREC 04/72).
This study was funded by an unconditional grant to the Institute for Environmental Medicine, Karolinska Institute, from Pharmacia UpJohn, Helsingborg, Sweden at the time marketing authorisation holder of Cyklokapron® (tranexamic acid).
Prof emeritus R Farmer is acknowledged for fruitful discussions on initial study design and protocol and Dr TJ Williams for invaluable assistance to L.A. and A.S. in the introduction to the GPRD and for the data extraction.