Oral contraceptives and venous thromboembolism

BMJ 2009; 339 doi: http://dx.doi.org/10.1136/bmj.b3164 (Published 13 August 2009) Cite this as: BMJ 2009;339:b3164
  1. Nick Dunn, senior lecturer in medical education
  1. 1Biomedical Sciences Building, University of Southampton Medical School, Southampton SO16 7PX
  1. nick.dunn{at}soton.ac.uk

    Pills containing either levonorgestrel or norethisterone with the lowest possible dose of oestrogen are advised as first choice.

    More than 100 million women use the oral contraceptive pill worldwide.1 Many types of pill are available—26 types are listed in the British National Formulary—and the choice of which one to use is important to the women who use them and their doctors. Two linked studies assess the risk of venous thromboembolism in women taking the combined oral contraceptive.2 3

    All oral contraceptives are effective in preventing pregnancy if they are taken correctly, so the choice of which one to use rests on the profile of side effects. Venous thromboembolism is one of the most serious side effects, and although it is rare, it can cause death (in about 1-2% of all cases of venous thromboembolism in women taking the pill).

    Van Hylckama Vlieg and colleagues (doi:10.1136/bmj.b2921) present a case-control study from six anticoagulation clinics in the Netherlands, and Lidegaard and colleagues (doi:10.1136/bmj.b2890) describe a cohort study in Danish women followed for 10 years from January 1995. Despite their different designs they produce remarkably similar results and confirm past studies of the risk of venous thromboembolism with the pill. Progestagen is important—risk is lowest for levonorgestrel and norethisterone, and relative risk is 1.5-2 for gestodene, desogestrel, and norgestimate compared with levonorgestrel.

    These studies provide reliable estimates of the risk of venous thromboembolism associated with the newer progestagen drosperinone. This progestagen is said to have pharmacological properties that closely resemble natural progesterone and not to increase the risk of venous thromboembolism.4 5 They also provide estimates of the risks associated with cyproterone acetate, an anti-androgen used to treat acne vulgaris, and with norgestimate, a progestagen similar to desogestrel. Drosperinone and cyproterone acetate were found to have similar risks to desogestrel and gestodene (about 1.6-1.8 greater than with levonorgestrel), whereas the risk with norgestimate was similar to that with levonorgestrel in the Danish study (relative risk 1.19, 95% confidence interval 0.96 to 1.47)3 and was reported to be rarely used in the Dutch study.2

    Another finding common to both studies is that reducing the dose of oestrogen from 30 μg to 20 μg of ethinylestradiol seems to reduce the risk of venous thromboembolism further; in the Danish study the relative risk for desogestrel and gestodene type pills fell from 1.8 to 1.5 times the risk for levonorgestrel users. The Dutch study broadly concurred.

    Both of these studies were observational and were therefore prone to confounding and bias. One weakness with the Danish study, carried out by merging national databases, was that the authors could not control for a family history of venous thromboembolic disease, particularly as a result of the presence of factor V Leiden. Carriers of the mutation who use the oral contraceptive pill have a 35 times higher risk of venous thromboembolism than non-carriers who are not taking the pill.6 This mutation has a prevalence of about 5% in white Europeans. Thus, this factor is an important potential confounder because doctors prescribing the pill are likely to take family history into account and choose what they perceive to be a safer product (for example, one containing the newly marketed drosperinone). The Dutch study dealt with this problem by asking participants for a family history of thrombotic disease. However, this may not have provided enough information to allow complete statistical adjustment, and the potential influence of this genetic predisposition in both these studies is unclear. However, the remarkable concordance of the main results from these studies supports their validity.

    What do the results mean for clinical practice? A clinical review also published in this issue assesses all methods of female contraception (doi:10.1136/bmj.b2895),7 and when discussing oral contraceptives it recommends those containing levonorgestrel or norethisterone, with as low a dose of oestrogen as possible. All of the more recent progestogens, possibly except norgestimate, now seem to be at a disadvantage with regard to venous thromboembolism. However, the absolute risk of having venous thromboembolism is low—the baseline risk is five per 100 000 person years, and this increases to about 15-25 per 100 000 person years when taking the pill.1 This incidence is low enough to enable some negotiation when dealing with individual patients, for whom personal experiences or prejudices about side effects should be considered, and a pill containing a recent progestagen or a higher dose of oestrogen may be more appropriate. Patients with a personal or family history of venous thromboembolism should not take combined oral contraceptives. Contrary to popular belief, smoking is not a risk factor for venous thromboembolism, and obesity (despite being associated with venous thromboembolism) is not a contraindication to using the pill, although choosing low risk varieties is sensible.


    Cite this as: BMJ 2009;339:b3164