Intended for healthcare professionals


Third generation oral contraception and venous thromboembolism

BMJ 1996; 312 doi: (Published 13 January 1996) Cite this as: BMJ 1996;312:68
  1. Klim McPherson
  1. Professor of public health epidemiology Health Promotion Sciences Unit, Department of Public Health and Policy, London School of Hygiene and Tropical Medicine, London WC1E 7HT

    The published evidence confirms the Committee on Safety of Medicine's concerns

    Eighteen weeks ago, Britain's Committee on Safety of Medicines raised concerns about the newest brands of oral contraceptive pill,1 responding to new evidence that pills containing desogestral and gestodene conferred a two times greater risk of venous thromboembolism than pills containing other progestagens. At the time, none of the data on which the committee's announcement was based had been published. Dissatisfied doctors, anxious patients, and hundreds of column inches in the press were the natural consequence of this imposed uncertainty.2 Data from three case control studies, one of them nested in a cohort study, have since been published,3 4 5 6 and this week's BMJ carries two papers reporting data from a third case control study (pp 83, 88).7 8

    The studies published in December were a subanalysis of data from the large WHO study of women in 10 countries exposed to third generation oral contraceptive pills3 4; a casecontrol study of current users of the oral contraceptive pill from the British general practice research database5; and a reanalysis of the Leiden thrombophilia study.6 All studies indicated a statistically significant doubling of the adjusted odds ratios for venous thromboembolism in patients taking third rather than second generation oral contraceptive pills. These results are consistent with those from the transnational study published in this issue of the BMJ. The increased risk cannot be explained by known or expected bias or confounding.

    The first law of epidemiology is that if a causal effect is large enough, it will show up despite all the problems of performing, analysing, and interpreting observational studies on real people. The recent studies are a case in point. Despite their different designs (with different funding arrangements) and different populations, they show a similar size and direction of effect. The doubling of risk of venous thromboembolism in users of third generation pills is important when the baseline risk in users of the pill is already three times greater than in non-users. Some studies have reported a relative risk of venous thromboembolism of about 9 for users of third generation pills compared to women using non-hormonal contraception, ignoring high risk subgroups.4 5 7 Fortunately, venous thromboembolism remains rare among young women, but any added risk is crucial when treating healthy women and when balancing risks against possible benefits such as protection against myocardial infarction.

    The effects of third generation oral contraceptive pills on the incidence of myocardial infarction remain uncertain and will now be difficult to study. The paper by Lewis et al shows that the reduction in incidence of nearly threefold among users of third compared to second generation oral contraceptive pills is not statistically significant, and there is no apparent difference with third generation pills compared to non-hormonal contraception.8 These results suggest that third generation pills may, in the long term, be beneficial, but such preliminary data cannot provide a reliable basis for policy decisions on safe contraception. The finding was not confirmed by Jick et al,5 and in view of the committee's pronouncement, further studies of third generation oral contraceptive pills in unselected women seem unlikely.

    Now that the data are published, it seems clear that the Committee on Safety of Medicines did what it had to do. Delaying its announcement until all the studies had been published would have incurred a further 500000 uninformed woman years of use of the third generation oral contraceptive pills, which might have resulted in 80 new cases of venous thromboembolism and possibly one death. We have no clear idea how many women responded to the committee's advice and in what way, so these numbers are speculative, but the number of additional unwanted pregnancies, and related cases of venous thromboembolism, will appear in official statistics in a few months' time. The cost in terms of anxiety and the new side effects from switching over to second generation pills cannot yet be measured. However, the committee's advice should have included, for all the data available, simple standardised point estimates of risk (with corresponding confidence intervals) for second and third generation oral contraceptive pills. People would then have been able to assess the nature and consistency of the evidence for themselves.

    From the four studies we now have a pooled estimate of relative risk of venous thromboembolism in unselected women of around 2.0 (95% confidence interval 1.4 to 2.7). We also have important data, from the second paper published in this issue8 and from other studies,9 that the greatest risk to health comes from smoking while taking the pill, rather than from the type of pill being used. Young women smokers who use the pill are 10 times more likely to suffer myocardial infarction than users who don't smoke (odds ratio 10.1; 5.7 to 17.9).8 This risk is higher among users of second than third generation pills, but the difference, unfortunately for those who believe that third generation pills improve the lipid profile, is not statistically significant. During the next century, large numbers of young women will certainly die from smoking related diseases; about half of regular smokers will die prematurely from their addictive habit.10 Although the secretary of state for health, Stephen Dorrell, felt unable to ignore the expert advice from the Committee on Safety of Medicines, he continues to ignore the published advice of his own department's report.11 By his prompt action, he may have prevented one death from venous thromboembolism, but he will not be around to count the cost of his government's inaction in vital areas of tobacco policy.

    Now that the dust has settled, the message for doctors seems clear: when advising patients on which pill to take, a careful personal and family history for increased risk of venous thromboembolism is essential, followed where appropriate by screening for thrombophilia. Women with risk factors for venous thromboembolism should probably not start taking third generation pills; but once informed of the small excess risk, women already taking them satisfactorily may choose to continue using them. Let us hope that any disproportionate fear of litigation soon gives way to intelligent and well informed collaborative decision making.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    9. 9.
    10. 10.
    11. 11.
    View Abstract