Data from transnational study of oral contraceptives have been misused

BMJ 1995; 311 doi: https://doi.org/10.1136/bmj.311.7013.1162 (Published 28 October 1995) Cite this as: BMJ 1995;311:1162
  1. Walter O Spitzer
  1. Strathcona professor of epidemiology and of medicine McGill University, Montreal, Canada

    EDITOR,--The unpublished data from the transnational study of oral contraceptives have been misused and misinterpreted in Britain because of inexplicable haste and urgency not thought necessary by other major regulatory agencies. My colleagues and I do not agree that our findings justify the drastic actions taken in Britian, whereby a “Dear Doctor” letter was sent to doctors, warning that oral contraceptives containing desogestrel and gestodene are associated with an increased risk of thromboembolism (pp 1112, 1117, and 1172).1 Doctors should be vigilant but not stampeded. Patients need not panic. In particular, the risk estimates in Britain are systematically lower than those in the four other European countries that participated in the study.

    On 8 October my colleagues and I completed the analysis of venous thromboembolism in this study (the two other components examine myocadial infarction and stroke). Voluntarily, we briefed medical officers at the Medicines Control Agency and a subcommittee of the Committee on Safety of Medicines within two days. As we did with the German drug safety authorities, we shared the data in confidence, believing that this was in the public interest. To our dismay, without consulting the scientists who had provided the data after five years' fieldwork, the Medicines Control Agency and Committee on Safety of Medicines invoked our findings four working days later to justify sending out the “Dear Doctor” letter.

    Unchallenged, that letter will effectively eliminate half of the oral contraceptives now on the British market. This effectively bans the drug while letting the committee off the hook about a banning decision. Despite Professor Michael Rawlins's attempt to mitigate the negative impact, its message seems to have caused a “pill scare,” given the newspaper headlines the morning after. As a pharmacoepidemiologist and principal investigator of the study I believe that the associations shown do not justify haste in decision making. After one week's reflection my colleagues and I are uncertain whether the modest increases in the odds ratio for venous thromboembolism can be best explained by bias or as causally significant. Our statistical advisory group believes that five unavoidable biases might be driving the risks up spuriously. Moreover, conventional epidemiological wisdom holds that when drug safety is being considered, relative risks (such as odds ratios) should give way to absolute risks or attributable risks when they are available or can be derived. We presented attributable risks to the Medicines Control Agency and Committee on Safety of Medicines. They were very low, and yet the “Dear Doctor” letter states “twice the risk,” an expression of relative risk that is both misleading and alarming.

    Our results for venous thromboembolism are equivocal, being neither fully reassuring nor alarming. But I thought that the absence of danger or imminent danger afforded the opportunity to evaluate the results carefully and deliberately, to present them to scientific critique and feedback, and to submit them to a peer reviewed journal. The Medicines Control Agency and Committee on Safety of Medicines chose to short circuit the formal peer review process to make decisions about third generation oral contraceptives in the absence of a community health emergency. This is an unacceptable precedent, and I vigorously object to having our unpublished data used thus without the benefit of the investigators' interpretation of the findings or the assessment of disinterested peers.

    Conversely, preliminary estimates of attributable risk in the myocardial infarction component of our study suggest that the newer progestogens may convey protection. Nevertheless, because of the adverse publicity we may not be able to complete the fieldwork to determine this. We intend to present our definitive findings on venous thromboembolism to scientific peers at the next annual meeting of the British Pharmacological Society.


    View Abstract

    Sign in

    Log in through your institution

    Free trial

    Register for a free trial to thebmj.com to receive unlimited access to all content on thebmj.com for 14 days.
    Sign up for a free trial