Re: Venous thrombosis in users of non-oral hormonal contraception: follow-up study, Denmark 2001-10
19th of May 2012
To the Editor of BMJ
We read the article “Venous thrombosis in users of non-oral hormonal contraception: follow-up study, Denmark 2001-10” by Lidegaard et al (1) with great interest. The conclusion that non-oral hormonal contraception increases the risk of venous thrombosis is surprising and has been covered extensively by the Danish media. On a global scale, however, the results of the study can influence not only prescription habits of physicians, but also impact the lives of numerous women due to unintended and therefore unsafe pregnancies. For obvious reasons we are very happy with the Danish register policy and the research possibilities linked to the data sources, although registers have unavoidable limitations, as stated in the recent comments (2,3,4) and therefore clinical translations should be done with the utmost prudence. We have some additional questions related to the presented results:
Firstly, we find it difficult to realise how some of the incidence ratios were calculated. In example, the 5287 diagnoses of venous thrombosis recorded in 9.429.128 woman years corresponds to an incidence rate of 5.6 per 10.000 woman years, not 8.1 per 10.000 woman years as reported in the results section. In table 2, 6.748.273 woman years are divided into 5.892.182 years of non-use of contraception and a total of 856.091 years of use of various types of contraception. If the 5.287 events had occurred in these 6.748.273 woman years, the event rate would indeed be close to 8 per 10.000 woman years, but only 2.826 events of venous thrombosis are recorded in the table, corresponding to an event rate of 4.2 per 10.000 woman years (reduced to 2.3 per 10.000 woman years for confirmed events). Could the authors please explain how the rate of 8.1 per 10.000 woman years was calculated?
Secondly, the number of events occurring in women using transdermal patches and implants are very low. Interestingly, table 3 shows that the relative risk for a confirmed event (n = 6) in users of transdermal patches is significantly higher compared with users of COC with levonorgestrel and 30-40 μg oestrogen. The relative risk is 2.5, with a wide confidence interval of 1.1 to 5.6. Nevertheless, the relative risk of a venous thrombosis not confirmed through prescription of anticoagulant medication (n = 7) is not significantly higher in users of transdermal patches compared with users of COC with levonorgestrel and 30-40 μg oestrogen. Here, the relative risk is 1.9 with a confidence interval of 0.9 to 3.9. This example clearly illustrates the risk of arriving at a fragile conclusion when working with numbers of this size. What is the margin of error regarding the calculation of 1250 women who should shift from a transdermal patch to COCs in order to prevent one event (discussion section)? Do the authors firmly believe that the findings of the study support the conclusion that “women are generally advised to use COC rather than transdermal patches”?
We furthermore agree with the concerns about the biological plausibility of the results and the preferred prescription of non-oral contraception to women at higher risk raised by A. Szarewski et al as one of the rapid responses (5). We find no reason to repeat the clear and concise arguments, but eagerly await the response of the authors to her letter and the above questions.
Competing interests:
Jane Skov: The author has no competing interests to declare
Johannes Sidelmann: The author has no competing interests to declare
Jørgen Jespersen: The author has no competing interests to declare
Sven O. Skouby: The author presently consults and in the past has consulted with manufacturers of hormonal contraceptives, including NuvaRing.
19 May 2012
Jane Skov
Postdoctoral researcher, PhD
Johannes Sidelmann, Associate Professor, PhD(1), Jørgen Jespersen, Professor, MD, Dsc (1) Sven O. Skouby, Professor, MD, Dsc (1,2) 1:Unit for Thrombosis Research, Institute of Public Health, University of Southern Denmark, Department of Clinical Biochemistry, Hospital of South West Denmark Denmark, 2:Department of Gynecology And Obstetrics, Herlev Hospital, Faculty of Health Sciences, University of Copenhagen, Denmark
Unit for Thrombosis Research, Institute of Public Health, University of Southern Denmark, Department of Clinical Biochemistry, Hospital of South West Denmark, Denmark
Rapid Response:
Re: Venous thrombosis in users of non-oral hormonal contraception: follow-up study, Denmark 2001-10
19th of May 2012
To the Editor of BMJ
We read the article “Venous thrombosis in users of non-oral hormonal contraception: follow-up study, Denmark 2001-10” by Lidegaard et al (1) with great interest. The conclusion that non-oral hormonal contraception increases the risk of venous thrombosis is surprising and has been covered extensively by the Danish media. On a global scale, however, the results of the study can influence not only prescription habits of physicians, but also impact the lives of numerous women due to unintended and therefore unsafe pregnancies. For obvious reasons we are very happy with the Danish register policy and the research possibilities linked to the data sources, although registers have unavoidable limitations, as stated in the recent comments (2,3,4) and therefore clinical translations should be done with the utmost prudence. We have some additional questions related to the presented results:
Firstly, we find it difficult to realise how some of the incidence ratios were calculated. In example, the 5287 diagnoses of venous thrombosis recorded in 9.429.128 woman years corresponds to an incidence rate of 5.6 per 10.000 woman years, not 8.1 per 10.000 woman years as reported in the results section. In table 2, 6.748.273 woman years are divided into 5.892.182 years of non-use of contraception and a total of 856.091 years of use of various types of contraception. If the 5.287 events had occurred in these 6.748.273 woman years, the event rate would indeed be close to 8 per 10.000 woman years, but only 2.826 events of venous thrombosis are recorded in the table, corresponding to an event rate of 4.2 per 10.000 woman years (reduced to 2.3 per 10.000 woman years for confirmed events). Could the authors please explain how the rate of 8.1 per 10.000 woman years was calculated?
Secondly, the number of events occurring in women using transdermal patches and implants are very low. Interestingly, table 3 shows that the relative risk for a confirmed event (n = 6) in users of transdermal patches is significantly higher compared with users of COC with levonorgestrel and 30-40 μg oestrogen. The relative risk is 2.5, with a wide confidence interval of 1.1 to 5.6. Nevertheless, the relative risk of a venous thrombosis not confirmed through prescription of anticoagulant medication (n = 7) is not significantly higher in users of transdermal patches compared with users of COC with levonorgestrel and 30-40 μg oestrogen. Here, the relative risk is 1.9 with a confidence interval of 0.9 to 3.9. This example clearly illustrates the risk of arriving at a fragile conclusion when working with numbers of this size. What is the margin of error regarding the calculation of 1250 women who should shift from a transdermal patch to COCs in order to prevent one event (discussion section)? Do the authors firmly believe that the findings of the study support the conclusion that “women are generally advised to use COC rather than transdermal patches”?
We furthermore agree with the concerns about the biological plausibility of the results and the preferred prescription of non-oral contraception to women at higher risk raised by A. Szarewski et al as one of the rapid responses (5). We find no reason to repeat the clear and concise arguments, but eagerly await the response of the authors to her letter and the above questions.
References:
1. Lidegaard O, Nielsen LH, Skovlund CW, Lokkegaard E. Venous thrombosis in users of non-oral hormonal contraception: follow-up study, Denmark 2001-10. BMJ 2012; 344:e2990
2. Shapiro S, rapid response to (1), 16th of May 2012 : http://www.bmj.com/content/344/bmj.e2990/rr/584950
3. Gaffield M et al, rapid response to (1), 16th of May 2012: http://www.bmj.com/content/344/bmj.e2990/rr/584992
4. Chandler et al, rapid response to (1), 17th of May 2012 http://www.bmj.com/content/344/bmj.e2990/rr/585117
5. Szarewski A, Mansour D, rapid response to (1), 14th of May: http://www.bmj.com/content/344/bmj.e2990/rr/584507
Competing interests: Jane Skov: The author has no competing interests to declare Johannes Sidelmann: The author has no competing interests to declare Jørgen Jespersen: The author has no competing interests to declare Sven O. Skouby: The author presently consults and in the past has consulted with manufacturers of hormonal contraceptives, including NuvaRing.