Venous thromboembolism occurs rarely in young women. 5 to 10 events per 100,000 woman-years can be expected in under-45 year olds not using oral contraceptives (1). Taking the "Pill" increases the risk. Several epidemiological studies have identified a particular danger with 3rd generation oral contraceptives containing the progestogens desogestrel (e.g. MARVELON) and gestodene (e.g. FEMOVAN) compared to second-generation pills e.g. containing levonorgestrel (e.g. in MICROGYNON) (a-t 1996; No. 2: 17-8; 1995; No. 11: 105-6). An incidence of thromboembolic complications of 20/100.000 woman-years is estimated for second-generation oral contraceptives and of 30-40/100.000 woman-years for those of the third generation (1). For oral contraceptives containing the progestogen drospirenone, which was licensed for the first time in 2000 (e.g. YASMIN; a-t 2000; 31: 103-4), there are so far no valid data on thrombogenicity, but there were risk signals from Norway, Sweden and Great Britain (a-t 2006; 37: 94). In two studies, contraceptives containing drospirenone and desogestrel produced similar procoagulatory effects in the coagulation system (2,3), while the changes were slighter with pills containing levonorgestrel (2). The EURAS* (4) and INGENIX* (5) studies, already announced by Schering and Jenapharm (a-t 2006; 37: 121), are now said to provide "exoneration" for drospirenone.
58,674 women taking an oral contraceptive for the first time or changing the product are included in the prospective European cohort study EURAS (4). They are asked every six months by questionnaire about side effects (average observation time 2.4 years). If serious adverse effects are suspected and in case of any uncertainties, additional information is obtained from the treating doctors. The study is intended to show the non-inferiority of drospirenone-containing compared to levonorgestrel-containing and other oral contraceptives with regard to the rate of venous thromboembolism and to be big enough to be able to rule out a doubling of risk. In the view of the authors, this aim has been achieved: the incidence associated with drospirenone is 91/100,000 woman-years compared to 80/100,000 associated with levonorgestrel (adjusted hazard ratio [HR] 1.0; 95% confidence interval [CI] 0.6-1.8) and 99/100,000 associated with the other preparations not described in more detail (adjusted HR 0.8; 95% CI 0.5-1.3) (4).
However, the safety of drospirenone-containing contraceptives is not proven; on the contrary, the upper value of the confidence interval of 1.8 in the comparison with levonorgestrel means that the risk of venous thromboembolic events with drosperinone can be up to 80% higher than with levonorgestrel - unacceptably high for assuming non-inferiority. Without adjustment, the upper value of the confidence interval is 2.0 (4).
This is neither the sole nor the most serious deficiency of the EURAS study. It can be inferred from the study protocol (6) that it was originally planned to analyse separately women taking oral contraceptives for the first time and those having used a different preparation previously. This is relevant as the thrombosis risk is greatest in the first year of use (1). In the publication (4), however, not even the number of first-time users in the individual groups is provided. Moreover, the rate of venous thromboembolism with levonorgestrel-containing contraceptives is more than four times higher than in earlier studies on which, for instance, the statements hitherto on the incidence by European and American drug authorities are based (1,7). Besides improved diagnosis, the authors refer to their own publications, including an article published in the same edition of Contraception (8), in which they conclude that even the risk in women not taking oral contraceptives is markedly higher than hitherto assumed.
The EURAS study was sponsored by Schering. In a meta-analysis of epidemiological studies of the thrombosis risk of oral contraceptives, the increase in risk due to 3rd generation preparations compared to 2nd generation preparations turns out markedly lower in studies (co-)financed by pharma manufacturers than in independent studies (a-t 2001; 32: 84). Statements on the authors' conflicts of interest are lacking: the first author, J. DINGER, was director of the gynaecology division at Schering while the study was being conducted (9). HEINEMANN's closeness to the industry is documented in numerous publications through expressions such as "unnecessary pill crisis" (10) or playing down the thromboembolic risk associated with oral contraceptives by comparing it with the risk of cycling or driving (11).
The INGENIX study, in which American health insurance data are analysed, is even more inadequate than EURAS (5). Although the increased risk of venous thromboembolism associated with third generation oral contraceptives compared to second generation pills is sufficiently documented, the comparative preparations are neither described nor distinguished in the study financed by Berlex, the supplier there of YASMIN. Therefore the study is not valid.
For oral contraceptives containing the comparatively little tested progestogen drospirenone (in YASMIN etc.) valid data on thrombogenicity are still lacking. Two recent cohort studies, EURAS and INGENIX, remain without validity because of massive method-related deficiencies.
We therefore advise against the use of drospirenone-containing contraceptive pills for the time being.