Oral contraceptives containing the newer progestogen types desogestrel, drospirenone, and gestodene have been associated with twice the risk of venous thromboembolism as those containing levonorgestrel, an older agent.
Women taking combination oral contraceptives (OCs) with levonorgestrel were at a threefold increased risk for confirmed venous thromboembolism (VTE), compared with women not taking any contraceptive pills, while users of OCs containing desogestrel, drospirenone, and gestodene saw a sixfold increased risk. The risk for drospirenone was seen as comparable to that of desogestrel and gestodene.
However, a lower dose of estrogen (20 mcg instead of 30 mcg) was not seen to lessen the risk associated with drospirenone-containing pills, the investigators found, while with desogestrel and gestodene, lower-estrogen preparations were associated with lower risk. (The OCs with desogestrel or gestodene and 20 mcg of ethinyl estradiol implied relative risks of VTE that were 23% and 17% lower, respectively, than the risks for the same progestogens with 30 mcg of ethinyl estradiol.)
The results, published in BMJ (2001 [doi: 10.1136/bmj.d6423]), confirm those of a previous study by the same Denmark-based team of investigators, who in 2009 reported a significantly higher risk of VTE for OCs containing these three progestogens than for OCs with levonorgestrel (BMJ 2009;339:b2890 [doi: 10.1136/bmj.b2890]).
Both studies were led by Dr. Øjvind Lidegaard, professor of obstetrics and gynecology at the University of Copenhagen, and his colleagues, who used the same registry-based cohort of all Danish women aged 15-49 years with no history of VTE and who were not pregnant. The new study followed the women from 2001 to 2009, 4 years longer than the previous study (though their prescription information was collected from 1995), and the study collected more detailed information on OC use and VTE events.
During the study period, which comprised more than 8 million woman-years of observation, 4,246 first episodes of VTE occurred. After adjustment for age, calendar year, education, and length of OC use, the relative risk of VTE in women who used pills with desogestrel, drospirenone, or gestodene was found to be twice that of women on levonorgestrel-containing pills.
Dr. Lidegaard and his colleagues found that, compared with women not using hormonal contraception, the relative risk of confirmed VTE in users of OCs containing 30-40 mcg of ethinyl estradiol with levonorgestrel was 2.9 (95% confidence interval, 2.2-3.8), compared with 6.6 in women using OCs containing desogestrel (95% CI, 5.6-7.8), 6.2 in users of gestodene (5.6-7.0), and 6.4 in women taking drospirenone (5.4-7.5).
With users of OCs containing levonorgestrel as a reference, and after adjustment for length of use, the rate ratio of confirmed VTE for users of OCs with desogestrel was 2.2 (95% CI, 1.7-3.0), with gestodene it was 2.1 (95% CI, 1.6-2.8), and with drospirenone it was 2.1 (1.6-2.8).
To prevent one VTE per year, approximately 2,000 women would need to switch from a pill containing desogestrel, gestodene, or drospirenone to one with levonorgestrel, the researchers concluded.
Critics of Dr. Lidegaard and colleagues' previous study, which revealed similar differences in risk between levonorgestrel and the newer progestogens, argued that because no declining risk was seen after the first few months of use for women using levonorgestrel-containing pills, as would be expected, left-censoring bias might have occurred, making the risk of VTE associated with levonorgestrel seem artificially low compared with that seen with drospirenone, which was introduced in 2001.
Dr. Lidegaard and his colleagues described several changes in their new study's design to strengthen it, noting that they had eliminated left-censoring bias by letting the new study period begin in 2001, which marked the introduction of drospirenone-containing OCs in Denmark, but still collected the women's full OC exposure history for the previous 6 years.
In addition, they defined length of OC use more precisely than in the previous study, stratified analyses into confirmed and unconfirmed VTE events, and better excluded women predisposed to VTE, they said.
The investigators acknowledged as a weakness of the study that they could not control for family disposition and body mass index. However, they noted, other studies had not shown a strong confounding effect for those factors even when data were available.