Risk of VTE with Oral Contraceptives
Abstract & Commentary
By Jeffrey T. Jensen, MD, MPH, Editor, Leon Speroff Professor of Obstetrics and Gynecology, Vice Chair for Research, Oregon Health and Science University, Portland, is Editor for OB/GYN Clinical Alert.
Synopsis: The risk of venous thrombosis associated with oral contraceptive use correlated with ethinyl estradiol dose and was lowest in users of levonorgestrel combination pills.
Source: van Hylckama Vlieg A, et al. The venous thrombo-tic risk of oral contraceptives, effects of oestrogen dose and progestogen type: Results of the MEGA case-control study. BMJ 2009;339:b2921.
In a population-based case-control study from the Netherlands, combined oral contraceptives increased the risk of venous thrombosis (VTE) fivefold compared with non-use (odds ratio [OR], 5.0; 95% confidence interval [CI], 4.2-5.8). The risk of VTE was lowest in users of oral contraceptives containing levonorgestrel (OR, 3.6; 95% CI, 2.9-4.6), and highest in users of products containing gestodene (OR, 5.6; 95% CI, 3.7-8.4), desogestrel (OR, 7.3; 95% CI, 5.3-10.0), cyproterone acetate (OR, 6.8; 95% CI, 4.7-10.0), and drospirenone (OR, 6.3; 95% CI, 2.9-13.7). The highest risk was seen in the first 3 months of use and risk was positively correlated with estrogen dose.
Synopsis: Combination oral contraceptives containing desogestrel, gestodene, or drospirenone were associated with a significantly higher risk of venous thrombosis than oral contraceptives containing levonorgestrel.
Source: Lidegaard O, et al. Hormonal contraception and risk of venous thromboembolism: National follow-up study. BMJ 2009;339:b2890.
A cohort study using the population-based Danish National Registry showed that the risk of venous thrombosis in current users of combined oral contraceptives decreases with increasing duration of use and decreasing estrogen dose. The overall absolute risk of venous thrombosis per 10,000 woman-years in non-users of oral contraceptives was 3.01 and in current users was 6.29. Compared with non-users of combined oral contraceptives the rate ratio of venous thromboembolism in current users decreased with duration of use (< 1 year, 4.17 [95% CI, 3.73-4.66]; 1-4 years, 2.98 [95% CI, 2.73-3.26]; and > 4 years, 2.76 [95% CI, 2.53-3.02]) and with decreasing dose of estrogen. Compared with oral contraceptives containing levonorgestrel (LNG; RR = 1), an increased risk was seen with products containing desogestrel: 1.82 (95% CI, 1.49-2.22); gestodene: 1.86 (95% CI, 1.59-2.18); drospirenone: 1.64 (95% CI, 1.27-2.10); and cyproterone acetate: 1.88 (95% CI, 1.47-2.42). Compared with non-users of oral contraceptives, the rate ratio for venous thromboembolism was not increased in users of progestogen-only oral contraceptives (levonorgestrel or norethisterone: RR 0.59 [95% CI, 0.33-1.03]; 75 µg desogestrel: RR, 1.12 [95% CI, 0.36-3.49]; and for users of the LNG-intrauterine system [IUS]: RR, 0.90 [95% CI, 0.64-1.26]).
The simultaneous publication of these 2 epidemiologic studies of the risk of venous thrombosis in oral contraceptive users in a leading general European medical journal, received considerable press in the United Kingdom but less notice here. The results deserve scrutiny as we review this important topic once again.
Oral contraceptives (OCs) will celebrate 50 years of market approval in the United States next year, and arguably represent one of the most important breakthroughs in human social biology. While a variety of excellent contraceptive choices currently exist, the availability and acceptance of oral contraception paved the path toward the improved status and equality of opportunity that women enjoy today. OCs represent the gold standard by which other highly effective reversible methods are measured. Combined OCs also provide a myriad of non-contraceptive health benefits and are broadly useful in female heath care.
Decisions in clinical medicine (as in life) require consideration of both benefit and risk. One of the important risks of combined OCs is thrombosis. Consistent evidence over the last 40 years has established that the risk of venous and arterial thrombosis is increased among users of combined OCs compared to non-pregnant non-users. Most of us probably quote the textbook statistic that the baseline risk is about 2 cases of VTE/10,000 women, with 4 cases/10,000 among users of OCs and 6 cases/10,000 in pregnant women. However, the literature reviewed by Heinemann and Dinger shows 2 levels of VTE incidence rates: one for community/cohort studies and one for database studies.1 The estimated overall VTE incidence rates for women of reproductive age ranges between 5.5-13.5 and 3.8-12.2 per 10,000 women-years in community and cohort studies, respectively, but only 0.7-3.8/10,000 in database studies. This difference is probably attributable to methodological problems associated with some database studies; principally exclusions and selection of controls and verification of disease. Combined oral contraceptives increase the risk whatever baseline is used. OC use also provides significant protection against pregnancy, a condition associated with a substantially higher risk of thrombosis. So as always, we must consider risks, benefits, and alternatives.
The van Hylckama Vlieg analysis was performed using data from the MEGA (Multiple Environmental and Genetic Assessment of risk factors for venous thrombosis) database, a large, population-based, case-control study on risk factors for venous thrombosis that includes data from 6 participating anticoagulation clinics in the Netherlands (Amersfoort, Amsterdam, The Hague, Leiden, Rotterdam, and Utrecht) collected between March 1999 and September 2004.2 Cases had a confirmed first episode of deep venous thrombosis (leg or arm) or pulmonary embolism. Controls were age-matched individuals without history of malignancy or cardiovascular disease. Women up to age 50 were studied, and the long period of data collection allowed the authors to collect information concerning length of exposure to oral contraceptives in large numbers of women.
Overall, compared to non-users current oral contraceptive use was associated with a crude fivefold increased risk of venous thrombosis (OR, 5.0; 95% CI, 4.2-5.8), with the highest risk during the first 3 months of use (OR, 12.6; 95% CI, 7.1-22.4) The risk estimate was lowest for levonorgestrel (OR, 3.6; 95% CI, 2.9-4.6) products, and highest for products containing desogestrel (OR, 7.3; 95% CI, 5.3-10.0), cyproterone acetate (OR, 6.8; 95% CI, 4.7-10.0), and drospirenone (OR, 6.3; 95% CI, 2.9-13.7). As can be noted by the confidence intervals, relative to levonorgestrel, the risk of VTE was significantly increased only for desogestrel (OR, 2.0; 95% CI, 1.4-2.8) and cyproterone acetate (OR, 2.0; 95% CI, 1.3-3.0), but not drospirenone (OR, 1.7; 95% CI, 0.7-3.9). Restricting the analysis to monophasic preparations with levonorgestrel, gestodene, or desogestrel demonstrated a nonsignificant trend toward a reduced thrombotic risk with ethinyl estradiol (EE) doses of 20 µg (OR, 0.8; 95% CI, 0.5-1.2) and increased risk (OR, 1.9; 95% CI, 1.1-3.4) for 50 µg pills relative to 30-35 µg. Based upon these data, the authors conclude that levonorgestrel products containing the lowest dose of estrogen should be the preferred product.
The Danish study made use of outstanding national databases that link to the Central Person Registry (information on address and vital status that is updated daily) using the personal identification number of all Danish citizens given at birth or immigration.3 The National Registry of Medicinal Products Statistics has recorded all redeemed prescriptions on Danish citizens since 1994, and the National Registry of Patients has collected discharge diagnoses and surgical codes from all Danish hospitals since 1997. Use of the registry eliminates recall bias for OC type. The study looked at incidence rates of VTE in users of combined OCs, progestin-only pills, and the LNG-IUS.
The analysis included an impressive 3.4 million woman-years of current use, 2.3 million woman-years of former use, 4.8 million woman-years of never use, or a total of 10.4 million woman-years of observation. The crude incidence of venous thromboembolism among non-users (never or former use) of hormonal contraceptives was 3.01/10,000 woman-years, and among current users of oral contraceptives was 6.29/10,000 woman-years. Consistent with other studies, the risk among women using combined OCs decreased with duration of use, from 4.17 (95% CI, 3.73-4.66) during the first year of use to 2.76 (95% CI, 2.53-3.02) after more than 4 years of use. The study also confirmed the trend toward a nonsignificant reduction in risk with lower EE dose to 20 µg. As was reported in the Dutch study, using levonorgestrel-containing products as a reference, the risk of VTE was significantly higher in OCs containing desogestrel (1.82; 95% CI, 1.49-2.22), cyproterone acetate (1.88; 95% CI, 1.47-2.42), and drospirenone (1.64; 95% CI, 1.27-2.10). An important contribution of this study was the lack of any apparent increased risk of thrombosis in women using the progestin-only pill and the LNG-IUS (adjusted rate ratio, 0.89; 95% CI, 0.64-1.26).
So how should clinicians view these new studies? Following the "pill scare" in the 1990s that came after an apparent increased risk of VTE with third generation progestogens was widely reported, the importance of selective prescription of new OCs to high-risk women, and the healthy user effect seen in women continuing use of existing products, has been discussed. Both of the studies reported here have the advantage of data collection occurring after the "pill scare," and the authors assume that selective prescription is an unlikely source of bias in the results, at least for desogestrel. However, important differences in users and non-users of oral contraceptives make up sources of bias due to confounding factors not easily adjusted for in epidemiologic studies.
Two large scale population-based phase IV postmarketing studies of drosperinone were commissioned by Schering AG (now Bayer Healthcare) after the introduction of the novel progestogen in the United States and Europe, to avoid a comparable "pill scare" situation to what occurred with the introduction of the third generation progestogens in the 1990s. Both used a prospective cohort design, and utilized independent research organizations, data safety monitoring boards, and data analysis teams that created a wall between corporate and public health interests in the data.
The American study was performed using cohorts established to investigate potential hyperkalemic complications possibly related to the potassium-sparing diuretic effects of drospirenone (DRSP), an agreed upon postmarketing study required for FDA approval of Yasmin® (30 µg EE, 3 mg DRSP).4 Women initiating Yasmin and medically similar initiators of other oral contraceptives were identified using a HIPAA-compliant proprietary research database built from electronically captured provider, facility, and pharmacy claims at United Healthcare-affiliated health plans and large employer groups, with follow-up in the written medical records. All initiators of Yasmin were successfully matched in the database to 2 initiators of comparable OCs. There were 18 confirmed instances of thromboembolism among the Yasmin initiators, for an incidence rate of 13/10,000 woman-years, and 39 among other oral contraceptive initiators, for an incidence rate of 14/10,000 woman-years (RR, 0.9; 95% CI, 0.5-1.6). The authors calculated absolute incidence rates. Based upon their data, a clinician can expect to observe similar numbers of cases of thromboembolism (1/769 drospirenone users, 1/714 other OC users) over the course of 1 year of use. But, since both of these estimates (about 14/10,000) sound higher than what we commonly quote patients (e.g., 4/10,000 for OC users), should we be concerned? Also, this study simply compared drospirenone to other OCs, and we have no data as to what these OCs were (how many contained desogestrel?), so what do we know about relative safety with respect to levonorgestrel?
The EURAS study provides some of these answers.5 The European Active Surveillance study (EURAS) was a multinational, prospective, non-interventional cohort study of new users of DRSP, levonorgestrel, and other progestin-containing OCs. The Center for Epidemiology and Health Research in Berlin performed the study, which was supervised by an independent advisory board.
The objectives of the study were to characterize and compare the risks of short- and long-term use of DRSP-containing OCs and established OCs in a study population representative of actual OC users. In other words, since only real world OC users were studied, incidence rates between preparations are directly comparable. The main clinical outcome of interest was to document the occurrence of uncommon cardiovascular events (e.g., arrhythmias, myocardial infarction, stroke, VTE, and sudden death) and compare the incidence rates between users of DRSP-containing OCs and users of established OCs. Based on the fact that European regulatory authorities considered the progestogen LNG to have the least impact on VTE risk, it was decided that the primary cardiovascular outcome of interest should be the VTE hazard ratio (HR) between users of DRSP-containing OCs and users of LNG-containing OCs. To study these events, large numbers of subjects were needed, so the study planned to enroll 50,000 women at multiple sites in Europe. Recruitment of the cohort members was conducted via a network of physicians who prescribed OCs in 7 European countries: Austria, Belgium, Denmark, France, Germany, the Netherlands, and the United Kingdom. At the participating centers, all women who received a prescription for a new OC were asked by their physicians whether they were willing to participate. All subjects who were either OC starters (first ever users) or OC switchers were eligible for enrollment in the study provided that they signed the informed consent and data privacy form. Importantly, more specific inclusion or exclusion criteria were not applied because of the non-interference approach of the study design. Cardiovascular outcomes were adjudicated by 3 independent medical experts specializing in radiology/nuclear medicine, cardiology, and internal medicine/phlebology who were blind to treatment group.
After baseline assessments and data collection, follow-up was scheduled every 6 months to assess adverse events or discontinuation. A total of 58,674 study participants were followed for 142,475 woman-years of observation. An array of comprehensive multi-tiered follow-up interventions kept loss-to-follow-up for the entire cohort to an impressively low 2.4% and did not differ between groups. Since some subjects were followed for as long as 4 years, many switched to another OC, to a non-oral hormonal method, or became non-users during the follow-up interval, providing data for these additional groups. Both intention-to-treat and as-treated analyses were performed, but since the groups were not randomized at enrollment, the distinction is not important, and the results were not affected.
The investigators found no difference in the overall incidence of serious adverse events (SAEs) between users of any of the oral contraceptive groups. Underlying the health benefits of contraception, the highest rate of SAEs occurred among non-users, primarily due to pregnancy-related events. The EURAS study was powered to show non-inferiority of DRSP-containing OCs regarding VTE risk in comparison to LNG and other OCs. A similar incidence of VTE (per 10,000 women-years of exposure) was found in all cohorts: DRSP 9.1 (95% CI, 5.9-13.3); LNG 8.0 (95% CI, 5.2-11.7); and other OCs 9.9 (95% CI, 7.4-13). Note the precise confidence intervals around the point estimates produced due to the large numbers of women studied. The adjusted HRs for DRSP vs LNG was 1.0 with a CI of 0.6-1.8. Thus, a twofold higher risk of VTE during DRSP use compared to LNG use was excluded, and non-inferiority according to the study objectives was demonstrated. All other outcomes were similar.
Like the American study, the big surprise of the EURAS study was the higher incidence of VTE seen in all groups. In the no-use cohort, 12 VTEs were observed. This corresponds to an overall incidence of 4.7/10,000 woman-years. Seven of these occurred in the course of pregnancy (19.4/10,000 woman-years) and 5 (2.3/10,000 woman-years) among non-pregnant non-users. The authors were careful to point out that these results are not necessarily representative for non-users who have never used OCs, and must be considered tentative.
So has the incidence of VTE gone up? Risk of VTE in the EURAS study was highest in the first 3 months of use. Another finding was that obese women (BMI > 30.0 kg/m2) had an approximately threefold higher VTE risk compared to women with normal weight (BMI, 20.0-24.9 kg/m2). This may explain the higher incidence seen in the American cohorts. The diagnosis of VTE is considerably easier and more precise today than it was 20 years ago, and no one can pass through an emergency room without the opportunity for advanced imaging. We are able to see more and smaller clots. It is controversial whether all of these diagnoses are significant.
Taking all of this evidence together, I think we can safely put the results of the BMJ articles in perspective. VTE risk is greatest during the early months of use of a new product, so stick with a winner. Obesity increases health risks, and magnifies the risk of VTE threefold. Use of lower-dose estrogen pills may increase safety, and evidence is beginning to accumulate that there may be additional safety as we reduce to 20 µg. The Danish study also provides epidemiologic data to support the safety of the LNG-IUS. Prospective trumps epidemiology in most cases, so the differences between type of progestogen appear to be less robust, and we can base our decisions for oral contraceptive prescriptions on other features besides VTE risk. Price and side-effect profiles matter most. Drospirenone has recently joined the ranks of generic progestogens, and this increases consumer choice. As we know that the highest risk of VTE exists with pregnancy, we need to do whatever we can to make sure our patients tolerate their pills well, comply with the regimen, and continue with the method. That's the bottom line.
- Heinemann LA, Dinger JC. Range of published estimates of venous thromboembolism incidence in young women. Contraception 2007;75:328-336.
- van Hylckama Vlieg A, et al. The venous thrombotic risk of oral contraceptives, effects of oestrogen dose and progestogen type: Results of the MEGA case- control study. BMJ 2009;339:b2921.
- Lidegaard O, et al. Hormonal contraception and risk of venous thromboembolism: National follow-up study. BMJ 2009;339:b2890.
- Seeger JD, et al. Risk of thromboembolism in women taking ethinylestradiol/drospirenone and other oral contraceptives. Obstet Gynecol 2007;110:587-593.
- Dinger JC, et al. The safety of a drospirenone-containing oral contraceptive: Final results from the European Active Surveillance Study on oral contraceptives based on 142,475 women-years of observation. Contraception 2007;75:344-354.