Chapter 23 Vascular Risks of Combined Estrogen-Progestin Oral Contraceptives Diana B. Petitti Table Of Contents |
|
Diana B. Petitti, MD, MPH |
Combined estrogen-progestin oral contraceptives have been linked with an increased risk of vascular disease since they were first marketed in 1960. It was quickly recognized that the vascular risks of oral contraceptives were related to estrogen dose, and formulations were altered to lower risk. In the 1970s, it became apparent that some women, namely heavy smokers and women older than 40 years, were at particularly high risk of some vascular diseases when using oral contraceptives. Recent, carefully conducted studies of oral contraceptives and the risks of myocardial infarction (MI), stroke, and venous thromboembolism clarify the risks of the low-estrogen-dose oral contraceptives now in widespread use in the United States. This review discusses these recently conducted studies with an emphasis on data that show which women are most likely to have a vascular event while using oral contraceptives and evidence on the effect of progestin type in modifying the risks of vascular disease in oral contraceptive users. |
HISTORICAL OVERVIEW |
Reports linking combined oral contraceptives with an increased risk of venous thrombosis and pulmonary embolism appeared soon after the products were first marketed in 1960.1 Following closely on reports of venous thromboembolism in healthy young women using oral contraceptives were reports of thromboembolic stroke2 and coronary thrombosis.3 The descriptions of these arterial events as thrombotic reflected the then-prevailing view that thrombosis was an important mechanism in the pathogenesis of both arterial vascular disease and venous thrombosis. When first observed, it was believed that the common link between these three vascular events in oral contraceptive users was a prothrombotic effect of estrogen. Studies of the effects of oral contraceptives on various clotting factors were contradictory, difficult to interpret, and not unambiguously indicative of a prothrombotic state in oral contraceptive users, as this state could be defined using tests available in this era. In addition, the notion that thrombosis was a factor in arterial vascular disease fell out of favor in the 1970s. For many years, the lack of a mechanistic explanation for the effect of oral contraceptives in increasing the risk of venous thromboembolism was used to cast doubt on the epidemiologic link. The field has achieved some measure of coherence recently, ironically through a return to the belief that there exists a state of thrombophilia, which can be due to genetic defects in clotting factors or exogenous factors.4 It is now postulated that oral contraceptives can cause this state or increase the chances of venous thrombosis in women with thrombophilia because of genetic defects in clotting factors or for other reasons.5 The importance of thrombosis as a mechanism in arterial vascular disease has been reacknowledged.6 Studies in the 1970s showed that the vascular risks of oral contraceptives are not the same in all women. Thus, it was noted that women who smoke cigarettes were at much higher risk of MI when using oral contraceptives compared with women who did not smoke cigarettes.7,8 A study’s estimate of the overall relative risk in oral contraceptive users is the average of the relative risk on the subgroups of women in whom the effects of use may be different. For example, a statement that the overall relative risk of MI in current oral contraceptive users is 2.0 obscures the fact that the relative risk in smokers is 4 and that in nonsmokers is 1.2.9 Many descriptions of the vascular risks of oral contraceptives fail to describe fully how various factors, such as smoking and hypertension, modify the effect of oral contraceptive use on vascular disease. Recognition of how various factors modify the effect of oral contraceptives is important because it provides the basis for identification of women who may be at particularly high risk of a vascular event when using oral contraceptives. These women can then be offered alternative contraceptive methods that are safer than oral contraceptives for them. Epidemiologic studies of oral contraceptive use and vascular disease published in the 1960s, 1970s, and even the 1980s do not contribute much information to the question of the effect of oral contraceptives on vascular disease in users in 2002 for several reasons. First, in early studies, many of the oral contraceptive formulations contained doses of estrogen that are much higher than the dose of estrogen in currently marketed oral contraceptives. Second, the methods used on many epidemiologic studies are flawed, and few studies until the 1990s explore the important issue of effect modification. Third, medical practices with regard to screening and monitoring of oral contraceptive users are more careful now than in the past. This review discusses recently conducted studies with an emphasis on the identification of the women who are at particularly high risk of vascular disease while using oral contraceptives and evidence for a possible effect of progestin type in modifying the risks of vascular disease in oral contraceptive users. |
VENOUS THROMBOEMBOLISM | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Overall Estimates of Relative Risk Table 1 shows estimates of the overall relative risk of venous thromboembolism in current oral contraceptive users from studies published since 1994.10,11,12,13,14,15,16 These studies are consistent in finding a twofold to fourfold elevation in the risk of venous thromboembolism in current oral contraceptive users compared with that in nonusers.
CI, confidence interval; GPRD, General Practice Research Database. NR, not reported; RR, relative risk; WHO, World Health Organization. * Fatal pulmonary embolism.
Two studies have found an increase in the risk of cerebral vein thrombosis in oral contraceptive users.17,18 Progestin Type In 1995, three simultaneously published reports linked combined oral contraceptives containing desogestrel and gestodene with an increase in the risk of venous thromboembolism that was greater than the increase in the risk of venous thromboembolism in formulations containing levonorgestrel and other progestins.19,20,21 A fourth report, published 1 month later, also found an increased in the risk of venous thromboembolism for formulations containing desogestrel or gestodene.15 Since these reports, much attention has been paid to the possibility that certain progestins might modify the effect of oral contraceptive use on venous thromboembolism. Table 2 shows estimates of the risk of venous thromboembolism in current users of oral contraceptives containing desogestrel or gestodene compared with users of oral contraceptives containing levonorgestrel from a recent systematic review and meta-analysis.22 The summary estimate of the relative risk (RR) of venous thromboembolism derived in this analysis was 1.7 (95% confidence interval [CI], 1.3–2.1). A second meta-analysis reported an identical estimate of the RR of venous thromboembolism in users of oral contraceptives containing desogestrel or gestodene compared with levonorgestrel (RR, 1.7; 95% CI, 1.4–2.0).23
TABLE 2. Estimated Relative Risk (95% Confidence Interval) of Venous
Thromboembolism in Current Users of Oral Contraceptives With Less
Than 50 μg of Ethinyl Estradiol Plus Gestodene or Desogestrel
Compared With Levonorgestrel*
CI, confidence interval; GPRD, General Practice Research Database; RR, relative risk; WHO, World Health Organization. *p for heterogeneity = .09. Modified from Hennessy S, Berlin JA, Kinman JL, et al: Risk of venous thromboembolism from oral contraceptives containing gestodene and desogestrel versus levonorgestrel: A meta-analysis and formal sensitivity analysis. Contraception 64:125, 2001
Various methodologic problems and biases have been suggested as explanations for the elevation in the risk of venous thromboembolism in oral contraceptives with these progestins.24,25 None accounts entirely for the difference in the risk of venous thromboembolism between oral contraceptive formulations containing desogestrel or gestodene and those containing levonorgestrel.11,24,26,27 Effect Modification Some studies show that the risk of venous thromboembolism is elevated more in the first year after initiation of oral contraceptive use.28,29 The mechanism for this early increase in risk is unknown. The RR of venous thromboembolism remains elevated even after the first year of use.28 Activated protein C inhibits clotting by cleaving factors Va and VIIIa. Resistance to the anticlotting effect of activated protein C (APC resistance) increases the risk of venous thrombosis. APC resistance can be caused by a point mutation, a G1A substitution at nucleoside position 1691, in the factor V gene, called factor V Leiden mutation. In a study in the Netherlands, the RR of venous thromboembolism was 34.7 (95% CI, 7.8–154) in oral contraceptive users who were homozygous or heterozygous for the factor V Leiden mutation compared with 3.8 (95% CI, 2.4–6.0) for oral contraceptives overall.13 Martinelli and colleagues17 reported that the RR of cerebral vein thrombosis in oral contraceptive users with a point mutation in the prothrombin gene (a G1A substitution at nucleoside position 20210) was 149.3 (95% CI, 31.0–711.0) compared with non−oral contraceptive users without this mutation. It appears that oral contraceptive use in women with inherited thrombophilia increases the risk of venous thromboembolism more than in other women. However, because these conditions are rare, the proportion of all cases of venous thromboembolism in women with thrombophilia is small. Population screening for thrombophilia based on family has also been evaluated and is not recommended because screening would have low sensitivity.30 |
MYOCARDIAL INFARCTION AND ISCHEMIC STROKE | |||||||||||||||||||||||||||||||||||||||||||||
Overall Estimates of Relative Risk Table 3 shows RR estimates for MI in current oral contraceptive users based on studies published since 1994.31,32,33,34,35. In a pooled analysis of data from two large case-control studies done in the western United States,34 the RR of MI was 1.0 (95% CI, 0.4–2.3) in current oral contraceptive users. The Myocardial Infarction Causality (MICA) study,31 done in England, Scotland, and Wales, reported an RR of MI in current oral contraceptive users of 1.4 (CI, 0.8–2.5). In contrast, the World Health Organization (WHO) study reported a fivefold increase in the RRs of MI in current oral contraceptive users.35
KPSC, Kaiser Permanente Southern California; MICA, Myocardial Infarction Causality; RATIO, Risk of Arterial Thrombosis in Relation to Oral Contraceptives; WHO, World Health Organization.
Table 3 also shows RR estimates for ischemic stroke in current oral contraceptive users based on studies published since 1994.36,37,38,39. In a pooled analysis of data from two large case-control studies done in the western United States, the RR of ischemic stroke was 1.1 (95% CI, 0.5–2.2) in current oral contraceptive users.38 The Risk of Arterial Thrombosis in Relation to Oral Contraceptives (RATIO) study36 reported an RR of ischemic stroke in current oral contraceptive users of 2.1 (CI, 1.5–3.1). In the WHO study, the RR of ischemic stroke was 3.0 in European centers (95% CI, 1.7–5.4) and 2.9 in developing country centers (95% CI, 2.2–4.0).39 A recent meta-analysis of studies reported a summary RR estimate for ischemic stroke of 2.8 (95% CI, 2.2–3.4).40 This estimate of RR may be biased upward because the meta-analysis included early studies of high–estrogen dose oral contraceptives and studies that did not adjust or otherwise account for differences in cardiovascular disease risk factors between oral contraceptive users and nonusers. Effect Modification Table 4 shows the RRs of MI and ischemic stroke in current oral contraceptive users according to whether the oral contraceptive user had another major cardiovascular risk factor (smoking, hypertension, diabetes, hypercholesterolemia).32,34,35,36,38,39 With the exception of the U.S. studies, the studies are consistent in finding very large increases in the RRs of MI in oral contraceptive users who smoked (RR estimates, 7–87) or who had hypertension (RR estimates, 7–68). The RRs of ischemic stroke were also high in oral contraceptive users who smoked (RR estimates, 4–7) or who had hypertension (RR estimates, 8–15). The RATIO study32 is the only one that reported risk estimates for MI and ischemic stroke in oral contraceptive users with diabetes or hypercholesterolemia. In this study, the RRs of MI and ischemic stroke were also very high in oral contraceptive users with these conditions.
TABLE 4. Estimated Relative Risk (95% Confidence Interval) of Myocardial
Infarction and Ischemic Stroke in Current Oral Contraceptive Users According
to Cardiovascular Risk Factors: Studies Since 1994 The apparent discrepancy in the overall RR estimates for MI and ischemic stroke between studies done in the United States and the WHO study may be due to differences in the frequency of oral contraceptive use in women with hypertension in the underlying populations included in the studies. Thus, in the U.S. studies, only 2 of 993 of MI controls (0.2%) and 3 of 1259 stroke controls (0.2%) both used oral contraceptives and had hypertension. In the WHO study, the percentage of controls who both used oral contraceptives and had hypertension was 0.6% for MI and stroke controls in the European centers and 1.6% and 0.4% in the MI and stroke controls, respectively, in the developing country centers. In addition, hypertension may have been more likely to be diagnosed among women in the U.S. studies because routine measurement of blood pressure prior to initiation of oral contraceptive use has been a guideline for many years there. Assessment of blood pressure and identification of hypertension was clearly not routine in the settings in which the WHO study was done. Thus, only 49% of current oral contraceptive users who were European center stroke controls and 48% who were developing country center stroke controls had not had their blood pressure checked39; only 33% of current oral contraceptive users who were European center MI controls and 22% who were developing country center MI controls had had their blood pressure checked.35 Two studies published since 1994 reported that oral contraceptive use increases the risk of ischemic stroke more in women with migraine.41,42 These findings are consistent with findings in the Collaborative Study of Stroke in Young Women.43 Risks According to Progestin Type Observations in the late 1970s and early 1980s that combined oral contraceptives alter lipids and affect carbohydrate metabolism provided a mechanistic basis for an effect of oral contraceptives on MI. The recognition that different progestins had different effects on lipids and carbohydrate metabolism at the same dose of estrogen led to development of progestins that, when combined with estrogen, led to more favorable effects on lipids. The favorable lipid profile in these new products led to claims that these oral contraceptives would not increase, and might even decrease, the risk of arterial vascular disease in oral contraceptive users. Table 5 shows the RRs of MI and ischemic stroke in current oral contraceptive users compared with nonusers according to progestin type. In the RATIO study,32 the Transnational study,33 and the WHO study,35 but not the MICA study,31 the risk of MI was increased more for oral contraceptives containing levonorgestrel than for those containing desogestrel or gestodene. The possibility that the higher risk of venous thromboembolism in users of oral contraceptive formulations containing desogestrel or gestodene might be offset by a lower risk of MI is neither established nor ruled out by studies published through 2002.
TABLE 5. Estimated Relative Risk (95%
Confidence Interval) of Myocardial Infarction and Ischemic Stroke in Current
Oral Contraceptive Users According to Progestin Type: Studies Since 1994 The risk of ischemic stroke is not different between oral contraceptives containing desogestrel or gestodene compared with those containinglevonorgestrel.36,37,44 |
HEMORRHAGIC STROKE | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Overall Estimate of Relative Risk In 1973, the Collaborative Study of Stroke in Young Women’45 reported a twofold increase in the risk of hemorrhagic stroke in current oral contraceptive users. Studies subsequent to the Collaborative Study have been mixed in their results.46 Table 6 shows the RR of hemorrhagic stroke in current oral contraceptive users in the three studies published since 1994.47,48,49. The overall RR is different from 1.0 only in the developing country centers in the WHO study.
CI, confidence interval; KPSC, Kaiser Permanente Southern California; RR, relative risk; WHO, World Health Organization.
Effect Modification Only the WHO study had numbers of oral contraceptive users who smoked, had hypertension, or were age 35 years or older large enough to estimate with precision the risk of hemorrhagic stroke in oral contraceptive users with these conditions.49 Table 7 shows that the RR of hemorrhagic stroke was considerably higher in oral contraceptive users in association with each of these factors, but especially in women who had hypertension. Taken as a whole, the recent studies show that the risk of hemorrhagic stroke is not increased in oral contraceptive users who are young, who do not have hypertension, and who do not smoke.
CI, confidence interval; HBP, hypertension; OC, oral contraceptive; RR, relative risk.
Risk According to Progestin Type The data on progestin type and the risk of hemorrhagic stroke are very limited. They do not permit any conclusion about whether oral contraceptives with different progestins might carry different risks of hemorrhagic stroke in high-risk women. |
DISCONTINUED USE |
The possibility that oral contraceptive users might continue to be at risk of vascular disease after discontinuation of use has been examined in both past and recent studies. Two reviews present extensive data on this topic.26,50 The studies are consistent in showing that women are not at increased risk of venous thromboembolism, MI, or stroke after they cease oral contraceptive use. |
ABSOLUTE RISK OF VASCULAR DISEASE IN ORAL CONTRACEPTIVE USERS |
The incidence rates of venous thromboembolism, MI, and ischemic stroke in women of reproductive age are each about 5 per 100,000 per year.50 With conservative (high) RR estimates for venous thromboembolism, MI, and ischemic stroke of 4, 2, and 2, respectively, the number of excess cases of vascular disease in current oral contraceptive users would be 25 per 100,000 users per year. It has become common to express absolute risks in terms of number needed to harm, which is calculated as the reciprocal of the absolute risk of an exposure. Expressed as a number needed to harm, 4000 women would need to take oral contraceptives for one to cause one of these vascular event in that year. With estimates of the RR of venous thromboembolism, MI, and ischemic stroke of 3, 1.2, and 1.2, which are the estimates that likely apply to nonsmokers without hypertension, the number of excess cases vascular disease in current oral contraceptive users would be 12 per 100,000 users per year. This is 8333, expressed as a number needed to harm. The incidence of venous thromboembolism, MI, and ischemic stroke increases sharply with age, and smoking and hypertension increase the risks of MI and ischemic stroke, even in the absence of oral contraceptive use. Because the oral contraceptive use interacts with smoking and hypertension and because the incidence of vascular disease increases with age, the vascular risks of oral contraceptive use are much greater in older women, women with hypertension, and women who smoke. |
SUMMARY AND CONCLUSIONS |
Combined estrogen-progestin oral contraceptives increase the risk of venous thromboembolism by a factor of 2 to 4. The risk is higher in the first year after initiating use, but risk persists beyond the first year. The risk of venous thromboembolism is increased more in users of oral contraceptives that contain desogestrel or gestodene. Women with inherited thrombophilia are at higher risk of venous thromboembolism when using oral contraceptives, but these conditions account for only a small percentage of all cases of venous thromboembolism in oral contraceptive users. The risks of MI and ischemic stroke are not elevated in women who do not smoke and do not have hypertension or other risk factors for arterial vascular disease. The possibility that oral contraceptives containing desogestrel or gestodene carry a lower risk of MI compared with those that contain levonorgestrel is neither proved nor ruled out by studies published through 2002. The risk of ischemic stroke does not differ between oral contraceptives containing different progestins. The risk of hemorrhagic stroke is not increased in oral contraceptive users overall, but women with hypertension and smokers are at increased risk of hemorrhagic stroke when using oral contraceptives. The absolute risk of vascular disease—venous thromboembolism, MI, or ischemic stroke—is very low in oral contraceptive users who are young, do not smoke, and do not have hypertension. |