Initially, oral contraceptives included at least 50 µg of ethinylestradiol or mestranol and a progestogen, typically norethindrone. Because estrogen was suspected of increasing risk for thromboemboli, contraceptives that contained less than 50 µg of estrogen and a new progestogen, levonorgestrel, were introduced. More recently, the newest progestogens (desogestrel, gestodene, and norgestimate) in combination with no more than 35 µg of ethinylestradiol have become available. Because the new progestogens have fewer androgenic metabolic effects and do not adversely affect lipid levels, they may be associated with lower risk for cardiovascular disease. As a result of these marked changes in formulation, the findings of early epidemiologic studies may not be fully generalizable to today's preparations.

The population using oral contraceptives has also changed. Women begin use at younger ages and continue for longer periods. Although serious cardiovascular disease is uncommon among women in their 20s and 30s, it is not rare in older premenopausal women, especially smokers. We review epidemiologic studies of myocardial infarction, stroke, and thromboembolic disease in relation to oral contraceptives, especially those that contain less than 50 µg of estrogen.

Methods

Top Methods Author & Article Info References

We searched the MEDLINE database for all English-language epidemiologic studies of oral contraceptives and cardiovascular disease published between 1967, when contraceptives containing low doses of estrogen (low-dose oral contraceptives) were introduced in the United States, and June 1997. Animal and metabolic studies were reviewed only if they were relevant to the mechanism of action of oral contraceptives. We also searched textbooks, meeting proceedings, and reference lists from identified articles. We did not consider abstracts because these typically present preliminary data and provide insufficient detail. We retrieved 374 relevant studies. Those without a quantitative estimate of effect were excluded, as were those that did not satisfy the following date and dose criteria.

By 1976, low-dose contraceptives dominated the U.S. market. In 1988, 82% of prescriptions for oral contraceptives were formulations that contained less than 50 µg of estrogen [1]. Contraceptives that contained more than 50 µg of estrogen were withdrawn in the United States in 1989 and in the United Kingdom in 1984 [2]. We included studies published since low-dose contraceptives were introduced in the United States in 1967 (they were introduced in the United Kingdom in 1973). For studies that did not specify the dose of estrogen used, we inferred that after 1988 (in the United States) and 1984 (in the United Kingdom), the study population predominantly used low-dose contraceptives. Formulations that contain 50 µg of mestranol, a less potent type of estrogen, were not considered to be high-dose oral contraceptives because they are equivalent to formulations that contain 35 µg of ethinylestradiol. Data were organized by cardiovascular end point, study design, estrogen dose, and type of progestogen.

Data on relative and absolute risk are presented to address current prescription guidelines. Relative risks are useful for etiologic study but provide little guidance in clinical or public health settings. When the baseline risk for disease is low, the absolute excess risk may be small even if the relative risk is high.

Data Synthesis

Myocardial Infarction

Current Use

The current use of oral contraceptives increases risk for myocardial infarction, but most of the excess risk is attributable to an interaction with cigarette smoking. For women who smoked 25 or more cigarettes per day and used oral contraceptives compared with women who did neither, Shapiro and colleagues [3] reported a relative risk of 39 and Rosenberg and coworkers [4] found a relative risk of 30. These extreme relative risks contrast with the four- to eightfold excess risks associated with smoking and the far lower risks seen in oral contraceptive users who do not smoke. Hence, studies of oral contraceptives and risk for myocardial infarction that do not evaluate smokers and nonsmokers separately are of limited use. Even those that adjust for smoking may not yield reliable estimates because of the interaction.

Taken together, case–control and cohort studies suggest that current users of oral contraceptives who are younger than 40 years of age and do not smoke have little or no increase in risk for myocardial infarction. Table 1 shows data from studies done since 1967 in nonsmokers [3, 5-12]. The more recent studies tend to find even smaller relative risks for current oral contraceptive use (regardless of smoking) than the earlier studies. Table 2 shows data from studies of low-dose contraceptives published since 1967 that did not stratify by smoking [13-23]. Among cohort studies, the results were as follows: No cases were seen in the Group Health Cooperative study [20, 24], a relative risk of 0.2 (95% CI, 0.05 to 0.7) was seen in the Finnish study by Hirvonen and Idanpaan-Heikkila [21], and a relative risk of 1.1 (CI, 0.5 to 2.5) was seen in the Copenhagen City Heart Study [19].

Most studies have been too small to address whether the risk for myocardial infarction from oral contraceptives differs according to coronary risk factors other than smoking. A recent hospital-based case–control study [18] found significantly increased risks for myocardial infarction in users of low-dose contraceptives in Europe and developing countries (4.69 and 2.93, respectively). These relative risks were higher than those seen in most previous studies, and the authors suggest that coexistent risk factors and inadequate screening among contraceptive users were responsible for the differences. In nonsmoking women who reported having their blood pressure checked before the current episode of contraceptive use, relative risks (1.34 [CI, 0.34 to 5.34] in Europe and 2.66 [CI, 0.81 to 8.79] in developing countries) were not statistically significant. The possibility of a synergistic effect between high blood pressure and oral contraceptive use has been suggested [11, 15].

The use of oral contraceptives is concentrated among young, fertile women with low incidence of myocardial infarction. In the United States, the baseline risk for fatal ischemic heart disease ranges from 1 to 2 per 100 000 persons in women younger than 35 years of age to 4.1 per 100 000 persons in women 35 to 39 years of age and to 10 to 21 per 100 000 persons for women in their 40s [25]. The bulk of the evidence suggests that nonsmokers have little or no increase in risk.

New Progestogens

Oral contraceptives that contain the new progestogens may not carry the same risk as earlier formulations, but their effect on the incidence of myocardial infarction or stroke has not been fully examined. A recent review [26] concluded that contraceptives containing desogestrel and gestodene have fewer adverse effects on carbohydrate and lipoprotein metabolism. Case-control studies of oral contraceptives that contain the newest progestogens suggest that these preparations carry lower relative risks for myocardial infarction than earlier preparations, but the data are sparse and based on fewer than 10 exposed cases [18, 22, 23] (Table 2 and Table 3). Moreover, because the data in these studies were not stratified by smoking status, the findings are difficult to interpret. The results suggest that the newer oral contraceptive preparations have an interaction with cigarette smoking but that the combined effect may be attenuated compared with that seen with earlier preparations.

Past Use

Because oral contraceptives can adversely affect coronary risk factors, there is considerable concern that women in their 50s and 60s would have delayed effects from the use of oral contraceptives. The data consistently show that past use is not associated with increased risk. In a meta-analysis of 13 studies of past use of oral contraceptives and risk for myocardial infarction [3, 5-711-13, 27-32], Stampfer and coworkers [33] estimated that past users had a pooled relative risk of 1.01 (CI, 0.91 to 1.13).

These findings suggest that any increase in risk for myocardial infarction due to oral contraceptive use occurs only with current use and probably acts through an acute prothrombotic interaction with cigarette smoking. Angiographic studies of young women with myocardial infarction tend to show an absence of atherosclerosis in cases associated with current or recent use of oral contraceptives [34-36]. In monkeys who were fed an atherogenic diet, oral contraceptives markedly decreased the extent of atherosclerosis compared with placebo [37, 38].

Stroke

Studies of oral contraceptives and stroke are often small, typically do not discriminate between hemorrhagic and thromboembolic stroke, and often do not control for major risk factors for stroke [39-41]. Unlike early studies, most recent prospective studies have not shown an increased risk for stroke among either past users of oral contraceptives or persons who had ever used oral contraceptives [20, 28, 42, 43]. Studies of stroke in current users have yielded inconsistent results. Table 4 shows data from studies of stroke in users of low-dose contraceptives [20, 21, 42-50].

Total Stroke

Previous studies of total stroke often combined users of high- and low-dose formulations. Because the risks for both ischemic and hemorrhagic stroke were higher among users of high-dose preparations, this practice inflated overall relative risks [51, 52]. Among recent cohort and nested case–control studies, one case was seen in the Group Health Cooperative study [20], a relative risk of 0.6 (CI, 0.1 to 2.9) was seen in the Royal College of General Practitioners Study [45], and no statistically significant increase in risk was seen in the study by Petitti and associates [46] (Table 4). A hospital-based case–control study found a nonsignificant relative risk of 1.41 (CI, 0.9 to 2.2) for all strokes in Europe but a relative risk of 1.86 (CI, 1.49 to 2.33) for all strokes in developing countries [44]. The study design, however, produced some limitations, including the possibility of recall bias and the difficult problem of proper selection of controls with diseases unrelated to use of oral contraceptives. This potential bias is not a concern for prospective studies, such as that by Petitti and associates [46].

Several studies of stroke suggest a positive interaction between oral contraceptives and smoking, although it is not as striking as the interaction between oral contraceptives and smoking associated with myocardial infarction. Hannaford and colleagues [45] stratified oral contraceptive use by smoking history, assessed only at baseline. Risk for total stroke among current users of oral contraceptives did not differ by smoking status. However, risk for fatal stroke was higher among current users with a history of smoking (relative risk, 7.1 [CI, 1.5 to 33.0]) than among nonsmokers (relative risk, 0.3 [CI, 0.0 to 3.0]). In former users, risk for fatal stroke appeared to be higher only among smokers. Petitti and associates [46] found a positive interaction between current use and smoking (P = 0.04) for hemorrhagic but not ischemic stroke. Current smokers who used contraceptives had a nonsignificant relative risk for hemorrhagic stroke of 3.6 (CI, 0.95 to 13.87) compared with a relative risk of 0.74 (CI, 0.17 to 3.25) for nonsmoking current users. The World Health Organization case–control study [49] also suggested that an interaction between smoking and contraceptives was associated with ischemic stroke.

The United Kingdom General Practice Research Database showed no substantial difference in risk for fatal stroke when the new progestogens were compared with earlier progestogens (two cases per 184 536 person-years for levonorgestrel, one case per 135 567 person-years for desogestrel, and two cases per 105 201 person-years for gestodene) [53].

No evidence supports the existence of an increased risk for stroke among past users of oral contraceptives, with the exception of smokers (relative risk, 1.8 [CI, 1.1 to 2.8]) in the Royal College of General Practitioners' study [45]. The Nurses' Health Study [28] found no statistically significant increase in risk for stroke among past users (ischemic stroke and subarachnoid hemorrhage were combined in this study).

Hemorrhagic Stroke

Oral contraceptives are associated with little or no increase in risk for subarachnoid and cerebral hemorrhage. Several studies have reported an increased risk for subarachnoid hemorrhage among past users, but these studies were based on few cases and included the use of high-dose preparations [7, 42, 54-58]. More recently, Hirvonen and Idanpaan-Heikkila [21] noted nine exposed cases in 485 000 person-years of use.

Five recent case–control studies (two of which were nested in prospective studies) found consistently nonsignificant relative risks for hemorrhagic stroke (Table 4) [43-47]. Petitti and associates [46] found a nonsignificant relative risk of 1.14 (CI, 0.60 to 2.16) for hemorrhagic stroke in a recent population-based, nested case–control study. These results are compatible with those of the recent case–control study by the World Health Organization [44], which found that current users of low-dose oral contraceptives had a nonsignificantly increased risk for hemorrhagic stroke in Europe and a moderately increased risk in developing countries. Women in developing countries were older, however, and had significantly fewer blood pressure checks before contraceptive use than women in Europe. The relative risk for women younger than 35 years of age was not statistically significant (relative risk, 1.05 [CI, 0.62 to 1.80]). The study by Petitti and associates, on the other hand, was conducted among carefully screened healthy women, and most current users in this study were younger than 35 years of age.

The incidence of hemorrhagic stroke in women 15 to 44 years of age is about 5.6 per 100 000 person-years [46]. In the United States, the incidence of fatal hemorrhagic stroke ranges from less than 1 per 100 000 persons for women younger than 35 years of age to 2.5 per 100 000 persons for women younger than 45 years of age [25]. Current studies provide no persuasive evidence of any increase in risk for hemorrhagic stroke among young women without risk factors who use low-dose contraceptives.

Ischemic Stroke

Ischemic stroke is uncommon in young women. Indeed, two cohort studies found too few cases to allow the calculation of relative risks for ischemic stroke [21, 42]. Several recent case–control studies (one nested in a prospective study) have suggested a small increase in risk [43, 46, 48-50].

In a Danish case–control study, Lidegaard [48] found that current users of contraceptives containing 30 to 40 µg of estrogen had a relative risk for ischemic stroke of 1.8 (CI, 1.1 to 2.9) compared with persons who had never used contraceptives. Surprisingly, former users had a 50% decreased risk compared with those who had never used contraceptives (CI, 0.4 to 0.7). Complicating the study was the use of a noncurrent control group and the inclusion of transient cerebral ischemia. A recent hospital-based case–control study examining the association between migraine and ischemic stroke [50] reported a relative risk of 2.7 with oral contraceptives containing 30 to 40 µg of estrogen and 1.7 with those containing 20 µg of estrogen. Whether these results were statistically significant is unclear because CIs were not reported. Furthermore, the study did not account for smoking and 45% of participants were smokers. In a study of fatal occlusive stroke, Thorogood and coworkers [43] found a nonsignificant relative risk of 4.4 (CI, 0.8 to 24.4) for persons who had ever used oral contraceptives. However, the number of case–control sets was too small to allow an estimate of relative risk associated with current use. Two recent large studies [46, 49] that adjusted for major risk factors for stroke, including smoking, found increased risks that were not statistically significant. In both studies, the CIs were wide and could not reliably differentiate between a true null and a small-to-moderate increase in risk.

Taken together, studies of low-dose oral contraceptives suggest that these drugs produce little increase in risk for ischemic stroke, but a two- to threefold increase in risk cannot be ruled out. Occlusive stroke in young women has an estimated rate of 5.4 per 100 000 person-years [46]. Fatal occlusive stroke is even rarer, with rates less than 0.5 per 100 000 for women younger than 45 years of age [25]. Therefore, any attributable risk for death from occlusive stroke associated with the use of oral contraceptives is small at most. Smokers may be particularly susceptible [46].

Venous Thrombosis and Pulmonary Embolism

Soon after oral contraceptives were introduced, their use was associated with a substantial increase in risk for thromboembolic disease [51, 52]; however, the early studies that reported these increases primarily examined products that contained 50 µg of estrogen or more. The study of thromboembolic disease presents major difficulties. Only a fraction of cases are diagnosed, and diagnosis is more likely in users of oral contraceptives because the association is so well known. Thus, equivocal cases in users may be more likely to have further diagnostic workup, and a history of oral contraceptive use may be sought more strenuously (and may be better documented) for this than for other illnesses. These difficulties are avoided when rates of thromboembolism are compared among women using different formulations. By comparing oral contraceptives that contain high levels of estrogen (>50 µg) with oral contraceptives that contain intermediate levels of estrogen (50 µg) and low levels of estrogen (<50 µg), Gerstman and colleagues [59] showed a dose-response relation between estrogen and venous thromboembolism. Of note, the only randomized, controlled trial showed no difference in risk for venous thromboembolism in 4965 users of oral contraceptives (containing 75 µg of mestranol) 15 to 40 years of age and 4933 controls who were using vaginal barrier methods of contraception (relative risk, 1.1 [CI, 0.4 to 2.9]) [60]. A further difficulty in the study of thromboembolic disease is the criteria used to define the diagnosis. Several older investigations used only a clinical diagnosis of thromboembolic disease that is now known to be unreliable, with misclassification rates as great as 50% [61, 62].

Low Estrogen Dose

Efforts to reduce risk for thromboembolic disease by decreasing estrogen content have proven successful. Bottiger and colleagues [63] noted a marked decline of approximately 80% in reports of nonfatal thromboembolism per 100 000 users when low-dose estrogen oral contraceptives replaced high-dose preparations in Sweden.

Recent studies continue to be limited by diagnosis bias and small numbers, and they often fail to provide information on dose. Table 5 summarizes information on studies of low-dose oral contraceptives [14, 20, 21, 24, 59, 64-67]. In a case–control study, Helmrich and coworkers [64] found a relative risk for venous thromboembolism of 11 (CI, 3.7 to 32) based on five cases with use of low-dose oral contraceptives. This risk was higher (although not significantly so) than the relative risk seen with use of high-dose contraceptives (relative risk, 5.5 [CI, 2.1 to 15]). Among the hospital controls, however, the age-standardized proportion of current oral contraceptive use was only 9%, lower than that for the general population; this finding suggests possible selection bias that may have yielded an overestimate of the relative risk. The World Health Organization hospital-based case–control study noted a relative risk of 2.9 (CI, 1.4 to 6.1) among current users of oral contraceptives but did not specify dose [14]. Thorogood and associates [65] found a nonsignificant relative risk for fatal venous thromboembolism of 1.6 (CI, 0.7 to 3.4) among users of predominantly low-dose oral contraceptives. Beaumont and coworkers [67], in a case–control study of hyperhomocysteinemia and thrombosis in 200 users of oral contraceptives, found that users of preparations containing 30 µg of estrogen had a nonsignificant reduction of 40% in risk for thrombosis relative to those using preparations that contained 50 µg of estrogen.

Data from prospective studies are sparse, reflecting the rarity of thromboembolism in young women (Table 5). Three cases were seen in the Group Health Cooperative study [20, 24], four deaths from thrombosis occurred in the study by Hirvonen and Idanpaan-Heikkila [21], and lower incidence rates were noted among users of low-dose contraceptives (39 per 100 000 person-years) compared with users of high-dose contraceptives (62 per 100 000 person-years) in the Oxford Family Planning study [68].

New Progestogens

The newest progestogens were introduced in an attempt to further decrease risk for cardiovascular side effects. Publicity surrounding an early pharmokinetic study of 22 women in Germany suggested an adverse effect for gestodene (but not desogestrel) [69, 70] that was not confirmed [71, 72]. However, German regulatory authorities alerted physicians to the possible danger of drugs containing gestodene and required an epidemiologic study of the relation between the newest progestogens and vascular disease. On the basis of several studies in progress, in October 1995 the Committee on Safety of Medicines in the United Kingdom warned physicians of a potentially increased risk for thromboembolic disease in users of contraceptives that contained desogestrel or gestodene. Many women either switched formulations or discontinued the use of oral contraceptives, leading to increased rates of unplanned pregnancies and abortions [73, 74]. The British Pregnancy Advisory Service, which provides nearly 20% of abortions in the United Kingdom, reported a 10% increase in the number of abortions performed after the government's warning. Of women surveyed, 40% reported discontinuing oral contraceptive use immediately after the warning [73]. At least one research study was discontinued because of publicity bias [75].

Recent studies [53, 76-78] suggest an increased risk for users of the most recent progestogens compared with persons who do not use oral contraceptives; this is approximately double the risk for users of levonorgestrel (Table 6). The increase in risk is similar for users of desogestrel and users of gestodene; in most (but not all) studies, the increase persists after adjustment for several risk factors for venous thromboembolism that may have influenced choice of preparation (for example, age, body weight, smoking habit, parity, and varicose veins).

Spitzer and colleagues [76] conducted a matched case–control study in 10 centers in Germany and the United Kingdom and found a modest increase in risk with products containing the newest progestogens compared with products containing earlier progestogens (relative risk, 1.5 [CI, 1.1 to 2.1]). The World Health Organization found a higher increase (relative risk, 2.6 [CI, 1.4 to 4.8]) in a hospital-based case–control study in nine countries, including the United Kingdom, where the most users of the newest progestogens were found [77, 79]. However, with general practitioner-based controls (community controls) in the United Kingdom study site, the relative risks were smaller and not statistically significant; the relative risk was 1.8 (CI, 0.7 to 4.8) for users of desogestrel and 0.9 (CI, 0.3 to 2.8) for users of gestodene. These varying results highlight the importance of appropriate selection and recruitment of controls and the biases that can occur with incomplete case or control series.

In a community-based, nested case–control study, Farmer and associates [78] reported age-adjusted relative risks of 1.76 (CI, 0.91 to 3.48) and 1.32 (CI, 0.70 to 2.49) for users of desogestrel and gestodene, respectively, compared with users of levonorgestrel. However, after possible confounders were controlled for, the relative risks were attenuated to 0.84 (CI, 0.38 to 1.85) and 0.87 (CI, 0.41 to 1.83). The authors attribute the decreased risk to residual confounding by age; however, they report crude and age-adjusted relative risks of 1.6 and 1.68, respectively, for users of the newest progestogens compared with users of the earlier progestogens. This negligible change in the point estimate provides strong evidence against confounding by age. No information was provided about concurrent disease, the only variable that significantly affected the regression model.

Overall, the data are consistent with a pattern of increased risk, but confounding and bias may account for some of the apparent differences in incidence of venous thromboembolism [75, 80]. One criticism of the findings has been that prescription of newly introduced drugs is overrepresented by first-time users, women who did not tolerate their previous oral contraceptives, and women with a family history of thrombosis. Thus, by comparing a new product with an old, one might be comparing women at higher risk with women at reduced risk. In support of this argument, Lewis and colleagues [75] identified the year during which the oral contraceptives that women 25 to 44 years of age were using had been introduced to the market. They found that risk successively increased as formulations became more recent. Jick and associates [53] found that the risk relative to users of levonorgestrel was higher in women who used desogestrel-containing oral contraceptives for less than 6 months (relative risk, 9.2 [CI, 0.9 to 36.3]) than in women who had used them for more than 6 months (relative risk, 1.8 [CI, 0.9 to 3.5]). However, these results may have been due to chance; the CIs were wide and overlapped. Bloemenkamp and colleagues [81] found no excess risk in women 15 to 19 years of age who, as young users, are more likely to receive newer products. Overall, this argument may explain part of the increased risk but is supported only by sparse data.

If these results are not due to bias, the thrombogenic activity of these progestogens may be different. Increases in progestational potency have been associated with increased risk for arterial disease [82, 83] but not venous thromboembolism [51, 84]. Reviews have concluded that oral contraceptives containing the newest progestogens do not differ from earlier low-dose combinations in their effect on hemostatic variables [26, 85, 86]. However, Rosing and coworkers [87] measured the effects of activated protein C on thrombin generation in plasma from 166 women and found that users of new progestogens were significantly less sensitive to activated protein C than were users of levonorgestrel or lynestrenol (P < 0.001). Indeed, the sensitivity of the new progestogen users was similar to that of heterozygous carriers of the factor V Leiden mutation who did not use oral contraceptives. The authors, however, did not distinguish among the three new progestogens.

Carriers of the factor V Leiden mutation (about 3% of the general population) have a sevenfold increase in thrombosis [66, 88] compared with noncarriers. Similarly, users of the newest progestogens have about a sixfold increased risk for thrombosis (a baseline relative risk of 2 to 4 for the older progestogens, multiplied twofold for the newest progestogens). This suggests that the newest progestogens may induce a prothrombotic state (through resistance to activated protein C) similar to that induced by factor V Leiden; this may, in part, explain the higher risk for thrombotic disease in women taking these progestogens.

The increased risk for thromboembolic disease with the newest compared with older progestogens is modest. The baseline rate of thromboembolic disease in the United States is 5 or less per 100 000 person-years among women of reproductive age [20, 24, 53], and the rate with the newest progestogens is approximately 30 per 100 000 person-years. Additional clinical and epidemiologic studies are needed to explore the thrombogenic potential of these contraceptives. Overall, the older high-dose oral contraceptives (containing 50 µg of estrogen or more) are associated with the highest risk for thromboembolic disease among all oral contraceptive formulations. Even for women with genetic coagulation disorders, the consequences of prescribing less reliable birth control methods should be carefully considered, given the fivefold increase in risk for thrombosis during pregnancy [89, 90].

Conclusions

Studies of the cardiovascular effects of combined oral contraceptives are hampered by two factors: the rarity of these events in young women and the changes in oral contraceptive formulations. Because cardiovascular events are uncommon, few occur even in large cohorts of women; this leads to statistically unstable estimates of effect. The case–control studies provide an efficient way to study more cases but may yield results that are difficult to interpret depending on control selection and participation.

Current use of contemporary oral contraceptive preparations by nonsmokers does not seem to increase risk for myocardial infarction. Lack of an effect of past use for coronary disease is now well documented [33]. Several factors may be responsible for the decline in absolute and relative risks for myocardial infarction associated with oral contraceptive use. Increased knowledge among health professionals of the effects of oral contraceptives has led to better selection of potential oral contraceptive users on the basis of medical history and better supervision of users during follow-up. In addition, the decline could be due in part to decreased use of oral contraceptives by smokers.

Current information is insufficient to permit a definitive statement about risk for stroke in women who use modern, low-dose contraceptives, but a large increase in risk can be excluded. There may be a small to moderate increase in risk for ischemic stroke, although no evidence supports an increased risk among past users.

Oral contraceptives that contain the newest progestogens are less androgenic, have less effect on carbohydrate and lipid metabolism, and may be associated with stronger suppression of ovarian activity. These properties have allowed a further reduction in the estrogenic content of oral contraceptives to 20 µg; no data show that further decreases in the estrogen dose confer any added benefit. However, the latest results provide some evidence that users of these progestogens have a higher risk for thromboembolic disease than do users of other oral contraceptives. This apparent increase in risk cannot be dismissed as a result of bias, but neither has it been confirmed. However, the findings of Rosing and colleagues [87] lend biological plausibility to this finding [91]. The effect of the newest progestogens on risk for myocardial infarction and stroke remains to be fully explored.

Finally, it is important to consider absolute rates of disease. Current preparations, including those containing the newest progestogens, are associated with a lower risk for cardiovascular disease relative to the risk associated with the original high-dose estrogen contraceptives and the risk associated with full-term pregnancy. In addition, cardiovascular diseases are becoming less prevalent. Cardiovascular diseases occur mainly among oral contraceptive users who smoke or have predisposing factors. Every effort should be made to encourage smoking cessation among potential users of oral contraceptives.

Dr. Stampfer: Channing Laboratory, 3rd Floor, 181 Longwood Avenue, Boston, MA 02115.