Correspondence
Oral Contraceptives and the Risk of Myocardial Infarction
N Engl J Med 2002; 346:1826-1829June 6, 2002
- Article
To the Editor:
Chasan-Taber and Stampfer (Dec. 20 issue)1 state that there is increasing evidence that third-generation oral contraceptives are safer than previous formulations in terms of cardiovascular risk. Part of the evidence provided by Tanis et al. in the same issue2 is a nonsignificant reduction in the risk of myocardial infarction among women who used third-generation oral contraceptives, as compared with those who used second-generation oral contraceptives (odds ratio, 0.5; 95 percent confidence interval, 0.2 to 1.1). However, the wide confidence interval highlights the low statistical power of the study and indicates that random variation may be an alternative explanation for the observation.
Indeed, every study of this topic thus far has had low power — a consequence of the attempt to investigate an uncommon disease in premenopausal women. The small number of women using different formulations in each study has resulted in a range of unstable estimates of risk (odds ratios from 0.3 to 1.8), with only one reaching statistical significance.
It has been argued that a reduced risk of myocardial infarction among women who use third-generation oral contraceptives offsets any increased risk of venous thromboembolism. Such reasoning is sophistry. Every woman who uses third-generation oral contraceptives is at risk for venous thromboembolism since, as compared with arterial events, this disease is relatively common among women in their 20s and 30s. On the other hand, women tend to stop using oral contraceptives well before they are at risk for myocardial infarction. Furthermore, arterial disease among women who use oral contraceptives occurs mainly in those with cardiovascular risk factors, such as current smoking and hypertension. Therefore, most women who use oral contraceptives have little if any risk of myocardial infarction, regardless of the preparation used. Decisions about what brand of oral contraceptive to use should not be based on speculative arguments about cardiac benefits.
2 ReferencesPhilip Hannaford, M.D.
University of Aberdeen, Aberdeen AB25 2AY, Scotland
p.hannaford@abdn. ac. uk 1
Chasan-Taber L ,Stampfer M . Oral contraceptives and myocardial infarction -- the search for the smoking gun. N Engl J Med 2001;345:1841-1842
Full Text | Web of Science | Medline2
Tanis BC ,van den Bosch MAAJ ,Kemmeren JM , et al. Oral contraceptives and the risk of myocardial infarction. N Engl J Med 2001;345:1787-1793
Full Text | Web of Science | Medline
To the Editor:
Chasan-Taber and Stampfer cite a sponsored journal supplement to support their statement that “preferential prescribing and selective referral contributed to the apparent association” of third-generation oral contraceptives with venous thromboembolism. They disregard a meta-analysis that includes a careful discussion of bias and confounding,1 a review article in the Journal about the hemostatic effects of various contraceptives,2 a World Health Organization report,3 and a report from the European Agency for the Evaluation of Medical Products.4 Chasan-Taber and Stampfer state that “four of five prior studies” found a lower risk of myocardial infarction with third-generation oral contraceptives. Two of these studies analyzed the same data. The authors of the other studies interpreted their own findings differently.
The conclusion of the article by Tanis et al. is guarded. The only other large study found that the risk of myocardial infarction with third-generation preparations was double that with second-generation preparations.5 Chasan-Taber and Stampfer dismiss these results as due to “recall bias.” If anything, such bias was excluded: patients were recruited before October 1995 (when discussions about third-generation oral contraceptives started), and the type of contraceptive used was ascertained from independent sources (one being general practitioners' records).5
In view of reports that the risk of venous thrombosis is higher with third-generation oral contraceptives than with earlier contraceptives, as found in epidemiologic studies as well as studies of hemostasis,1-4 and in view of the uncertainty about any difference between generations of contraceptives with respect to the risk of myocardial infarction, the editorialists' conclusion that “increasing evidence suggests that third-generation oral contraceptives are indeed safer than previous formulations” is unfounded.
5 ReferencesJan P. Vandenbroucke, M.D., Ph.D.
Leiden University Medical Center, 2300 RC Leiden, the Netherlands
j.p. vandenbroucke@lumc. nl 1
Kemmeren JM ,Algra A ,Grobbee DE . Third generation oral contraceptives and risk of venous thrombosis: meta-analysis. BMJ 2001;323:131-134
CrossRef | Web of Science | Medline2
Vandenbroucke JP ,Rosing J ,Bloemenkamp KWM , et al. Oral contraceptives and the risk of venous thrombosis. N Engl J Med 2001;344:1527-1535
Full Text | Web of Science | Medline3
Cardiovascular disease and steroid hormone contraception: report of a WHO scientific group. WHO Tech Rep Ser 1998;877:1-89
Web of Science4
CPMP public assessment report: combined oral contraceptives and venous thromboembolism. London: European Agency for the Evaluation of Medical Products, 2001. (Document EMEA/CPMP/2201/01/en/Final.) (Accessed May 17, 2002, at http://www.emea.eu.int/pdfs/human/regaffair/0220101en.pdf.)
5
Dunn N ,Thorogood M ,Faragher B , et al. Oral contraceptives and myocardial infarction: results of the MICA case-control study. BMJ 1999;318:1579-1583
CrossRef | Web of Science | Medline
To the Editor:
In their article on oral contraceptives and the risk of myocardial infarction, Tanis et al. report that the use of any oral contraceptive increases a woman's risk of myocardial infarction (odds ratio for the comparison with nonuse, 2.0; 95 percent confidence interval, 1.5 to 2.8). This conclusion is probably an underestimate, because the study suffers from a “stack effect.” Tanis et al. used a population-based, case–control study design in which 248 women 18 to 49 years old were enrolled between 1990 and 1995. The authors state that controls were matched to patients with myocardial infarction according to age, calendar year of the index event, and area of residence. However, the authors failed to match the controls and the patients according to the frequency of their age distribution — a problem that leads to a fundamental design error that I refer to as a stack effect.1 A stack effect is an error that occurs when investigators “overmatch” young control subjects to young case subjects. This error may result in an underestimate of the true odds ratio in studies of diseases for which the risk factors — in this instance, the use of oral contraceptives — varies in relation to the age of the cohort.
Specifically, Tanis et al. show in Table 2 of their article that 327 controls and 37 patients 18 to 34 years old were enrolled, subgroups that correspond to 35.7 percent and 15.1 percent of the total number of controls (916) and patients (245). In addition, 252 controls and 117 patients (27.5 percent and 47.8 percent, respectively) 45 to 49 years old were enrolled. However, the authors also clearly show in this table that the frequency of oral-contraceptive use among women 18 to 34 years old (both controls and patients) was 66 percent but among women 45 to 49 years old was 20 percent. By including a higher relative percentage of younger controls than of younger patients and a lower relative percentage of older controls than of older patients, Tanis et al. in fact incorporated a stack effect. Thus, the reported odds ratio for myocardial infarction in association with the use of oral contraceptives is probably an underestimate of the true odds ratio, since younger women (who have a higher frequency of oral-contraceptive use than older women) were overrepresented in the control group.
1 ReferencesChris Kahlenborn, M.D.
Altoona Hospital, Altoona, PA 16601
kahlen@alt3.com To the Editor:
Tanis et al. show an interesting correlation between the use of oral contraceptives and myocardial infarction. However, they did not assess an important potential confounder: cocaine use. Cocaine use is an important risk factor for myocardial infarction, particularly in young patients, such as those who constitute a large proportion of the patients in this study. It would be interesting to know the prevalence of cocaine use in the two study groups.
Siu Fai Li, M.D.
Jacobi Medical Center, Bronx, NY 10461
souffle@banet.net To the Editor:
In their study of oral contraceptives and the risk of myocardial infarction, Tanis et al. suggest that information bias was unlikely because their questionnaires covered many issues and because the women were not informed about the primary objective of the research. However, the women (particularly the patients with myocardial infarction) may have been alert to the hypothesis because they were sent color photographs of oral contraceptives. In addition, since the study covered the period from 1990 to 1995 and since the questionnaires were mailed after 1995 (the exact date is not mentioned), the women had to recall accurately their use of oral contraceptives during a period one to six years previously. Information bias could have resulted in overestimation of the magnitude of the overall association between oral-contraceptive use and myocardial infarction.
This possibility can be explored by evaluating the well-documented interaction between oral-contraceptive use and heavy smoking (summarized in the editorial by Chasan-Taber and Stampfer). It is unlikely that a more than 30-fold increase in the risk (if present) of myocardial infarction among women who take oral contraceptives and who also smoke heavily can be fully accounted for by information bias. In addition, since the risk among women who used levonorgestrel was about twice that among women who used gestodene and desogestrel, it would be informative to determine whether the combined effects of heavy smoking and the use of these compounds were also different.
Samuel Shapiro, M.B.
Columbia University, New York, NY 10032
sydzeev@aol.com Author/Editor Response
The authors reply:
To the Editor: Dr. Hannaford points out that the lower risk of myocardial infarction that we observed in association with third-generation contraceptives (those containing desogestrel or gestodene) as compared with second-generation contraceptives (those containing levonorgestrel) may be explained by random variation around equivalence. We concur that this is the most likely explanation, given that the only other large study, one performed in the United Kingdom, found that the risk associated with third-generation oral contraceptives was 1.8 times that associated with second-generation contraceptives.1 Combining the evidence from these two studies to increase power, we find that the overall precision-weighted odds ratio for myocardial infarction among women who used third-generation contraceptives is 1.7 (95 percent confidence interval, 1.1 to 2.8) for the comparison with nonusers and 0.8 (95 percent confidence interval, 0.4 to 1.5) for the comparison with women who used second-generation contraceptives. So, third-generation contraceptives increase the risk of myocardial infarction, and the difference from second-generation contraceptives, if any, is small.
Third-generation oral contraceptives double the risk of venous thrombosis associated with second-generation oral contraceptives, which are already associated with a fourfold increase in risk relative to nonuse of oral contraceptives.2 Among young women, the incidence of venous thrombosis is much higher than the risk of arterial disease,3 and there are no obvious measures to reduce the risk of venous thrombosis (such as measurement of blood pressure or advice on smoking cessation), so physicians prescribing oral contraceptives should focus most on prescribing those associated with the lowest risk of venous thrombosis.
Perhaps we should focus not just on statistics but also on biology. There is clear evidence that during oral-contraceptive use there are prothrombotic, hemostatic changes that are more pronounced with third-generation preparations than with older preparations.4 There are also effects on lipids, but these effects probably do not explain the role of oral contraceptives in myocardial infarction, since the increased risk does not vary according to the duration of use and does not last after discontinuation.1 From the viewpoint of hemostasis, it is difficult to imagine how a substance would be prothrombotic in the veins and antithrombotic in the arteries.
Dr. Shapiro asks about the possibility of recall bias. Previous studies have shown that women are able to recall accurately their oral-contraceptive use.5 We agree that the strong interaction with smoking renders the possibility of recall bias unlikely.
The stack effect described by Dr. Kahlenborn is better known as confounding. Indeed, if a risk factor related to oral contraceptive use, such as age, is distributed unevenly among patients and controls, a crude odds ratio would be confounded. Therefore, we presented our results after stratification according to age and adjusted all estimates of risk according to age, an approach that fully takes care of this confounding.
Dr. Li raises the issue of cocaine use. Contrary to popular belief, drug use is not widespread in the Netherlands. In our study, only 5 of 1173 participating women (0.4 percent) reported the use of any illicit drug in the year preceding the myocardial infarction (or a similar date, among the controls).
5 ReferencesFrits R. Rosendaal, M.D.
Bea C. Tanis, M.D.
Leiden University Medical Center, 2300 RC Leiden, the Netherlands
f.r. rosendaal@lumc. nl Yolanda van der Graaf, M.D.
University Medical Center, 3508 GA Utrecht, the Netherlands1
Dunn N ,Thorogood M ,Faragher B , et al. Oral contraceptives and myocardial infarction: results of the MICA case-control study. BMJ 1999;318:1579-1583
CrossRef | Web of Science | Medline2
Kemmeren JM ,Algra A ,Grobbee DE . Third generation oral contraceptives and risk of venous thrombosis: meta-analysis. BMJ 2001;323:131-134
CrossRef | Web of Science | Medline3
Rosendaal FR . Thrombosis in the young: epidemiology and risk factors: a focus on venous thrombosis. Thromb Haemost 1997;78:1-6
Web of Science | Medline4
Rosing J ,Middeldorp S ,Curvers J , et al. Low-dose oral contraceptives and acquired resistance to activated protein C: a randomised cross-over study. Lancet 1999;354:2036-2040
CrossRef | Web of Science | Medline5
Norell SE ,Boethius G ,Persson I . Oral contraceptive use: interview data versus pharmacy records. Int J Epidemiol 1998;27:1033-1037
CrossRef | Web of Science | Medline
Author/Editor Response
The editorialists reply:
To the Editor: Our editorial accompanies an article on oral contraceptives and the risk of myocardial infarction, and therefore our comments regarding “cardiovascular disease” refer to arterial disease. Dr. Hannaford denounces the sophistry of (unnamed) critics who claim to balance any increased risk of venous disease with a reduced risk of myocardial infarction. We made no comments on this controversy. Venous disease is associated with a different set of risk factors and is not the focus of our editorial.
Both Dr. Hannaford and Dr. Vandenbroucke note the limited statistical power of studies that compare third-generation and second-generation oral contraceptives with respect to the risk of myocardial infarction. The study by Tanis et al. included substantially more women who used third-generation oral contraceptives than had previous studies. Although women who used third-generation oral contraceptives appeared to have about half the risk of myocardial infarction that women who used second-generation agents had (odds ratio, 0.5; 95 percent confidence interval, 0.2 to 1.1), this difference was not statistically significant. However, these findings were consistent with the direction of the effect observed in all but one prior study.
Dr. Vandenbroucke states that we dismissed the results of Dunn and colleagues1 as due to recall bias and goes on to declare that “if anything, such bias was excluded.” However, we wrote only that the findings “may have been due in part to recall bias,” in agreement with the views offered by Lidegaard2 in an editorial accompanying that study. Furthermore, Dunn and colleagues state that “recall bias may have occurred because we asked interviewees to recall contraceptive habits. The adverse publicity about third-generation oral contraceptives . . . may have biased responses to our questions.”1 Dr. Vandenbroucke also disapproves of our references. An editorial has limited space for references and cannot provide a comprehensive literature review. We cited the article by Heinemann as one recent review,3 one that is representative of the substantial literature in which these issues have been debated.4
We agree with Dr. Shapiro's useful suggestion that the interaction between smoking and oral-contraceptive use be examined for each generation of oral contraceptives separately. Most studies have been too small to address this issue, but taken together, they suggest a possible attenuation of the synergy between smoking and oral contraceptives with respect to the risk of myocardial infarction among users of second-generation and third-generation preparations.
It would be a shame if the main conclusion of our editorial were lost in this debate. We made no recommendations regarding switching brands of oral contraceptives, nor did we compare the relative risks of arterial disease and venous disease. Instead, the overriding point is that smoking is by far the most important risk factor for myocardial infarction in young and middle-aged women. Priorities should therefore include improvements in smoking-cessation programs and strategies to prevent young people from starting to smoke.
4 ReferencesLisa Chasan-Taber, Sc.D.
University of Massachusetts, Amherst, MA 01003
lct@schoolph.umass. edu Meir J. Stampfer, M.D., Dr.P.H.
Harvard School of Public Health, Boston, MA 021151
Dunn N ,Thorogood M ,Faragher B , et al. Oral contraceptives and myocardial infarction: results of the MICA case-control study. BMJ 1999;318:1579-1583
CrossRef | Web of Science | Medline2
Lidegaard O . Oral contraceptives and myocardial infarction: reassuring new findings. BMJ 1999;318:1583-1584
Web of Science | Medline3
Heinemann LA . Emerging evidence on oral contraceptives and arterial disease. Contraception 2000;62:Suppl:29S-36S
CrossRef | Web of Science | Medline4
Chasan-Taber L ,Stampfer MJ . Epidemiology of oral contraceptives and cardiovascular disease. Ann Intern Med 1998;128:467-477
Web of Science | Medline
- Citing Articles (4)
Citing Articles
1
Ira Wolinsky, Dorothy Klimis-Zacas, Anastasia Kalea. 2003. Nutritional Issues of Cardiovascular Disease in Women. , 257-293.
CrossRef2
(2002) Current Awareness: Pharmacoepidemiology and Drug Safety. Pharmacoepidemiology and Drug Safety 11:8, 727-742
CrossRef3
&NA;. (2002) Debate on cardiovascular risks with OCs continues. Reactions Weekly &NA;:906, 4
CrossRef4
&NA;. (2002) Debate on cardiovascular risks with OCs continues. Inpharma Weekly &NA;:1342, 20
CrossRef
