|
|||||
|
Oral Contraceptives and the Risk of Myocardial
Infarction
Bea C. Tanis, M.D., Maurice A.A.J. van den Bosch, M.D.,
Jeanet M. Kemmeren, Ph.D., Volkert Manger Cats, M.D., Frans M. Helmerhorst,
M.D., Ale Algra, M.D., Yolanda van der Graaf, M.D., and Frits R. Rosendaal,
M.D.
|
ABSTRACT
Background An
association between the use of oral contraceptives and the risk of
myocardial infarction has been found in some, but not all, studies.
We investigated this association, according to the type of
progestagen included in third-generation (i.e., desogestrel or
gestodene) and second-generation (i.e., levonorgestrel) oral
contraceptives, the dose of estrogen, and the presence or absence of
prothrombotic mutations.
Methods In a
nationwide, population-based, case–control study, we identified and
enrolled 248 women 18 through 49 years of age who had had a first
myocardial infarction between 1990 and 1995 and 925 control women
who had not had a myocardial infarction and who were matched for
age, calendar year of the index event, and area of residence.
Subjects supplied information on oral-contraceptive use and major
cardiovascular risk factors. An analysis for factor V Leiden and the
G20210A mutation in the prothrombin gene was conducted in 217
patients and 763 controls.
Results The odds
ratio for myocardial infarction among women who used any type of
combined oral contraceptive, as compared with nonusers, was 2.0 (95
percent confidence interval, 1.5 to 2.8). The adjusted odds ratio
was 2.5 (95 percent confidence interval, 1.5 to 4.1) among women who
used second-generation oral contraceptives and 1.3 (95 percent
confidence interval, 0.7 to 2.5) among those who used
third-generation oral contraceptives. Among women who used oral
contraceptives, the odds ratio was 2.1 (95 percent confidence
interval, 1.5 to 3.0) for those without a prothrombotic mutation and
1.9 (95 percent confidence interval, 0.6 to 5.5) for those with a
mutation.
Conclusions The risk of
myocardial infarction was increased among women who used
second-generation oral contraceptives. The results with respect to
the use of third-generation oral contraceptives were inconclusive
but suggested that the risk was lower than the risk associated with
second-generation oral contraceptives. The risk of myocardial
infarction was similar among women who used oral contraceptives
whether or not they had a prothrombotic mutation.
The first report of coronary thrombosis
associated with the use of oral contraceptives appeared in 1963.1
Later studies established the use of oral contraceptives as a risk
factor for venous as well as arterial thrombosis.2,3,4,5,6,7
Various modifications were made in an attempt to lower these risks,
including a reduction in the estrogen dose and changes in the progestagen
compound. Oral contraceptives containing an estrogen and the
progestagen desogestrel or gestodene, available since the 1980s, are
associated with at least a doubling of the risk of venous thrombosis
as compared with other combined oral contraceptives.8,9,10,11,12
It has been suggested that these third-generation contraceptives protect
against myocardial infarction by having a favorable effect on the
lipid profile,13,14,15
because studies showed that women who used these types had a slight
increase in the level of high-density lipoprotein cholesterol.15,16
Only a few studies of the association between oral contraceptives
and myocardial infarction have included a direct comparison of
third- and second-generation progestagens, and the results have been
contradictory.17,18,19,20,21
We investigated whether the use of low-dose combined oral contraceptives
affects the risk of myocardial infarction. We assessed the effect of
the type of progestagen included in the oral contraceptive (levonorgestrel
as compared with gestodene or desogestrel), the dose of estrogen,
and the presence of the G1691A mutation in the factor V gene (factor
V Leiden) and the G20210A mutation in the prothrombin gene, which
have been associated with myocardial infarction in young women22,23
as well as with a particularly high risk of venous thrombosis in
women who use oral contraceptives.24
Methods
Study Design
The Risk of Arterial Thrombosis in Relation to Oral Contraceptives
(RATIO) study is a population-based case–control study of the
relation of arterial disease to the use of oral contraceptives among
women 18 to 49 years of age in the Netherlands. The study protocol
was approved by the ethics committees of the participating hospitals
(see the Appendix). Oral informed consent was obtained from all
participants.
Identification of Women with Myocardial
Infarction
Eligible patients were women 18 to 49 years of age who were hospitalized
for a first myocardial infarction between January 1990 and October
1995. Myocardial infarction was defined by the presence of symptoms,
elevated cardiac-enzyme levels, and electrocardiographic changes.25
The patients were identified through a search of computerized
hospital data bases for International Classification
of Diseases, 9th Revision, Clinical Modification codes
for acute myocardial infarction. Of the 321 women who were admitted
to the 16 participating centers during this period, 29 (9 percent)
were excluded: 19 died during admission, 9 died between discharge
and the start of the study, and 1 was unable to participate. The
medical records of all patients were reviewed by one investigator.
Of the 292 remaining patients, 21 could not be located and 23
declined to participate (response rate, 85 percent).
Control Women
The study was designed to investigate three types of arterial
disease: myocardial infarction, ischemic stroke, and peripheral arterial
disease; the results for each type are reported separately. We
identified and recruited one large control group through random-digit
dialing.26
In this method, private telephone numbers randomly generated by a
computer are dialed until someone answers or at least seven attempts
have been made at various times of the day and the week, including
the weekend. We reached someone at 98 percent of the numbers after a
total of 15,725 telephone calls. Once it was ascertained that a
household included a woman who was eligible for the study, she was
asked to participate. We recruited control women from the six
geographic areas where the patients lived, and using questionnaires,
we assigned each an index year corresponding to the one of the six
years (1990 to 1995) in which the patients had had an index event.
Therefore, a control woman randomly received one of six
questionnaires concerning one of the index years. All questions
elicited information about either the index date (in the case of
questions about the body-mass index, menopausal status, level of
education, and family history), the year before the index date (in
the case of questions about a history of hypertension, diabetes,
hypercholesterolemia, alcohol use, and smoking), or the month before
the index date (in the case of questions about the use of oral
contraceptives). The index date was the date of the myocardial
infarction in the patients and midyear in the controls. To minimize
age differences between the patients and the controls, control women
in the older age groups were oversampled by increasing the age limit
of eligibility criteria during recruitment. The control group
therefore was a population sample stratified according to age (in
five-year categories), area of residence, and calendar year.
Eligible controls were women 18 to 49 years of age who had no
history of coronary, cerebral, or peripheral arterial disease. A
total of 1259 eligible women were reached by random-digit dialing,
925 of whom agreed to participate and returned the questionnaire (73
percent).
Data Collection
The standardized questionnaire that was mailed to patients and
controls included questions about demographics, use of oral contraceptives,
reproductive history, height and weight, and the presence or absence
of a history of hypertension, diabetes, hypercholesterolemia, and
cigarette smoking and a family history of cardiovascular disease.
Color photographs of all oral-contraceptive pills marketed in the
Netherlands during the study period were included to help women
recall the formulations they might have used. Oral contraceptives
were divided into four groups according to the type of progestagens
included: first-generation formulations containing lynestrenol or
norethindrone, second-generation formulations containing
levonorgestrel, third-generation formulations containing desogestrel
or gestodene, and oral contraceptives containing an estrogen and
other progestagens (cyproterone or norgestimate) or a progestagen
alone. We also classified oral contraceptives according to the dose
of estrogen. Women were categorized according to their use of oral
contraceptives (never, former, or current). The level of education
was categorized as primary school or less, secondary school, or
higher education or university. Obesity was defined as a body-mass
index (the weight in kilograms divided by the square of the height
in meters) of at least 27.3.23
Women were classified as having hypertension, diabetes, or
hypercholesterolemia when they reported that the condition had been
diagnosed by a physician or that they had been taking medication for
the condition before the index date. Smoking status was categorized
as never, former, or current. Current smokers were those who reported
smoking in the year before the index date. Alcohol use was
categorized as none, 1 to 15 drinks per week, and more than 15
drinks per week. A family history of cardiovascular disease was
defined as the occurrence of myocardial infarction, stroke, or
peripheral arterial disease before the age of 60 years in one or
more first-degree relatives.
Samples of venous blood or buccal swabs were obtained from 217
patients (88 percent) and 763 controls (82 percent) who consented to
undergo DNA analysis for factor V Leiden and the G20210A mutation in
the prothrombin gene. The polymerase chain reaction was used for the
analysis.27,28
Statistical Analysis
We used unconditional logistic-regression analyses to calculate
odds ratios for the relation between the use of oral contraceptives and
myocardial infarction, and we derived confidence intervals from the
model. We adjusted for the three stratification factors — age (in
five-year categories), area of residence, and calendar year — and
for putative confounding factors (smoking status; presence or
absence of hypercholesterolemia, diabetes, hypertension, obesity,
and a family history of cardiovascular disease; level of education;
and alcohol intake). To exclude an effect of the dose of estrogen in
the analyses that were focused on the type of progestagen included
in the oral contraceptive, we excluded women who used formulations
other than those containing 30 µg of ethinyl estradiol: 28 women (13
patients and 15 controls) used second-generation oral contraceptives
containing 50 µg of ethinyl estradiol, 67 women (13 patients and
54 controls) used triphasic second-generation oral contraceptives, 3
women (all controls) used triphasic third-generation oral contraceptives,
18 women (2 patients and 16 controls) used third-generation oral
contraceptives containing 20 µg of ethinyl estradiol, and in 6 women
(1 patient and 5 controls) the dose of ethinyl estradiol was
unknown. In a further effort to minimize the possibility of
confounding, in particular by the presence of preexisting disease,
we repeated the analysis after excluding women with major
cardiovascular risk factors. We also directly investigated whether
confounding was present, in particular prescription bias, by
analyzing risk factors and oral-contraceptive use in the control
women. Analyses of the dose of ethinyl estradiol were restricted to
women who used oral contraceptives containing 50 µg of ethinyl
estradiol and 150 µg of levonorgestrel or 30 µg of ethinyl estradiol
and 125 µg of levonorgestrel. Finally, we assessed the effect of
combinations of risk factors: the use of oral contraceptives and
conventional risk factors (current smoking, hypercholesterolemia,
diabetes, and hypertension), as well as factor V Leiden and the
G20210A mutation in the prothrombin gene.
Results
Table
1 shows the characteristics of the 248 women who had had a
myocardial infarction and the 925 control women. Patients ranged in
age from 24 to 49 years (mean, 43), and controls ranged in age from
18 to 49 years (mean, 38). Patients had a higher prevalence than
controls of major risk factors for cardiovascular disease, such as
hypertension (24 percent vs. 6 percent), hypercholesterolemia (11
percent vs. 3 percent), diabetes (6 percent vs. 1 percent), and
current smoking (84 percent vs. 43 percent). Patients also had a
lower level of education than controls (11 percent vs. 27 percent
with post–secondary-school education).
|
The risk of myocardial infarction among users of any type of oral contraceptive
was twice that of nonusers (95 percent confidence interval, 1.5 to
2.8), after adjustment for age, calendar year, and area of residence
(Table 2).
Additional adjustment for putative confounding factors increased the
odds ratio in most age categories, and the overall risk remained
doubled (Table 2).
Women with no conventional risk factors (hypertension,
hypercholesterolemia, diabetes, or smoking) who used oral
contraceptives had a relative risk of myocardial infarction of 3.1
(95 percent confidence interval, 1.0 to 9.2). The duration of
oral-contraceptive use did not differ significantly between patients
and controls (median, 10 years).
|
Second-generation oral contraceptives containing levonorgestrel were
used by 24 percent of the patients and 19 percent of the controls (Table 3).
Third-generation oral contraceptives containing desogestrel or
gestodene were used by 8 percent of the patients and 12 percent of
the controls. The odds ratio for myocardial infarction was 2.8 (95
percent confidence interval, 1.3 to 6.3) for women who used
first-generation contraceptives, as compared with those who had not
used oral contraceptives; 2.4 (95 percent confidence interval, 1.6
to 3.6) for women who had used second-generation contraceptives; and
1.3 (95 percent confidence interval, 0.8 to 2.3) for women who had
used third-generation contraceptives (Table 3). When
we restricted this analysis to users of second-generation oral
contraceptives (37 patients and 94 controls) and third-generation oral
contraceptives (18 patients and 91 controls) that contained 30 µg of
ethinyl estradiol, the odds ratios did not change substantially: 2.7
for users of second-generation oral contraceptives (95 percent
confidence interval, 1.6 to 4.3) and 1.6 for users of
third-generation oral contraceptives (95 percent confidence interval,
0.9 to 2.9). A direct comparison of oral contraceptives containing
30 µg of ethinyl estradiol and levonorgestrel, desogestrel, or
gestodene revealed an odds ratio for myocardial infarction of 0.5
(95 percent confidence interval, 0.2 to 1.1) for third-generation as
compared with second-generation oral contraceptives (after adjustment
for stratification variables). The odds ratios were similar for
third-generation brands containing desogestrel or gestodene. Further
adjustment for confounding did not affect these estimates (Table 3).
|
In an analysis that was restricted to the 41 patients and 104 controls
who had used contraceptives with a second-generation progestagen, as
compared with those who had not used oral contraceptives, the risk
of myocardial infarction was similar for oral contraceptives with
different doses of estrogen. The odds ratio was 2.0 (95 percent
confidence interval, 0.6 to 7.3) for brands containing 50 µg of
ethinyl estradiol with levonorgestrel and 2.6 (95 percent confidence
interval, 1.6 to 4.2) for brands containing 30 µg of ethinyl
estradiol with levonorgestrel. A direct comparison of oral
contraceptives containing levonorgestrel and ethinyl estradiol
revealed an odds ratio of 1.7 (95 percent confidence interval, 0.4
to 7.9) for all brands that contained less than 50 µg of ethinyl
estradiol as compared with brands that contained 50 µg of ethinyl
estradiol or more.
We analyzed the effect of other cardiovascular risk factors in
women who used oral contraceptives, as compared with the reference
category of women who had not used oral contraceptives and who did
not have the given risk factor (Table 4). The
adjusted odds ratios for myocardial infarction among women who had
not used oral contraceptives were 7.9 (95 percent confidence
interval, 4.9 to 12.9) for those who smoked, 5.1 (95 percent
confidence interval, 2.9 to 8.8) for those with hypertension, 3.3
(95 percent confidence interval, 1.6 to 6.8) for those with
hypercholesterolemia, 4.2 (95 percent confidence interval, 1.6 to
10.9) for those with diabetes, and 3.4 (95 percent confidence
interval, 2.2 to 5.3) for those who were obese. Among women who had
used oral contraceptives, the risk of myocardial infarction was
highest among those who smoked (odds ratio, 13.6), those who had
diabetes (odds ratio, 17.4), and those who had hypercholesterolemia
(odds ratio, 24.7).
|
Factor V Leiden or a G20210A mutation in the prothrombin gene was
present in 18 of 214 patients (8 percent) and 58 of 760 controls (8
percent). Two control women carried both mutations. The odds ratio
for myocardial infarction among women with a prothrombotic mutation
was 1.1 (95 percent confidence interval, 0.6 to 1.9), as compared
with women without a mutation. In the subgroup of smokers the
presence of one of these mutations increased the risk of myocardial
infarction by 1.6 (95 percent confidence interval, 0.8 to 3.3).
Among women younger than 35 years of age who had a prothrombotic
mutation, the odds ratio was 1.6 (95 percent confidence interval,
0.4 to 5.8), and among those who were at least 35 years old it was
0.9 (95 percent confidence interval, 0.5 to 1.7). The use of oral
contraceptives doubled the risk of myocardial infarction among women
without a prothrombotic mutation (odds ratio, 2.1; 95 percent
confidence interval, 1.5 to 3.0) and among women with a
prothrombotic mutation (odds ratio, 1.9; 95 percent confidence
interval, 0.6 to 5.5).
Discussion
In this case–control study we found that the use of currently
available combined oral contraceptives increased the overall risk
of a first myocardial infarction. As compared with nonusers, women
who used first- and second-generation oral contraceptives had a
significantly increased risk, but the results were inconclusive for
women who used third-generation oral contraceptives. The risk was
increased in all age groups except for the small group of women who
were 18 to 24 years old, and there were no significant differences
in the odds ratios between the age categories or between the doses
of estrogen. The risks were highest among users of oral
contraceptives who smoked, who had diabetes mellitus, or who had
hypercholesterolemia, but they were not affected by the presence of
factor V Leiden or the G20210A mutation in the prothrombin gene.
The use of second-generation oral contraceptives increased the
risk of myocardial infarction by a factor of 2.5. The use of third-generation
oral contraceptives did not increase the risk significantly (odds
ratio, 1.3). The direct comparison of second- and third-generation
oral contraceptives suggested that the use of third-generation
agents was associated with a lower risk of myocardial infarction,
but the confidence interval was wide and therefore a definite
conclusion could not be reached.
Five studies, including ours, have directly compared the effect
of the use of second- and third-generation oral contraceptives on
the risk of myocardial infarction,17,18,19,20,21
with reported odds ratios that ranged from 0.318
to 1.8.21
Only the study by Dunn et al.21
and our study were designed to assess whether the use of
third-generation oral contraceptives has a different effect on the
risk of myocardial infarction than does the use of second-generation
agents and included a sufficient number of women who used
third-generation oral contraceptives to allow conclusions to be
drawn. Dunn et al. suggested that the risk is higher with
third-generation than with second-generation oral contraceptives
(odds ratio, 1.8; 95 percent confidence interval, 0.7 to 4.8),
whereas we found the reverse (odds ratio, 0.5; 95 percent confidence
interval, 0.2 to 1.1). As can be seen from the confidence interval,
the study by Dunn et al. also did not permit a definite conclusion
to be reached.
Our study was designed as a nationwide, population-based,
case–control study, with patients recruited from all eight academic
centers in the Netherlands and eight surrounding hospitals. One of
the strengths of our study is that the use of both second- and
third-generation oral contraceptives is widespread in the
Netherlands, thus providing a large population of potential study
subjects. In the evaluation of our results, we also need to address
the possibility of bias. Because all patients with known myocardial
infarction were hospitalized and the patients were selected entirely
on the basis of the discharge diagnosis, selection bias is
improbable. The rate of nonresponse was fairly low and was unlikely
to have been associated with the use of oral contraceptives or the
type of agent used. Information bias was unlikely, because the women
were not told about the primary objective of the study and the questionnaire
elicited information about many issues. The subjects' recall was
optimized by the inclusion in the questionnaire of color photographs
of all available oral contraceptives.29
However, the possibility of recall bias cannot be excluded. Patients
who died after a myocardial infarction were not included in the
study, but it is unlikely that the use of oral contraceptives would
be a specific contributing factor to the case fatality rate.
Selective prescription following screening for risk factors may
affect the risks associated with the use of oral contraceptives. We
therefore investigated risk-factor status according to the use of
oral contraceptives in the control women and found little difference
in the prevalence of cardiovascular risk factors between those who
used oral contraceptives and those who did not (Table 5). There
were small differences in the incidence of hypercholesterolemia and
diabetes and in body-mass index, which were in part explained by the
younger age of oral-contraceptive users. To minimize the likelihood
of confounding, we also conducted an analysis restricted to women
with no cardiovascular risk factors and still found that women who
used oral contraceptives had a risk of myocardial infarction that
was three times the risk among nonusers.
|
Although the risk of myocardial infarction in users of oral contraceptives
is small in absolute terms, it has an important effect on women's
health, since 35 to 45 percent of women of reproductive age use oral
contraceptives.30
Because all combined oral contraceptives are equally effective means
of birth control, the issue of safety is paramount. Since the
absolute risk of myocardial infarction is highly age-dependent, the
risk associated with the use of oral contraceptives will have the
greatest effect in older women. A large number of women who were 35
years of age or older still used oral contraceptives (26 percent).
This finding, however, may be specific to the Netherlands (the rate
is 24 percent in national statistics).30
Before prescribing oral contraceptives, clinicians should screen
women for conventional risk factors for cardiovascular events, and
they should remember that the most important advice they can give
these women remains to quit smoking.
Supported by a grant (97-063) from the Netherlands Heart
Foundation. Dr. Helmerhorst has supervised research studies
sponsored by multiple pharmaceutical companies that manufacture
oral-contraceptive agents.
We are indebted to Dr. Bruno Stricker for advice during the planning
of the study, to Dr. Jan Vandenbroucke and Dr. Tim Farley for critical
reading of and advice on the analysis and writing, to Annemieke van
Dam for her work in contacting patients and controls as well as for
general data management, to Esther van Lunteren and Marjon de Boer
for their assistance in recruiting controls, to Tineke
Krommenhoek-van Es for the DNA analyses, to Dr. Hans Vos for
supervision, and to all the women who participated in this project.
Source Information
From the Thrombosis and Hemostasis Research Center, Department of
Hematology (B.C.T., F.R.R.), and the Departments of Cardiology (V.M.C.),
Obstetrics, Gynecology, and Reproductive Medicine (F.M.H.), and Clinical
Epidemiology (F.R.R.), Leiden University Medical Center, Leiden; and the Julius
Center for General Practice and Patient-Oriented Research (M.A.A.J.B., J.M.K.,
A.A., Y.G.) and the Department of Neurology (A.A.), University Medical Center,
Utrecht — both in the Netherlands.
Address reprint requests to Dr. Rosendaal at the Department of
Clinical Epidemiology, Leiden University Medical Center, Bldg. 1, C9-P, P.O.
Box 9600, 2300 RC Leiden, the Netherlands, or at [log in to unmask].
References
Appendix
The following investigators and centers in the Netherlands
participated in the study: Leiden University Medical Center, Leiden
— E.E. van der Wall; Sint Antonius Hospital, Nieuwegein — N.M.
van Hemel; Academic Medical Center, Amsterdam — R.J.G. Peters;
Rijnstate Hospital, Arnhem — H.A. Bosker; Medical Center Haaglanden,
Westeinde Hospital, The Hague — J. Kolf; University Medical Center,
Nijmegen–St. Radboud — F.W.A. Verheugt; Leyenburg Hospital, The
Hague — B.J.M. Delemarre; University Medical Center,
Rotterdam–Dijkzigt — F.A.M. Jonkman; Academic Hospital, Maastricht —
F. Vermeer; Rijnland Hospital, Leiderdorp — C. van Rees; Medical
Center Free University, Amsterdam — O. Kamp; University Medical
Center, Utrecht — E.O. Robles de Medina (deceased); Academic
Hospital, Groningen — M. van den Berg; Bronovo Hospital, The Hague —
P.R.M. van Dijkman; Sint Franciscus Hospital, Rotterdam — A.
Schelling; and Diaconessenhuis Leiden — S.A.G.J. Witteveen.
Edward E.
Rylander, M.D.
Diplomat American
Board of Family Practice.
Diplomat American
Board of Palliative Medicine.