Third generation oral contraceptives and risk of myocardial infarction: an international case-control study

Introduction

The aim of the transnational project was to examine the safety of the third generation combined oral contraceptives, which contain the progestogens gestodene and desogestrel. At the outset gestodene was of special interest because of concerns in the European regulatory agencies and prominent discussion in the medical and lay press. The project incorporated three matched case-control studies with virtually identical methods1 2 for which the exposure factor of particular interest was use of third generation oral contraceptives. The outcomes for the three studies were venous thromboembolism, myocardial infarction, and ischaemic (thrombotic) stroke. Results of the study testing an association between the third generation products and venous thromboembolism are reported in the accompanying article.3 We report here the initial results of the case-control study assessing and contrasting the relation between second and third generation oral contraceptives and myocardial infarction in young women. Case recruitment continues. Identification and quantification of publicity bias will be incorporated in an amended protocol if extended field work proves feasible in the United Kingdom.

Results

We enrolled 153 cases of myocardial infarction (11 of them fatal) and 498 controls. Eighty two cases were identified in the United Kingdom, 47 in Germany, five in Switzerland, six in Austria, and 13 in France. When we compared current use of third generation with current use of second generation products as risk factors for myocardial infarction in all 651 women the odds ratio was 0.36 (0.1 to 1.2) (P=0.1; table 1). When current use of third generation products was compared with no previous use of oral contraceptives the odds ratio was 0.3 (0.1 to 1.0) (P=0.06). When we excluded the three countries with small sample sizes (Austria, Switzerland, and France) the estimates and confidence intervals became 0.45 (0.1 to 1.8) (P=0.26) for use of third generation v second generation products; 1.0 (0.3 to 3.5) (P=0.96) for current use of third generation products v no current use of oral contraception; and 2.2 (1.0 to 5.0) (P=0.07) for current use of second generation products v no current use.

When we compared women who currently used third generation oral contraceptives with those who currently used second generation products we considered the controls as two groups matched for hospital or community. The odds ratio was 0.91 (0.2 to 4.7) (P=0.9) for cases and hospital controls and 0.25 (0.06 to 1.0) (P=0.05) for cases and community controls.

In matched analyses the odds ratio was 0.40 (0.1 to 1.6) (P=0.19) in the comparison of third v second generation oral contraceptives. The odds ratio for second generation products v no use, was 3.1 (1.4 to 6.8) (P=0.01) and for third generation products v no current use 1.2 (0.34 to 4.4) (P=0.76). When current smoking was adjusted for use of oral contraceptives the odds ratio for the risk of myocardial infarction was 10.1 (5.7 to 17.9) (P<0.001) among our study subjects. Among women who used third generation products the crude odds ratio for current smoking was 3.1 (0.5 to 19.8) (P=0.23). Among those who used second generation products it was 11.1 (3.0 to 40.2) (P<0.001), and for women who were not current users of oral contraceptives the equivalent risk estimate was 7.7 (4.0 to 14.7) (P<0.001).

Given the prevalence of use of third and second generation preparations in the controls of this study, the observed odds ratio of 0.36 (0.1 to 1.2) for third compared with second generation products is consistent with a switch from second to third generation oral contraceptives, resulting in 12 fewer deaths from acute myocardial infarction per year in England and Wales. The confidence interval for the number of deaths gives a range from 18 fewer to three more deaths from acute myocardial infarction per year in England and Wales. In Germany the range is from 65 fewer deaths to 14 more; the point estimate is consistent with 46 fewer deaths annually with a switch from second to third generation oral contraceptives.

Discussion

Odds ratios for the risk of myocardial infarction ranged from 0.30 to 0.45 when we compared third generation with second generation oral contraceptives. These results are based on only six cases of acute myocardial infarction and 34 controls exposed to third generation oral contraceptives and on 23 cases and 45 controls exposed to second generation products. This is reflected in the crude odds ratio for the individual study components, which ranged from 0.06 (0.01 to 0.76); for one case and six controls) in the southern region (Austria, France, Switzerland) to 1.07 (0.25 to 4.55); for four cases and 18 controls) in the United Kingdom, with Germany having intermediate values (0.18 (0.02 to 1.62); for one case and 10 controls). The combined odds ratio for continental Europe was 0.15 (0.03 to 0.72); for two cases and 16 controls). The overall, fully adjusted odds ratio for all five countries was 0.36 (0.1 to 1.2). The point estimates are consistent with a benefit in the range of twofold to fourfold for those who use third generation rather than second generation oral contraceptives. This finding is from an interim analysis so should be interpreted with caution. If the findings are confirmed by subsequent work, it would translate into an important public health message. The excess risk of venous thromboembolism associated with the use of third generation oral contraceptives may be offset by the reduced risk of acute myocardial infarction when compared with second generation products.

As for the accompanying study, this study may be biased in terms of diagnosis, exposure selection, as well as attrition of susceptible subjects (a potential bias described in the accompanying article).3 Adjustment for duration of use of current oral contraceptives reduced the odds ratios further for the comparisons between third and second generation products. Matched analyses suggest strongly that overmatching did not occur. We showed that cigarette smoking was associated with a much higher risk of heart attack than the use of either generation of oral contraceptives.

These results are from only one study. Experimental, clinical, and laboratory studies both support and contradict the epidemiological findings.8 9 They cannot and should not be invoked to indict second generation oral contraceptives. The benefits of the well established low dose second generation oral contraceptives clearly outweigh their risk, particularly in developed and developing countries where third generation preparations have not been approved or are not available. They have dropped to half or less of the odds ratios observed in the epidemiological studies of oral contraceptives in the mid-1970s.10 11 The accompanying paper shows that products containing levonorgestrel, in particular, had low profiles of risk.3 The risks associated with pregnancy still heavily outweigh those associated with the use of any oral contraceptive. Women incur many risks, including smoking, with higher likelihood of harm than the probabilities of myocardial infarction associated with the use of oral contraceptives reported here.

Our findings on the potential benefits of third generation oral contraceptives with respect to the risk of myocardial infarction against the backdrop of potential harm with respect to risk of venous thromboembolism show that all known risks and benefits should be evaluated simultaneously when decisions about such products are taken. Both generations of oral contraceptives can be recommended on the basis of judicious clinical advice by the doctor and the informed choice of the woman using contraception.

The investigators were accountable only to the scientific reference board (members listed at the end of the accompanying paper3), which approved the protocol, received periodic reports, and conducted audits on the field and of the data before submission. The board was also advised on statistical issues by the statistical advisory group (members listed in the accompanying paper3).