Risk of myocardial infarction, angina and stroke in users of oral contraceptives: an updated analysis of a cohort study

Transcription

1 British Journal of Obstetrics and Gynaecology August 1998, Vol. 105, pp Risk of myocardial infarction, angina and stroke in users of oral contraceptives: an updated analysis of a cohort study Jonathan Mant Clinical Lecturer (Public Heulth Medicine), Rosemary Painter Computer Scientist, Martin Vessey Professor (Public Heulth) Division of Public Health and Primary Health Cure, University of Oxford, Institute of Health Sciences Objectives To investigate risk of myocardial infarction, angina and stroke in users of contraceptive pills compared with users of other methods of contraception. Design Prospective cohort study, with recruitment between 1968 and 1974 and annual follow up until the age of 45 years. After this age, only women who had never used oral contraception or those who had used it for eight or more years continued to be followed up annually until July Setting Seventeen family planning clinics in England and Scotland. Population 17,032 women aged between 25 and 39 years at entry to the study. Main outcome measures Occurrence of angina, myocardial infarction or stroke that was associated with either hospital admission or outpatient referral to hospital or death. Results Increased risk of myocardial infarction in oral contraceptive users was observed only in women who were heavy smokers at entry to the study. In this subgroup the relative risk of a myocardial infarction was 4.2 (95% CI ) in ever users of oral contraception compared with non-users, 4-9 ( ) in current users, and 4-0 ( ) in ex-users. In all current users the relative risk of angina was 0.5 ( ), and the relative risk of ischaemic stroke was 2.9 ( ). The increased risk of ischaemic stroke did not persist in ex-users. Conclusions Use of oral contraception is associated with increased risk of ischaemic stroke and increased risk of myocardial infarction (only in heavy smokers), but no increased risk of angina. These increased risks need to be considered within the context of the very low absolute risks of cardiovascular disease in this population women need to take oral contraception for one year to cause one extra stroke, and 1060 women who are heavy smokers need to take it for one year to cause one extra myocardial infarction. INTRODUCTION The Oxford Family Planning Association (Oxford- FPA) study was established in 1968 to monitor the effects on health of different methods of contraception. In 1984 a significantly increased risk of nonhaemorrhagic stroke was reported in current users of oral contraceptives in this cohort, and in 1987 a non-significant twofold increase in risk of myocardial infarction was reported in ever-users of oral contraception. No association was observed between use of oral contraception and angina or sub-arachnoid haemorrhage in these analyses, but at that time there were only 29 new cases of angina in the study population and 13 cases of sub-arachnoid haemorrhage. An analysis of mortality after 20 years of follow up found non-significant increases in risk of Correspondence: Dr J. Mant, Department of General Practice, The Medical School, Edgbaston, Birmingham B15 2TT, UK. 890 death from ischaemic heart disease [relative risk (RR) 3.3, 95% confidence interval (CI) 0.9 to 17.9) and cerebrovascular disease (RR 1.4, 95% CI 0.3 to 8-5) in oral contraceptive users compared with users of other forms of contraception3. The Oxford-FPA study is one of five prospective studies to have investigated possible links between method of contraception and cardiovascular disease4. Caseecontrol studies nested within the prospective Royal College of General Practitioners Oral Contraception Study (RCGP Study) found an increased risk of stroke and myocardial infarction in current users of oral contraceptive^^.^. The association with myocardial infarction was observed only in women who smoked. The Nurses Health Study found no association between past use of oral contraception and coronary heart disease or ~troke~-~. Neither the Walnut Creek Study nor the Puget Sound study found any significant association 0 RCOG 1998 British Journal of Obstetrics and Gynaerology

2 ORAL CONTRACEPTION AND CARDIOVASCULAR DISEASE 891 between oral contraceptive use and stroke or myocardial infarction'0,''. However, there were few cases in either study, so such associations could not be ruled out. Since cardiovascular events are rare in the age group of women who might take oral contraceptives, most of the epidemiological evidence of an association between oral contraception and cardiovascular disease comes from retrospective case control studies. These have tended to demonstrate an increased risk of stroke and myocardial infarction in current users of oral contra~eptives~j*-'~. However, some recent case-control studies have suggested that this increased risk is less (or even non-existent) with low oestrogen and third generation The aim of this report is to provide updated information concerning the relation between the oral contraceptive pill and cardiovascular disease (excluding venous thromboembolism) as observed in the Oxford-FPA Study. Specifically, we wished to test whether the associations between oral contraception and cardiovascular disease that have consistently been demonstrated in case control studies would be observed in a cohort study where there have now been sufficient numbers of cardiovascular events for statistically significant differences to be detected. METHODS The Oxford-FPA Study is a cohort study of 17,032 women who were recruited at 17 large family planning clinics in England and Scotland between 1968 and All participants were British, married, caucasian and aged between 25 and 39 years at entry to the study18. At recruitment, information was collected concerning method of contraception, past medical history (including hypertension and diabetes), social history, height and weight. The women were followed up annually until July Follow up was either by interview at the family planning clinic, or by postal questionnaire for women who had stopped attending the clinic. Women who did not return the questionnaire were approached either by telephone or home visit. Follow up information included changes in contraceptive practices, and reasons for any referrals that might have been made to hospital. Each hospital admission was followed up with the relevant consultant, and a copy of the discharge summary obtained. The women were flagged at the NHS central registers. The annual loss to follow up because women could not be traced or refused to co-operate was 0.4%; 15,292 women were still participating in the study at the age of 45, at which point they were sent a further questionnaire which, amongst other items, asked for information about smoking status. At this age, women were divided into three groilps: never-users of the oral contraceptive pill (n = 5881); users of the oral contraceptive pill for eight or more years (n = 3520); and the remainder (n = 5891). Only women in the first two groups were followed up after they reached 45 years of age. Analysis The conditions examined were myocardial infarction [International Classification of Disease 8th revision (ICD-8) code 410; angina (ICD-8 code 413); ischaemic stroke (ICD-8 codes , ); subarachnoid haemorrhage (ICD-8 code 430); intracerebral haemorrhage (ICD-8 code 43 1); and transient ischaemic attack (ICD-8 code 435). Each event was categorised according to contraceptive status at the time it occurred. Events occurring within 12 months of oral contraception being stopped were included within the 'current users' category. This is a routine practice in the Oxford-FPA study in order to minimise the distorting effects which might otherwise uccur if women who develop symptoms possibly attributable to cardiovascular disease stop taking the pill before an event occurs. Event rates were computed per woman years at risk in the study population and adjusted for relevant potential confounders (age, social class, smoking and obesity for all diagnoses; parity for myocardial infarction and angina only) using an indirect method of standardisation18. It was recognised that an important source of bias might be that women with increased risk of an event because of hypertension, diabetes or hyperlipidaemia were less likely to be prescribed oral contraception. Therefore, the analysis was repeated excluding women once they had one of these risk factors recorded. Restriction was the strategy chosen to overcome this potential bias in preference to other methods since few women were prescribed oral contraception in this study with such risk factors, and therefore there were very few outcome events in the subgroup of oral contraceptive users with known risk factors. Indeed, in the subgroup of current users of oral contraception, there were less than 1000 years of observation of women with any of these risk factors, and no cases of myocardial infarction, angina or stroke were recorded. It was anticipated that even with this restriction, there might be some residual confounding. Risk factors for cardiovascular disease which developed after entry to the study in patients that were managed in primary care without referral to hospital will not have been identified. Knowledge of these risk factors might have influenced choice of contraception. To assess whether the selective follow up introduced after women had reached the age of 45 might 0 RCOG 1998 Br J Obstet Gynaecol 105, 89G896

3 ~~ 892 J. MANT ET AL. have influenced the association with myocardial infarction, two separate analyses were performed. In the first analysis only woman years at risk and events that had accrued between women reached the age of 45 were taken into account (in other words, limiting the analysis to the period before the selective follow up had any effect). In the second analysis only years at risk and events after women had reached the age of 45 were included. RESULTS Myocardial infarction In total 85 women have been diagnosed as having myocardial infarction in the study population. We looked at the effect of various risk factors on the risk of myocardial infarction after adjustment for each of the other factors (except height and weight). Details concerning the categories used in making these adjustments are given in the footnote to Table 1. Risk of myocardial infarction was significantly positively associated with age, smoking and obesity (Quetelet index) and negatively associated with height. No significant association was found with weight, social class or parity. An unexpected finding was that the effect of height was essentially confined to women under 5, who had a twofold increase in risk of myocardial infarction compared with taller women. The relative risk of a myocardial infarction in women over 5 4 compared with women under 5 was 0.44 (95% CI ). Table 1 shows the effect of duration of use and time since last use of the oral contraceptive pill on risk of myocardial infarction in the whole study population, and in a subgroup which excludes women once they have been diagnosed as having hypertension, hyperlipidaemia or diabetes. This restriction reduced the total period of woman years of observation from 310,565 to 303,373. There are no significant effects of the oral contraceptive pill on risk of myocardial infarction in the whole population, but there is a significantly increased risk in women who had used the pill for up to eight years in the subgroup of women without hypertension, hyperlipidaemia or diabetes. As evidenced by the differences between the crude and adjusted rates, confounding, particularly by age, is having a major effect on the unadjusted results. In women younger than the age of 45, the adjusted relative risk of a myocardial infarction was 3.1 (95% CI ) in ever uses of oral contraception, whereas in women older than 45, the adjusted relative risk was 1-4 ( ). Table 2 shows the interaction between use of the oral contraceptive pill, smoking at entry to the study, and risk of myocardial infarction in the subgroup of women without hypertension, hyperlipidaemia or diabetes. There is no increased risk attached to use of the oral contraceptive pill in non-smokers or light smokers (up to 14 cigarettes a day). The excess risk of myocardial infarction in users of the oral contraceptive pill is only evident in heavy smokers. In this group there is a fourfold increase in risk of myocar- Table 1. The oral contraceptive pill (OCP) and risk of myocardial infarction. Rates and relative risks (RR) are adjusted for age, parity, social class, smoking, and quetelet index. Relative risks are as compared to never-users of the OCP. Restricted population excludes women once they have been given a diagnosis of hypertension, hyperlipidaemia, or diabetes. WYAR = woman years at risk. Crude rate Adjusted rate Adjusted RR No. of cases No. Woman (per 1000 (per 1000 Adjusted RR after restriction after of cases years at risk WYAR) WYAR) (95%) CI (95% CI) restriction Never-user of OCP o 1.o 31 Duration of use of OCP (years) Up to ( ) 1.9 ( )* 18 > l.o(o.61.6) l.o(o6-1.8) 21 Time since last use of OCP (years) Current ( ) 1.5 (0.63 2) (0 62.3) 1.2 ( ) 12 > (0 62.0) 1.4 ( ) 19 TOTAL Categories usedfor calculating adjusted rates: Age 25-39; 4W, 4549; 50-54; 55+. Parity: Cl; 2-3; 4+. Social class: I-11,111; IV-VI (VI: unemployed, students, and armed forces). Smoking: never; ex-; 1-14; 15+. Quetelet Index (kg/m2): ; ; ; ; ; RCOG 1998 Br J Obstet GynaecollO5,

4 ORAL CONTRACEPTION AND CARDIOVASCULAR DISEASE 893 Table 2. Interaction between smoking, use of oral contraceptive pill (OCP), and risk of myocardial infarction. The values given show the relative risk of myocardial infarction in ocp users compared with never-users (95% CI). Excludes women once they have been given a diagnosis of hypertension, hyperlipidaemia, or diabetes. Relative risks are adjusted for age, parity, sociadl class and quetelet index (categories as shown in footnote to Table 1). Relative risks are comparing risk in ocp users with never users within the relevant smoking stratum. Current use of ocp includes up to one year after stopping. No. of Smoking status at entry to the study cases Ever users of OCP Current user of OCP Ex-user of OCP Non-smoker ( ) only 1 case 0.6 ( ) Ex-smoker 4 all cases were in non-users Current smoker (1-14) (042.2) 0.6 ( ) 1.O (04-2.4) (15+) (1+16.6)* 4.9 ( )* 4.0 ( )* Table 3. The oral contraceptive pill (OCP) and risk of angina. Rates and relative risks (RR) are adjusted for age, parity, social class, smoking, and quetelet index. Relative risks are as compared to never-users of the OCP. Restricted population excludes women once they have been given a diagnosis of hypertension, hyperlipidaemia, or diabetes WYAR = woman years at risk. Crude rate Adjusted rate Adjcsted RR No. of cases No. Women (per 1000 (per 1000 Adjusted RR after restriction after of cases years at risk WYAR) WYAR) (95%) CI (95% CI) restriction Never-user of OCP so '.O 52 Duration of use of OCP (years) Up to ( ) 1.1 ( ) (04-0.9)* 0.7 (041.i) 23 Time since last use of OCP (years) Current ( )* 9.5 ( ) (0.5-1,3) 1.0 (0.61.9) 16 > (04-1.0) 0.7 ( ) 16 TOTAL dial infarction if the oral contraceptive pill is taken, from 0.24 (95%, CI ) per 1000 woman years at risk in heavy smokers who have never used oral contraception to 0.96 (95% CI ) per 1000 women years at risk in ex-users of oral contraception and to 1.18 (95% CI ) per 1000 woman years at risk in current users of oral contraception. Angina One hundred and four women were diagnosed as having angina during the study. After adjustment for each of the other factors (except height and weight; see footnote to table I), risk of angina was significantly positively associated with age, smoking and obesity and negatively associated with height and social class. There was no significant association with weight or parity. The relation to height was similar to that observed for myocardial infarction. The risk of angina was generally lower in users of oral contraception compared with non-users. Currents users of oral contraceptives had a 70% lower risk of angina than never users, but there were only three cases in current users, so the confidence intervals are wide (Table 3). In the subgroup of women without a diagnosis of diabetes, hyperlipidaemia or hypertension, there were no significant differences in risk of angina between users and non-users of oral contraception. Stroke During the study 28 women were diagnosed as having a subarachnoid haemorrhage, 27 women a transient ischaemic attack, 44 women an ischaemic stroke, and 7 an intracerebral haemorrhage. After adjustment for other factors (see footnote to Table 1) positive associations were observed between age and smoking and risk of ischaemic stroke. No significant associations were observed between use of the oral contraceptive pill and risk of subarachnoid haemorrhage or transient ischaemic attack. Table 4 shows the effects of the oral contraceptive 0 RCOG 1998 Br J Obstet GynaecollO5, 89&896

5 894 J. MANT ET AL. pill on risk of ischaemic stroke. Restricting the population as described above reduces the woman years at risk from 310,564 to 303,297. This is slightly different from Table 1, since a different end-point has been used. There is no significant relation between duration of use and risk of stroke, but there is an excess risk of ischaemic stroke in current users. This risk disappears once women stop taking the pill. The relative risk of an ischaemic stroke in current users was 3.1 (95% CI j in non-smokers, 1.5 (95% CI ) in ex-smokers, 1.5 (95% CI ) in moderate smokers (1-14 per day), and 6.5 (%YO CI ) in heavy smokers (1 5+ per day). Changes in smoking status between entry to study and age 45 The analyses performed above were all adjusted for smoking status at entry. At this time 17% of oral contraceptive users were heavy smokers (over 14/day), and 19% were moderate smokers (less than 15/day)18. With regard to users of other forms of contraception, 9% were heavy and 17% moderate smokers. The smoking status of 80% of women did not change between entry to the study and age 45. At this age 16% of women who had used oral contraception were heavy smokers, and 10Yo moderate smokers. Eleven percent of women who had never used oral contraception were heavy smokers and 8Yo moderate smokers. DISCUSSION In this study use of the oral contraceptive pill was associated with an increased risk of myocardial infarction in women who smoked more than fifteen cigarettes per day. This increase in risk was observed both in current and past users of oral contraception. However, women who did not smoke, or who smoked less than 15 cigarettes per day, did not increase their risk of myocardial infarction by taking the oral contraceptive pill. A different relationship was observed with angina. Users of oral contraception were at lower risk of angina than non-users, though this reduction in risk was not significant when the analysis was restricted to women who did not have hypertension, diabetes, or hyperlipidaemia. Current users of oral contraceptives were observed to be at increased risk of ischaemic stroke, but this increased risk did not persist in past users. Possible effects of bias The analyses were adjusted to take account of the possible effects of differences in age, parity, social class, obesity, and smoking between oral contraceptive pill users and non-users. There may have been some minor residual confounding with regard to smoking, since the results were adjusted for smoking status at entry to the study, and could not take into account changes in smoking habits over the years. However, we found that the majority of women had not changed their smoking habits by the age of 45, and the changes that had occurred did not appear to be closely related to oral contraceptive use. Another potential confounder is comorbidity, in particular risk factors for cardiovascular disease, such as hypertension, diabetes, and hyperlipidaemia. In order to control for this the analysis was restricted to women without these diagnoses. The effect of restriction was to increase the size of the relative risk of myocardial infarction and stroke in oral contraceptive users, and to reduce the apparent protection offered against angina. This is as one might have anticipated, since women with risk factors for cardiovascular disease are less likely to be prescribed Table 4. The oral contraceptive pill (OCP) and risk of ischaemic stroke. Rates and relative risks (RR) are adjusted for age, social class, smoking, and quetelet index. Relative risks are as compared to never-users of the OCP. Restricted population excludes women once they have been given a diagnosis of hypertension, hyperlipidaemia. or diabetes. WYAR = women years at risk. Crude rate Adjusted rate Adjusted RR No. of cases No. Women (per 1000 (per 1000 Adjusted RR after restriction after of cases years at risk WYAR) WYAR) (95%) CI (95% CI) restriction Never-user of OCP Duration of use of OCP (years) o 1.o 12 Up to ( ) 1.2 (042.9) 10 >8 Time since last use of OCP (years) ( ) 2.0 (0.946) 15 Current (l.l-s,l)* 2.9 ( )* ( ) 0.7 ( ) 4 > (042.8) 1.2 (043.5) 6 TOTAL RCOG 1998 Br J Obstet Gynaecol105,

6 the oral contraceptive pill. Even with restriction, it is conceivable that there was residual confounding, since cases of hypertension, hyperlipidaemia or diabetes that became apparent after entry to the study which were managed entirely in primary care will not have been identified. This residual confounding will have reduced the size of any positive association between oral contraception and cardiovascular disease. Other potential confounders include aspects of lifestyle such as use of alcohol, diet and exercise. In the Walnut Creek study non-users of oral contraception were more likely to be heavy drinkers or non-drinkers than users'". Given the U-shaped relation that exists between alcohol and risk of cardiovascular di~ease,'~ if alcohol is a confounder of the relation between oral contraceptive use and cardiovascular disease, it is likely to act by masking any adverse effect. It is difficult to predict in which direction other aspects of lifestyle might confound the association between oral contraception and cardiovascular disease. Selection bias is an important consideration, but on balance, it seems more likely that the effect of such bias would be to reduce any positive association between oral contraceptives and cardiovascular disease. Therefore, selection bias may explain the apparent lower risk of angina in oral contraceptive users, but is an unlikely explanation for the increased risks of myocardial infarction and stroke that were observed. Loss to follow up was low in this study. At the age of 45 a selective follow up was initiated in that only women who had never used oral contraception or who had used it for eight or more years were still followed up. This makes the observed difference in risk of myocardial infarction of women younger and older than the age of 45 in relation to oral contraceptive use difficult to interpret. It may be simply be an artefact of the selective follow up, since women at highest risk of myocardial infarction, i.e. those who had used oral contraception for between one and eight years, were not followed up beyond the age of 45. Alternatively, it is plausible that the effect of oral contraception is modified by age. Whatever the explanation, it seems likely that if selective follow up did introduce bias, then its effect would have been to mask any increased risk of myocardial infarction in users of oral contraception. It has been suggested that part of the explanation for the apparent link between the oral contraceptive pill and cardiovascular disease is ascertainment bias; that there might be more thorough diagnosis and surveillance of users than non-users.20 We believe this to be an unlikely explanation for the associations observed in the Oxford-FPA study. Only hospitalconfirmed diagnoses of cardiovascular events were recorded. Stroke and myocardial infarction diag- 0 RCOG 1998 Br J Obstet GynaecollO5, ORAL CONTRACEPTION AND CARDIOVASCULAR DISEASE 895 nosed in hospital are reasonably 'hard' end-points. Transient ischaemic attack was excluded from the stroke analyses, since this is arguably a 'soft' diagnosis that might be prone to such bias. Comparison with other studies The results of the Oxford-FPA study concur with the results of the RCGP study, which is the only other prospective study of oral contraception to have been carried out in the UK. Both studies observed increased risk of myocardial infarction in oral contraceptive users which was almost entirely confined to heavy srnokers.'j In the RCGP study the risk was much more marked in current users than in past users, but in the Oxford-FPA study, the risk appears to persist. However, a clinically important persistence of risk is unlikely. The large Nurses' Health Study in the United States observed no increased risk of myocardial infarction associated with past use.' Furthermore, our understanding of the pathology of oral contraceptive related heart disease is that it is connected with changes in coagulation rather than to atheroma formation,21 so it would be difficult to explain why an increased risk might persist after oral contraception ceases. Both the RCGP study and the Oxford-FPA study found an increased risk of ischaemic stroke in current users of oral contraception, but not in ex-users, though there was some evidence in the RCGP study that the risk persisted in smoker^.^ We are not aware of any published studies that have explored the possible relationship between oral contraceptive use and angina. The Oxford-FPA study found some evidence that the oral contraceptive pill might protect against angina, but it is possible that this reflects selection bias. In contrast to myocardial infarction, it is interesting to note that there is no evidence from these data of a positive association of oral contraceptive use with angina. The differing effects of oral contraception on these two manifestations of coronary heart disease may reflect the underlying pathological mechanisms: changes in coagulation might be expected to increase risk of myocardial infarction, but not angina. The results are also consistent with the recent WHO Collaborative case control ~tudies,'~.~~ These reported an odds ratio of 2-99 for ischaemic stroke, and 5.01 for acute myocardial infarction, with an important interaction with smoking. Many of the oral contraceptive users in the WHO study had other risk factors for heart disease. Therefore, the higher odds ratio in the WHO STUDY for myocardial infarction probably reflects the interaction of oral contraception with these risk factors.

7 896 J. MANT ET AL. Height and risk of ischaemic heart disease In this study short women (ie, under 5 tall) are at increased risk of angina and myocardial infarction compared with taller women. A similar association has been reported in other studies that have investigated the relation between height and risk of heart disease.22 Implications for clinical practice Sixty-eight percent of the woman years of exposure to oral contraception in this study was to pills with 50 pg of oestrogen, which is higher than the standard doses in use today. By 1987 only 2.7% of oral contraceptives prescribed in the United Kingdom contained 50 pg of ~estrogen.~~ Nevertheless, there are lessons that can be drawn for current clinical practice. Even with these older dosages, the absolute risks of cardiovascular disease in oral contraceptive users are low and could be reduced further if women who take oral contraception are actively encouraged not to smoke. From the data in the Oxford-FPA study 5880 women need to take the oral contraceptive pill for one year to cause one extra stroke, and 1060 women who are heavy smokers need to take it for one year to cause one extra myocardial infarction. However, the biases at play in the Oxford-FPA study might have underestimated the risks of oral contraception with regard to cardiovascular disease. Also, the healthier characteristics of the study population compared with the general population in terms of smoking habits, weight, and social class24 will have lowered the absolute risks of disease. Nevertheless, these factors will have been counterbalanced by the apparent greater safety of the lower dose pills that are now in common ~se. ~. ~ Overall, the data provide reassurance that the risks of cardiovascular disease are low in women who are prescribed the pill appropriately. Acknowledgements We would like to thank Mrs J. Winfield and the staff at the participating clinics for their assistance with this project. The Oxford-FPA study is funded by grants from the Medical Research Council and the Knott Family Trust. References Vessey MP, Lawless M, Yeates D. Oral contraceptives and stroke: findings in a large prospective study. BMJ 1984; 289: Mant D, Villard-Mackintosh L, Vessey MP, Yeates D. Myocardial infarction and angina pectoris in young women. J Epidemof Comtnun Health 1987;41: Vessey MP, Villard-Mackintosh L, McPherson K, Yeates D. Mortality among oral contraceptive users: 20 year follow up of women in a cohort study. BMJ 1989; 299: Thorogood M. Oral contraceptives and cardiovascular disease: an epidemiologic overview. Pharmacoepidemiol Drug Sa+ty 1993; 2: Hannaford PC, Croft PR, Kay CR. Oral contraception and stroke: evidence from the Royal College of General Practitioners Oral Contraception Study. Stroke 1994; 25: Croft P, Hannaford PC. Risk factors for acute myocardial infarction in women: evidence from the Royal College of General Practitioners oral contraception study. BMJ 1989; 298: Stampfer MJ, Willett WC, Colditz GA, Speizer FE, Hennekens CH. A prospective study of past use of oral contraceptive agents and risk of cardiovascular diseases. N End J Med 1988; 319: Stampfer MJ, Willett WC, Colditz GA, Speizer FE, Hennekens CH. Past use of oral Contraceptives and cardiovascular disease: a meta-analysis in the context of the Nurses Health Study. Am J Obstet Gynecoll990; 163: Colditz GA for the Nurses Health Study Research Group. Oral contraceptive use and mortality during 12 years of follow up: the Nurses Health Study. Ann Intern Med 1994; 120: Ramcharan S, Pelligrin FA, Ray R, Hsu JP. The Walnut Creek Contraceptive Study. Center for Population Research Monograph No National Institute of Health Washington DC, Porter JB, Hunter JR, Jick H, Stergachis A. Oral contraceptives and non-fatal vascular disease. Ohstet Gynecol1985; 66: WHO Collaborative Study of Cardiovascular Disease and Steroid Hormone Contraception. Ischaemic stroke and combined oral contraceptives: results of an international, multicentre. case-control study. Luncet 1996; 348: WHO Collaborative Study of Cardiovascular Disease and Steroid Hormone Contraception. Haemorrhagic stroke, overall stroke risk, and combined oral contraceptives: results of an international, multicentre, casecontrol study. Luncet 1996; 348: WHO Collaborative Study of Cardiovascular Disease and Steroid Hormone Contraception. Acute myocardial infarction and combined oral contraceptives: results of an international multicentre caseecontrol study. Lancet 1997; 349: Petitti D, Sidney S, Bernstein A, Wolf S, Quesenberry C, Ziel HK. Stroke in users of low-dose oral contraceptives. N Eng/ J Med 1996; 335: Sidney S, Petitti DB, Quesenberry CP, Klatsky AL, Ziel HK, Wolf S. Myocardial infarction in users of low-dose oral contraceptives. Obstet Gynecoll996; 88: Lewis MA, Spitzer WO, Heinemann AJ, Macrae KD, Brupacher R, Thorogood M on behalf of Transnational Research Group on Oral Contraceptives and the Health of Young Women. Third generation oral contraceptives and risk of myocardial infarction: an international casecontrol study. BMJ 1996; Vessey MP, Doll R, Peto R, Johnson B, Wiggins P. A long term follow up study of women using different methods of contraception: an interim report. J Biosoc Sci 1976; 8: Doll R, Peto R, Hall E, Wheatley K, Gray R. Mortality in relation to consumption of alcohol: 13 years observations on male British doctors. BMJ 1994; 309: Katerndahl DA, Realini JP, Cohen PA. Oral contraceptive use and cardiovascular disease: is the relationship real or due to study bias? J Fam Pract 1992; 35: Godsland IF, Crook D. Pathogenesis of vascular disease in oral contraceptive users. B r J Curdiol 1996; 3: Nwasokwa ON, Weiss M, Gladstone C, Bodenheimer MM. Higher prevalence and greater severity of coronary disease in short versus tall women. Am Heart J 1997; Thorogood M, Vessey MP. Trends in use of oral contraceptives in Britain. Br JFumily Plan 1990; Vessey MP, McPherson K, Johnson B. Mortality among women participating in the Oxford/FPA contraceptive study. Lancet 1977; Received26 September 1997 Returned for revision 4 February 1998 Accepted30 April RCOG 1998 Br J Obstet Gynaecol 105, 89C896