Elsevier

The Lancet

Volume 348, Issue 9026, 24 August 1996, Pages 498-505
The Lancet

Articles
Ischaemic stroke and combined oral contraceptives: results of an international, multicentre, case-control study

https://doi.org/10.1016/S0140-6736(95)12393-8Get rights and content

Summary

Background

The association between use of oral contraceptives (OCs) and cerebral infarction was established in studies from northern Europe and the USA during the 1960s and 1970s. Since then, the constituents of hormonal OCs have changed and now contain lower doses of oestrogen and progestagen. Current recommendations restrict OC use to younger women who do not have other risk factors for cardiovascular disease. In this international study we assessed the risk of OC-associated first stroke in women from Europe and other countries throughout the world.

Methods

In this hospital-based, case-control study, we assessed the risk of ischaemic stroke in association with current use of combined OCs in 697 cases, aged 20-44 years, and 1962 age-matched hospital controls in 21 centres in Africa, Asia, Europe, and Latin America. The diagnosis of ischaemic stroke was almost exclusively based on computed tomography (CT), magnetic resonance imaging (MRI), or cerebral angiography carried out within 3 weeks of the clinical event. All cases and controls were interviewed while in hospital with the same questionnaire, which included information on medical and personal history, details of lifetime contraceptive use, and blood-pressure measurements before the most recent episode of OC use.

Findings

The overall odds ratio of ischaemic stroke was 2·99 (95% CI 1·65–5·40) in Europe and 2·93 (2·15–4·00) in the non-European (developing) countries. Odds ratios were lower in younger women and those who did not smoke, and less than 2 in women who did not have hypertension and who reported that their blood pressure had been checked before the current episode of OC use. By contrast, among current OC users with a history of hypertension, the odds ratio was 10·7 (2·04–56·6) in Europe and 14·5 (5·36–39·0) in the developing countries. In Europe, the odds ratio associated with current use of low-dose OCs (<50 μg oestrogen) was 1·53 (0·71–3·31), whereas for higher-dose preparations it was 5·30 (2·56–11·0). In the developing countries, there was no significant difference between overall estimates of risk associated with use of low-dose or higher-dose OCs (3·26 [2·19–4·86] vs 2·71 [1·75–4·19]). This differential effect of dose in Europe and the developing countries is likely to be due to different levels of other risk factors among users of low-dose and higher-dose OCs in the two groups of countries. There was no significant increase in odds ratios with increasing duration of OC use among current users; odds ratios were not significantly increased after cessation of OC use.

Interpretation

The incidence of ischaemic stroke is low in women of reproductive age and any risk attributable to OC use is small. The risk can be further reduced if users are younger than 35 years, do not smoke, do not have a history of hypertension, and have blood pressure measured before the start of OC use. In such women OC preparations with low oestrogen doses may be associated with even lower risk.

Introduction

The possibility that cerebrovascular accidents (strokes) might be induced by oral contraceptives (OCs) was first raised by Lorentz1 in 1962, shortly after OCs became available. Subsequent case-reports and reviews of temporal changes in clinical experience of stroke among young women2, 3 suggested an association between stroke and OC use, a finding that was consistent with the results of several case-control”3 and cohort studies.14, 15, 16, 17, 18 Concern about the cardiovascular side-effects of OCs has limited their use world wide, although an estimated 60·70 million women are OC users.19

Most studies of the cardiovascular side-effects of OCs were carried out during the 1960s and 1970s, and there are four reasons to re-examine this association. First, OCs contain lower doses of oestrogen and progestagen than they did in the past, and current recommendations are to restrict use to younger women who do not have other risk factors for cardiovascular disease.20 Although these changes seem appropriate, few data are available to confirm whether there has been a reduction in cardiovascular risk associated with OC use. Second, most previous studies have been done in northern Europe and the USA, and extrapolation of these findings to populations with different incidence rates and prevalence of risk factors for stroke may not be appropriate. Third, although use of OCs seemed to be associated with ischaemic stroke in previous studies, the association with haemorrhagic stroke is less certain—perhaps partly because of inadequate differentiation of stroke type. Finally, previous studies did not have sufficient power to assess effects of duration and past use of OCs, and any interaction between OC-associated risk and other risk factors, such as hypertension and smoking.

This paper reports the association between ischaemic stroke and use of combined OCs, in a multicentre, hospital-based, case-control study of stroke, acute myocardial infarction, and venous thromboembolism, in Africa, Asia, Europe, and Latin America (including the Caribbean). The primary aim of the stroke component of the study was to assess whether current OC use was associated with increased risk of a first stroke in women from Europe and from the other three regions combined (the developing countries). Secondary aims were to evaluate the risk of first stroke in each of the four regions, and assess whether the risk differed among subgroups of women—such as smokers or women with hypertension— or according to type, duration, and past use of OCs.

Section snippets

Patients and methods

A detailed description of study methods has been published elsewhere.21, 22 In summary, this hospital-based, case-control study was undertaken in 21 centres in 17 countries subdivided into four regions. Each centre recruited cases and controls from a variable number of collaborating hospitals. Eligible cases were women aged 20·44 years (or 15·49 years in three centres), who had been admitted to a collaborating hospital between Feb 1, 1989, and Jan 31, 1993, and had, in the opinion of the

Results

703 (31·4%) of the 2242 strokes studied were ischaemic. Overall, 456 (20·3%) cases were unclassified, although this proportion was more than 80% in Africa because of inadequate diagnostic facilities (table 1). Four cases were excluded from subsequent analyses because their OC status was not known, and two were excluded because no matched controls had been recruited. 1952 controls were matched to the remaining 697 cases—an average of 2·6 controls per case in Europe and 2·8 per case in the

Discussion

Overall, current use of combined OCs was associated with a significantly increased risk of ischaemic stroke. However, risks were substantially lower among important subgroups of users. The overall odds ratios are similar to those reported from Denmark,13 but lower than those in most earlier studies.5, 8, 9, 12, 16 The only identifiable confounders were number of livebirths and a history of hypertension, in Europe, and a history of hypertension, rheumatic heart disease, or both in the developing

Age, blood pressure, and OC-associated risk

Adjusted odds ratios associated with OC use were greater in older than younger women in Europe and the developing countries. The effect of age on OC-associated risk of ischaemic stroke showed a continuous trend in the developing countries, but this trend was not so clear in Europe. An earlier study reported an effect of age on OC-associated risk for combined circulatory diseases15 but not specifically for ischaemic stroke, which may reflect the limited size of most previous studies of stroke.

Smoking and duration of OC use

Current smoking was not a substantial confounder (ie, did not change risk estimates by ≥5%) of the overall association between OC use and ischaemic stroke in Europe or the developing countries. However, among some subgroups, current smoking was found to increase odds ratios by more than 5%, and hence all risk estimates were adjusted for smoking. In addition, OC-associated odds ratios were higher among current smokers than non-smokers, and the patterns of risk associated with OC use and smoking

Risks by type of OC

In Europe, OC-associated risk of ischaemic stroke was largely confined to the use of higher-oestrogen-dose OCs (table 6). By contrast, in the developing countries, both low-dose and higher-dose preparations were associated overall with about a three-fold increase in odds ratios. However, this difference in risk was affected by women who reported that they had not had their blood pressure measured before they started to use OCs containing second-generation progestagens; among those women use of

Validity and limitations of study

The reference group used in most previous studies of cardiovacular risk associated with OC use was women who had never used OCs. However, in our study the preferred reference group was non-users because risk was almost exclusively associated with current OC use, and never-users, particularly in Europe, represented a small and possibly atypical subset of non-users.

A unique feature of this study was the high proportion of cases that had an accurate diagnosis of ischaemic stroke. Before the

Conclusion

Although this study shows that current OC use was associated with about a three-fold increase in risk of ishaemic stroke, the odds ratios were lower for younger women and those who did not smoke, and less than 2 for those who did not have a history of hypertension and had had their blood pressure checked before the current episode of OC use. Smoking appeared to have a synergistic effect on risk of ischaemic stroke associated with OC use. A history of hypertension among current OC users was

References (36)

  • Oral contraceptives and stroke in young women

    JAMA

    (1975)
  • H Jick et al.

    Oral contraceptives and non-fatal stroke in healthy young women

    Ann Intern Med

    (1978)
  • KL Mettinger et al.

    Stroke before 55 years of age at Karolinska Hospital 1973–77: a study of 388 well-defined cases

    Acta Neurol Scand

    (1984)
  • M Thorogood et al.

    Fatal stroke and use of oral contraceptives: findings from a case-control study

    Am J Epidemiol

    (1992)
  • O Lidegaard

    Oral contraception and risk of a cerbral thromboembolic attack: results of a case-control study

    BMJ

    (1993)
  • S Ramcharan et al.

    The Walnut Creek contraceptive study. Center for Population Research Monograph Number 81–564

    (1981)
  • Further analysis of mortality in oral contraceptive users

    Lancet

    (1981)
  • PM Layde et al.

    Incidence of arterial disease among oral contraceptive users

    J R Coll Gen Practitioners

    (1983)
  • Cited by (0)

    Writing Committee, study organisation, and participants given at end of article

    View full text