Elsevier

The Lancet

Volume 352, Issue 9145, 19–26 December 1998, Pages 1965-1969
The Lancet

Articles
Cardiovascular and cancer morbidity and mortality and sudden cardiac death in postmenopausal women on oestrogen replacement therapy (ERT)

https://doi.org/10.1016/S0140-6736(98)05066-1Get rights and content

Summary

Background

Advantages and disadvantages of postmenopausal oestrogen replacement therapy (ERT) are still not clear. We aimed to analyse the relation between postmenopausal oestrogen replacement therapy (ERT), cardiovascular disease, and cancer.

Methods

We examined 7944 women born between 1923 and 1930, who participated in a mammography screening for breast cancer, and who were followed up from 1987 to 1995. The follow-up consisted of 53 305 person-years. 988 women were current users and 757 were former users of ERT. Information about hormone use and health events was obtained through biennial questionnaires and recording and linking information from the hospital discharge registers of the region, the national cancer register, the social insurance reimbursement register, and the national death register. We used proportional-hazards models to calculate risk ratios and 95% Cls, adjusted for eight confounding variables.

Findings

Current ERT was associated with decreased cardiovascular mortality and a decrease in sudden cardiac death. Adjusted risk ratio (RR) for cardiovascular mortality in current users was 0·21 (95% Cl 0·08–0·59) and in former users 0·75 (0·41–1·37). Absolute risk per 1000 person-years for deaths from acute myocardial infarction (AMI) was 1·1 in never users, 1·2 in former users, and 0·45 in current users (p=0·197). Corresponding absolute risk for other coronary-artery-disease (CAD) deaths was 1·0, 0·81, and 0 (p=0·009), and for deaths from stroke 1·2, 1·0, and 0·15 (p=0·012). Absolute risk for sudden cardiac death was 1·6 in never users, 1·0 in former users, and 0 in current users (p<0·001). Cardiovascular morbidity was not decreased by ERT: the RR for current use was 1·07 (0·86–1·32) and for former use 1·11 (0·89–1·39). Incidence of cardiovascular disease per 1000 person-years was 24·9 in never users, 23·4 in former users, and 20·9 in current users (p=0·153). Breast-cancer morbidity did not increase with current ERT—the RR was 0·57 (0·27–1·20). Incidence of breast cancer was 1·8, 1·6, and 1·0 in never, former, and current users (p=0·242). Endometrial cancer increased with current ERT—the RR was 5·06 (2·47–10·41). Incidence of endometrial cancer was 0·52 in never users, 0·51 in former users, and 2·1 in current users (p<0·001).

Interpretation

Current ERT reduced primarily sudden cardiac death and predicted reduced cardiovascular mortality, but did not reduce morbidity. ERT did not increase the risk of breast cancer, but was associated with increased risk of endometrial cancer.

Introduction

Observational studies have shown a reduction in cardiovascular-disease risk and mortality in postmenopausal women on oestrogen replacement therapy (ERT), but to what extent this reduction results from selection bias is contested.1, 2, 3, 4, 5, 6 Weaknesses in the studies have been insufficient recording of cardiovascular disease, and lack of information on sudden cardiac death. The increased risk of endometrial cancer has been greatly decreased or almost eliminated by the addition of progestogen to ERT.7 However, opinions are still divided on the association between ERT and increased risk of breast cancer.8

The rapidly increasing use of ERT makes the question of its advantages and disadvantages highly pertinent. We report results from a community-based prospective cohort study. The study is based on questionnaires and repeated mammographies, and on the linking of existing registers in Finland. The registers contain information on individuals' health events during their lifespan.

Section snippets

Study population and design

During 1987–88, all women born between 1923 and 1930 living in Turku were asked to participate in a free mammography screening for breast cancer organised by the city health authorities. Of 8164 invited, 7944 women participated in the study. A validated questionnaire was filled in by participants with the help of trained nurses who confirmed and checked answers. The questionnaire contained detailed inquiries about occupation, height, weight, smoking habits, and former and present use of hormone

Results

At baseline there was a significant difference between never users and current users of ERT in age (p<0·001), social class (p<0·001), BMI (p<0·001), hypertension (p<0·001), and diabetes (p<0·003); but not in smoking, CAD, or cardiac failure (table 1). Among current users, only endometrial cancer increased (p<0·001; table 2). Total mortality decreased among current users (p<0·001), as did stroke mortality (p=0·012) and other CAD deaths other than acute-myocardial-infarction (AMI) deaths

Discussion

As Grodstein5 points out, women who take hormones are a self-selected group and may have healthier lifestyles with fewer risk factors than women who do not. The occurrence of risk factors at the starting point of this study was different between current, former, and never users of oestrogen, but there was no significant difference between the groups in relation to smoking, occurrence of CAD, or cardiac failure. In some studies, women with CAD were excluded, as they were for instance, in the

References (25)

  • L Teppo et al.

    Data quality and quality control of a population-based cancer registry: experience in Finland

    Acta Oncol

    (1994)
  • D Collett

    Modeling survival data in medical research

    (1994)
  • Cited by (164)

    • Adjuvant aromatase inhibitor therapy: Outcomes and safety

      2010, Cancer Treatment Reviews
      Citation Excerpt :

      In these patients, studied in the MA.17 trial, CVD in the LET group was not statistically different from that in the placebo group after 30 months of follow-up, and the incidence of MI was also similar (Table 2).12 The MI data are similar to what would be expected in an age-matched non-breast cancer population,70,71 and data from other databases, such as the UK General Practice Research Database or the Swedish MI register, also support this conclusion.72 Additional post-unblinding data from MA.17 have demonstrated a similar incidence of CVD in women electing to switch to LET after placebo and those continuing on placebo (4.2% vs 3.1%; p = 0.17); MI occurred in six and five patients in the respective groups.19

    View all citing articles on Scopus
    View full text