In[. J . Cancer: 46, 366-373 (1990) 0 1990 Wiley-Liss, Inc.

Publication of the International Union Against Cancer Publication de I’Union lnternationale Contre 1e Cancer

ORAL CONTRACEPTIVES AND RISK OF BREAST CANCER Charlotte PAUL^, D.C.G. SKEWand G.F.S. SPEARS Department of Preventive and Social Medicine and the Hugh Adam Cancer Epidemiology Unit, University of Otago, Dunedin. New Zealand. A national population-based case-control study was conducted in New Zealand to assess the effects of hormonal contraception on breast-cancer risk. A total of 891 Women aged 25 to 54 with a first diagnosis of breast cancer, and 1864 control subjects, randomly selected from the electoral rolls, were interviewed. The relative risk of breast cancer for women who had ever used oral contraceptives was 1.0 (95% confidence interval 0.82-1.3). There was no increase in risk with duration of use, even among women who had continued to use oral contraceptives for 14 or more years (relative risk = 1.1, 95% confidence interval 0.78-1.7). The risk of breast cancer was not increased by use of oral contraceptives for long periods before the first pregnancy or by starting use at a young age. Parity, age at menarche, family history of breast cancer, or history of benign breast disease did not modify the effect of oral contraceptives on breast-cancer risk. Relative risk estimates were slightly, although not significantly, increased during the first few years after starting oral contraception and in women under 35 years of age at diagnosis.

The question of whether oral contraceptives increase the risk of breast cancer remains unresolved, despite the publication of at least 7 cohort studies and 25 case-control studies over 2 decades (Schlesselman, 1989; Thomas, 1989; Population Information Program, 1988). Although most studies have found no overall association, a disturbing number have shown increased risks in certain sub-groups of women who reported use of oral contraceptives. The sub-groups involved, however, have tended to vary from study to study, and it is difficult to detect any consistent pattern. One explanation suggested for some of the negative studies was that they involved populations who had not used oral contraceptives frequently enough at young ages, sufficiently long ago, for a latent adverse effect to emerge (McPherson and Drife, 1986). New Zealand is a particularly suitable place for study, because New Zealand women have been exceptionally high users of oral contraceptives since the 1960s. By 1966 it was estimated that 20% of women of reproductive age in New Zealand and Australia were using the pill, as compared with 8% in the USA and 3% in the UK (Guttmacher, 1966). National population surveys conducted in New Zealand and Great Britain in 1976 showed that a higher proportion of young New Zealanders (aged 15-24) were using oral contraception (Paul ef al., 1990). In 1983 we started a national case-control study of breast cancer in New Zealand women under 55 years of age. Following an interim report, based on the first 2 years of data collection (Paul et al., 1986), we now report the findings after completion of this 4-year study. SUBJECTS AND METHODS

In this population-based study, all New Zealand women aged 25-54 in whom breast cancer had been diagnosed were considered for inclusion as cases; control subjects were selected at random from the electoral rolls. The study population was confined to women whose names were in a current electoral roll and whose telephone number could be found. Cases The National Cancer Registry (or, in the Auckland region, the Auckland Breast Cancer Study Group) identified 1452

women aged 25-54 in whom a diagnosis of breast cancer (International Classification of Diseases, rubric 174) had been confirmed histologically between 1 July 1983 and 30 June 1987. Permission for the release of information was given by all superintendents of public hospitals and by all but 7 of the specialists responsible for private patients. Patients were excluded from the study if breast cancer had been diagnosed previously, if their names were not in a current electoral roll, or if their telephone number could not be found. After these criteria had been applied, 1126 women remained. We also required that patients be interviewed between 4 and 8 months after diagnosis, extended in the 3rd and 4th years to 12 months after diagnosis; this resulted in the exclusion of a further 104 patients, of whom 17 had died and 87 had been identified too late to be interviewed within 8 or 12 months after diagnosis. Table I shows other reasons why potentially eligible cases were not interviewed. Interviews were completed with 901 (95%) of the 944 suitable women whom we approached. After the interview, a further 10 women were excluded because they were found to have had breast cancer diagnosed previously or because the final diagnosis was not breast cancer (one case). Thus we finally studied 891 cases. Controls In New Zealand, electoral registration is compulsory for adults aged 18 and over. Control subjects were randomly selected from the electoral rolls, and women whose telephone numbers could not be found were excluded. As age is not specified in the electoral rolls, we had to choose more women than were required for the study. We wrote to these women requesting information about their age and seeking their participation. If they did not reply, we telephoned them. All controls were written to between 1 November 1983 and 31 October 1987. Of 4,725 women to whom we sent letters, all except 65 provided information regarding eligibility. A total of 1,898 women were found to be outside the age range 25 to 55. We randomly excluded half the potential controls under the age of 35 (560 women) to approximate more closely the age distribution of the cases. The outcome for the remaining 2,267 potential controls is shown in Table I. Interviews were completed with 1,915 (90%) of 2,134 suitable subjects. For each control a “reference date” was calculated by subtracting six months from the date of interview, to correspond to the mean time between diagnosis and interview among the cases. After the interview, 51 women were excluded because of a history of breast cancer or because they were younger than 25 or older than 54 on the reference date. The final control group thus comprised 1,864 women. Data collection Information was collected using identical methods for cases and controls. Women were interviewed by telephone after an

~

‘To whom correspondence and reprint requests should be sent, at Department of Preventive and Social Medicine, University of Otago Medical School, Dunedin, New Zealand.

Received: March 3, 1990 and in revised form May 11, 1990.

367

ORAL CONTRACEPTIVES AND BREAST-CANCER RISK

TABLE I - NUMBERS OF CASE AND CONTROL SUBJECTS INTERVIEWED Potential case and control subjects

Cases

Interviewed: Interviewed and eligible Excluded after interview because ineligible Reason not interviewed Died Identifield too late Previously interviewed as a control Doctor refused Subject refused Too ill

Not traoed Absent overseas Languag,edifficulties Total potentially eligible subjects

Controls

89 1 10

1,864 51

17 87 6 51 43 8

9 219 8 71’

4 5 4 1,126

23 22 2,261

’About half of these untraced women would have been outside the age range for the study.

initial approach by letter. The interview took about 20 min and sought details of medical, social, menstrual, and reproductive histories. Both the letter and interview were designed so that they did not disclose the study hypotheses. All interviews were conducted by 2 nurse-interviewers (94%)or by one of us (CP). Most began with the interviewer being blind as to whether the subject was a case or a control; although case patients generally reported their illness during the interview, this usually occurred after the contraceptive history had been taken. To help recall of past contraceptive use, our initial letter was accompanied by a calendar, 011 which women were asked to mark important dates such as marriages and births of children for reference during the interview. As a check on the interview method, we also wrote to the general practitioners of women who reported recent use of prescribed contraceptives. Replies were received from 363 (95%) out of 382 physicians. There was close agreement between the contraceptive histories received from the general practitioners and from the women themselves. Thus there was agreement as to whether oral contraceptives had been used in the last 5 years for 92% of cases and 91% of controls. Among women who had used oral contraceptives in the last 5 years, there was agreement on the duration of such use (within 2 years) for 100% of cases and 96% of controls. Definition t$ terms Oral contraceptive use for 1 month or longer was categorized as ever-use. A full-term pregnancy was a pregnancy lasting 7 months or more. Women were classified as post-menopausal if they had not menstruated in the 12 months before the date of interview. Menopause was further classified as artificial, if menstruation ceased because of surgery or radiation, or else as natural. A ffew women who had not menstruated in the previous 12 months were classified as pre-menopausal because the reason for the cessation of their periods was pregnancy and lactation or prolonged progestogen use. A family history of breast cancer was classified as first degree if the subject’s mother, sister, or daughter had breast cancer; as second degree if an aunt or grandmother had breast cancer; as unknown if the history in lirst-degree relatives was not known and no seconddegree relatives were known to have breast cancer; otherwise, as no family history.

Statistical analyses For cases, contraceptive use before the date of diagnosis was considered in the analyses. For controls, the reference date was used. We iused standard statistical methods for the analysis of

case-control studies. Relative risks were estimated by calculating odds ratios, adjusted for age (in 5-year groups) by the method of Mantel and Haenszel (1959). Confidence intervals were estimated by the procedure of Cornfield (1956) as programmed by Thomas (1975). When simultaneously adjusting for several risk factors, we applied the multiple logistic model (Breslow and Day, 1980) using the computer program GLIM (Baker and Nelder, 1978). Age adjustment was again in 5 year groups, except for the analysis reported in Table 111, in which age was adjusted for as a continuous variable within each 10 year age group. For tests of trend, the actual duration of exposure (rather than the category value) was used as the independent variable. RESULTS

Characteristics of case and control subjects are summarized in Table 11. The average age was higher among the cases, so the table gives age-adjusted relative risks for all other variables. The relative risk of breast cancer was higher among women with an early menarche, with no full-term pregnancies, with low parity, who did not breast-feed, with a family history of breast cancer, or with a history of benign breast disease. Maori women were also at higher risk of breast cancer. Other potentially confounding variables examined included menopausal status, body-mass index, area of residence, social class, education, cigarette smoking, alcohol consumption, year of interview, and use of depot medroxyprogesterone acetate (DMPA, “Depo-Provera”). The proportions of cases and controls who had used oral contraceptives at some time were 76.9% (685/891) and 82.5% (1 538/1864) respectively. Three control subjects were unable to recall their duration of oral contraceptive use, or had missing information for other confounding variables, so further analyses were restricted to the remaining 1,861 control women. The relative risk for women who had ever used oral contraceptives, adjusted for potential confounding variables by logistic regression, was 1.0 (95% confidence interval 0.82-1.3). Table 111 shows the risk of breast cancer according to age at diagnosis and duration of oral contraceptive use, relative to women in each age group who had never used oral contraceptives. There was no overall increase in risk with increasing duration of use, even for use for 14 years or longer. Nor did risk increase with duration of use among women aged 35 or over at diagnosis. In young women, aged 25 to 34, oral-contraceptive users had a relative risk of 1.2 (95% confidence interval 0.44-3.4), compared with women who had never used oral contraceptives. Women in this age group who had used oral contraceptives for 10 to 13 years and 14 or more years had relative risks of 2.3 and 1.7 respectively, although the 95% confidence intervals were wide and included 1.O in each case. There was no statistically significant trend in the relative risk of breast cancer according to duration of use (p = 0.42). The relative risk associated with duration of oral contraception was examined by age at first use, time since first use, and time since last use of oral contraceptives. Table IV shows the relative risks according to age at first use of oral contraceptives. Use starting very young, at ages 17 to 19 or even before age 17, was not associated with any consistent increase in risk. Similarly, use of oral contraceptives beginning a long time in the past (Table V) was not associated with an increased risk, even after long durations of use. The relative risk of breast cancer in women who had used oral contraceptives for 10 or more years starting 15 or more years earlier was 1.0 (95% confidence interval 0.74-1.3). The risk of breast cancer was slightly, although not significantly, elevated among women who had first used oral contraceptives more recently. The rel-

368

PAUL ET AL.

TABLE I1 - CHARACTERISTICS OF 891 CASES AND Characteristic

25-29 30-34 35-39 4 w 45-49 50-54 Ethnic group Non-Maori Maori Age at menarche

Oral contraceptives and risk of breast cancer.

A national population-based case-control study was conducted in New Zealand to assess the effects of hormonal contraception on breast-cancer risk. A t...
813KB Sizes 0 Downloads 0 Views