ORIGINAL INVESTIGATION Induced and Spontaneous Abortion and Incidence of Breast Cancer Among Young Women A Prospective Cohort Study Karin B. Michels, ScD, PhD; Fei Xue, MD; Graham A. Colditz, MD, DrPH; Walter C. Willett, MD, DrPH Background: Induced abortion has been inconsistently ported a history of spontaneous abortions. The hazard associated with breast cancer risk in case-control studies. ratio for breast cancer among women who had 1 or more Retrospective cohort studies using registry information in induced abortions was 1.01 (95% confidence interval, Scandinavia have not suggested an increase in the inci- 0.88-1.17) after adjustment for established breast can- dence of breast cancer, although data on individual re- cer risk factors; among women with 1 or more sponta- productive factors were not accounted for. neous abortions, the covariate-adjusted hazard ratio was 0.89 (95% confidence interval, 0.78-1.01). The relation Methods: We examined the association between in- between induced abortion and the incidence of breast can- duced and spontaneous abortion and the incidence of breast cer did not differ materially by number of abortions cancer in a prospective cohort of young women, the Nurses’ (P for trend=.98), age at abortion (P for trend=.68), par- Health Study II. The study included 105 716 women 29 to ity (P for interaction=.54), or timing of abortion with re- 46 years old at the start of follow-up in 1993. Information spect to a full-term pregnancy (P for interaction=.10). on induced or spontaneous abortions was collected in 1993 and updated biennially. During 973 437 person-years of fol- Conclusion: Among this predominantly premeno- low-up between 1993 and 2003, 1458 newly diagnosed cases of invasive breast cancer were ascertained. pausal population, neither induced nor spontaneous abor- tion was associated with the incidence of breast cancer. Results: A total of 16 118 participants (15%) reported a history of induced abortion, and 21 753 (21%) re- Arch Intern Med. 2007;167:814-820 FULL-TERM PREGNANCY BE- Given the personal and sensitive nature of fore the age of 35 years re- information on abortion, reporting bias may duces lifetime risk of breast be different for cases and controls, with case cancer,1 possibly by accel- patients being more likely to volunteer this erating breast-cell differen- information, leaving case-control studies Atiation. An incomplete pregnancy may not difficult to interpret.3 result in sufficient differentiation to counter In 2 retrospective cohort studies with the high levels of pregnancy hormones that prospectively recorded abortion data, reg- may foster proliferation.2 However, these istries available in Scandinavian countries biological mechanisms are uncertain, and were used, and registry data on induced a prematurely terminated pregnancy may abortion were linked with cancer registry not affect breast cancer risk at all. information. In a Swedish study by Linde- Author Affiliations: Obstetrics 5 and Gynecology Epidemiology Induced abortion has been associated fors-Harris et al, induced abortion was as- Center, Department of with an increased risk of breast cancer in sociated with a reduced risk of breast can- Obstetrics and Gynecology some retrospective case-control studies, cer. In a study4 that included 1.5 million (Drs Michels and Xue), and whereas no association has been found in Danish women, no association was found Channing Laboratory, studies with information on abortion re- between induced abortion and breast can- Department of Medicine corded prospectively.3,4 Most epidemio- cer incidence. Similarly, in 3 prospective (Drs Michels, Colditz, and logic studies3 of spontaneous abortion (mis- case-control studies that used registry in- Willett), Brigham and Women’s carriage) and breast cancer have indicated formation on induced abortion assessed be- Hospital, Harvard Medical no association. Most epidemiologic evi- fore breast cancer developed, no associa- School, and Departments of Epidemiology (Drs Michels, dence on the relation between induced tion or an inverse association between Xue, Colditz, and Willett) and abortion and breast cancer is from retro- induced abortion and the risk of breast can- 6-8 Nutrition (Dr Willett), Harvard spective case-control studies, the data on cer was observed. Registry studies, how- School of Public Health, abortion having been gathered from women ever, generally do not capture the entire re- Boston, Mass. with breast cancer and healthy controls. productive history of all women included (REPRINTED) ARCH INTERN MED/ VOL 167, APR 23, 2007 WWW.ARCHINTERNMED.COM 814 ©2007 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 10/02/2021 and are unable to control for several breast cancer risk fac- years, and 35 years or older. On each subsequent biennial ques- tors, such as age at menarche, menopausal status, body mass tionnaire, participants were asked whether they had been preg- index, and alcohol intake. In the 3 prospective cohort stud- nant in the previous 2-year period and, if they had, whether ies on postmenopausal women conducted to date, the Iowa pregnancies that lasted less than 6 months resulted in miscar- Women’s Study,9 the Shanghai Textile Workers Study,10 riages or induced abortions. Response options provided al- lowed participants to report miscarriages and induced abor- and the European Prospective Investigation Into Cancer 11 tions for each calendar year since the previous questionnaire and Nutrition, information on induced abortion was ob- was sent (questionnaires are mailed in June); response op- tained directly from the participants, and no association be- tions did not include “no.” tween induced abortion and breast cancer incidence was Of the 107 721 study participants who returned the 1993 found in these studies. In prospective cohort studies in questionnaire, 106 804 (99.1%) answered the question on spon- which abortion information is obtained from the women taneous and induced abortion; 16 359 (15.2%) of them re- directly, reproductive history is generally assessed retro- ported that they had experienced 1 or more induced abor- spectively at baseline among disease-free women, and par- tions, and 22 236 (20.6%) reported 1 or more spontaneous ticipants are followed up prospectively until they develop abortions. Among the 105 716 women included in the present the outcome of interest. This approach may underesti- analysis, 16 118 (15.2%) reported 1 or more induced abor- tions and 21 753 (20.6%) reported 1 or more spontaneous abor- mate abortions, and this nondifferential misclassification tions. Among all 116 671 women in the NHSII, 115 754 pro- may bias the effect estimate toward the null. Further- vided information on abortion or miscarriage at least once in more, in all 3 cohorts, information on abortion was as- follow-up questionnaires between 1993 and 2003. Among them, sessed only once, at baseline, and was not updated during 16 916 (14.6%) reported ever having had an induced abortion follow-up, which may add misclassification. and 25 837 (22.3%) reported ever having had a miscarriage. We examined the associations between induced and spontaneous abortion and the incidence of breast cancer RISK FACTORS FOR BREAST CANCER among the participants in the Nurses’ Health Study II (NHSII). In this cohort, information on induced and spon- Data on established or suspected risk factors for breast cancer taneous abortion was updated throughout follow-up. used for the present analyses were collected in 1989 and on biennial questionnaires through 2003. Date of birth, height, fam- ily history of breast cancer, age at menarche, and alcohol con- METHODS sumption were reported on the 1989 questionnaire. Informa- tion on weight, history of benign breast cancer, parity, age at STUDY POPULATION first birth, and menopausal status was reported in 1989 and up- dated with every biennial questionnaire. The NHSII is an ongoing prospective cohort study designed to examine associations between lifestyle factors, reproductive fac- ASCERTAINMENT OF BREAST CANCER tors, and the occurrence of breast cancer and other major ill- nesses. In 1989, a total of 116 671 female registered nurses 25 At each follow-up cycle, we ask whether breast cancer has been to 42 years of age who were free of cancer (except nonmela- diagnosed and, if so, the date of diagnosis. We routinely search noma skin cancer) and living in 1 of 14 US states responded to the National Death Index for women who did not respond to a baseline questionnaire. Follow-up questionnaires were mailed the questionnaires. We ask all women who reported breast can- biennially, updating information on reproductive factors, life- cer (or the next of kin for those who have died) for permission style factors, and newly diagnosed diseases. In 1993, informa- to review the relevant medical records and confirm the diag- tion on induced and spontaneous abortion was first assessed, nosis. Self-reported breast cancer was histologically con- which represented the baseline for the present analysis. Ex- firmed for 99% of women whose medical records could be ob- cluding women who had been diagnosed as having cancer tained. Because the accuracy of participants’ reports was (n=4065), who did not answer the question on abortion extremely high among those whose records were obtained, we (n=958), or who did not respond in 1993 (n=8955) left a total included self-reported breast cancer for those whose medical of 105 716 women to be included. Of these women, 92% were records were not available. We censored cases of carcinoma in non-Hispanic white. During follow-up, we censored women who situ (n=399) from the primary analyses, but results including were newly diagnosed as having breast cancer, had died, or were in situ cases were comparable to those for invasive cases only. lost to follow-up. Women who skipped questionnaires or did not respond to the questions on induced or spontaneous abor- STATISTICAL ANALYSIS tion in any given cycle were censored only for those periods.