Correspondence
Exercise and Breast Cancer
N Engl J Med 1997; 337:708-709September 4, 1997
- Article
To the Editor:
We wish to underscore Dr. McTiernan's cautionary comments in her editorial on the potential bias in the study of physical activity and the risk of breast cancer by Thune et al. (May 1 issue),1,2 since we find serious flaws in the analysis presented.
Our main objection is that the women studied were not classified according to total amount of physical activity, both during leisure time and at work. Therefore, no dose–response relation can be inferred between a woman's total physical activity and her risk of cancer. It is possible, for example, that some of the 66 women with breast cancer who were sedentary during their leisure time were physically active at work and that some of the 62 women with breast cancer who were sedentary at work were physically active during their leisure time. Indeed, this must be so, since the authors note that there were so few women with cancer who were sedentary both during their leisure time and at work that “the effect of this combination could not be analyzed.” The number of women in the study who were totally sedentary is not stated.
The basic information is there, in data carefully collected on more than 25,000 women over an average of 14 years. Perhaps a second look could make this enormous effort more fruitful.
2 ReferencesAgnes Berger, Ph.D.
Carol A. Bodian, Dr.P.H.
Mount Sinai Medical Center, New York, NY 100291
McTiernan A . Exercise and breast cancer -- time to get moving? N Engl J Med 1997;336:1311-1312
Full Text | Web of Science | Medline2
Thune I ,Brenn T ,Lund E ,Gaard M . Physical activity and the risk of breast cancer. N Engl J Med 1997;336:1269-1275
Full Text | Web of Science | Medline
To the Editor:
Thune et al. performed an analysis of covariance of their data but do not mention adjustments for daily smoking. Their Table 1 lists the percentage of daily smokers among the women studied according to their level of activity, and ostensibly shows a very strong correlation between the activity level and the proportion of smokers. Was this correlation adjusted for in the analysis and simply not mentioned? If not, could the lower rates of cancer be attributable to the lower rates of smoking among women who exercised?
Jonathan A. Jackson, Ph.D.
Georgetown University, Washington, DC 20007To the Editor:
In the article by Thune et al. there may be a potentially important form of ascertainment bias that could distort the observed association between exercise and the rates of breast cancer.
Many of the cancers were undoubtedly diagnosed by screening mammography. If the intensity of screening differed systematically among the exercise groups, the observed association between exercise and breast cancer might be biased toward the group undergoing the more intense surveillance. For example, sedentary women were most likely to smoke. Bronchitis or other smoking-related illnesses may result in more contact with health care providers and therefore more opportunity to be referred for mammography. Thus, enhanced scrutiny rather than a reduced level of exercise could account for the increased incidence of breast cancer.
Alternatively, it is possible that for women with a higher level of activity, preventive health care such as regular mammography may be part of a healthy lifestyle. In this scenario, the observed results would underestimate the magnitude of the “protective” relation between exercise and breast cancer. Knowing either the rates of screening mammography or the distribution of cancers according to stage in the exercise groups would help address concern about biased ascertainment as an explanation for the findings.
Steven Woloshin, M.D.
Lisa Schwartz, M.D.
Veterans Affairs Medical Center, White River Junction, VT 05009To the Editor:
Thune et al. report a 37 percent reduction in the risk of breast cancer among women who exercise regularly as compared with those who are sedentary. Although it is important for physicians to be aware of population-based evidence, it may be equally important that patients be given an individual perspective. For example, the large relative reduction of 37 percent translates into a much smaller absolute reduction of only about 0.5 percent (from 1.5 to 1.0 percent); this absolute reduction can only be estimated, since the data in the article are adjusted for a number of variables.
From another perspective, regular exercise increases the chance of not getting breast cancer from 98.5 percent to 99 percent. It is premature to conclude how regular exercise reduces the risk of breast cancer.1 Caution should be used in presenting large relative reductions in risk without the accompanying small reductions in absolute risk.
1 ReferencesKim Goldenberg, M.D.
Wright State University School of Medicine, Dayton, OH 45401-09271
McTiernan A . Exercise and breast cancer -- time to get moving? N Engl J Med 1997;336:1311-1312
Full Text | Web of Science | Medline
Author/Editor Response
The authors reply:
To the Editor: We appreciate the question Berger and Bodian ask about the relation between overall physical activity and breast cancer. In an early version of our manuscript, work and leisure-time activity were combined and analyses stratified according to the level of physical activity at work, but they were not presented in the published article. Among women who reported that their work involved walking and who also exercised regularly during their leisure time, the risk of breast cancer was reduced by 52 percent (relative risk, 0.48; 95 percent confidence interval, 0.27 to 0.86; P for trend = 0.01) (Table 1Table 1
Adjusted Relative Risk of Breast Cancer According to the Combined Level of Physical Activity at Work and during Leisure Time among Women Whose Work Involved Walking, in the 1977–1983 Survey.). These findings underscore the importance of the combined effect of physical activity at work and during leisure time on the risk of breast cancer.Jackson questions whether we adjusted our analysis for daily smoking, because the women who exercised smoked less than the sedentary women. Univariate, age-adjusted analyses revealed no association between smoking and the risk of breast cancer (relative risk, 0.94; 95 percent confidence interval, 0.73 to 1.20). Furthermore, adjustments for smoking habits did not affect the estimates of risk in our multivariate models, and we therefore excluded this factor from the final model. In most studies, smoking has not been observed to influence the risk of breast cancer.1
Woloshin and Schwartz ask whether screening mammography entails a potentially important form of ascertainment bias. During the follow-up period, no organized screening mammography took place in Norway. The recording of all new cases of breast cancer in Norway was based on compulsory reporting by hospital departments and pathology laboratories and on the submission of death certificates. This method ensures a very high rate of case ascertainment. In addition, the distribution of the stages of cancer among the exercise groups showed that the early stages were overrepresented in women who exercised regularly; stage I breast cancer was observed more often in women who exercised regularly than in sedentary women (30 percent vs. 18 percent). This should mean that the observed results would underestimate the magnitude of the protective relation between exercise and breast cancer.
Associations between exposure and disease are important in elucidating the pathogenesis of the disease, which was the aim of our study. In searching for possible risk factors for a disease, calculations of absolute risks are premature. Our knowledge is based on the average patient in a cohort, not the unique patient at a consultation who is asking for advice. Despite this concern,2 the time has come to investigate further the relation between physical activity and the risk of breast cancer. This need is based on the observation that physical activity may influence both weight gain and the cumulative exposure to ovarian hormones — factors that affect carcinogenesis of the breast.3,4 Thus, emerging knowledge points to physical activity as a modifiable factor affecting the risk of breast cancer that is of great importance for public health and for women in general, but there is a need for further study.
4 ReferencesInger Thune, M.D., Ph.D.
Eiliv Lund, M.D., Ph.D.
University of Tromsø, N-9037 Tromsø, Norway1
Vatten LJ ,Kvinnsland S . Cigarette smoking and risk of breast cancer: a prospective study of 24,329 Norwegian women. Eur J Cancer 1990;26:830-833
CrossRef | Web of Science | Medline2
McTiernan A . Exercise and breast cancer -- time to get moving? N Engl J Med 1997;336:1311-1312
Full Text | Web of Science | Medline3
Bernstein L ,Ross RK . Hormones and breast cancer risk. Epidemiol Rev 1993;15:48-65
Web of Science | Medline4
Ziegler RG ,Hoover RN ,Nomura AMY , et al. Relative weight, weight change, height, and breast cancer risk in Asian-American women. J Natl Cancer Inst 1996;88:650-660
CrossRef | Web of Science | Medline
- Citing Articles (2)
Citing Articles
1
Zsolt Radak, Dezso Gaal, Albert W. Taylor, Takao Kaneko, Shunichi Tahara, Hideko Nakamoto, Sataro Goto. (2002) Attenuation of the Development of Murine Solid Leukemia Tumor by Physical Exercise. Antioxidants & Redox Signaling 4:1, 213-219
CrossRef2
Christine M. Friedenreich, Inger Thune, Louise A. Brinton, Demetrius Albanes. (1998) Epidemiologic issues related to the association between physical activity and breast cancer. Cancer 83:S3, 600-610
CrossRef
