Rapid Responses to:

PAPERS: Naja Rod Nielsen, Zuo-Feng Zhang, Tage S Kristensen, Bo Netterstrøm, Peter Schnohr, and Morten Grønbæk Self reported stress and risk of breast cancer: prospective cohort study BMJ 2005; 331: 548 [Abstract] [Full text]

Rapid Responses: Submit a response to this article

Rapid Responses published:

Being alive is the condition for diagnosis

Wenbin Liang   (16 August 2005)

High stress and breast cancer

Ellen C G Grant   (12 September 2005)

The Question of Defining Stress

Archita Gulati   (12 September 2005)

Stress and Risk of Breast Cancer: Confounding, Interactions, and Other Methodologic Issues in Analysis

Eric L. Ding   (12 September 2005)

Understanding behaviour: what is more appropriate? Quantitative vs qualitative research

Tolulola Taiwo   (14 September 2005)

Exeperience suggests the opposite

Sidha Sambandan   (16 September 2005)

Stress and Breast Cancer: Another Smoke Screen?

Lynn Howard Ehrle   (21 September 2005)

Self-perceived stress and breast cancer risk

Kefah Mokbel, Mohamed Salhab   (21 September 2005)

Does stress protect against breast cancer? Authors’ response

Naja Rod Nielsen, Zuo-Feng Zhang, Tage S. Kristensen, Bo Netterstrøm, Morten Grønbæk   (23 September 2005)

Re: High stress and breast cancer

Ellen C G Grant   (27 September 2005)

Nielsen et el. do not evaluate stress and breast cancer, but rather, socioeconomic status and breast cancer

Candyce H. Kroenke, Ichiro Kawachi, Laura D. Kubzansky   (7 December 2005)

Being alive is the condition for diagnosis 16 August 2005
Wenbin Liang, taking master of public health Curtin University of Technology Send response to journal: Re: Being alive is the condition for diagnosis	Dear Editor, In the article “ self reported stress and risk of breast cancer: prospective cohort study”, it appears that stress may reduce the risk of breast cancer.[1] However it is possible that high level of stress “kills” quite a few number of women who would have become a case by other disease such as MI during the tumor induction period or even latent period, and further reduces the number of detected cases among the “stress group”. Moreover, As it is showed in the article at base line women who have higher stress level seems to have higher level of alcohol consumption, to be more likely to be physical inactive. While the mean age of the study population is 57, at which people are likely at high risk of many kinds of diseases, furthermore 2224 deaths were observed and only 251 breast cancers were diagnosed and only 19 cases are among high stress group.[1] The result could easily be biased due to lost of “cases”(case in induction or latent period) among the deaths. It would be interesting to see the difference of stress level among the deaths and the life span among different “stress group” in the cohort. 1. Nielsen, N. at al 2005 "Self reported stress and risk of breast cancer: prospective cohort study" BMJ 2005; 0: bmj.38547.638183.06v1 Competing interests: None declared
High stress and breast cancer 12 September 2005
Ellen C G Grant, physician and medical gynaecologist Kingston-upon-Thames, KT2 7JU, UK Send response to journal: Re: High stress and breast cancer	The conclusion of the epidemiological study by Neilson and colleagues that high stress reduces the risk of breast cancer defies common sense.1 The study is seriously flawed by confounding and interactions.2,3 A higher percentage of women in the high stress group died (39.3%) during follow-up and more had received hormone therapy and had high alcohol intake than women in the medium or low stress groups. Breast cancer incidence relates strongly to current and longer use of progestogens but these factors were not specified separately by the authors. Amenorrhoea was three times more common in ex-takers of oral contraceptives than in controls in a large UK study.4 Is there evidence that oestrogen levels are lower in current takers of HRT if they have high stress? Most women with breast cancer believe that high stress has contributed to the development of their disease. The incidence if breast cancer has increased dramatically since women were expected to go out to work but also house keep and bring up children, often as single parents. 1 Nielsen NR, Zhang Z-F, Kristensen TS, et al. Self reported stress and risk of breast cancer: prospective cohort study.BMJ 2005; 331: 548 2 Liang W. Being alive is the condition for diagnosis. http://bmj.com/cgi/eletters/331/7516/548#114779, 16 Aug 2005 3 Ding E L. Stress and Risk of Breast Cancer: Confounding, Interactions, and Other Methodologic Issues in Analysis. http://bmj.com/cgi/eletters/331/7516/548#116156, 9 Sep 2005 [Editorial note--corrected and reposted as http://bmj.bmjjournals.com/cgi/eletters/331/7516/548?ck=nck#116390, 12 Sep 2005] 4 Royal College of General Practitioners. Oral Contraceptives and Health. London: Pitman Medical, 1974 ,pp 73-74. Competing interests: None declared
The Question of Defining Stress 12 September 2005
Archita Gulati, Senior SHO General Surgery Royal Oldham Hospital,Rochdale Road, Oldham, Ol1 2JH. Send response to journal: Re: The Question of Defining Stress	In their study assessing the relationship between stress and the incidence of primary breast cancer, Nielson et al1 use self reported ‘perceived daily stress’ as their main variable.Surely, a self reported measure of intensity and frequency of stress would vary over a period of time. Stress level at the time of initial participation whilst completing the questionnaire would be greatly influenced by personal factors and life events existing around the participant at that time. For example, a major life event in the proximity of administration of the questionnaire could have influenced the initial response adversely and contributed to recording of high stress levels. We agree with the authors that ‘how stress is defined is and measured remains a point of debate’(1) . Would it therefore be fair to base the conclusions of decreased risk of breast cancer on a point measurement of a variable which would be likely to change widely over a 16 to 18 year period ? Reference: 1. Nielsen NR,ZhangZF, Kristensen TS, Netterstrom B, Schnohr P, Gronbaek M.BMJ 2005; 331:548-550.Self reported stress and risk of breast cancer: prospective cohort study. Competing interests: None declared
Stress and Risk of Breast Cancer: Confounding, Interactions, and Other Methodologic Issues in Analysis 12 September 2005
Eric L. Ding, Doctoral candidate Department of Epidemiology, Harvard School of Public Health, Boston, Massachusetts, 02115 USA Send response to journal: Re: Stress and Risk of Breast Cancer: Confounding, Interactions, and Other Methodologic Issues in Analysis	Stress and Risk of Breast Cancer: Confounding, Interactions, and Other Methodologic Issues in Analysis To the Editor, The study by Nielsen et al.[1] observed that women with higher level of stress have lower risk of breast cancer. While their prospective study is well designed, this study has several limitations in its analysis. First, it is surprising that the authors did not adjust for smoking nor justify their reason not to account for smoking. Long term smoking and exposure to passive smoking may be related to increased risk of breast cancer,[2] while smoking may also suppress estrogen,[3] an established risk factor for breast cancer.[4] However, stress is also known to be associated with smoking,[5] a relationship which may be bidirectional. Thus, smoking is plausibly both an intermediate and a confounder, in which case inverse-probability-weighting [6] is likely necessary to obtain unbiased causal relative risk estimates. Additionally, a previous review of the relationship between stress and disease suggests confounding, particularly by smoking, is likely responsible for reported associations.[7] Secondly, as the authors recognized, the study also did not adjust for age of menarche, age of first birth, and family history of breast cancer. Additionally, for breast cancer risk, there are also known interactions between parity and age at first birth,[8] smoking and hormone use,[9] and BMI and menopausal status,[10] accounted for in previous analyses but not accounted for in this study. Though the authors argue not all these variables may be true confounders— not accounting for the important interaction of such unadjusted variables with established confounders may still lead to residual confounding. Finally, there is also possibly inadequate adjustment for dietary variables, particularly alcohol. Though the study adjusted for alcohol consumption, alcohol was modeled using broad categories 0, 1-14, and >14 drinks/week. However, a previous pooled analysis of 6 large international cohorts of breast cancer incidence found a strong linear dose-response relationship between alcohol and breast cancer,[11] suggesting adjustment using 3 broad categories of alcohol intake may lead to residual confounding. Additionally, multiple studies also report an important effect modification of the association between alcohol and breast cancer by folate intake[12-14], again suggesting not accounting for such interaction may also lead to inadequate control for confounding by alcohol. Therefore, in addition to previous analyses suggesting confounding problems in studies of psychosocial stress,[7] the inadequate control of residual confounding and the possible dual intermediate-and-confounder role of smoking may have led to biased estimates in this study of stress and risk of breast cancer. However, further careful investigations of the role of stress is still highly encouraged due to importance of identifying modifiable risk factors to prevent breast cancer. Eric L. Ding Doctoral Candidate Department of Epidemiology Harvard School of Public Health Boston, Massachusetts 02115 United States eding@hsph.harvard.edu 1. Nielsen NR, Zhang Z-F, Kristensen TS, Netterstrom B, Schnohr P, Gronbaek M. Self reported stress and risk of breast cancer: prospective cohort study. BMJ 2005;331(7516):548-. 2. Terry PD, Rohan TE. Cigarette Smoking and the Risk of Breast Cancer in Women: A Review of the Literature. Cancer Epidemiol Biomarkers Prev 2002;11(10):953-971. 3. Baron JA, La Vecchia C, Levi F. The antiestrogenic effect of cigarette smoking in women. Am J Obstet Gynecol 1990;162(2):502-14. 4. Clemons M, Goss P. Estrogen and the Risk of Breast Cancer. N Engl J Med 2001;344(4):276-285. 5. Heslop P, Smith GD, Carroll D, Macleod J, Hyland F, Hart C. Perceived stress and coronary heart disease risk factors: the contribution of socio-economic position. Br J Health Psychol 2001;6(Pt 2):167-78. 6. Robins JM, Hernan MA, Brumback B. Marginal structural models and causal inference in epidemiology. Epidemiology 2000;11(5):550-60. 7. Macleod J, Davey Smith G, Heslop P, Metcalfe C, Carroll D, Hart C. Are the effects of psychosocial exposures attributable to confounding? Evidence from a prospective observational study on psychological stress and mortality. J Epidemiol Community Health 2001;55(12):878-884. 8. Tworoger SS, Missmer SA, Barbieri RL, Willett WC, Colditz GA, Hankinson SE. Plasma Sex Hormone Concentrations and Subsequent Risk of Breast Cancer Among Women Using Postmenopausal Hormones. J Natl Cancer Inst 2005;97(8):595-602. 9. Mueck AO, Seeger H. Smoking, estradiol metabolism and hormone replacement therapy. Arzneimittelforschung 2003;53(1):1-11. 10. van den Brandt PA, Spiegelman D, Yaun SS, Adami HO, Beeson L, Folsom AR, et al. Pooled analysis of prospective cohort studies on height, weight, and breast cancer risk. Am J Epidemiol 2000;152(6):514-27. 11. Smith-Warner SA, Spiegelman D, Yaun SS, van den Brandt PA, Folsom AR, Goldbohm RA, et al. Alcohol and breast cancer in women: a pooled analysis of cohort studies. JAMA 1998;279(7):535-40. 12. Baglietto L, English DR, Gertig DM, Hopper JL, Giles GG. Does dietary folate intake modify effect of alcohol consumption on breast cancer risk? Prospective cohort study. Bmj 2005. 13. Sellers TA, Kushi LH, Cerhan JR, Vierkant RA, Gapstur SM, Vachon CM, et al. Dietary folate intake, alcohol, and risk of breast cancer in a prospective study of postmenopausal women. Epidemiology 2001;12(4):420-8. 14. Zhang S, Hunter DJ, Hankinson SE, Giovannucci EL, Rosner BA, Colditz GA, et al. A prospective study of folate intake and the risk of breast cancer. JAMA 1999;281(17):1632-7. Competing interests: None declared
Understanding behaviour: what is more appropriate? Quantitative vs qualitative research 14 September 2005
Tolulola Taiwo, Medical resident University of Alberta Send response to journal: Re: Understanding behaviour: what is more appropriate? Quantitative vs qualitative research	Dear Editor, The article by Nielsen et al showed an inverse relationship between stress and breast cancer. However, the findings also raise an important question in the appropriateness of quantitative research in studying human behavior and characteristics. There are inherent difficulties in defining stress - one is considered as stressful by one person may not be perceived as such by others. Other potentially confounding factors and interacting variables, such as poverty, unemployment that have been shown to significantly increase stress levels were not accounted for in this study. Secondly, the challenge lies in generalising these findings to other populations due to socioeconomic and cultural variations. Tolulola Taiwo M.B.,B.S, M.P.H Competing interests: None declared
Exeperience suggests the opposite 16 September 2005
Sidha Sambandan, General Practitioner Yare Valley Medical Practice, 202 Thorpe Road, Norwich NR1 1TJ Send response to journal: Re: Exeperience suggests the opposite	Over 15 years of analysing and trying to understand the personality and psyche of patients who are diagnosed with Breast cancer, I have observed that the majority have a tendency to suppress their emotions. They tend to be introverted and more importantly the majority have had some major life event in the previous two years. The first documentation of this observation was made by Sir James Paget in his Textbook around 1875. Competing interests: None declared
Stress and Breast Cancer: Another Smoke Screen? 21 September 2005
Lynn Howard Ehrle, Senior Biomedical Policy Analyst, Organic Consumers Assoc. 8888 Mayflower Dr., Plymouth, MI 48170 Send response to journal: Re: Stress and Breast Cancer: Another Smoke Screen?	Nielsen et al (10 Sept) cite stress as an important factor in breast cancer and they believe it[breast cancer] is "a hormone dependent disease with a clear positive relation to high endogenous concentrations of oestrogen." The emphasis upon stress, diet, exercise, yoga, family history, etc., tend to cloud the issue (intentionally?). It has long been known that breast tissue is highly effected by ionizing radiation. This fact was demonstrated by Ian MacKenzie in 1965 when he found that women who had fluoroscope examinations for tuberculosis had a 24-fold greater risk of breast cancer as compared to others who were not irradiated(1). This study was confirmed in 1969 by Myrden and Hiltz(2). Then, in 1970, a paper by Tamplin and Gofman(1963-associate director, Lawrence Livermore Natl. Lab and director of its Biomedical Research Div) quantified the dose -response for radiation-induced breast cancer(3). In 1971, Segaloff and Maxfield, in a radiation/oestrogen experiment, concluded, "radiation alone could be the causative agent in the rising incidence of breast carcinoma"(4). This was the first study to document a synergistic response but it was ignored by cancer "experts" who were playing ball with the embryonic nuclear power industry. Despite the best efforts by independent investigators (Nobel laureate H.J. Muller,Ernest Sternglass, John Gofman,Rosalie Bertell,Alice Stewart,Brian MacMahon,Baruch Modan,Rudi Nussbaum,Chris Busby,Inge Schmitz -Feuerhake, Alexey Yablokov, to name a few) to focus on radiation health effects, the radiation connection was too hot to handle. How could the National Cancer Institute(NCI) distribute a booklet to its cancer centers in 2003 (What You Need to Know About Breast Cancer)proclaiming "No one knows the exact causes of breast cancer"? How could the NCI and cancer specialists around the world ignore the work of these respected scientists or later papers published in major medical journals that disproved this claim? A 1996 paper by Mary S. Wolff(Mt. Sinai School of Medicine)that stated, "In women, strong links have been established between breast cancer risk and ionizing radiation"(5). Or the commentary by five cancer experts on risk factors-- "With the notable exceptions of ionizing radiation and inherited genetic damage, none of the established risk factors for breast cancer directly causes the disease"(6). John W. Gofman published an extensively referenced book(340 references)in 1996 that identified medical X rays as the primary cause of the skyrocketing incidence of breast cancer(7), representing the best historical account of the disease. Incredibly, on 27 July 1996, NCI director Richard Klausner, speaking at Congresswoman Nancy Pelosi's town hall meeting, noted, "As far as I am aware, all of the data we have to date on low-level ionizing radiation does not demonstrate an increased risk"(transcript sent to this writer). He must have known of the preliminary results of the NCI-funded U.S. Scoliosis Cohort Study. Published in the low circulation journal Spine(8),it found an average of 24.7 X rays and a mean cumulative dose of 10.8 cGy(a very low dose) resulted in a 69% increase in breast cancer mortality, a jarring revelation to the radiation community. In 2004, the Breast Cancer Fund published a booklet on the primary causes of cancer, i.e., radiation and its synergistic co-action with chemicals. Six prominent scientists reviewed the document, the first one to reference the Segaloff-Maxfield study and another later paper demonstrating synergism by Calef and Hei(Establishment of a radiation and estrogen-induced breast cancer model. Carcinogenesis 2000;21:769-776). The Fund's conclusion: "Exposure to ionizing radiation is the best-established environmental cause of human breast cancer"(9)and it urged an investigation into the synergistic interaction of radiation and chemicals, a necessary component in cancer prevention, risk analysis, and The Precautionary Principle. References 1. MacKenzie I. Breast cancer following multiple fluoroscopies. Brit J Cancer 1965;19:1-8. 2. Myrden JA, Hiltz JE. Breast cancer following multiple fluoroscopies during artificial pneumothorax treatment of pulmonary tuberculosis. Canadian Med Assoc J 1969;100:1032-1064. 3. Tamplin AR, Gofman JW. Radiation-induced breast cancer. Lancet 1970;1:297. 4. Segaloff A, Maxfield WS. The synergism between radiation and oestrogen in the production of mammary cancer in the rat. Cancer Research 1971;31:166-168. 5. Wolff Ms, Collman GW, Barrett JC, Huff J. Brreast cancer and environmental risk: epidemiological and experimental findings. Annu Rev Pharmacol Toxicol 1996;36:573-596. 6. Davis DL, Axelrod D, Bailey L, Gaynor M, Sasco AJ. Enviro Health Perspect 1998;9:523-529. 7. Gofman JW. Preventing Breast Cancer: The Story of a Major,Proven, Preventable Cause of this Disease(2nd ed). San Francisco:Committee for Nuclear Responsibility,1996. 8. Doody MM,Lonstein JE, Stovall M, Hacker DG, Luckyanov N, Land CE. Breast cancer mortality after diagnostic radiography:findings from the U.S.Scoliosis Cohort Study. Spine 2000;25:2052-2063. 9. Evans N(ed). State of the Evidence:What Is the Connection Between the Environment and Breast Cancer? San Francisco: Breast Cancer Fund/Breast Cancer Action, 2004 Competing interests: None declared
Self-perceived stress and breast cancer risk 21 September 2005
Kefah Mokbel, Professor at Brunel Institute of Cancer Genetics,Consultant Breast & Endocrine Surgeon. The Princess Grace Hospital, 42-52 Nottingham Place, London W1M 3FD, Mohamed Salhab Send response to journal: Re: Self-perceived stress and breast cancer risk	Dear Sirs The finding by Nielsen et al (1) that women with high levels of self- perceived stress had a lower risk of developing breast cancer, than those with low levels is very interesting. However the authors’ conclusion lacks support from their own observations and from other experimental studies. The authors suggested that the risk reduction observed was most likely caused by decreased endogenous oestrogen production, due to the effect of chronic stress on the hypothalamic-pituitary-adrenal axis. They did not report the hormone receptor status of the tumours in the 251 women who developed breast cancer during follow-up. If their hypothesis were correct, then one would expect to observe a reduction in the risk of developing oestrogen-receptor positive, but not oestrogen-receptor negative, breast cancer in those women with high stress levels. Most women who participated in the study were post-menopausal at base-line, with a mean age of 57 years and, in this population, the main source of oestrogen production is peripheral tissue aromatase activity rather than the ovaries as in pre-menopausal women (2.3). The various cytokines known to be elevated in chronic stress states can influence aromatase activity (3). Furthermore, the risk reduction was observed mainly in women taking hormone replacement therapy, whose serum oestrogen levels are determined mainly by exogenous rather than endogenous sources. Finally, there is experimental evidence that chronic stress increases oestrogen and other steroid levels in serum (4). Therefore, it is difficult to provide a biological explanation for the observations reported by the authors. Further prospective studies with an objective stress assessment as base- line and at regular intervals during follow-up combined with the measurement of biological factors known to influence the risk of breast cancer are required. Professor Kefah Mokbel. MS, FRCS Professor at Brunel Institute of Cancer Genetics, Consultant Breast & Endocrine Surgeon. The Princess Grace Hospital, 42-52 Nottingham Place, London W1M 3FD e-mail: kefahmokbel@hotmail.com Mr Mohamed Salhab MRCS Research Fellow, St Georges Hospital,London, SW17 0QT References: 1. Nielsen et al, Self-reported Stress and Risk of Breast Cancer: Prospective cohort Study BMJ 2005, 331:548-550 2. Mokbel K. The evolving role of aromatase inhibitors in breast cancer. Int J Clin Oncol. 2002 ;7(5):279-83. Review. 3. Singh A, Purohit A, Duncan LJ, Mokbel K, Ghilchik MW, Reed MJ. Control of aromatase activity in breast tumours: the role of the immune system. J Steroid Biochem Mol Biol. 1997;61(3-6):185-92 4. MacNiven E et al. Chronic stress increases estrogen and other steroids in inseminated rats. Physiol Behav. 1992; 52(1):159-62. Competing interests: None declared
Does stress protect against breast cancer? Authors’ response 23 September 2005
Naja Rod Nielsen, PhD candidate National Institute of Public Health, Zuo-Feng Zhang, Tage S. Kristensen, Bo Netterstrøm, Morten Grønbæk Send response to journal: Re: Does stress protect against breast cancer? Authors’ response	Several readers of the BMJ have responded to our paper, which showed a lower incidence of breast cancer among women with high levels of perceived daily stress compared to women reporting no stress.1 We appreciate this interest in our work. Wenbin Liang points out that being alive is a condition for being diagnosed with breast cancer and that women with high levels of stress may die of other diseases during the tumor induction period and therefore never get diagnosed with breast cancer. Of the 6,689 women participating in the study, 2,224 died during follow-up and we understand Wenbin Liangs and others concern about the impact of this censoring on the results. As mentioned in the paper, more women in the high stress group (39.3 %) died during follow-up than women in the medium (30.4 %) and low (35.1 %) stress group. When using the Cox proportional hazards models we assume independent censoring, which in this case means that censoring is assumed to be independent of breast cancer risk within levels of perceived stress. It is, however, possible that competing causes that could lead to censoring, such as death of cardiovascular disease during follow-up, could be associated with risk of breast cancer within strata of stress due to other common risk factors. So the problem is not that stress affects cardiovascular disease, but that cardiovascular disease and breast cancer may be associated within strata of stress, so that the average risk of breast cancer among women censored in each stress-stratum is not equivalent to what it would have been if no such censoring had occurred. Some of the common risk or protective factors for both cardiovascular disease and breast cancer, such as moderate alcohol consumption and socio- economic status, have opposite effects on the two diseases, while other risk factors, such as low physical activity and postmenopausal obesity, increase the risk of both diseases. Women at high risk of cardiovascular disease could therefore either be at slightly lower or slightly higher risk of breast cancer depending on their risk factor profile. Thus, while we agree that bias from non-independent censoring by competing causes may have partially influenced our observation, some of the bias may level out and we find this kind of bias unlikely to fully explain the observed association between daily stress and breast cancer. Further, even though more women in the high stress group die during follow-up compared to the low stress group, stress does not seem to be a strong predictor for death and there is no dose-response effect in deaths similar to the one we find for stress and risk of breast cancer. Combined, this provides some confidence that differential censoring due to competing risks has not artificially created the observed inverse association between stress and breast cancer. Dr. Grant finds that our results defy common sense, because as she states “most women with breast cancer believe that high stress contributed to their disease”. This may indeed be the belief of some women and we hope that our research may help counteract such self-blaming. Though Dr Grant finds the fact, that the incidence of breast cancer has increased after women were included in the work force, to support her view, there is no clear consensus about how stress affects breast cancer risk,2-9 and only few studies have addressed the association between stress in everyday life and risk of breast cancer. Neither is it clear that working women should have higher stress-levels than non-working women. And let us just emphasise once more that stress cannot be considered a healthy response and that our results indicate that stress leads to disturbances in homeostasis and in normal body function. These disturbances may lead to a lower risk of breast cancer but may also increase the risk of a range of other diseases. Dr. Grant finds our study to be seriously flawed by confounding and interactions. It is true that higher proportion of women who reported high levels of stress also reported higher alcohol consumption and received hormone therapy. However, since both high alcohol consumption and hormone therapy may increase the risk of breast cancer, possible confounding from these factors would lead to an underestimation of the observed association. In fact, after adjusting for these factors we observed a slightly stronger inverse association between stress and breast cancer. As Eric L. Ding also emphasises, there may still be some residual confounding from both factors, and we can therefore not exclude that the observed association is an underestimate the true relation between stress and breast cancer. Dr. Grant also ask if there is any evidence that estrogen levels are lower in women who receive hormone therapy. The mechanisms behind the observed lower risk of breast cancer among stressed women remain unknown. As we state in the paper, one hypothesis is that prolonged activation of stress hormones can lead to lower production of estrogens, which is a main risk factor for breast cancer. This hypothesis is only based on laboratory-based experimental data and remains to be confirmed in humans. However, in the Nurses Health Study, where caregiving was used as an indicator of chronic stress among female nurses, significantly lower levels of bioavailable estradiol was found among caregivers than among non-caregivers.8 How stress is defined and measured remains a point of debate and both Dr. Gulati and Dr. Taiwo question the validity of our stress measure. We agree with Dr. Gulati that ones stress level may be greatly influenced by personal factors and life events at the time of participation, - the same will be true for measures of alcohol consumption, level of physical activity, use of medication, blood pressure and all the other thing we measure in epidemiology. A one-time measure of a risk factor over long follow-up should therefore never be over-interpreted and the potential impact of long follow-up ought to be evaluated. In order to do this we assessed the associations between stress and incidence of breast cancer in the first and last nine years of follow-up and found the estimates to be very similar. This indicates that the effect of stress is relatively stable over time and the long follow-up is thereby unlikely to have resulted in a great deal of exposure misclassification. In fact, potential misclassification bias from using only a baseline measure of stress would most likely result in underestimation of the association under study. Dr. Taiwo finds it problematic that a situation, which is considered as stressful by one person, not necessarily is perceived as such by others. This is a very natural consequence of the fact that we all have different capacity and ways to cope with a potential stressful situation, which in our opinion is a valid reason for preferring a self-reported measure of stress as opposed to a count of potential stressful situations defined by the researcher. Dr. Taiwo further claims that we did not take confounding from such factors as poverty and unemployment into account. While this may be true, we adjusted our analyses for education and we have previously included income in our analyses, but neither factor changed the risk estimates. We appreciate that Eric L. Ding finds our study well designed. In our opinion, tobacco smoking is not a well-established risk factor for breast cancer, but we are aware that some studies have suggested a positive association.10 More women with high levels of stress also engaged in tobacco smoking and Eric L. Ding is right to accentuate that this may be a potential source of bias. In fact, we addressed potential confounding from tobacco smoking in our early analyses. Smoking was not associated with breast cancer risk in our study population and adjustment for tobacco smoking did therefore not change our risk estimates. Had tobacco smoking in fact been positively associated with breast cancer in this study, confounding from this factor would still not have been able to explain the observed inverse association between stress and breast cancer because it would create a spurious positive association. We tested for interactions between the variables included in our model, but we were naturally not able to account for interactions between unmeasured variables, and we cannot exclude the possibility for some residual confounding from these variables, as we also discussed in the paper. Dr. Sambandan finds our results to be contrary to her 15 years of experience with breast cancer patients, many of whom experienced stressful life events in the two years prior to diagnosis. Death of a spouse or near relative or friend is a major acute stressor, while the stress we experience in daily life is more moderate and chronic in nature. The physiological stress response is highly dependent on the type and the timing of the stressor,11 which makes it vital to distinguish between different types of stress. Thus, as we also emphasise in the paper, a greater risk of breast cancer associated with stressful life events is not necessarily in contrast with a lower risk of breast cancer associated with daily stress. We solely report what we observe in our study, and we repeat that we are not sure about the mechanisms behind the observed lower risk of breast cancer among stressed women. In the paper we suggest a hormonal hypothesis and Dr. Mokbel finds this hypothesis unlikely to explain our results. We agree that other mechanisms may be in play and that further epidemiologic and experimental studies are needed to settle this issue. An obvious next step in this line of research is to see if these results can be replicated in other populations and with other measures of chronic stress and to assess the validity of the estrogen hypothesis in humans. References 1. Nielsen NR, Zhang ZF, Kristensen TS, Netterstrom B, Schnohr P, Gronbaek M. Self reported stress and risk of breast cancer: prospective cohort study. BMJ 2005;331:548. 2. Achat H, Kawachi I, Byrne C, Hankinson S, Colditz G. A prospective study of job strain and risk of breast cancer. Int J Epidemiol 2000;29:622-8. 3. Duijts SF, Zeegers MP, Borne BV. The association between stressful life events and breast cancer risk: a meta-analysis. Int J Cancer 2003;107:1023-9. 4. Lillberg K, Verkasalo PK, Kaprio J, Teppo L, Helenius H, Koskenvuo M. Stress of daily activities and risk of breast cancer: a prospective cohort study in Finland. Int J Cancer 2001;91:888-93. 5. Lillberg K, Verkasalo PK, Kaprio J, Teppo L, Helenius H, Koskenvuo M. Stressful life events and risk of breast cancer in 10,808 women: a cohort study. Am J Epidemiol 2003;157:415-23. 6. Protheroe D, Turvey K, Horgan K, Benson E, Bowers D, House A. Stressful life events and difficulties and onset of breast cancer: case- control study. BMJ 1999;319:1027-30. 7. Johansen C,.Olsen JH. Psychological stress, cancer incidence and mortality from non-malignant diseases. Br J Cancer 1997;75:144-8. 8. Kroenke CH, Hankinson SE, Schernhammer ES, Colditz GA, Kawachi I, Holmes MD. Caregiving stress, endogenous sex steroid hormone levels, and breast cancer incidence. Am J Epidemiol 2004;159:1019-27. 9. Schernhammer ES, Hankinson SE, Rosner B, Kroenke CH, Willett WC, Colditz GA et al. Job stress and breast cancer risk: the nurses' health study. Am J Epidemiol 2004;160:1079-86. 10. Terry PD, Rohan TE. Cigarette smoking and the risk of breast cancer in women: a review of the literature. Cancer Epidemiol Biomarkers Prev 2002;11:953-71. 11. McEwen BS, Wingfield JC. The concept of allostasis in biology and biomedicine. Horm Behav 2003;43:2-15. Competing interests: None declared
Re: High stress and breast cancer 27 September 2005
Ellen C G Grant, physican and medical gynaecologist kingston-upon-Thames, KT2 7JU, UK Send response to journal: Re: Re: High stress and breast cancer	The baseline data in Table 1 in the study by Nielsen and colleagues showed that 69% (717/1045) of the women currently taking HRT in 1981-3 reported high or medium stress compared with 56% (3179/5644).1 The data in Table 5 reveals that by 1999 twice as many of the women taking HRT in 1981-3 had developed breast cancer compared with women who were not current HRT users at baseline (6.4% compared with 3.2%). It is therefore difficult to understand how the authors failed to conclude from this basic data that HRT use combined with higher stress awareness increases the risk of breast cancer. 1 Nielsen NR, Zhang Z-F, Kristensen TS, et al. Self reported stress and risk of breast cancer: prospective cohort study. BMJ 2005; 331: 548 Competing interests: None declared
Nielsen et el. do not evaluate stress and breast cancer, but rather, socioeconomic status and breast cancer 7 December 2005
Candyce H. Kroenke, RWJ Health and Society Scholar 3333 California St. Ste 465, San Francisco, CA 94118, Ichiro Kawachi, Laura D. Kubzansky Send response to journal: Re: Nielsen et el. do not evaluate stress and breast cancer, but rather, socioeconomic status and breast cancer	Nielsen and colleagues recently published results (1) showing an impressive, graded inverse association between stress, measured as the combination of tension, anxiety, and depression, and risk of breast cancer over 18 years. Although many believe stress may play a causal role in the development of breast cancer, evidence to date has been difficult to decipher. Thus, we support publication of findings in this area, and the continued discussion of the role of factors such as social support, stress, and mental health in chronic disease outcomes. Publication of this article may stimulate further thought in this area which is needed. In our previous work, we reported a lack of association between caregiving stress and breast cancer over an 8-year time period. (2) In an additional cross-sectional analysis, women reporting the highest number of caregiving hours had lower levels of estradiol and bioavailable estradiol. Though we did not find an association with breast cancer, it is possible that with additional follow-up, we would have seen an inverse finding similar to that in Nielsen’s study. The authors’ findings are plausible in the context of our findings. Unfortunately, we do not believe that the findings reported in the article constitute a fair evaluation of the relationship between stress and breast cancer. We believe that the finding Nielsen and colleagues report is a reiteration of the relationship between socioeconomic status (SES) and breast cancer. SES is well known to be inversely related to stress and positively related to breast cancer. Thus, unlike many SES- disease gradients, women with lower SES are at decreased risk for breast cancer. Cumulative exposure to estrogen (3-5) is hypothesized to be the key mechanism that explains this relationship. There are a variety of important factors contributing to cumulative exposure to estrogen related to SES and to breast cancer including age at menarche, age at first birth, number of births, and age at menopause. A strength of our work was the ability to adjust fully for cumulative exposure to estrogen. However, in the present study, adjustments were made only for number of births. The authors imply adequate adjustment for cumulative estrogen exposure because “adjustment for number of children, which also tends to be correlated with age at first pregnancy….only slightly changed the estimates.” However, adjustment for parity alone may be insufficient adjustment. In the authors’ attempt to adjust for SES, they adjusted for women’s education using the following groupings: <8, 8-11, and 12+ years and indicated that adjustment for education did little to alter the estimates. However, this categorization may not enable adequate adjustment for women’s education as this grouping leaves room for residual confounding in the 12+ years’ group. Individuals who have completed high school only may differ in many ways from people who have university-level education. In the article, the authors do not clearly present their rationale behind the choices of categorization of potential confounding variables and so it is not clear whether the choice was made to ensure an adequate distribution across stress categories or for another reason. Importantly, women’s SES may also be strongly influenced by whether or not she is married (6) and if so, her husband’s education and income. Though low SES is related to higher stress and this may theoretically lead to a reduction in cumulative exposure to estrogen, it would be necessary to adequately adjust for these known etiologic factors which strongly influence breast cancer, or it is not possible to conclude that stress had an influence on breast cancer in their study population. And if those particular etiologic factors are related to stress, it is important to specify and investigate the nature of those relationships. It is critical to understand how psychosocial factors may contribute to health outcomes. In order to do so, it will be necessary to consider more upstream factors like socioeconomic status in order to think fully about the mechanisms through which psychosocial factors affect health outcomes. Related to this, it will be important to carefully consider those mechanistic relationships in light of important and known etiologic factors to ensure that findings truly contribute to our scientific understanding of psychosocial influences on health. 1) Nielsen NR, Zhang ZF, Kristensen TS, Netterstrom B, Schnohr P, Gronbaek M. Self reported stress and risk of breast cancer: prospective cohort study. BMJ. 2005 Sep 10;331(7516):548. 2) Kroenke CH, Hankinson SE, Schernhammer ES, Colditz GA, Kawachi I, Holmes MD. Caregiving stress, endogenous sex steroid hormone levels, and breast cancer incidence. Am J Epidemiol. 2004;159(11):1019-27. 3) Endogenous sex hormones and breast cancer in postmenopausal women: reanalysis of nine prospective studies. J Natl Cancer Inst 2002;94:606-16. 4) Bernstein L, Ross RK. Endogenous hormones and breast cancer risk. Epidemiol Rev 1993;15:48-65. 5) Thomas HV, Reeves GK, Key TJ. Endogenous estrogen and postmenopausal breast cancer: a quantitative review. Cancer Causes Control 1997;8:922-8. 6) Morgan LA. Economic well-being following marital termination: a comparison of widowed and divorced women. J Fam Issues 1989; 10:86-101. Competing interests: None declared