Fax: (916) 734-5641
Preferences of Women Evaluating Risks of Tamoxifen (POWER) study of preferences for tamoxifen for breast cancer risk reduction†
Article first published online: 11 APR 2005
Copyright © 2005 American Cancer Society
Volume 103, Issue 10, pages 1996–2005, 15 May 2005
How to Cite
Melnikow, J., Paterniti, D., Azari, R., Kuenneth, C., Birch, S., Kuppermann, M., Nuovo, J., Keyzer, J. and Henderson, S. (2005), Preferences of Women Evaluating Risks of Tamoxifen (POWER) study of preferences for tamoxifen for breast cancer risk reduction. Cancer, 103: 1996–2005. doi: 10.1002/cncr.20981
Presented in part at the annual meeting of the North American Primary Care Research Group, Halifax, Nova Scotia, Canada, October 13–16, 2001; and the annual meeting of the Society for General Internal Medicine, Vancouver, British Columbia, Canada, April 30–May 3, 2003.
- Issue published online: 28 APR 2005
- Article first published online: 11 APR 2005
- Manuscript Accepted: 27 DEC 2004
- Manuscript Revised: 15 DEC 2004
- Manuscript Received: 18 AUG 2004
- California Breast Cancer Research Program. Grant Number: 5PB-0110
- National Cancer Institute. Grant Number: R01 CA86043
- breast cancer;
- prevention and control;
- decision making
The objective of this study was to understand the attitudes and preferences of risk-eligible women regarding use of tamoxifen for breast cancer risk reduction.
A cross-sectional, mixed-methods interview study was conducted at a university medical center and at community sites. Participants were women who had an estimated 5-year breast cancer risk ≥ 1.7% and no prior breast cancer. Interviews were conducted in English or Spanish. The interview included a 15-minute, standardized educational session on the potential benefits and harms of tamoxifen followed by close-ended and open-ended questions about participants' inclinations to take tamoxifen and factors important to their decision. A demographic questionnaire, a test on knowledge of potential benefits and harms of tamoxifen, and an interview evaluation were included.
Two hundred fifty-five women completed interviews. Their estimated mean 5-year breast cancer risk was 2.8%; and their mean self-perceived 5-year risk was 32.7%. After the educational intervention, 45 women (17.6%) were inclined to take tamoxifen. Very high risk women (> 3.5%) were no more inclined to take it than women with lower risk (1.7–3.5%). In a multivariable analysis, lower income, confidence in the effectiveness of tamoxifen, and concern about fractures were associated with being inclined to take it; concern about pulmonary embolism, dyspareunia, cataracts, and low self-perceived breast cancer risk were associated negatively with taking tamoxifen. Participants expressed concerns about adverse effects.
Less than 20% of women were interested in tamoxifen after education about potential benefits and harms, despite a very high self-perceived breast cancer risk. Candidate chemoprevention agents must have few potential adverse effects to achieve widespread acceptance. Cancer 2005. © 2005 American Cancer Society.
Since the approval of tamoxifen for breast cancer risk reduction by the United States Food and Drug Administration (FDA) in 1998, its use for this purpose has been controversial. Tamoxifen, which is a synthetic estrogen receptor modulator, has been used for many years in the treatment of breast carcinoma.1 The largest randomized trial to evaluate risk reduction, the National Surgical Adjuvant Breast and Bowel Project (NSABP) P-1 trial, found a 49% reduction in the incidence of breast cancer.2 Analyses of individual trials conducted in Europe showed no significant risk reduction3, 4 or found significant but smaller risk reductions.5 A recent overview of all risk reduction trials, including European trials, estimated a 38% reduction in breast cancer incidence confined to estrogen receptor-positive carcinomas.6 Significant adverse effects that were attributable to tamoxifen were noted both in the trial overview and in the NSABP P-1 trial, including increased risks of endometrial carcinoma, deep venous thrombosis, pulmonary embolism, and cataracts requiring surgery.2, 6 Subsequent long-term follow-up of the NSABP P-1 participants revealed increased risks of stroke and endometrial sarcoma among tamoxifen users, and tamoxifen prescribing information was amended to include these potential risks in 2002.7
In 2001, the Canadian Task Force on Preventive Health Care recommended that physicians assess breast cancer risk and counsel women about the potential benefits and harms of tamoxifen if it was estimated that their 5-year breast cancer risk was > 1.66% (the eligibility threshold for the NSABP P-1 trial).8 In 2002, the U.S. Preventive Services Task Force followed, recommending that clinicians discuss chemoprevention with women at high risk for breast cancer and at low risk for the adverse effects.9 Counseling all women at this risk threshold has the potential to consume substantial time and resources. It is estimated that more than 10 million women have a breast cancer risk ≥ 1.67%.10
Differing perspectives on the relative balance of potential benefits and harms for healthy women taking tamoxifen fuel the controversy about its use for breast cancer risk reduction. According to a recent estimate by Freedman et al., about 2 million women in the United States would experience a net benefit from taking tamoxifen.10 Women's use of tamoxifen for risk reduction has not been widespread, however. It has been estimated that only 5% of Astra Zeneca's tamoxifen sales derive from its use for risk reduction (unpublished results), implying that women's uptake for this purpose is very low.
How women weigh the potential harms and benefits of tamoxifen and make decisions about risk reduction therapy is not well understood. We evaluated women's preferences for tamoxifen and related outcomes. In a structured interview that included a standardized educational session about the potential benefits and harms of tamoxifen, we used quantitative and qualitative methods to understand the attitudes and preferences of a diverse group of women who met the criteria to take tamoxifen for breast cancer risk reduction. This mixed-method approach provides a rich context for understanding the process through which evidence-based recommendations become practice.11
MATERIALS AND METHODS
This study was reviewed and approved by the University of California–Davis Office of Human Research Protection. After obtaining written, informed consent from participants, a structured, 1-hour, in-person interview in English or Spanish was conducted that combined qualitative and quantitative interview items. After preliminary questions, a standardized educational session was used to describe the potential beneficial and harmful outcomes of taking tamoxifen for breast cancer risk reduction to all participants by trained nurse or physician assistant interviewers.
Participants were women who lived in Sacramento and surrounding communities. Women were eligible to participate if they had not already had breast cancer, if they spoke either English or Spanish, and if they had an estimated 5-year breast cancer risk using the Breast Cancer Risk Assessment Program, version 112 ≥ 1.7%. This program, which was based on the Gail model, estimates 5-year and lifetime breast cancer risk based on age, race, age at menarche, age at first live birth, family history, and history of previous breast biopsies. To achieve the risk threshold, recruitment focused on identifying participants age > 50 years, but younger potential participants also were screened. Participants were recruited from Sacramento and surrounding northern Central Valley communities through contacts with community and church groups, presentations at health fairs, print and radio advertising, and by a direct mailing to participants enrolled in the observational arm of the Women's Health Initiative University of California–Davis study site. Interviews were conducted at a university medical center and occasionally on site at churches, participant's homes, and community health clinics.
A 15-minute, standardized educational session on the potential benefits and harms of tamoxifen for breast cancer risk reduction was part of the interview. The session presented the overall benefits and adverse outcomes as rates per 1000 women over 5 years found in the initial results of the NSABP P-1 trial.2 Breast carcinoma, osteoporotic fractures, endometrial carcinoma, deep venous thrombosis, pulmonary embolism, and cataracts requiring surgery (considered severe outcomes) as well as hot flashes, vaginal discharge, and dyspareunia were described in lay terms. Information on rates was presented in multiple formats: verbally, in tables, in color-coded charts, and in a display of color-coded beads representing women affected by each of the severe outcomes against a background of clear beads representing women not experiencing any severe outcomes. The risk of endometrial cancer was omitted from the educational session for participants who had undergone a hysterectomy. During and after the educational session, interviewers answered questions from participants. Immediately after the educational session, interviewers administered a true-false written test to assess comprehension of the basic information presented in the educational session.
The interview was developed in English, piloted for clarity and face validity, revised, translated into Spanish, and piloted again in Spanish. Most interview items and the educational content were developed specifically for the study based on focus groups with risk-eligible African-American, Latina, and white women. Questions on which factors were important in the decision about whether to take tamoxifen were derived from an iterative review of focus group transcripts by the research team. Ranked response options for close-ended questions spanned a five-item scale, from “not very inclined to take it” to “very inclined to take it” or from “not important at all” to “very important.” After revisions were made based on pilot testing, final versions of the interview instrument and the true-false test were reviewed by the research team for face validity. Two native Spanish-speaking members of the research team who were not involved with the original translation reviewed the Spanish-language instruments.
Initial study queries focused on concern about breast cancer, estimated self-perceived 5-year and lifetime breast cancer risk, sources of information about breast cancer, familiarity with tamoxifen, and, for those women who had heard of tamoxifen prior to the interview, willingness to consider taking it for breast cancer risk reduction. After the educational session, the interview continued with questions about confidence in tamoxifen to reduce the risk of breast cancer, inclination toward or against taking tamoxifen, and factors important to the decision, including concern about adverse outcomes; side effects and cost; and use of hormone replacement therapy, other medications, herbs, and supplements. Questions on trust in the health care system were obtained from a previous study of decision-making regarding prenatal testing.13 Women were given their 5-year and lifetime National Cancer Institute breast cancer risk assessment program estimates at the end of the interview.
The interview included 13 open-ended questions. Seven follow-up questions were used to collect additional information about the answers to close-ended items (e.g., if a woman reported that she did not have enough information to make a decision about taking tamoxifen, then she was probed regarding the kind of information she needed). Six open-ended questions sought information about the overall context of women's knowledge and behaviors related to decision-making (e.g., “What are you currently doing to reduce your risk of breast cancer?”). Open-ended questions were divided into subsets, and every fourth participant was asked to respond to one of the question subsets.
Additional data on participants' sociodemographic characteristics and a subjective evaluation of the interview experience were collected by questionnaire at the time of the interview. Interviewers rated the level of difficulty participants experienced in understanding the potential benefits and harms of tamoxifen. Participants also rated the difficulty of understanding these benefits and harms and the level of difficulty in making the decision about whether or not they would take tamoxifen.
Quantitative analyses were conducted in SAS software (version 8.2; SAS Inc., Cary, NC). A McNemar test was used to evaluate change in the inclination to take tamoxifen before the educational session compared with afterward for women who had heard of tamoxifen previously. Bivariate chi-square analyses compared women who were inclined to take tamoxifen with all other participants for a range of potentially associated demographic and decision-making variables. Logistic regression modeling was conducted to measure the association between women's attitudes and beliefs about the benefits and harms of tamoxifen and their inclination to try tamoxifen. To construct the model, women's decision-making about tamoxifen was grouped into conceptual domains based on information derived from the initial focus groups. These domains were used to formulate a set of potential predictors for the logistic regression model. Variables from these domains were selected for the initial model because they were determined, a priori, to be important or because the odds ratios were significant in bivariate analysis. Adjusted odds ratios were computed for those factors affecting decision-making that were shared by at least five women. Variables were evaluated in a manual, iterative, model-building process. Colinearity diagnostics indicated that each of the variables in the final model provided unique information. The Akaike Information Criterion was used to evaluate model fit.14
The qualitative analysis developed a coding scheme for characterizing responses to open-ended questions based on an iterative comparison of individual responses. Members of the project team reviewed all open-ended responses and created a list of the range of responses provided in the data. Next, they derived a list of all key elements by inductively assessing the recurring responses in the interview transcripts. Reviewers noted key elements and grouped those elements that were similar categorically. Components of each element were then classified into subcategories (e.g., the element “reducing risk” contained the subcategories “clinical visits” and “self-efficacy”).
Reviewers came to a mutual agreement about the classification of categories. Because one objective of the analysis was to create an exhaustive list of the range of responses, classification disagreement resulted in splitting (rather than lumping) key elements into separate groupings. Reviewers then developed a code book that included definitions of elements that were important in assessing perceptions of risk (e.g., reducing the risk of breast cancer), related substantive subcategories (e.g., clinical visits, self-efficacy, family history), and recurring contexts in which these elements occurred. Results of the coding process were reviewed with the multidisciplinary team of coinvestigators and were assessed for clinical relevance.
Seven hundred ninety-eight women were evaluated for enrollment in the current study. Twenty-six women were ineligible, mostly because of a prior history of breast cancer. Of the remaining 771 women, 341 women had an estimated 5-year breast cancer risk ≥ 1.7%. One woman was diagnosed with breast cancer in the interval between screening and the interview. Eighty-five women did not keep interview appointments, despite reminders. Of the 341 eligible women, 255 women completed the interview (Fig. 1). Greater than 90% of the interviews were conducted at the university medical center. Table 1 shows the characteristics of the final sample.
|Characteristic||No. of patients||%|
|≥ 75 yrs||62||24.6|
|< 200% federal poverty threshold||33||14.7|
|< High school||12||4.8|
|High school graduate||31||12.3|
|On hormone replacement therapy||154||62.4|
Participants' mean, self-perceived, 5-year breast cancer risk was 32.7%, which was > 10 times the mean 5-year risk estimated by the Breast Cancer Risk Assessment Program (2.8%). Despite this substantial overestimation of numeric risk, 70.9% of participants described their risk as low or average. One hundred ninety-one participants (75%) had heard of tamoxifen before the interview. Among these women, the proportion inclined to take tamoxifen (19%) did not change after the educational session. Ten women who had been inclined against tamoxifen for risk reduction before the intervention were inclined to try it after the intervention, and 9 women who had been inclined to try it subsequently reported being against it (P = 0.82). Although there was little change in the combined number of women who were either inclined against tamoxifen or neutral, the number who were neutral fell from 101 before the educational intervention to 55 after the intervention.
Participants performed well on the test after the educational session, with 80% answering all questions correctly and < 5% answering more than 1 question incorrectly. Five participants (2%) felt that it was difficult or very difficult to understand the risks and benefits of tamoxifen. Interviewers found that 7 participants (2.8%) had a difficult or very difficult time understanding.
Among all 255 participants, 45 women (17.6%) were inclined to take tamoxifen after the educational session. Fewer participants who rated their self-perceived risk for breast cancer as “low” (compared with those who rated their risk as “average” or “high”) were inclined to take it (P < 0.01), but we found no association between estimated risk (according to the Breast Cancer Risk Assessment Program) and the inclination to take tamoxifen (P = 0.98). Higher risk women (5-year risk > 3.5%) were no more inclined than the overall group to take tamoxifen (data not shown).
Greater than 50% of participants listed the following risks as “very important” in the decision whether or not to take tamoxifen: breast cancer (68.8%), pulmonary embolism (67.2%), endometrial carcinoma (62.7% among women without hysterectomy), and deep venous thrombosis (58.4%). Thirty-seven participants (15.1%) indicated it was difficult or very difficult to decide whether or not to try tamoxifen for breast cancer risk reduction. Interviewers observed only 17 women (6.8%) who had a difficult or very difficult time with the decision. The results of stratified bivariate analysis on the inclination to take tamoxifen are shown in Table 2.
|Characteristic||Willing to try tamoxifen (n = 45)||Not willing to try tamoxifen (n = 206)||P value|
|< 65 yrs||16||35.6||64||31.5|
|≥ 75 yrs||10||22.2||49||24.1|
|Less than high school||3||6.8||8||3.9|
|High school graduate||5||11.4||26||12.7|
|Health status||< 0.05|
|< 200% FPL||10||26.3||23||12.3|
|≥ 200% FPL||28||73.7||164||87.7|
|Family history of breast cancer (sister, mother, other relative)||0.22|
|Confusion about tamoxifen risks/benefits||0.84|
|Paying $5/month for tamoxifen||0.97|
|Paying $100/month for tamoxifen||0.98|
|Prior knowledge of tamoxifen||0.61|
|Women's Health Initiative||19||61.3||124||72.5|
|Importance of fractures||< 0.0001|
|Very important/moderately important||41||91.1||116||56.6|
|Somewhat/not very/not at all Important||4||8.9||89||43.4|
|Importance of dyspareunia||< 0.01|
|Very important/moderately important||6||13.3||75||63.0|
|Somewhat/not very/not at all important||39||86.7||128||37.0|
|Importance of hot flashes||< 0.01|
|Very important/moderately important||10||22.2||90||43.9|
|Somewhat/not very/not at all important||35||77.8||115||56.1|
|Importance of cataracts||< 0.0001|
|Very important/moderately important||10||22.2||119||58.1|
|Somewhat/not very/not at all important||35||77.8||86||42.0|
|Importance of pulmonary embolism||< 0.0001|
|Very important/moderately important||27||60.0||181||88.3|
|Somewhat/not very/not at all important||18||40.0||24||11.7|
|Confidence in tamoxifen to reduce breast cancer||< 0.0001|
|Somewhat/not very/not at all confident||8||17.8||123||59.7|
|Self-perceived breast cancer risk||< 0.01|
Logistic regression analysis emphasized the importance of benefits and harms to participants' decision-making (Table 3). Age, race/ethnicity, the use of hormone-replacement therapy, and having undergone a hysterectomy were not associated significantly with the inclination to take tamoxifen in bivariate analysis and did not improve the model fit. Education was not associated with the inclination to take tamoxifen but was included in the model to ensure that the effect of household income was not related to education. We explored interaction terms for low income with other variables, including education and health status, but because of small cell sizes, these interaction terms were unstable, and the odds ratios were not significant. The model fit was not improved by the inclusion of interaction terms.
|High school graduate or less||1.11||0.28–4.34|
|At least some college||1.00|
|< 200% FPL||4.71b||1.11–19.94|
|≥ 200% FPL||1.00|
|Good, fair, or poor||0.67||0.23–1.94|
|Very good or excellent||1.00|
|Self-perceived breast cancer risk|
|Medium or high risk||1.00|
|Somewhat/not very/not at all important||1.00|
|Somewhat/not very/not at all important||1.00|
|Blood clot in the lung|
|Somewhat/not very/not at all important||1.00|
|Confidence in tamoxifen to reduce breast cancer|
|Somewhat/not very/not at all confident||1.00|
|Important/somewhat/not very/not at all important||1.00|
Needing to pay a substantial amount ($100) or a small amount ($5) for tamoxifen was not associated with the inclination to take tamoxifen and was not included in the model. Household income < 200% of the federal poverty level, feeling that tamoxifen's effect on fractures was important, and being confident that tamoxifen could reduce breast cancer risk all were associated significantly with the inclination to take tamoxifen. Significant negative associations with the inclination to take tamoxifen were found for low self-perceived breast cancer risk and concern about dyspareunia, cataracts, and pulmonary embolism.
The effect of concern about endometrial carcinoma in the current analysis was explored among the participants who had a uterus (n = 146 women) and thus, would have to consider this potential harm. Among these participants, concern about endometrial carcinoma was not associated with the inclination to take tamoxifen in bivariate or multivariable analysis. With our sample size, we had 80% power to find a 19% difference in concern about endometrial carcinoma among women inclined to take tamoxifen compared with those who were not.
Every woman who participated in the interview was asked several open-ended questions. With rotation of the questions in the interview, about 20% of the total sample responded to each of the open-ended questions. In general, participants attributed their risk of breast cancer to their own overall health and family history of cancer. Although age or health status were not associated with the inclination to take tamoxifen in the multivariable analysis, detailed qualitative responses indicated that consideration of age and overall health may have figured in participants' thinking about their susceptibility to breast cancer or tamoxifen-related side effects.
The role of self-efficacy, including current activity level and health state, and the use of preventive measures in lowering the risk of breast cancer or breast cancer mortality was mentioned prominently. Participants described clinical visits for breast examinations and mammograms and self-preventive strategies as ways to reduce risk. Self-prevention ranged from self-examination to changes in diet and exercise, specifically reducing or eliminating alcohol, caffeine, and tobacco. In contrast, some participants said they were “doing nothing” to reduce their risk of breast cancer. Overall, qualitative responses suggested that participants balanced their perceived risk factors against personal preventive strategies to determine their own susceptibility to breast cancer before making their decisions about the use of tamoxifen for risk reduction.
After the educational intervention, participants were asked to rate how confident they were about the ability of tamoxifen to reduce the risk of breast cancer. Women's qualitative responses to the open-ended question, “How confident do you feel about the ability of tamoxifen to reduce the risk of breast cancer?” are reported in Table 4.
|Level confidence and open-ended response||Important decision-making elements|
|Perceived risk||Information source|
|…My sister has it [tamoxifen], and she doesn't show any side effect||Family history (breast cancer risk)||Family|
|I suppose the fact that it has not been around for that long. I don't know if there have been other studies.||—||Research results|
|The effectiveness [makes me confident]; my concern would be the blood clots and the eye problems.||Side effects of tamoxifen||—|
|It has improved prevention of breast cancer, but the side effects concern me.||Side effects of tamoxifen||Research results|
|The main thing that I would consider is the hormones. I take the estrogen and progesterone and I feel good taking them.||Current health state/activity||Self|
|I know this lady who has been taking tamoxifen for 3 years, and she swears by taking it. I think she told me the right information.||Other's health history (side effects of tamoxifen)||Friend|
|[I worry about] the side effects…since I've had problems with two of them it might persuade me not to take it, although I am somewhat confident that it can reduce the rate of cancer. I have had phlebitis, and my mother died of a blood clot of the lung.||Own health history (breast cancer risk); family history (side effects of tamoxifen)||Self|
|I'm looking more at the negatives.||Side effects of tamoxifen||—|
|I think there are too many unknowns.||General risks with tamoxifen use||—|
|Just the statistics, that there was the difference between 9 and 17 women who had to deal with breast cancer…almost twice as many women.||General risk of breast cancer||Statistics|
|I weigh my current state of health and that I still have my uterus, also considering my family health.||Current health history (side effects of tamoxifen)||Self|
|When you look at the numbers it has some effect, but statistically I don't know if it really has an effect.||—||Statistics|
|Very concerned about the serious side effects, endometrial cancer is as bad as breast cancer.||Side effects of tamoxifen||—|
|…When I look at the other side effects, the blood clotting and the eye problems, I can't see that there is a big advantage in the breast cancer, I'm not sure if it's worth the risk. I'm not one to solve things with medication. I don't run to the doctor every time I have a pain.||Side effects of tamoxifen||Self|
|…Where it has some lowering of breast cancer, it seems to increase a bunch of other risks, I'm not sure I want to risk the exposure.||Side effects of tamoxifen||Self|
Participants related confidence in tamoxifen to their specific concerns about the potential harms of taking tamoxifen and their own health history. Many expressed concern about disrupting their current state of health and “trading one thing (breast cancer) for another (potential adverse effects from tamoxifen).” “Own health history” figured prominently as a source of information for participants at all levels of confidence. Few described their level of confidence in tamoxifen's ability to reduce the risk of breast cancer without mentioning its potential adverse effects. Many participants reported being concerned about blood clots as an adverse effect. Several related this concern to their family history or their current health state. Participants who focused on research results that were presented in the educational session weighed both positive (e.g., statistical significance) and negative (e.g., length of study, occurrence of side effects) aspects of study design and tamoxifen-related outcomes. Nonwhite women expressed concern about research results less often than white women.
The results of the current study demonstrate three main points. First, among women age ≥ 50 years who potentially were eligible to take tamoxifen for breast cancer risk reduction, only 17.6% expressed an inclination to take it after they participated in an educational session that described the benefits and harms of this medication. The educational session was based on trial results from the most favorable study2 and presented overall risk, as opposed to age-specific risks for adverse outcomes, so that harms, in general, were underestimated for participants in the study, who tended to be older.15 Second, among those women who had heard of tamoxifen, few women initially were inclined to try it, and very few changed their minds in favor after the educational session. Consideration of possible benefits, in particular breast cancer risk reduction and reduction of osteoporotic fractures, was related to women's willingness to try tamoxifen. Third, having a low self-perceived breast cancer risk, in addition to concern about adverse side effects from taking tamoxifen, strongly influenced the decision to decline tamoxifen prophylaxis.
The same pattern of preferences in women with and without a uterus indicated that concern about endometrial carcinoma did not play a major role in the decision-making process. In this instance, raloxifene, which currently is being studied for breast cancer risk reduction in the Study of Tamoxifen and Raloxifene (STAR) trial,16 may not find a more favorable reception than tamoxifen if it is shown to be effective, despite its lack of effect on the endometrium.
Women's responses to open-ended questions confirmed that concern about the risks of taking tamoxifen and perceived susceptibility to breast cancer and to tamoxifen-induced adverse effects were major considerations in the decision-making process. Participants appeared to balance the risk of breast cancer against the risk of potential adverse effects from tamoxifen; they then weighed perceived risk factors against personal preventive strategies, factors they felt were protective against risk.
Our findings are consistent with the other published research indicating that high-risk women are reluctant to consider tamoxifen for risk reduction. In response to a general question on a community survey about breast cancer chemoprevention without specific information on risks and benefits, 23% of women expressed an interest.17 Among 43 eligible patients who were identified from surgical practices, only 2 patients elected to take tamoxifen after they were given information about its potential benefits and harms. Educational sessions had no influence on patient decision-making.18 For these women, concerns about adverse effects, particularly thromboembolic events and endometrial carcinoma, were the reasons cited most commonly for not wanting to take tamoxifen.
The relation of having an income that is < 200% of the federal poverty level to the greater odds of being inclined toward tamoxifen for risk reduction is intriguing. This relation was independent of education, race/ethnicity, or age. Unlike income, the need to pay a substantial amount ($100) or a small amount ($5) for tamoxifen was not associated with an inclination to take it. Little research has been done on the relation between medical decision-making and income. This finding may be explained by differences in the way individuals in different incomes groups value the timing of outcomes.19
The current study was limited by the recruitment of participants as a cross-sectional convenience sample. Only women over age 50 years were interviewed. Compared with the overall community of Sacramento, participants were older (31.8% ages 50–64 years vs. 52.7% ages 50–64 years in Sacramento), more likely to be white (77% vs. 70%), more educated (45% college educated vs. 20.8%), and fewer had a household income under $25,000 (20.6% vs. 32.6%). Conversely, we have no information on the demographic profile of women in the Sacramento community who have sufficient breast cancer risk to consider taking tamoxifen for risk reduction. We have no information on whether women who indicated that they were inclined to try tamoxifen for breast cancer risk reduction actually sought a prescription from their health care providers or whether women who said they were not inclined to take it later obtained a prescription. Neither do we know whether those who may have obtained a prescription actually took the medication.
Since its approval by the FDA, controversy has continued regarding the appropriate selection of women for counseling on breast cancer risk reduction using tamoxifen.6, 10, 20, 21 The current results indicate that many high-risk women are unwilling to consider tamoxifen, even with extensive education about its potential benefits and harms. Some authors have concluded that a higher risk threshold for breast cancer should be used, particularly among women over age 50 years, who are at higher risk for adverse effects from tamoxifen.22 Counseling women about the benefits and harms of tamoxifen is a time-consuming process for health care providers. In our study, a thorough explanation of potential benefits and adverse effects took at least 15 minutes. Given the low level of interest in tamoxifen for risk reduction among potentially eligible women in studies performed to date and the very small number of women whose decision was influenced by the educational session, we question the recommendation that all women with an estimated 5-year breast cancer risk ≥ 1.67% should be counseled about tamoxifen for risk reduction. Although information about tamoxifen for risk reduction should be widely available, setting a higher breast cancer risk threshold for individual counseling would target a smaller group of women who would be more likely to experience substantial benefit. Unlike treatment, preventive therapies are offered to those who currently are healthy to reduce the risk of illness in the future. During the evaluation of new candidates for breast cancer chemoprevention, including raloxifene and aromatase inhibitors,16 their potential adverse effects will have a powerful influence on women's acceptance of and adherence to these agents.
The authors thank John Robbins, M.D., the staff of the University of California–Davis Women's Health Initiative, and the Health Education Council for invaluable assistance with recruitment of study participants; Michael DeGregorio, Pharm.D., for his extensive knowledge about tamoxifen and creative ideas about how to explain potential benefits and harms; Anna Romero, F.N.P., Laura Gutierrez, P.A., and Gwen Moore for conducting interviews; Daniel Tancredi, M.S., for advice on the statistical analysis; and Mairin Rooney for coordinating the study.
- 8Chemoprevention of breast cancer. A joint guideline from the Canadian Task Force on Preventive Health Care and the Canadian Breast Cancer Initiative's Steering Committee on Clinical Practice Guidelines for the Care and Treatment of Breast Cancer. CMAJ. 2001; 164: 1681–1690., , , et al.
- 12Breast Cancer Risk Assessment Program. Available from URL: http://bcra.nci.nih.gov/brc/ [accessed April 2005].
- 14Model selection and multi-model inference. A practical information-theoretic approach, 2nd ed (2002), 3rd printing. New York: Springer, 2004., .