JOURNAL OF WOMENS HEALTH Volume 15, Number 7, 2006 Mary Ann Liebert, Inc.

Healthy Women with a Family History of Breast Cancer: Impact of a Tailored Genetic Counseling Intervention on Risk Perception, Knowledge, and Menopausal Therapy Decision Making
ELLEN T. MATLOFF, M.S.,1 ANNE MOYER, Ph.D.,2 KRISTEN M. SHANNON, M.S.,3 KRISTIN B. NIENDORF, M.S.,3 and NANANDA F. COL, M.D.4

ABSTRACT Background: Women with a family history of breast cancer have several menopausal therapy options, including tamoxifen, hormone therapy (HT), alternative medications, or no treatment. This complex decision should be based on each womans risk to develop breast cancer, menopausal symptoms, preferences, and risks for other conditions. The authors determined the effects of a personalized risk assessment and genetic counseling intervention on knowledge, risk perception, and decision making in a group of healthy women who had a first-degree relative with breast cancer. Methods: Forty-eight cancer-free menopausal women age 40 years who had at least one first-degree relative with breast cancer were randomized to a genetic counseling intervention or control. Intervention participants were given a personalized risk assessment for breast cancer, heart disease, osteoporosis, and uterine cancer based on family history and personal health data. Knowledge, risk perception, and medication usage were measured at baseline, 1 month, and 6 months. Results: Knowledge was higher in the intervention group at both follow-up time points postintervention. Perceived risk for developing breast cancer was significantly lower and more accurate in the intervention group at 1 and 6 months postintervention than at baseline, as was perceived risk of developing heart disease. Although the counseling intervention did affect both knowledge and risk perception, overall, both groups were reluctant to take any form of menopausal therapy. Conclusions: A personalized risk assessment and genetic counseling intervention improves patient knowledge and risk perception; however, it is unclear that the intervention influenced menopausal treatment decisions.

1Yale

University School of Medicine/Yale Cancer Center, New Haven, Connecticut. of Psychology, State University of New York at Stony Brook, Stony Brook, New York. 3Massachusetts General Hospital/Center for Cancer Risk Analysis, Boston, Massachusetts. 4Brown University Medical School and Rhode Island Hospital, Providence, Rhode Island. This study was funded by a grant from the Susan G. Komen Foundation.
2Department

843

844

MATLOFF ET AL.

INTRODUCTION

is pervasive in our society. It is well known that women overestimate their risk not only of developing the disease but also of dying from it.1 Highrisk women now have the option of taking the chemopreventive agent tamoxifen to reduce their risk of developing breast cancer.2 Tamoxifen has been shown to reduce the risk of incident breast cancer by 50% in healthy high-risk women.3 Tamoxifen carries its own risks, however, including an increased risk of endometrial cancer, thromboembolic events, hot flashes, and vaginal discharge. A second medication, raloxifene, has been evaluated for breast cancer prophylaxis through the STAR trial and is as effective as tamoxifen in reducing the risk of invasive breast cancer, but has a lower associated risk of thromboembolic events and cataracts.4 Aromatase inhibitors have also been considered for prophylaxis.5 All women approaching menopause have the option of taking menopausal hormone therapy (HT). HT reduces menopausal symptoms, such as hot flashes and night sweats, and helps to prevent osteoporosis.6 The topic of HT has become controversial since the results of the Womens Health Initiative (WHI) trial were released showing that combination HT (estrogen plus progestin) increased the risk of breast cancer and cardiovascular disease (CVD).7 However, the overall risks of breast cancer in women who take short-term HT are low, and this medication is still an option for menopausal women, particularly those who are experiencing severe menopausal symptoms that are unresponsive to nonhormonal therapies. The choice, therefore, of whether to take a medication that will reduce the risk of breast cancer or one that reduces menopausal symptoms or to take no intervention at all is complex for many women. Women with a family history of breast cancer are known to overestimate their own risks of developing the disease,8 although many of them are not at substantially elevated risk.9 This is especially true if they have had no abnormal biopsy findings and their family history does not include known hereditary cancer mutations, multiple affected relatives, early-onset disease, ovarian cancer, or Jewish ancestry.10 Women with a family history of breast cancer comprise a wide range of risk for developing the disease. Accurate risk assessment for breast cancer is complex yet crucial to decision making about appropriate
HE FEAR OF DEVELOPING BREAST CANCER

menopausal therapy. An informed decision should be based on each womans personal risks for breast cancer, heart disease, and osteoporosis, as well as her own preferences and menopausal symptoms.11 The minimum eligibility requirement for tamoxifen based on the National Surgical Adjuvant Breast and Bowel Project (NSABP) is a 5-year estimated risk of developing breast cancer of 1.67%.3 Using this threshold, 18% of women in the community setting and 37% of women between the ages of 60 and 70 have a risk high enough to make them eligible for prophylactic tamoxifen.12 However, only 1 in 4 women meeting these minimum eligibility requirements would likely have a favorable benefit/risk ratio for taking the medication.13 Each provider would need to screen 5 or 6 women between the ages of 40 and 70 years to find 1 woman at high enough risk to be eligible for counseling about breast cancer chemoprevention.13 This is an arduous task in a busy primary care setting, especially considering that counseling should include a detailed review of personal and family history as well as the risks, benefits, and limitations of each menopausal option. Several studies have examined the impact of cancer genetic counseling on knowledge, risk perception, and surveillance and risk reduction behaviors in patients known to be at high risk to develop breast cancer based on a strong family history of the disease.1416 Overall, data from these studies suggest that genetic counseling increases knowledge of cancer genetics but does not change risk perception in these high-risk women.16,17 In fact, women at high risk for hereditary breast and ovarian cancer have shown resistance to accurate risk information, even after receiving detailed genetic counseling.16,18 It appears that personal experiences and beliefs about cancer risk may carry more weight for these highrisk patients than the scientific data used by the healthcare providers in risk assessment.1921 Considering that most of these high-risk women have strong family histories of breast or ovarian cancer or both and have likely experienced cancer diagnosis, treatment, and death in family members, it is not surprising that their life experiences have had a profound and lasting impact on their beliefs about cancer and risk. Risk perceptions constructed from such experiences are often resistant to change.19 Women with striking family histories of breast or ovarian cancer, or both, who are at high risk

MENOPAUSAL THERAPY DECISION MAKING

845

of carrying a genetic mutation represent the minority of patients with a family history of breast cancer. We hypothesized that women who are at 10% risk to carry a genetic mutation and who are, therefore, more representative of the average woman with a family history of breast cancer may be more receptive to genetic counseling about menopausal therapy. This population of women would more likely be candidates for several menopausal therapy options compared with BRCA1/2 carriers, for whom the high risk of breast cancer would likely direct all treatment decisions. The theoretical framework used in this study draws from concepts described in the Health Belief Model (HBM)22 and the Theory of Planned Behavior (TPB).23 The HBM is an approach to understanding preventive health behavior based on an individuals perceived health beliefs.24 This model has been applied to genetic counseling and testing in previous studies,25,26 as well as to the use of HT.24 The HBM can help explain why healthy individuals may be resistant to preventive health behaviors even when presented with their risks and options24 and was, therefore, an appropriate framework for this study. The framing of the output of the Markov model drew upon concepts from the TPB23 and descriptive frameworks that use expectations, values, and other determinants of decision-making behavior to structure decision support.2729 There is also a well-described reluctance for patients to use medications, in general; it appears that many people have a negative perception of traditional medicines and believe them to be generally harmful substances that are overused by physicians.30,31 This reluctance has been shown to apply to both HT and tamoxifen use32,33 and is, therefore, a known barrier in presenting patients with a wellbalanced discussion of their menopausal therapy options. We sought to examine if a personalized risk assessment and genetic counseling intervention developed for this pilot study would affect knowledge, risk perception, and decision making in a group of women who had one first-degree relative with breast cancer, as compared with a control group (who received no intervention). We predicted that in this low-risk to moderate-risk cohort interested in menopausal therapy options, this intervention would result in increased knowledge and more accurate risk perception and would lead to less decisional conflict. We also

hypothesized that such an intervention would change decision intentions.

MATERIALS AND METHODS

This study was conducted with approval from the Institutional Review Boards of Yale School of Medicine and Partners Health Care System (MGH). Women who were 40 years of age, had at least one first-degree relative with breast cancer, had gone through natural menopause (no menses for 6 months or no menses plus menopausal symptoms), and were not currently taking a menopausal therapy were invited to participate in this randomized trial. Women with at least one first-degree relative with breast cancer are encouraged to seek more detailed risk assessment to see if they are candidates for the STAR trial34; therefore, by design, all women eligible for our study had a breast cancer risk high enough to warrant more detailed risk assessment and possible consideration for breast chemoprevention. Exclusion criteria included having had cancer, atypical hyperplasia, or lobular carcinoma in situ (LCIS); being a known carrier of a BRCA1 or BRCA2 mutation; and having heart disease. Women with a family history that placed them at high ( 10%) risk of carrying a mutation in a hereditary breast and ovarian cancer gene based on genetic counselor assessment were not eligible for this study and were referred for clinical genetic counseling. Participants were recruited through study advertisements in physician offices, institutional listservs (e-mail lists), and newspaper articles between August 2002 and January 2004 and were randomized to either an intervention or control arm upon recruitment (every other patient) (Fig. 1). The stated goal on recruitment materials was to examine if a personalized counseling session would help women at increased risk to develop breast cancer make a decision about which menopausal therapy (such as HT or tamoxifen), if any, would be best for them. The newspaper articles generated an overwhelming number of responses at the New Haven office, and interested parties contacted our program secretary directly. Those patients who were not deemed eligible through telephone screening were not tracked. Participants in the control arm received no intervention and were offered the counseling intervention after they completed all three questionnaires over 6 months.

846

MATLOFF ET AL.

FIG. 1.

This chart illustrates the flow of study participants for this genetic counseling trial.

Informed consent was obtained from all participants and each completed a detailed intake questionnaire. Intervention participants physicians completed an intake form with health information, including height, weight, pulse, blood pressure, bone mineral density (BMD), and lipid profile. The protocol included two 60-minute study visits. Participants handed in the baseline assessment at visit 1, and a detailed three-generation pedigree and additional health information were elicited from each woman by a certified genetic counselor. This counselor, using a standardized counseling flipbook developed for this study, reviewed detailed information about menopause, the risks and benefits of each menopausal ther-

apy option, and the factors that influence disease risk assessment. The associated side effects of each menopausal therapy included symptoms, chemopreventive effects, short-term and longterm risks, and magnitude of such risks. The health information was entered into a patientspecific decision analytic Markov model that included embedded regression equations predicting a womans future risks for breast cancer, heart disease, osteoporosis, and endometrial cancer.11,35 This decision model was used to calculate each individuals future risks (10-year and lifetime) for breast and endometrial cancer, heart disease, and osteoporosis, which was presented to the patient at visit 2, which was held an average of 34 weeks later. These risks were presented

MENOPAUSAL THERAPY DECISION MAKING

847

using text and simple graphics (Fig. 2). Absolute risks were provided, and a balanced discussion of all options was conducted.36 The risk assessment was tailored to each patient by the genetic counselor based on the three-generation pedigree, personal risk factors, and preferences. For example, a patient whose mother was diagnosed with breast cancer at age 80 and had many unaffected female relatives would be counseled that her actual risk to develop breast cancer was lower than that of a patient whose mother was diagnosed with breast cancer at age 40.37a A patient whose sister developed breast cancer after receiving radiation to the chest wall as treatment for lymphoma would likely have a lower risk to develop breast cancer than a patient whose sister developed breast cancer in the absence of outside risk factors. The GAIL model does not take such data into account, and for these reasons, using a genetic counselor to help frame risk was helpful in this setting.37b The counseling sessions were based on a counseling script but allowed flexibility based on the patients questions, concerns, and personal history. Each counseling participant received a personalized letter summarizing her data that was

copied to her physicians, with her permission. The genetic counselor was also available to discuss results with each physician by phone if the physician had questions. Counseling participants and controls received follow-up questionnaires at 1 month and 6 months after intervention.

Demographics
Age, marital status, education, ethnicity, and household income were assessed using self-reported measures. Ethnicity was obtained on participants, including absence or presence of Jewish ancestry, because these data are essential to hereditary breast cancer risk assessment.

Accuracy of knowledge about menopause and menopausal therapy

A 30-item true/false questionnaire was developed to assess knowledge of menopause; risk factors for breast cancer, heart disease, and osteoporosis; and the risks and benefits of HT, tamoxifen, raloxifene, and no treatment (available from the authors on request). Items were based on current information and statistics published by the Hormone Foundation (the public

FIG. 2. This type of graph was used to present each participant with her 10-year and lifetime risks for breast and endometrial cancer, heart disease (CHD), and osteoporosis (hip Fx).

848

MATLOFF ET AL.

education affiliate of the Endocrine Society) and the National Cancer Institute (NCI).38,39 Some items were also modified from a measure assessing knowledge of osteoporosis.40 Sample items included during menopause the ovaries stop producing estrogen and progesterone (true) and tamoxifen increases the risk of breast cancer (false). A not sure option was available to reduce the occurrence of guessing. The average internal consistency at the three time points in this sample was acceptable, 0.74.

Decisional conflict
Decision uncertainty and factors contributing to uncertainty regarding menopausal therapy decision making were assessed with the Decisional Conflict Scale.44 Items are rated on a 5-point scale, ranging from 1 (strongly agree) to 5 (strongly disagree). The scale includes subscales assessing decision uncertainty (e.g., Im unsure what to do in this decision), factors contributing to uncertainty (e.g., I feel I know the risks and side effects of menopausal therapy), and perceived effective decision making (assessed at 1-month and 6-month follow-ups only; e.g., I feel I have made an informed choice). Ratings are summed and divided by the number of items to form an average score out of 5. The scale has shown good testretest reliability (0.81), internal consistency (0.780.92), and predictive validity (discriminated significantly between those who accepted/rejected and those who delayed/were unsure regarding an invitation to be immunized or screened).44 The average internal consistency over the three time points in this sample was good, 0.80.

Perceived risk of breast cancer, heart disease, and osteoporosis

Participants were asked how likely they thought they were to develop breast cancer, heart disease, and osteoporosis some time during their life on a scale from 0% (not at all likely) to 100% (extremely likely) based on a measure developed by Valdimarsdottir et al.41,42 Participants were also asked how likely they thought the average woman was to develop one of these conditions at some time during her life on the same scale. The measure of Valdimarsdottir et al. has been used successfully in previous samples.41,42 The perceived risk measures have shown high test-retest reliability (0.85) and criterion validity (perceived risk mediated the effect of a parent dying of cancer on intrusive thoughts and avoidance about breast cancer).41 We constructed a self/average perceived risk discrepancy score by subtracting participants risk estimate for the average woman from the estimate they made for themselves. Thus, positive self/average perceived risk discrepancy scores indicated that participants felt at elevated risk relative to the average woman, and negative scores indicated that they felt at decreased risk relative to the average woman.

Medication usage
Participants self-reported on their questionnaires what medications they were currently taking, and decision intentions were measured by asking whether they thought they would be likely to take HT, tamoxifen, raloxifene, another medication, or no medication in the next 6 months (yes or no).

Satisfaction with the counseling intervention

We created questions to evaluate the satisfaction of participants with this particular counseling intervention. Satisfaction with the counseling intervention was measured by self-report on the questionnaires in terms of the information received, support received, the amount of time involved, and the convenience of the counseling session. We used a scale from 0 (not at all satisfied) to 100 (extremely satisfied) for each of these measures.

Worry about breast cancer, heart disease, and osteoporosis

How worried women were about developing breast cancer, heart disease, and osteoporosis was assessed with a single-item measure modified from Diefenbach et al.43 This assessment has been shown to predict mammography screening in women with a family history of breast cancer. Levels of worry during the past month were rated on a scale ranging from 1 (not at all) to 4 (almost all of the time).

Statistical analyses
Data were screened for extreme skewness and kurtosis ( 2), and transformations were applied where appropriate. t tests were used to test for

MENOPAUSAL THERAPY DECISION MAKING

849

equivalence between the randomized groups at baseline on demographic and outcome variables for continuous variables, and chi-square tests were used for categorical variables. The primary statistical procedure was analysis of covariance (ANCOVA), controlling for baseline values, on the outcomes of knowledge, risk perception, and decisional conflict conducted for 1-month and 6month outcomes separately. Baseline values were used as covariates to control for nonsignificant chance variation between groups at baseline for separate analyses of outcomes at 1-month and 6month follow-ups. Chi-square or Fishers exact test was used for categorical outcome variables. For the variable that did not have baseline assessments, perceived effectiveness of decision making, t tests were used to test for betweengroup differences. Paired t tests were used to analyze preintervention to postintervention changes in variables assessed in the intervention group only. Pearson correlation coefficients were used to examine interrelationships among study variables. To adjust for multiple between-group comparisons, a familywise Bonferroni-corrected alpha level of 0.008 was used for the analyses examining group differences in perceived risks, self/average risk discrepancy, and self/actual risk discrepancy for breast cancer, heart disease, and osteoporosis; medication usage; and decisional conflict. Analyses are confined to the 48 participants who completed assessments at all three time points (23 in the intervention group and 25 in the control group). Participants were excluded from analyses for which they had missing data, and no imputations were conducted.

ried (71%), well educated (83% had completed college, graduate school, or professional school), and affluent (58% had an annual household income, before taxes, of $75,000). In terms of ethnicity, 54% were of European, non-Ashkenazi Jewish ancestry, 21% were of Ashkenazi Jewish ancestry, 8%, were of both Ashkenazi Jewish and European non-Ashkenazi Jewish ancestry, 2% were of Latin American/Caribbean ancestry, and an additional 4% were of other ancestries, such as European and Native American. Ethnicity information was unavailable for 10% of participants, who either did not know or did not report their ancestry. No significant differences were noted between the counseling and control arms of this study in terms of demographics, knowledge, perceived risk, decisional intentions, or decisional conflict. The average lifetime risk for the intervention group for breast cancer was 20% (range 16%31%), for heart disease was 13% (range 4%22%), and for osteoporosis was 24% (range 11%54%).

Accuracy of knowledge about menopause and menopausal therapy

The intervention group had higher levels of knowledge about menopause and menopausal therapy at both follow-up time points, with significantly more questions out of 30 answered correctly (F[1, 44] 21.24, p 0.001, p2 0.33 and F[1,44] 23.83, p 0.001, p2 0.35, respectively) (Table 1). The proportion of these true/false questions answered correctly by the intervention group rose from 60% at baseline to 73% at follow-up. The proportion of questions answered not sure diminished from 32% to 13%.

RESULTS Characteristics of study participants

Of the 48 participants who completed all questionnaires and both counseling sessions, the mean age was 49 years (range 4155). Most were marTABLE 1. Outcome Number of knowledge questions answered correctly (out of 30) Counseling intervention Control ACCURACY
OF

Perceived risk of breast cancer, heart disease, and osteoporosis

Participants perceived risk for developing breast cancer was significantly higher in the control group at 1-month (F[1, 44] 14.08, p 0.008,
AND

KNOWLEDGE ABOUT MENOPAUSE

MENOPAUSAL THERAPY 1 month M (SD) 22.18 (3.4) 19.0 (3.8) 6 months M (SD) 22.5 (3.5) 18.6 (3.7)

Baseline M (SD) 17.5 (4.2) 18.0 (3.8)

850

MATLOFF ET AL.

0.24) and 6-month F[1, 44] 8.48, p 0.008, p2 0.16) follow-up (Table 2). Their perceived risk of developing heart disease was also significantly higher at 1 month (F[1, 44] 13.14, p 0.008, p2 0.23) and 6 months (F[1, 44] 9.41, p 0.008, p2 0.18). There were no significant group differences for perceived risk of developing osteoporosis at either time point (F[1, 44] 6.71, p 0.008, p2 0.13 and F[1, 44] 6.14, p 0.008, p2 0.12). At baseline, all participants thought they were more likely than the average woman to get breast cancer during their lifetime and less likely to develop heart disease or osteoporosis (Table 2). The intervention group was no less likely to rate their own risk for breast cancer as higher than that of the average woman than was the control group at 1 month (F[1, 43] 7.52, p 0.008, p2 0.15) and 6 months (F[1, 43] 4.43, p 0.008, p2 0.09). Conversely, the intervention group continued to estimate that they were at less than the average womans risk to develop heart disease at 1 month and 6 months, whereas the control group
p

did not (Table 3). This difference was significant only at the 1-month follow-up time point (F[1, 43] 7.78, p 0.008, p2 0.15; at 6 months 2 0.15). For osteoporosis, both groups conp tinued to estimate overall that they were less likely than the average women to develop this condition, and there were no significant group differences at either follow-up point ( p2 0.02 and 0.03). The self/average discrepancy risk scores measure the patients subjective feelings about risk, but not the accuracy of their risk perceptions. Among women in the counseling intervention, for whom actual risk estimates for breast cancer, heart disease, and osteoporosis were calculated, we tested whether womens estimates became closer to the risks that were conveyed to them during the counseling. Initially, participants substantially overestimated their risks, from 17% for osteoporosis to 37% for breast cancer (Table 2). Paired t tests indicated that the self/actual risk discrepancy for breast cancer was significantly lower than baseline at 1 month (t[21] 3.10, p
SELF/ACTUAL RISK DISCREPANCIES Baseline M (SD) 1 month M (SD) 6 months M (SD)

TABLE 2. Outcome

PERCEIVED RISKS, SELF/AVERAGE RISK DISCREPANCIES,

AND

Perceived risk of developing Breast cancer Counseling intervention Control Heart disease Counseling intervention Control Osteoporosis Counseling intervention Control Discrepancy between perceived risk for self and average woman fora Breast cancer Counseling intervention Control Heart disease Counseling intervention Control Osteoporosis Counseling intervention Control Discrepancy between perceived risk and estimated risk forb Breast cancer Counseling intervention Heart disease Counseling intervention Osteoporosis Counseling intervention
aPositive bPositive

57.8 (20.4) 57.7 (20.6) 49.3 (21.7) 46.1 (24.2) 43.0 (25.0) 41.2 (22.7) 16.3 (17.9) 22.3 (24.3) 4.7 (21.5) 2.3 (23.8) 11.3 (14.4) 4.0 (17.8) 36.9 (20.4) 35.0 (22.2) 17.6 (26.1)

39.3 (26.8) 63.7 (21.3) 33.6 (25.5) 52.1 (24.5) 34.7 (25.1) 45.4 (21.5) 0.8 (22.3) 21.1 (25.4) 14.9 (20.9) 2.2 (26.1) 13.9 (20.4) 5.9 (14.9) 18.9 (28.6) 16.0 (24.2) 9.2 (16.0)

41.0 (24.9) 58.8 (20.9) 37.2 (20.2) 48.4 (26.9) 37.0 (25.7) 46.0 (20.2) 3.6 (20.1) 18.3 (23.0) 14.2 (19.1) 1.3 (28.5) 13.7 (24.9) 1.4 (18.4) 17.1 (25.9) 12.3 (17.0) 6.4 (13.2)

values indicate higher perceived risk for self than for the average woman. values indicate higher perceived risk for self than the estimated risk for that patient.

MENOPAUSAL THERAPY DECISION MAKING

TABLE 3. Usage of Hormone replacement Yes Counseling intervention Control No Counseling intervention Control Tamoxifen Yes Counseling intervention Control No Counseling intervention Control Raloxifene Yes Counseling intervention Control No Counseling intervention Control USE
OF

851
AND

HT, TAMOXIFEN, Baseline n (%)

RALOXIFENE 1 Month n (%) 6 Months n (%)

0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0)

1 (4) 1 (4) 22 (96) 23 (96) 0 (0) 0 (0) 23 (100) 24 (100) 0 (0) 0 (0) 23 (100) 24 (100)

2 (9) 2 (8) 21 (91) 23 (92) 0 (0) 0 (0) 23 (100) 25 (100) 0 (0) 0 (0) 23 (100) 25 (100)

0.008, d 0.81) and at 6 months postintervention t[21] 3.10, p 0.008, d 0.77). This was also the case for heart disease (t[21] 3.1, p 0.008, d 0.67 and t[21] 3.39, p 0.008, d 0.57) but not for osteoporosis (t[21] 2.68, p 0.008, d 0.36 and t[21] 2.25, p 0.008, d 0.23) at both follow-up points. However, women in the intervention group continued to overestimate their risks in comparison to what was conveyed to them during counseling.

p 0.001, p2 0.31, controlling for baseline levels). For the subscale assessing perceived effective decision making, assessed only at the follow-ups, scores for the counseling group were superior only at the 6-month follow-up (M 2.0, SD 0.6, and M 2.7, SD 0.9; t[46] 2.87, p 0.008, d 0.35) but not at the 1-month follow-up (M 2.2, SD 0.8 and M 2.5, SD 0.9; t[43] 1.19, p 0.05, d 0.90).

Medication usage Decisional conflict

Scores on the subscales of the Decisional Conflict Scale are expressed so that lower values indicate less decisional conflict. For the Decision Uncertainty Subscale, there were no group differences, controlling for baseline levels, at either the 1-month (M 3.2, SD 1.1, and M 3.4, SD 0.9 for the counseling and control group, 2 respectively; F[1,40] 0.41, p 0.05, p 0.01) or the 6-month (M 3.0, SD 1.1, and M 3.1, SD 0.9; F[1, 41] 0.02, p 0.05, 2 0.00) follow-up point. For the Factors p Contributing to Uncertainty Subscale, the counseling group was superior at both the 1-month (M 1.8, SD 0.5, and M 2.7, SD 1.1; F[1, 40] 19.00 p 0.001, p2 0.32) and 6-month follow-up (M 1.9, SD 0.5; F[1, 42] 19.00, As shown in Table 3, overall, the sample was reluctant to take HT, tamoxifen, and raloxifene. However, they were somewhat more willing to take other medications (data not shown). These other medications, not all necessarily for reducing menopausal symptoms, were compounded progesterone, dehydroepiandrosterone (DHEA), antibiotics, agents to control lipid levels, alendronate, pain relievers, sleep aids, decongestants, antidepressants, diabetes medications, vitamin supplements, and such herbals as black cohosh. There were no group differences at either followup point in the proportion indicating that they were taking HT (Fishers exact test, p 0.98 and p 1.00), raloxifene, or tamoxifen (statistics not calculated because proportion indicating yes was 0 in each group).

852

MATLOFF ET AL.

Satisfaction with counseling intervention

On a self-reported scale of 0100, participants in the intervention group reported being very satisfied with the counseling they received in terms of the information they received (M 85.64, SD 16.93), the support they received (M 80.13, SD 21.87), and the time involved (M 80.39, SD 19.95), but somewhat less with the convenience of the session (M 76.61, SD 25.27). Forty percent of patients (10) in the control group opted to come in for the counseling intervention after their study questionnaires were completed.

DISCUSSION
The data from this randomized, controlled study indicate that a personalized risk assessment and counseling intervention offered by genetic counselors for menopausal women with a family history of breast cancer can significantly increase knowledge about menopause and menopausal therapy options, improve the accuracy of risk perception, decrease decisional conflict, and is associated with overall satisfaction with the intervention. This increased knowledge makes it possible for women to make a more informed decision about which menopausal therapy, if any, to use. The counseling intervention developed was flexible enough to accommodate integration of new data as they emerged, and the counseling script was updated when new information was published (but before our study began). This flexibility is crucial, as data in this area are emerging constantly. The development of new chemopreventive medications in the future may make the decision-making process even more complex, adding new options, each with its own risks and benefits. The flexibility of our counseling model adds to the utility of this approach by facilitating the rapid integration of new information into the model. Women in the counseling group had risk perceptions for breast cancer, heart disease, and osteoporosis that were closer to their own personalized risks at 1 month and 6 months postintervention than at baseline. However, it is noteworthy that although womens estimates of their own risks moved closer to their predicted risk estimates, the sample still overestimated their own risks of developing breast cancer. This resistance to risk information is similar to that published elsewhere.18 Although the intervention resulted in more accurate risk perceptions for breast cancer and heart disease at both 1-month and 6-months follow-up, this more accurate risk perception did not appear to change actual usage of menopausal therapy, at least within the 6 months after the counseling intervention. This is somewhat unexpected because these women continued to overestimate their own risks of breast cancer and may, therefore, have been predicted to be more likely to opt for tamoxifen or raloxifene. However, the sample as a whole was reluctant to use tamoxifen, raloxifene, or HT, even with more accurate informa-

Intercorrelations among study variables

Table 4 shows the correlations among perceived risk, actual risk, and worry for breast cancer, heart disease, and osteoporosis and use of medications at 6 months after the counseling intervention (because no participants were taking tamoxifen or raloxifene, no relationships could be calculated for these medications). The pattern of relationships is quite consistent across the three health conditions. Whereas perceived risk and worry were significantly correlated, actual risk and worry and perceived risk and actual risk were not. In addition, use of HT was not associated with any of these variables, with the exception of actual risk for osteoporosis. As this finding was only significant at the 0.05 level, it should be viewed cautiously in the presence of these multiple comparisons.

TABLE 4. INTERCORRELATIONS AMONG STUDY VARIABLES AT 6 MONTHS Outcome Breast cancer 1. Perceived risk 2. Actual risk 3. Worry 4. HT use Heart disease 1. Perceived risk 2. Actual risk 3. Worry 4. HT use Osteoporosis 1. Perceived risk 2. Actual risk 3. Worry 4. HT use *p 0.05; **p 1 2 0.16 3 0.45*** 0.01*** 0.69*** 0.29*** 0.70*** 0.00*** 4 0.08* 0.06* 0.17* 0.15* 0.24* 0.16* 0.14* 0.43* 0.22*

0.21

0.05

0.01; ***p

0.001.

MENOPAUSAL THERAPY DECISION MAKING

853

tion about their personal risks and the risks, benefits, and limitations of each intervention. Several studies have shown that uptake of tamoxifen, even in the high-risk setting, is relatively low.4547 Melnikow et al.47 found that only 17.6% of women eligible for tamoxifen were inclined to take the medication after receiving a 15-minute standardized educational intervention about its risks and benefits, even though these women had greatly overestimated their own risks to develop breast cancer within the next 5 years.47 Very highrisk women in this cohort were not more likely to choose tamoxifen than were low-risk women.47 Port et al.33 found that 95.3% of women who qualified for prophylactic tamoxifen declined to take the medication, even though this group of women also had elevated perceived risks of developing breast cancer. There was no increase in patient acceptance of the medication even after a balanced presentation of the risks and benefits from a physician.33 The most commonly cited reason for refusing tamoxifen was fear of side effects.33 The reluctance to take tamoxifen in the current study is not simply a reluctance to take medications, in general, as the sample was more willing to take other medications, including statins, bisphosphonates, antidepressants, and such herbals as black cohosh, for which limited safety and efficacy data are available. This may be because medication adherence is related to personal perception of the necessity of the medication, concerns about adverse side effects, and the patients perception of the risk/benefit ratio.30,4850 Perhaps patients who have personally had a test finding demonstrating high blood pressure or low bone density are more likely to take a medication to treat that finding than a patient considering prophylactic treatment based on a risk assessment. Patients may also feel that complementary treatments are more natural and are, therefore, safer30 and may be more likely to treat menopausal or other symptoms with herbal medications. It is possible that this reluctance was, in part, due to the fact that the personalized risk assessment findings were presented to participants in a nondirective manner by a genetic counselor and not as a recommendation for one menopausal therapy by a physician. It has been noted that women are more likely to take a chemopreventive medication when they receive a recommendation to do so from their physician.45,46 It is also important to note that none of the subjects in this

study had a personal finding (e.g., atypical hyperplasia) placing them at increased risk for breast cancer. In fact, such a personal finding would have biased their risk assessment toward one of the menopausal options, and these participants were, therefore, excluded from entry into the study. Bober et al.46 found that women who had had an abnormal breast biopsy were more likely to take tamoxifen. It was hypothesized that women who have an abnormal breast finding appear to feel vulnerable and may, therefore, be more open to chemoprevention.46 Concurring with this hypothesis, Port et al.33 reported that of the 2 women willing to take tamoxifen prophylactically in their study, both already had a personal finding (LCIS and atypical hyperplasia) placing them at increased risk for breast cancer. Reluctance to take HT has also been described in detail.32 Overall, it appears that patients perceptions of the benefits of HT are often countered by concerns about the potential risks, even when these beliefs are at odds with the clinical evidence.32 Patient reluctance to take medications for medical conditions with which they have been diagnosed has been well documented.48 On top of this baseline reluctance, the WHI findings demonstrating that the risks of HT outweighed the benefits for the average menopausal woman were published and reported during the recruitment phase of our study. We incorporated these data into our counseling protocol and flipbook. The WHI data were widely disseminated to patients by the media and the medical community and have had a significant impact on overall use of HT in postmenopausal women.51 It is possible that the WHI data and other outside influences played a role in patients knowledge, concerns, and opinions about the menopausal options discussed in this study. A model using genetic counselors to provide this type of in-depth risk assessment and counseling has both strengths and weaknesses. Genetic counselors are expert at eliciting and analyzing three-generation pedigrees, making it possible to assess the patients personal and hereditary risks for cancer, heart disease, and other genetic conditions that may impact decision making for menopausal therapy. The genetic counselor is in a position to estimate whether the risks generated by standard models, such as the GAIL model, may actually represent upperbound or lower-bound risk estimates based on family history. Genetic counselors are also able to

854

MATLOFF ET AL.

explain and recommend genetic testing, when appropriate, that may elucidate the actual risks for the patient. For example, one Jewish woman in this study reported a sister diagnosed with breast cancer at age 48 and no other significant family history. After the patient learned of the risk factors for hereditary breast cancer in visit 1, she further researched her family history and learned of a distant paternal history of breast cancer. The patients father agreed to have testing for the common Jewish BRCA1 and BRCA2 mutations and was found to be a BRCA1 carrier. Our patient was tested and did not carry this mutation, making her a true negative whose risks of breast and ovarian cancer were lowered to that of other women in the general population. This patient was counseled that she was no longer at increased risk to develop breast cancer and did not meet the minimum criteria for tamoxifen. The combination of using decision models, pedigree analysis, and genetic counseling and testing allowed the patient to receive the menopausal therapy most appropriate for her situation. A potential weakness of using genetic counselors in this model is that genetic counselors cannot prescribe medications; therefore, after the counseling session, the patients had to return to their referring physicians to receive a prescription. The counselor did send each patient a detailed summary letter outlining her risks and recommendations, copying the referring physician whenever permission was granted. However, this extra step in the process could potentially be burdensome to the patient and could be a barrier in healthcare service delivery. Patients may also be reluctant to follow the advice of a genetic counselor who does not have a medical degree and with whom they do not have a long-term relationship. For these reasons, it may be best if the genetic counselor and the referring physician use a team approach in counseling these patients. One limitation of our study is that our sample represents self-selected women who were highly motivated and actively seeking knowledge about menopausal therapy options. The majority of these women were white and affluent and likely had increased access to medical information and services as compared with the general population. Women with less education and exposure to this topic likely have even less knowledge about their risks and menopausal options. It is also possible that women attracted to our study were experiencing more menopausal symptoms than wo-

men in the general population and were, thus, more concerned about their options. The recruitment method we used (having interested participants contact the study) likely self-selected participants who were interested in this topic. The follow-up of only 6 months is another limitation, as is the measurement of intent to take medication instead of actual usage. High dropoff rates of prophylactic tamoxifen use have been demonstrated in other trials3; therefore, long-term actual usage would be an important end point in future studies and would require at least 5 years of clinical follow-up. The low dropout rate in our study (Fig. 1) is notable in that many participants traveled over an hour and took time off from work to participate in this study. The interest generated by this study suggests that menopausal decision making is of concern for educated women who have a first-degree relative with breast cancer. Although this counseling intervention did not appear to affect decision intentions, women in the counseling group were satisfied with the knowledge and support they received in the counseling intervention. Informed patients are more likely to participate actively in their care, make better decisions, come to a common understanding with their physicians, and adhere more fully to treatment.36 In our study, a high proportion (40%) of women in the control group opted to come in for the counseling intervention after their study questionnaires were completed. These two 60minute sessions represented a significant investment of time for all participants. These women reported that they were confused about which menopausal therapy would be best for them and desired more information before making such an important decision. It appears that this counseling intervention fulfilled an important need for these women, whether or not it influenced their menopausal therapy decision making. In summary, a personalized risk assessment and genetic counseling intervention increased knowledge and resulted in more accurate disease risk perception in menopausal women at increased risk to develop breast cancer. Satisfaction with this intervention was high, and decisional conflict was decreased. However, it is unclear if this intervention influences the decision to take tamoxifen, raloxifene, or HT. The most crucial finding from our study is that even women at relatively low risk to develop breast cancer overestimate their own risks of developing the disease and are yet reluctant to use

MENOPAUSAL THERAPY DECISION MAKING

855
8. Daly M, Lerman C, Ross E, Schwartz C, Masny A. Gail model breast cancer risk components are poor predictors of risk perception and screening behavior. Breast Cancer Res Treat 1996;41:59. 9. Claus E, Risch N, Thompson W. Autosomal dominant inheritance of early-onset breast cancer. Cancer 1994;73:643. 10. Couch F, DeShano M, Blackwood M, et al. BRCA1 mutations in women attending clinics that evaluate the risk of breast cancer. N Engl J Med 1997;336: 1409. 11. Col N, Eckman M, Karas R, et al. Patient-specific decisions about hormone replacement therapy in postmenopausal women. JAMA 1997;277:1140. 12. Brewster A, Christo D, Lai H, Helzlsouer K. Breast carcinoma chemoprevention in the community setting. Cancer 2005;103:1147. 13. Fabian C, Kimler B. Selective estrogen-receptor modulators for primary prevention of breast cancer. J Clin Oncol 2005;23:1644. 14. Lerman C, Narod S, Schulman K. BRCA1 testing in families with hereditary breast-ovarian cancer: A prospective study of patient decision making and outcomes. JAMA 1996;275:1885. 15. Kash K, Holland J, Halper M, Miller D. Psychological distress and surveillance behaviors of women with a family history of breast cancer. J Natl Cancer Inst 1992;84:24. 16. Lloyd S, Watson M, Waites B, et al. Familial breast cancer: A controlled study of risk perception, psychological morbidity and health beliefs in women attending for genetic counselling. Br J Cancer 1996; 74:482. 17. Braithwaite D, Emery J, Walter F, Prevost T, Sutton S. Psychological impact of genetic counseling for familial cancer: A systematic review and meta-analysis. J Natl Cancer Inst 2004;96:122. 18. Gurmankin A, Domcheck S, Stopfer J, Fels C, Armstrong K. Patients resistance to risk information in genetic counseling for BRCA1/2. Arch Intern Med 2005; 165:523. 19. dAgincourt-Canning L. The effect of experimental knowledge on construction of risk perception in hereditary breast/ovarian cancer. J Genet Counseling 2005;14:55. 20. Kelly K, Leventhal H, Andrykowski M, et al. Using the common sense model to understand perceived cancer risk in individuals testing for BRCA1/2 mutations. Psycho-Oncol 2005;14:3448. 21. Madlensky L. Is family history related to preventative health behaviors and medical management in breast cancer patients? Breast Cancer Res Treat 2005; 90:47. 22. Becker M. Health Belief Model and Personal Health Behavior. Thorofare: Charles S. Black, Inc., 1974. 23. Ajzen I. The Theory of Planned Behavior. Organ Behav Hum Decis Processes 1991;50:179. 24. McGinley A. Health beliefs and womens use of hormone replacement therapy. Holistic Nurs Pract 2004; 18:18.

breast cancer chemoprevention. These findings persist even after they receive a personalized risk assessment and genetic counseling. The relationship between increased cancer anxiety and decreased clinical examinations has been described before,15 as has the resistance to breast cancer risk information in high-risk cohorts.16,1820 Future studies should focus on the interrelationship between increased risk perception for breast cancer and, paradoxically, the hesitation to use breast cancer chemoprevention. The value of using genetic counselors for calculating risk, explaining this risk and the risks and benefits of associated menopausal options, and, importantly, addressing the patients personal perceptions of cancer risk and reluctance to use medications should also be investigated.

ACKNOWLEDGMENTS
We thank the Susan G. Komen Foundation for its support and all the patients who participated in this study. Thanks also to Jamie Fong, Sarah Knapp, Stephanie Sohl, Konstantin Lukin, Bradley Jerson, and Delene Etwaru for their help with this paper.

REFERENCES
1. Wilcox S, Stefanick M. Knowledge and perceived risk of major diseases in middle-aged and older women. Health Psychol 1999;18:346. 2. Chlebowski R, Collyar D, Somerfield M. American Society of Clinical Oncology technology assessment of breast cancer risk reduction strategies: Tamoxifen and raloxifene. J Clin Oncol 1999;17:1939. 3. Fisher B, Constantino J, Wickerman D. Tamoxifen for the prevention of breast cancer: Report of the National Surgical Adjuvant Breast and Bowel Project P-1 Study. J Natl Cancer Inst 1998;90:1371. 4. Chlebowski R, Kim J, Col N. Estrogen deficiency symptom management in breast cancer survivors in the changing context of menopausal hormone therapy. Semin Oncol 2003;30:776. 5. Cuzick J. Aromatase inhibitors for breast cancer prevention. J Clin Oncol 2005;23:1636. 6. Morrissey D, Kirchner J. Management of the climacteric. Options abound to relieve womens midlife symptoms. Postgrad Med 2000;108:85. 7. WHI. Risks and benefits of estrogen plus progestin in healthy postmenopausal women, principal results from the Womens Health Initiative randomized controlled trial. JAMA 2002;288:321.

856
25. Bunn J, Bosompra K, Ashikaga T, Flynn B, Worden J. Factors influencing intention to obtain a genetic test for colon cancer risk: A population-based study. Prev Med 2002;34:567. 26. Doukas D, Localio A, Li Y. Attitudes and beliefs concerning prostate cancer genetic screening. Clin Genet 2004;66:445. 27. Fischhoff B. What do patients want? Help in making effective choices. Effective Clin Pract Patient Decis Making 1999;2:199. 28. Llewellyn-Thomas H. Patients health-care decisionmaking: A framework for descriptive and experimental investigations. Med Decis Making 1995;15:101. 29. Mulley A. Outcomes research: Implications for policy and practice. In: Smith R, Delamother T, eds. Outcomes in clincal practice London: BMJ Publishing Group, 1995:13. 30. Horne R. Patients beliefs about treatment: The hidden determinant of treatment outcome? J Psychosom Res 1999;47:491. 31. Horne R, Graupner L, Frost S, Weinman J, Wright S, Hankins M. Medicine in a multi-cultural society: The effect of cultural background on beliefs about medications. Soc Sci Med 2004;59:1307. 32. Buick D, Crook D, Horne R. Womens perceptions of hormone replacement therapy: Risks and benefits (19802002). Climacteric 2005;8:24. 33. Port E, Mongomery L, Heerdt A, Borgen P. Patient reluctance toward tamoxifen use for breast cancer primary prevention. Ann Surg Oncol 2001;8:580. 34. Vogel V, Constantino J, Wickerman D, et al. Effects of tamoxifen vs raloxifene on the risk of developing invasive breast cancer and other disease outcomes; the NSABP study of tamoxifen and raloxifene (STAR) P-2 Trial. JAMA 2006, 295(23). 35. Col N, Pauker S, Goldberg R, et al. Individualizing therapy to prevent long-term consequences of estrogen deficiency in postmenopausal women. Arch Intern Med 1999;159:1458. 36. Epstein R, Alper B, Quill T. Communicating evidence for participatory decision making. JAMA 2004;291:2359. 37a.Claus E, Schildkraut J, Thompson W, Risch N. The genetic attributable risks of breast and ovarian cancer. Cancer 1996;77:2318. 37b.Gail MH, Brinton LA, Byar DP, et al. Projecting individualized probabilities of developing breast cancer for white females who are being examined annually. J Natl Cancer Inst 1989;81:1879. 38. Santen R, Borwhat M, Gleason S. Menopause: What every woman should know. The Hormone Foundation, 1997. Charlottesville, VA. 39. National Cancer Institute. Cancer facts: Questions and answers about the study of tamoxifen and raloxifene (STAR). 2000.

MATLOFF ET AL.
40. Ailinger RL, Harper DC, Lasus HA. Bone up on osteoporosis: Development of the Facts on osteoporosis quiz. Orthop Nurs 1998;Sept/Oct:66. 41. Valdimarsdottir HB, Bovbjerg DH, Kash KM, Holland JC, Osborne MP, Miller DG. Psychological distress in women with a familial risk of breast cancer. PsychoOncol 1995;4:133. 42. Zakowski SG, Valdimarsdottir HB, Bovbjerg DH, et al. Predictors of intrusive thoughts and avoidance in women with family histories of breast cancer. Ann Behav Med 1997;19:362. 43. Diefenbach MA, Miller SM, Daly MB. Specific worry about breast cancer predicts mammography use in women at risk for breast and ovarian cancer. Health Psychol 1999;5:532. 44. OConnor AM. Validation of a Decisional Conflict Scale. Med Decis Making 1995;15:25. 45. Vogel V III, Constantino J, Wickerham D. Re: Tamoxifen for prevention of breast cancer: Report of the National Surgical Adjuvant Breast and Bowel Project P-1 Study. J Natl Cancer Inst 2002;94:1504. 46. Bober S, Hoke L, Duda R, Regan M, Tung N. Decision-making about tamoxifen in women at high risk for breast cancer: Clinical and psychological factors. J Clin Oncol 2004;22:4951. 47. Melnikow J, Paterniti D, Azari R, et al. Preferences of women evaluating risks of tamoxifen (POWER) for breast cancer risk reduction. Cancer 2005;103: 1996. 48. Horne R, Weinman J. Patients beliefs about prescribed medicines and their role in adherence to treatment in chronic physical illness. J Psychsom Res 1999;47:555. 49. Grunfeld E, Hunter M, Sikka P, Mittal S. Adherence beliefs among breast cancer patients taking tamoxifen. Patient Educ Counseling 2005;59:97. 50. Dowell J, Hudson H. A qualitative study of medication-taking behavior in primary care. Fam Pract 1997; 14:369. 51. Barber C, Margolis K, Luepker R, Arnett D. The impact of the Womens Health Initiative on discontinuation of postmenopausal hormone therapy: The Minnesota Heart Survey (20002002). J Womens Health 2004;13:975.

Address reprint requests to: Ellen T. Matloff, M.S. Yale Cancer Genetic Counseling 55 Church Street, Suite 402 New Haven, CT 06520 E-mail: Ellen.Matloff@yale.edu