Understanding low chemoprevention uptake by women at high risk of breast cancer: findings from a qualitative inductive study of women’s risk-reduction experiences

For many women at high risk of breast cancer due to family history or known genetic mutation, managing that risk is a key psychological concern. Such women face a lifetime breast cancer risk of 20–80%, and are often motivated by fear of cancer to search out ways to reduce this risk [1‐3]. Chemoprevention is one of several risk-management methods developed for such women, and has been shown to decrease lifetime risk by approximately 50% for most subgroups of the high-risk population [4‐7]. Chemoprevention involves taking a 5-year course of daily antiestrogen pills (usually tamoxifen for pre-menopausal women, raloxifene for post-menopausal women) as an alternative to surgical prevention options (bilateral prophylactic mastectomy and/or oophorectomy) [8], but uptake remains very low. The proportion of high-risk women eligible for primary prevention who take antiestrogen medication is 5% or less, and these numbers are only slightly higher for carriers of pathogenic BRCA variants [9‐12].

Reasons for low use of chemoprevention are not yet fully understood, but prior studies have illuminated several important factors associated with low uptake of preventive medication or the rarer choice to use chemoprevention. One of the most important influences on these decisions is a healthcare provider’s recommendation for or against chemoprevention [11, 13‐17]; the recommendation of a medical oncologist seems particularly strongly associated with use of chemoprevention [18, 19]. A variety of provider-level barriers also impede chemoprevention usage, however, including lack of confidence in using risk-prediction models to identify high-risk patients, and lack of confidence discussing risk or prescribing chemoprevention medications [20‐23]. From the patient side of the equation, cancer worry/anxiety, perceived breast cancer risk, and objective cancer risk are positively associated with the use of chemoprevention [1, 11, 13, 15‐19, 24‐28], while underestimation of benefits and/or overestimation of risks, perception of (or worry about) low drug efficacy, and concerns about side effects are negatively associated with chemoprevention use [11, 20, 21, 27, 29‐33]. Decision aids may increase women’s understanding of the risks and benefits of chemoprevention, but may not be consistently associated with either higher or lower uptake [15, 34, 35].

The mechanisms of these associations, however, are underexplored. Study designs have usually involved explicitly educating women about chemoprevention within the context of the study. It is therefore unclear whether correlates of uptake affect women’s general engagement in breast cancer prevention, their likelihood of choosing to use chemoprevention when that specific option is offered, or their likelihood of choosing chemoprevention over other risk-management options [36]. In addition, while lack of awareness and insufficient knowledge are among the top known barriers to chemoprevention uptake [14, 30, 37], few studies have examined how women who are aware of chemoprevention differ from those who are not. The vast majority of chemoprevention studies have also relied on samples of predominantly White participants, underrepresenting or nearly excluding women of other racial/ethnic backgrounds [24, 33, 34, 38]. Awareness of this problem has begun to cause a shift, with a few studies beginning to enroll more multi-racial participant groups [19, 25, 35, 39].

The current paper takes advantage of data related to chemoprevention choices obtained from a parent study designed to use high-risk women’s own stories about coping with risk to understand the dynamics of women’s prevention decision making. The parent study generated 50 open-ended qualitative interviews with White and African American women at elevated risk of breast cancer. In this paper, we draw on these detailed narratives to step beyond the context of a chemoprevention trial and better understand the factors that shape women’s awareness of and choices about chemoprevention.

For the parent project upon which this paper is based, fifty semi-structured, in-depth interviews were conducted by the first author (Principal Investigator) between May 2015 and March 2016. The relevant interview protocol was developed specifically for this study (see Additional file 1 [40]). Participants were recruited from the High Risk Breast Cancer Program and Clinical Genetics Clinic at The Ohio State University (OSU) Comprehensive Cancer Center, and through the ResearchMatch national research volunteer database and Study Search online study listing tool at OSU. Eligible participants were at least 18 years old with an elevated risk of breast cancer but no personal history of cancer [41]. Interviews were conducted in-person or via telephone and lasted 22–120 min. Iterative data analyses (described below) occurred alongside data collection, and the decision to stop conducting interviews was made once it was clear that saturated information had been collected about all the core emergent interview themes [42]. Ethical approval was granted by the Institutional Review Board of The James Cancer Hospital, OSU Comprehensive Cancer Center, and the Ohio State University (Protocol # 2014C0101), and participants provided informed consent. Interviews covered a range of topics, including perceived breast cancer risk; sources and content of risk information; understanding and consideration of prevention options; decision-making processes and networks; and psychosocial well-being. Other findings from the parent project have been published and are under review elsewhere [3, 43‐45].

For the specific study presented in this paper, analyses were limited to the 47 participants whose breast cancer risk was sufficiently high to render them eligible to use chemoprevention. This was an observational study of women’s experiences and perspectives; no participants were offered chemoprevention in the course of the study. Thirteen percent of participants (6) were categorized at “severe” lifetime risk of breast cancer (diagnosed with a pathogenic BRCA variant) and 87% (41) were categorized at “high” risk (family history including multiple, young, and/or bilateral cases of breast or ovarian cancer) [41]. Women in these categories have a quantitative lifetime risk of breast cancer over 20%; clinical guidelines published by the National Comprehensive Cancer Network (NCCN) recommend that healthcare providers discuss chemoprevention with such women.

Transcribed data were checked by the first author for accuracy and then analyzed in five stages, using a constructivist grounded theory approach [42]. First, three separate coders used open coding to generate exhaustive lists of themes present in the first eight interviews; these lists were compared to generate one comprehensive list of emergent themes. Some themes were anticipated by original interview questions and others emerged inductively from women’s own accounts. Related themes were then organized into categories. Second, each interview was coded systematically by at least two coders, who assigned segments of women’s stories to as many themes and categories as relevant. NVivo 11 (qualitative data management software) was used throughout this and the following stage of analysis. Additional themes and categories emerged during systematic exploration of the data. Coders caught one another’s omissions and discussed any areas of disagreement until reliable coding principles were agreed upon. Transcription, coding, and memo-ing stages all began while interviews were still being conducted, so that insights gained from analysis of early interviews could be reflected in the content of later interviews. By the end of data collection, all themes were theoretically saturated [42, 46].

The third stage of analysis involved in-depth exploration of themes and categories created in previous stages, using Excel tables that mapped relationships among themes, categories, and groups of participants, as well as memos that explored the content and boundaries of themes and categories. One important distinction that emerged from this process was between those who were ‘aware’ and ‘unaware’ of chemoprevention—being ‘aware’ was identified as having expressed any pre-interview understanding (regardless of terminology used) that medications exist that can reduce breast cancer risk among high-risk women. We therefore sorted participants into aware and unaware groups, and then examined those two groups further to compare them with respect to other emergent themes and distinctions that had emerged from inductive analysis. These themes included: exposure to breast cancer of a loved one; cancer worry; race; financial issues and SES; the presence of comorbid conditions; interactions with healthcare professionals; preferences toward and experiences with gathering information; body image and sexuality; and perceptions of risk and health beliefs. Finally, we sorted women who had heard of chemoprevention into five emergent categories of reflecting their “disposition toward chemoprevention,” and explored the relationships among themes that might influence women’s dispositions. These disposition categories were: (1) chose it; (2) leaning or leaned toward it; (3) aware and considering, or hasn’t decided yet; (4) aware and would consider under certain conditions; and (5) aware and would not consider, not considering, wouldn’t do, or considered and ruled out.

The first author was involved in all parts of the project: study design, data collection, and all stages of analysis and theory building. This consistent engagement ensured the continuous presence of a high level of expertise relevant to both the substance and the methodology of the study. The inclusion of two additional research team members throughout the coding, analysis, and write-up phases created important opportunities for the introduction of alternative perspectives and interrogation of assumptions that may have been made by a single author [47].

Participants’ ages ranged from 21 to 69, with a mean of 44 years. This study intentionally recruited a sample for which racial comparisons could be made within the constraints of the attainable sample size; 40% (19) of participants identified as African American and 60% (28) were non-Hispanic Whites. The sample consisted mostly of well-educated women—64% had at least a college degree. Half (51%) of participants were categorized as high socioeconomic status (SES) and the other half as medium/low SES, based on composite information including educational level, occupational status, and household income. See Table 1 for sample demographics. Participants are identified when quoted below by pseudonym, race, and age range; the latter groups are as follows: early 20s = 20–24; late 20s = 25–29; early 30s = 30–34; late 30s = 35–39; etc.

Uptake of chemoprevention in this sample echoed findings of previous studies [48]. Of 47 women interviewed, only four had ever seriously considered chemoprevention; three of these women actually chose chemoprevention and took the prescribed medication. The inductive analyses described above generated insights about chemoprevention awareness as a necessary prerequisite for use of this risk-reduction method, about information access and personal characteristics associated with chemoprevention awareness, and about the concerns of women who are aware of chemoprevention but not inclined to consider it.

Low uptake of chemoprevention has been a perplexing phenomenon for breast cancer researchers, particularly given high-risk women’s expressed desire for risk-reduction options, and the considerable resources expended to develop effective chemoprevention regimens [21, 30, 48‐51]. With fewer than half of these chemoprevention-eligible participants having heard of the chemoprevention option, this study highlights the importance of limited chemoprevention awareness as a key feature of low uptake. Overall, these data expose three types of knowledge gaps: general lack of chemoprevention awareness among high-risk women, particularly common lack of awareness among African Americans, and less knowledge of chemoprevention even among high-risk women aware of other risk-reduction options. These gaps are not commonly addressed in the chemoprevention literature, but pose a critical challenge to the use of chemoprevention as a risk-management method. Low levels of awareness not only constrain women’s ability to choose this risk-reduction route, they also make it impossible to assess how high-risk women would respond if they were better informed.

As illustrated in Fig. 1, this study suggests two categories of dynamics that may both underlie low awareness of chemoprevention and directly affect disposition toward chemoprevention among women who know about it. First, awareness of chemoprevention is likely mediated through key features of women’s environment of risk-reduction information. Most importantly, access to specialists seems to have a strong influence on awareness of chemoprevention. Our data suggest that it may only be when they see a cancer, breast, or genetics specialist that high-risk women have the dedicated risk-reduction conversations in which chemoprevention is discussed. Consistent with prior research, we find that it is largely women whose physicians recommend chemoprevention who seriously consider the option [11, 13, 14, 16‐19].

Most women gather risk-reduction information from sources other than specialist health care providers. Other high-risk individuals, support groups designed for high-risk people, and family members with medical or professional expertise can be considered high-quality sources of information, in the sense that they are associated with knowing about chemoprevention. The information source most consistently likely to generate knowledge of chemoprevention as a risk-reduction option, however, seems to be conversations about risk and risk-management with any healthcare provider.

Given the important impacts of conversations with a healthcare provider on women’s attitudes toward chemoprevention, it is critical that clinicians be knowledgeable, confident, and effective in advising high-risk women about chemoprevention and other risk-reduction options. Unfortunately, recent studies document significant gaps in clinicians’ ability to provide such high-quality medical advice. The majority of healthcare providers—from generalists to breast surgeons—lack comfort and experience in identifying high-risk women using Gail or other risk assessment models [22, 52‐54]. Both general practitioners and familial cancer specialists also lack core knowledge about chemoprevention and experience prescribing it [23, 55, 56]. The most common barriers physicians identify in their own ability to advise patients about chemoprevention are lack of education on the topic and lack of time to discuss it [49, 54, 55]; many also remain unconvinced that chemoprevention can be effective [22, 57] or unaware that it is clinically recommended for some groups of patients [23]. Our findings—and the additional context provided by these studies of clinician-level barriers—suggest structural interventions that could assist high-risk women in becoming fully informed about their risk-reduction options. Specifically, these could include more consistent insurance coverage for specialist consultations and genetic testing and continuing medical education programs designed to improve the ability of primary care providers to identify, advise, and refer high-risk women.

Second, these data suggest that specific personal characteristics are associated with chemoprevention awareness. In general, we find that women who use more nuanced information gathering strategies are also more likely to know about chemoprevention, and that knowing basic prevention information enables women to ask more detailed questions that could lead to actually considering specific risk-reduction behaviors [43]. Future studies should explore access to and use of risk-management information in more detail, since these dynamics are critical to the ability of high-risk women to make health-protective decisions. In addition, while the causal direction of the relationship between higher cancer worry and higher awareness of chemoprevention could hypothetically go either way, this general association again suggests that it might be useful to educate and enable primary care providers to discuss risk-related information with all high-risk patients.

Finally, this study and others indicate that women’s concerns about chemoprevention are quite complex [15, 20, 58]. The few women in this study who had taken a chemoprevention regimen retrospectively assessed the decision as an obvious and clear-cut choice to dramatically reduce their breast cancer risk. The substantial proportion of women who knew about chemoprevention but were not inclined to pursue it, however, articulated a range of overlapping reasons for their reluctance, from lack of information (about side effects, long-term health risks, or effectiveness) to semi-confident preferences (to avoid medication or pursue alternate risk-reduction approaches). The reluctance to use any medication at all (see also [32, 33]) is not generally consistent with the use of medication in the U.S. overall, or with the far more substantial uptake of other preventive medications (such as statins). Furthermore, women’s assessment of chemoprevention as more drastic than prophylactic surgery and potentially having very acute and serious side effects suggests that perhaps women fear chemotherapy-like experiences while on chemoprevention, despite rare evidence that this occurs [58, 59]. If this is the case, it might have appeared in our data in the form of (a) explicit references to chemotherapy in women’s discussions of chemoprevention, and (b) associations between women reluctant to use chemoprevention and those who knew someone who had been on chemotherapy. Our data did not contain these patterns, but larger samples could be used to investigate this possibility more systematically. The terms used to describe chemoprevention may be critical—we used “medication that can reduce your risk of cancer” unless participants already knew the term “chemoprevention”, but it would be worth investigating the frequency with which the physicians who educate high-risk women use “chemoprevention” vs. the names of specific drugs or general descriptors like ours. Overall, the complex patterns surrounding women’s concerns both deepen the mysteries surrounding low chemoprevention uptake and suggest further research directions that could clarify the issue.

The main limitation of this study has to do with its sample. Forty-seven in-depth interviews is an appropriate number for inductive research seeking deep understanding of women’s experiences, and met the primary sample-size criterion of achieving theoretical saturation. A larger sample size of high-risk women—recruited to represent the broader population and not just those in high-risk care or in a chemoprevention trial—would allow confirmation of these findings, as well as quantitative analysis of predictors of chemoprevention awareness and disposition. In addition, it would be helpful for future samples to include women who are high-risk not only because of familial history but also due to personal history of atypical hyperplasia or lobular carcinoma in situ, since these patients are particularly likely to benefit from chemoprevention. Finally, some important outstanding questions can only be addressed with a larger sample, such as about the mechanisms that drive racial and SES disparities in chemoprevention awareness.

The results presented here illuminate previously unaddressed complexities in chemoprevention decision making. First, although most risk-reduction studies focus on each option separately, women do not, so incomplete information about risk-prevention methods or their consequences artificially constrains women’s choices. Second, the women in this sample do not simply use or reject chemoprevention; their opinions instead exist on a continuum of dispositions that could change with age, new information about hereditary risk, or physician advice. Taking a dispositional approach to prevention decisions, as some decision-aid developers have begun to do, may help us more thoroughly understand and support women’s decisions. Third, most women voice not one but multiple reasons they are reluctant to take antiestrogen medications. Some of these reasons reflect information gaps that could be filled to facilitate informed decision making, while others likely reflect more stable preferences of informed women.

These findings point in new scientific directions, but also have important clinical implications. It is essential that high-risk women be informed of all their risk-reduction options—and it is clear that this is not currently the case. Filling the critical gap in chemoprevention awareness will require improvements in healthcare and health systems serving this population. Primary care providers may play an integral role, by identifying high-risk women and giving them basic information about risk and prevention [34, 35], pointing women toward patient-information libraries and decision aids that may aid in self-education [15, 34, 35], and referring them to specialists who can discuss individual risk and risk-management options in detail. Given the formative role of specialist access, ensuring that women have health insurance and that their plans cover specialist care and genetic testing is also crucial. Finally, future attempts to facilitate informed risk-management decision-making will require closer attention to the nuances of women’s disposition toward their options and the multiple factors they consider when making risk-management decisions.