How do women at increased breast cancer risk perceive and decide between risks of cancer and risk‐reducing treatments? A synthesis of qualitative research

Abstract Objective Risk‐reducing procedures can be offered to people at increased cancer risk, but many procedures can have iatrogenic effects. People therefore need to weigh risks associated with both cancer and the risk‐reduction procedure in their decisions. By reviewing relevant literature on breast cancer (BC) risk reduction, we aimed to understand how women at relatively high risk of BC perceive their risk and how their risk perceptions influence their decisions about risk reduction. Methods Synthesis of 15 qualitative studies obtained from systematic searches of SCOPUS, Web of Knowledge, PsychINFO, and Medline electronic databases (inception‐June 2015). Results Women did not think about risk probabilistically. Instead, they allocated themselves to broad risk categories, typically influenced by their own or familial experiences of BC. In deciding about risk‐reduction procedures, some women reported weighing the risks and benefits, but papers did not describe how they did so. For many women, however, an overriding wish to reduce intense worry about BC led them to choose aggressive risk‐reducing procedures without such deliberation. Conclusions Reasoning that categorisation is a fundamental aspect of risk perception, we argue that patients can be encouraged to develop more nuanced and accurate categorisations of their own risk through their interactions with clinicians. Empirically‐based ethical reflection is required to determine whether and when it is appropriate to provide risk‐reduction procedures to alleviate worry.

also increases a woman's risk of developing a new BC. 7 Several riskreduction procedures are available, with well-understood risks and benefits. Screening, using radiographic mammography, ultrasound, or magnetic resonance imaging, 8 cannot prevent cancer but enables early detection, which improves prognosis. However, screening can miss cancers, and false positive results cause unnecessary alarm. 9 Chemoprevention, through selective estrogen receptor modulators or aromatase inhibitors, 10,11 can reduce incidence of estrogen receptor positive and negative cancer in postmenopausal women. 12 Selective estrogen receptor modulators increase risk of endometrial cancer, thromboembolic events and menstrual and skin complaints, and aromatase inhibitors may reduce bone growth and contribute to infertility and liver and kidney dysfunction. 13 Risk-reducing mastectomy (RRM) can improve life expectancy in women with BRCA mutations. In lower risk BC survivors, RRM may reduce risk of new BC but has not been shown to improve life expectancy because it cannot reduce the likelihood of metastatic disease. 14 Risk-reducing mastectomy is irreversible, carries surgical risks, may require follow-up surgery, and can cause physical discomfort and emotional distress linked to breast appearance and feelings of damaged femininity. 15,16 Bilateral salpingo-oophorectomy (surgical removal of the ovaries and fallopian tubes) can reduce the risk of BC by up to 50% and risk of ovarian cancer by 90% to 95 %, although the extent of BC risk reduction has recently been questioned. 17 Oophorectomy is irreversible, risks surgical complications, and causes infertility and premature menopause. 17,18 Therefore, women at high risk of BC face complex choices about risk mitigation. Practitioners caring for women face corresponding challenges around how to help women weigh risks and benefits in making risk-reducing decisions.
According to early "likelihood-value" theories of decision making, people should weigh risks and benefits of different decision options by estimating the likelihood and personal value of potential outcomes and select the option that offers the optimal combination of these. 19,20 However, people do not, in practice, report thinking about risk in continuous estimates of likelihood and value 21,22 and measures of these poorly predict behaviour. 23,24 The evidence suggests, instead, that people base risk perceptions and associated decisions on mental heuristics, ie, "rules of thumb" or approximations that allow easier decision making. 24 Examples are the "affect heuristic," whereby people's inferences of risk are guided by their emotional feelings, and the "availability heuristic," whereby people infer risk from the ease with which risk-related information can be recalled. 25 Whilst heuristics can reduce the extent to which people contemplate objective risks and logically integrate them into decision making, they might also improve decisions by allowing people to respond to complex information that they would otherwise be unable to assimilate in a more rational way. 26 A large and diverse range of heuristics have been documented, but many are specific to the demands and contexts of particular decisions. 27 Thus, women's risk perception and decision making in BC risk reduction cannot be understood through merely appreciating that they use heuristics but requires detailed understanding of the heuristics they use and the influence of these on decision making. We are aware of no work that reviews this evidence in BC risk reduction, and this was the aim of the present study.
The measurement procedures of quantitative research in this context necessarily presuppose the main ways in which risk perception and decision making vary. Consistent with our inductive aim, we therefore focused on qualitative research. Our specific aims were, first, to synthesise qualitative literature concerning how women with elevated risk for BC perceive risk and how these perceptions influence decision making about risk reduction and, second, to draw implications for how clinicians can help women make these decisions.

| Inclusion and exclusion criteria
We included peer-reviewed qualitative studies that examined risk perception or decision making in adult women at high risk for BC. We used a broad definition of risk perception as "an individual's personal understandings of BC risk and of the risks and benefits of risk-reduction options." Higher risk groups included women with established genetic mutations (eg, BRCA 1/2), or familial risk factors (affected first-degree relatives), higher scorers on predictive scales derived from epidemiologic analyses of risk factors, members of ethnic populations characterized by higher risk, and women who had previously been examined with BC. Studies of women currently under treatment for BC were excluded. Inclusion was limited to English language reports.
Databases (see below) were searched from inception to June 2015.

| Search strategy
Search terms and alternatives were initially identified using several reviews relevant to this area 22,23 and then augmented by scoping searches. Terms in the title, abstract, or keywords relevant to BC ("breast cancer" or "breast carcinoma" or "breast neoplasm") and risk perception ("risk perception" or "risk understanding" or "perceived risk" or worry or dread or "anticipated emotion" or "anticipatory emotion" or emotion* or vulnerability) and qualitative methodology (qualitative or "Grounded theory" or IPA or "interpretative phenomenological analysis" or "thematic analysis" or "content analysis" or "narrative Analysis" or "conversation analysis" or "discourse analysis" or interview* or "focus groups") were searched. searched. Reference lists from previous systematic reviews were also searched. Searches were combined, and duplicates removed before study selection. Hand searches were also conducted of the reference lists of the included articles.

| Study selection
Using the electronic databases, search terms were identified from titles, abstracts, and keywords. Following the search, identified studies were assessed for inclusion. Initially, HGF screened all identified titles and then the abstracts of selected titles for potential inclusion. Then all papers identified as potentially relevant were read by HGF who assessed whether they met the inclusion criteria. When this was unclear, SLB also read the study and a joint decision was reached.
Stage of exclusion and the reasons for exclusion were recorded (Table 1). Figure 1 describes study selection using the Preferred Reporting Items for Systematic reviews and Meta-Analysis 28 flow diagram.

| Data synthesis
We wanted to develop new theoretical insights, grounded in the findings of individual studies but with general applicability across those studies and, therefore, potentially beyond them. Thus, we took an inductive approach, drawing upon grounded formal theory. 29,30 This approach starts with a descriptive analysis of data from the reviewed studies (including the research question, sample description, inductive categories arising from the authors' analysis, and the illustrative data contained in the reports) but then progresses to a theoretical analysis.
The key method of analysis is constant comparison within and across studies to detect convergences and resolve inconsistencies. Our synthesis was at the level of reported findings rather than the authors' interpretations, and we sometimes drew different theoretical conclusions from the authors in the context of the developing analysis. HGF peformed a preliminary synthesis, first developing a descriptive analysis then a theoretical integration of this analysis. Then HGF and SLB interrogated this to identify consistencies and inconsistencies with the source data, and SLB provided a reformulated model. SLB and HGF then compared the reformulated model to the preliminary synthesis.
The final synthesis arose through discussion amongst all authors. Table 2

| Overview
Women generally perceived risk categorically rather than probabilistically, partly based on previous family experiences of BC. Decisions about risk reduction arose in 2 very different ways, depending on how intensely women worried about BC.   Grounded theory Understand women's decision-making processes and the social contexts that influence these processes.

| Women did not perceive risk probabilistically
(Continues)

| Women perceived risk categorically
Instead, when asked to describe personal risk, most women used verbal labels to describe risk categories to which they felt they belonged. [34][35][36][37][38][39] Categorisation was pervasive across the differing samples and analysis methods, and women were explicit about doing this.
Category labels were diverse. Some, such as "probable," "high risk," or certain (of BC) could be seen as ordinal points across a spectrum of likelihood. 34,38,39 Others described positions relative to population risk, such as "a bit higher than population "risk" or "no higher than anyone else." 34,36,39 Other labels encompassed qualitative categories such as "vulnerable" or "at risk." 39 Categorisations were generally realistic.
Almost all women acknowledged being at high risk in either an absolute sense or comparison with the wider population, and 1 study showed that women's self-categorisations were largely consistent with the categorical estimates that they had been given by professionals. 38 Women were explicit about the ways that they developed these categories and about using category labels to help them to think about risk. Some spontaneously assigned themselves to categories, based on either their personal or their family experiences related to cancer (see below), their theories about specific risk factors, 32,34 or their emotional responses to risk. 34 Others reported that health professionals introduced them to the categories that they used. 33,39 In 1 study, 38 BRCA1/2 mutation carriers wanted health professionals to provide them with risk "labels."

| Family experiences defined risk perceptions
Women's perceived risk categories were informed by family experiences of BC. 34,35,39 Some women were explicit about how their family histories led them to believe that they were in a high risk category. [39][40][41] However, in most women, the influence of family experiences was implicit. 33,[39][40][41][42][43] Women assumed that their futures would follow the path of a family member's illness with little consideration that their own cancer likelihoods or experiences might differ. 35,[39][40][41][42] The ages, sites, and stages at which relatives had been examined and the outcomes of relatives' illnesses defined expectations of their own fates, in that they expected to have cancers that would develop with the same trajectories [34,41,42] . Relatives' ages were particularly important. Dagan and Goldblatt 41 referred to this as the "family clock," Werner-Lin 40 as a "danger zone." Remaining healthy at the age at which their mothers developed BC provided hope for future health, 41 and women felt profound relief when these landmarks had passed. 34 Family experiences sometimes led to a sense of inevitability about BC, inducing a "labelling" error whereby BC risk was described as "certain." This was evident in those whose mothers had been examined with BC 44 and in some with affected first or second-degree relatives. 32,35 After positive BRCA tests, many women in Werner-Lin 40 study felt that they were on a "path towards cancer."

| Some women deliberated about risk, but worry impelled others toward aggressive risk-reduction procedures
Women made decisions in 2 ways, largely depending upon their level of worry. Where worry was not intense, women generally reported weighing their risk of BC and iatrogenic risks associated with risk-reduction procedures. They also considered other factors, such as how engaging in research might help other women, 34 how future childbearing and breast-feeding aspirations might militate against oophorectomy or RRM, and how current parental responsibilities militated against any such major surgery. 35,[39][40][41] Sometimes they postponed decisions if they felt unready to make them. 37 Unfortunately, the reviewed studies did not reveal how women made comparisons between different risks to make their decisions. Therefore, whilst we know that women largely thought about risk categorically, we do not know how they weighed different risks in reaching decisions. Nonetheless, women who weighed risks and benefits were generally content with their decisions. 37,[41][42][43] Often the decisions provoked strong emotions, but women did not describe these emotions as influencing their decisions. 44,45 Each study described women whose decisions were shaped by intense levels of worry about BC. Women described fears about BC as ever-present, intrusive, uncontrollable, and sometimes "intolerable." 31,32,36,41 The threat of BC was a "constant companion" for these women 31,40 and induced a persisting sense of threat. 44 No study explicitly examined why these women worried so intensely, and worry was not clearly explained by the risk category to which women allocated themselves. Whilst some women attributed worry to being in a high risk "category" or to recalling salient family experiences of BC, 33,36-38 women who saw themselves as at high risk did not all worry so intensely. 32,34 Believing that cancer was inevitable could even reduce worry where women with family history of BC resigned themselves to this. 39,45 Worry resisted reassurance from professionals' descriptions of objective risk or from good outcomes of clinical investigations. 32,34,37 Instead, minimising worry became the overriding and urgent goal that women pursued through their decisions about risk reduction. 32,33,36,[41][42][43]46,47 Women acknowledged that they were more worried than their objective risk warranted, but their decisions were nonetheless determined by worry. 33,38,46 The goal of minimising worry typically led women to eschew conservative options and choose the most aggressive available to them. Worry led women to prefer mammographic screening to self-care such as breast self-examination, 46 participation in a chemoprevention trial to screening, 36 and RRM to screening. 33,44 Many described needing to "do something" and feared missing opportunities to reduce risk, but there was little indication that women had closely considered the iatrogenic risks associated with However, wider research in social psychology shows categorisation to be a fundamental aspect of how people interpret information and use it to make decisions. 50,51 Categorisation is a heuristic that allows people to remember, retrieve, and use risk information more easily. 52 Thus, category labels such as "above average" or "at risk" provided easily accessible meanings that could help women to make decisions. Although categorisation of risk could be biased by family history, women wanted to use accurate categories. In particular, they sought accurate categorical information from health professionals 43 and were able to remember these when later asked. 38 Therefore, categorisation should not be regarded as an inherently inaccurate way for women to perceive risk.
Unfortunately, the reviewed studies did not illuminate how women compared different risks to make their decisions. Unlike probabilistic risk perceptions, where a common metric allows comparison of different outcome probabilities, categories used by women in the reviewed studies carry unique meanings that do not offer a common metric. 48 For example, patients considering surgery may use different types of category to describe risk of BC (eg, "I am vulnerable") and surgery (eg, "I am at moderate risk"). One possibility suggested by the broader psychological literature is that some risk categories, such as feeling "vulnerable," carry stronger emotional connotations than others, such as having "moderate" risk. The risk associated with the stronger emotional connotation might therefore drive the decision. 27 Research is needed to elucidate how women compare categorical risk perceptions in decisions about risk reduction.
The clearest evidence about decision making in the reviewed papers arose where it was influenced by worry. Where worry was intense, women did not consider risks and benefits. Instead, they chose aggressive risk reduction to reduce worry, and they paid little attention to iatrogenic risk. That is, worry "hijacked" decision making. At first sight, this is consistent with evidence for widespread use of a heuristic whereby people infer risk from their emotional responses. 53,54 However, our review emphasised another mechanism: that worry reduction became a decision-making goal in its own right. That is, women pursued an "emotion-focussed" coping strategy where they sought to reduce worry and were largely unconcerned with objective risk. 54 The influence of worry is harder to address. Where it is linked to perceiving oneself to be at high risk of BC, worry might be alleviated by helping women to question the risk perceptions that they have formed, as we describe above. However, cancer worry is not simply a product of high-risk perceptions and indeed appears to be largely insensitive to reassurance about risk. 54 It is partly a consequence of unrelated factors, including general emotional state, negative life experiences, and stressful or unsupportive environments. 58 Whilst counseling or other interventions for worry might therefore prove helpful in some instances, it is unrealistic to expect that worry can be completely banished in clinical contexts in which patients are faced with the mortal threat of serious illness. Therefore, particularly when the clinical benefits of interventions are unclear, clinicians will face the dilemma of how to reconcile the normative expectations for them to respect patients' requests, motivated by escape from worry, whilst simultaneously meeting their needs, which go beyond worry to encompass the balance between reduction of BC risk and avoidance of iatrogenic harm. The dilemma hinges on the ethical question as to whether and when it is appropriate for clinicians to provide invasive, and even surgical, responses to psychological need. Ethical analyses in the fields of cosmetic 59 and bariatric surgery 60 suggest surgical intervention for psychological benefit can be justified, but that a case always needs to be made that clinical and psychological benefits outweigh risk, that benefits are likely to occur and that benefits cannot be achieved with less risk.
Where solutions to ethical dilemmas cannot be derived from general normative principles, Kleinman 61 proposed that the starting point for reflecting on possible solutions should be detailed study of how practitioners and patients resolve these dilemmas in practice. Unfortunately, the studies that we reviewed provided little insight into this.
Future research that examines, not only women's accounts but also those of the clinicians caring for them, and which examines how decisions are negotiated between them in consultations, could provide evidence from which realistic and ethically robust solutions can be derived.