Talking numbers: how women and providers use risk scores during and after risk counseling – a qualitative investigation from the NRG Oncology/NSABP DMP-1 study

Abstract

Objectives

Little research exists on how risk scores are used in counselling. We examined (a) how Breast Cancer Risk Assessment Tool (BCRAT) scores are presented during counselling; (b) how women react and (c) discuss them afterwards.

Design

Consultations were video-recorded and participants were interviewed after the consultation as part of the NRG Oncology/National Surgical Adjuvant Breast and Bowel Project Decision-Making Project 1 (NSABP DMP-1).

Setting

Two NSABP DMP-1 breast cancer care centres in the USA: one large comprehensive cancer centre serving a high-risk population and an academic safety-net medical centre in an urban setting.

Participants

Thirty women evaluated for breast cancer risk and their counselling providers were included.

Methods

Participants who were identified as at increased risk of breast cancer were recruited to participate in qualitative study with a video-recorded consultation and subsequent semi-structured interview that included giving feedback and input after viewing their own consultation. Consultation videos were summarised jointly and inductively as a team.tThe interview material was searched deductively for text segments that contained the inductively derived themes related to risk assessment. Subgroup analysis according to demographic variables such as age and Gail score were conducted, investigating reactions to risk scores and contrasting and comparing them with the pertinent video analysis data. From this, four descriptive categories of reactions to risk scores emerged. The descriptive categories were clearly defined after 19 interviews; all 30 interviews fit principally into one of the four descriptive categories.

Results

Risk scores were individualised and given meaning by providers through: (a) presenting thresholds, (b) making comparisons and (c) emphasising or minimising the calculated risk. The risk score information elicited little reaction from participants during consultations, though some added to, agreed with or qualified the provider’s information. During interviews, participants reacted to the numbers in four primary ways: (a) engaging easily with numbers; (b) expressing greater anxiety after discussing the risk score; (c) accepting the risk score and (d) not talking about the risk score.

Conclusions

Our study highlights the necessity that patients’ experiences must be understood and put into relation to risk assessment information to become a meaningful treatment decision-making tool, for instance by categorising patients’ information engagement into types.

Trial registration number

NCT01399359.

STRENGTHS AND LIMITATIONS OF THIS STUDY

• Videos recorded individual consultations conveyed interactions between a provider and patient in primary cancer prevention settings.

• Interviews including a review of the own video consultation with participants enabled scrutiny of risk score meaning by participants.

• Risk assessments that were routinely conducted as part of breast cancer risk counselling in our sample and may not be applicable where a cancer prevention risk assessment is not routinely used or discussed.

• Need for future patient type validation with a wider range of cancer prevention counsellors in non-urban settings.

Introduction

Counselling on prevention and interventional strategies to reduce disease risk is considered an important aspect of clinical care from primary to specialty practice.^1–4 Treatment guidelines about prevention encourage the use of risk scores to identify individuals at increased risk and counsel them on the likelihood of developing a particular disease within a given time period.^5–8 As the first point of cancer prevention care, primary care providers and specialists, including family physicians, obstetricians/gynaecologists and internists, as well as nurse practitioners, are tasked with counselling on risk for disease across a whole spectrum of preventive medicine.^9–13

In risk counselling, the benefits of lowering the risk of developing breast cancer should be weighed against the intervention options, from lifestyle changes (such as lowering alcohol intake, maintaining a healthy body weight, limiting hormone exposure), to surgery (prophylactic mastectomy and oophorectomy), and taking an oral selective oestrogen receptor modulator (SERM).^{7 14 15} Standardised risk assessment instruments such as the Breast Cancer Risk Assessment Tool (BCRAT)¹⁴ and others^16–21 provide a base value of risk for individuals, which is presented as a percentage of risk over time; both 5 years and over a lifetime (up to the age of 90 years). This individually calculated risk can be used to initiate a discussion between providers and patients about risk option preferences. The BCRAT is particularly relevant for epidemiological and clinical risk factors outside of family history and is used to determine eligibility for SERM. SERM presents an option for individuals with a calculated BCRAT >1.66% for 5 years or 20% for a lifetime.^22–24

The National Surgical Adjuvant Breast and Bowel Project Decision-Making Project 1 (NSABP DMP-1) investigated social, cultural and psychological factors driving decision-making regarding SERM use in women counselled on breast cancer prevention options.²⁵ Physician recommendation was found to be the most important factor for SERM uptake,²⁵ but only if it aligned with the social and experiential factors of the counselled women.^25–28 Objective risk assessment was not found to be a decisive factor. Detailed investigation of decision-making processes showed the importance of perceived control in relation to perceived risk as a factor determining decision-making, as well as an understanding of the reversibility of the decision, the perception of medications, and how close the possibility of cancer felt to oneself.^{26 28}

Provider risk counselling is often the most important entry point for identifying women at high risk of developing breast cancer and providers are increasingly recognising the importance of (genetic) risk factors and counselling,^11–13 yet how frequently risk assessments are used depends on specialty and training.^{9 29–31} When risk assessments are conducted with patients, little is known about the communication strategies in practice. In this article, we aim to identify and describe how risk information and risk scores used in counselling is provided and worked with in the communication between provider and participant. We specifically explored: (a) how BCRAT scores are introduced during counselling; (b) the reactions of women during counselling sessions and (c) discussions of these scores afterward, as they pertain to an individual’s own breast cancer risk.

Materials and methods

NRG Oncology/NSABP DMP-1

This study used the data available from the qualitative arm of the NSABP DMP-1 to investigate the communication strategies and role of risk information in breast cancer risk counselling. The DMP-1 was a mixed-method study to investigate the social, environmental and psychological factors involved in decision-making about risk reduction strategies in women counselled on SERM use for breast cancer risk. It consisted of a survey arm and a qualitative, observational arm.²⁵ The qualitative arm recorded 30 breast consultation sessions of women who were identified prior to counselling as at increased risk of breast cancer by their provider from two DMP-1 study sites: a large comprehensive cancer centre serving a high-risk population and academic safety-net medical centre in an urban setting serving a large population of racial and ethnic minorities. In addition, in-depth interviews with participants from the recorded consultation sessions were conducted within 6 weeks of the consultation.^{25 26}

Data collection

Between April 2012 and August 2013, participants scheduled for appointments to discuss their breast health who were identified as at increased risk of breast cancer after a regularly scheduled mammogram or check-up, due to a family risk of breast cancer or to discuss biopsy results were contacted and recruited purposively before breast cancer risk counselling sessions appointments. Written consent was given prior to being video-recorded during their session. In order to capture regular care counselling, providers were not given counselling content outside of the eligibility criteria that they intended to discuss SERM use.²⁶ Subsequently, a qualitative interview was conducted with participants on-site by experienced and trained interviewers. Both interviewers were women, a health researcher (CMG) and clinical psychologist (PAP) who contacted participants for setting up interviews and explaining research background and goals before securing informed consent. Using a previously pilot-tested, semi-structured guideline that was tailored to the content of the individual consultation session, prior to the interview, after being viewed by members of the qualitative team (health researcher JWK, social scientist doctoral candidate SBB, PhD medical ethnographer and epidemiologist CH, CMG and PAP). Overall, the interviews addressed: the experience of the consultation and treatment options discussed; the experience of breast cancer risk and views of treatment options; and feedback and input after viewing their own consultation. As participants watched their own consultation video, they were encouraged to comment on the recorded consultation session content and to answer questions from the research team (JWK, SBB, CH, CMG and PAP). A follow-up telephone interview on the decision made by each participant finalised the data collection for the primary study.

Analysis

We used all 30 available NSABP DMP-1 data protected²⁶ consultations and interview transcripts. Consultation videos were summarised jointly and inductively as a team according to Schubert.³² Next, the joint summaries were coded thematically (JWK, SBB). Joint summary themes that discussed risk calculation and/or assessments were compiled deductively according to these themes and corresponding video segments reviewed for participant reactions. Each reaction was described in further analytical memos. The video data was then coded inductively according to presentation and interaction types that evolved from the analysis.

In addition, the interview material was searched deductively for text segments that contained the inductively derived themes related to risk assessment. Subgroup analysis according to demographic variables such as age and Gail score were conducted, investigating reactions to risk scores and contrasting and comparing them with the pertinent video analysis data. From this, four descriptive categories of reactions to risk scores emerged. The descriptive categories were clearly defined after 19 interviews; all 30 interviews fit principally into one of the four descriptive categories.

Analysis was done by the first author (SBB), in regular consultation with the last author, senior principle investigator (CH). Regular meetings and presentation of findings were discussed with a qualitative methods working group at the Institute for Public Health, Charité-Universitätsmedizin Berlin. Analysis was assisted and organised throughout in MAXQDA V.18,³³ reported using the Consolidated criteria for Reporting Qualitative research.³⁴

Patient and public involvement

No patients were directly involved in the design or recruitment of this study. However, our previous studies about patients’ priorities, experiences and preferences regarding breast cancer risk informed this current study, design and recruitment.^{26–28 35} Results of the study will be made available to study participants at the participating centres.

Results

Key demographics for the sample are summarised in table 1, and reported in extensive detail in previous publications.^26–28

Table 1.

Key participant demographics

Participant characteristics
Age range	37–73 years
Breast Cancer Risk Assessment Tool score range	N (%)*
5 years	1%–20%
Lifetime	10%–41%
Previous atypical cell biopsy findings
Yes	22 (73%)
No	8 (27%)
Biopsy with lobular carcinoma in situ (LCIS)	5 (17%)
Race/ethnicity
White	21 (70%)
Black/African American	6 (20%)
Latino/Hispanic/multiracial/  unknown	3 (10%)

*Total n=30.

Five providers conducted the 30 counselling sessions. All but one provider had extensive experience counselling on risk; one was new to risk counselling. The five medical providers included the following specialties: general internists with specialty training in breast health, nurse practitioner and oncologist. The total consultation length for each participant ranged from 11 to 37 min. Further characteristics can be found in a previous publication.²⁶ For 16 participants (53%), this was their first visit with this provider. There were 21 participants (70%) for whom SERM was recommended. Providers presented the risk score to all but one participant either as a printed handout or on the computer screen; the discussion of the risk score took place at the beginning of the consultation in nearly all consultations. The amount of time that providers and participants specifically discussed the risk score itself within the consultation session ranged from 13 s to 6 min, corresponding to 2%–24% of the consultation time. In two cases, no risk score was discussed. Almost all consultations closely followed the discussion topics about breast cancer risk as listed in existing breast cancer risk guidelines, from lifestyle changes for prevention, to screening and surgery, in addition to SERM.^{7 14} However, how systematically breast cancer risk factors were discussed varied according to the provider’s style and experience.

During counselling: providers’ personalised risk score numbers

During counselling, providers gave meaning to the risk score by: (a) presenting thresholds, (b) making comparisons and (c) emphasising or minimising risk and risk reduction.

Presenting thresholds

Risk levels, which are set as the minimum levels at which tamoxifen may be prescribed as a risk reduction therapy (5 years=1.66%; lifetime=20%), were introduced as a threshold at which a provider should discuss breast cancer prevention options. One provider referred to such thresholds as ‘magic numbers’ (table 2). In three cases, providers discussed the risk and SERM, but never cited a risk score to the patient during counselling (table 2).

Table 2.

Video analysis codes

Code system: video analysis—risk memo analysis	Coded segments (n)
Reactions to risk numbers
Participant risk score not given	3
Confused reaction to risk score	5
No reaction to risk score	4
Accepting positively framed numbers	4
Accepting negatively framed numbers	8
Rejection of risk assessment level	6
Negative reactions	11
Positive reaction to risk score information	10
Risk presentations
No numerical values discussed	3
Tailoring risk information	5
Risk number ambiguities/grey areas of risk numbers	5
Percentages	22
Relative risk reduction	19
Negative framing	15
Emphasis on negative	5
Frequencies	9
Combined positive/negative framing	11
Positive framing	10
Numerical benefits	1
Material presentations	5
Presentation of absolute numbers	0

Framing risk scores and risk reduction

The strength of the provider’s recommendation for SERM use for a participant influenced the way the risk score was explained during counselling. Most risk scores were clarified by an emphasis on the likelihood of developing breast cancer. For example, a provider who strongly recommended that a participant consider SERM as a result of atypical cell biopsy findings regularly cited a relative risk reduction of 86% based on findings from prevention clinical trials. When a provider clearly recommended against SERM use, risk scores were framed to illustrate how unlikely the person was to develop breast cancer. The residual risk, or remaining likelihood of developing breast cancer after SERM treatment, was discussed only with people demonstrating statistical proficiency during the office visit and showed eagerness to grasp risk statistics (table 2).

Making comparisons

Another important strategy that was used to make the risk score meaningful was to compare the patient’s risk score to that of the ‘average’ women (table 2). For example, providers highlighted how the participant’s risk score was double or three-times the average. The risk score of the participant as compared with that of the average woman was presented in a format using absolute numbers.

Minimising a risk score: recognition of risk assessment tool limitations

In some consultations, providers recognised the limitations and ambiguity of risk assessment tools. In these cases, the provider verbally minimised the meaning of the individual’s calculated risk score in order to incorporate other factors that would modify the meaning of a counselee’s individual risk (table 2). This happened particularly when the explanation of the provider highlighted that they felt the risk score did not adequately reflect the risk factors of a given participant. For those with complicating medical factors, such as a simultaneous family history of breast cancer and stroke, or for those who were young, providers emphasised that for such a combination of factors, the available risk score might not provide an appropriate risk estimate. Women who had a specific finding of atypical cells called lobular carcinoma in situ (LCIS) were presented with a percentage range because the BCRAT is unable to calculate an individual risk score. In such instances, providers expressed ambiguity, were careful not to give a precise risk estimate or did not use a risk score to underscore the recommendation.

During counselling: participants’ reactions

Overall, most consultations did not illicit reactions to the risk score information during the counselling session. Most participants simply listened or gave signs of affirmation while the providers presented the risk score information. Engaged discussions about risk scores were characterised by a high level of comfort with risk numbers on the part of the participant or their need for clarity about the presented information (table 3). A subset of the participants were prompted to engage in discussions about their risk when the way provider communicated or framed risk scores resulted in less clarity or when participants added their own information to what the provider presented.

Table 3.

During counselling—providers’ personalised risk score numbers and participants’ reactions

Providers’ personalised risk score numbers
Theme	Quote
Giving thresholds	Provider BA: The chemoprevention drugs are the next thing I want to talk about. That’s where this magic number comes in. So 1.66% makes you eligible to consider chemoprevention drugs. (Participant L)
Framing risk scores and risk reduction	Provider BC: If we say your lifetime risk is 36%, that means 36 women out of 100 will develop breast cancer. 1 in 3. That’s kind of high. If all 100 women take tamoxifen or raloxifene, that risk is reduced by 86%. Or there’s a residual 5% risk of getting breast cancer. So instead of 36 women out of that 100 getting breast cancer, only 5 out of 100 get breast cancer. Thirty-one out of that 100 don’t get told in their lifetime they have breast cancer. That’s big. (Participant Z)
Making comparisons	Provider BE: We use the [BCRAT] to help assess your risk—and I have the [BCRAT risk score], it’s all in that packet at the bottom. The (BCRAT) uses a variety of interchronologic history: your age, and whether you had atypical hyperplasia. The atypical hyperplasia lesion is the one that … is driving your risk the most. The [BCRAT] gives you a 5-year risk of breast cancer and a lifetime risk of breast cancer. As you see, your 5-year risk is estimated to be 1.8%, that’s compared with an average risk of 0.7%. Alternatively, the lifetime risk is 15.3% as compared to 7.5%. Basically, that says you're at approximately a double risk of developing breast cancer in your life. (Participant M)
Minimising a risk score: recognition of risk assessment tool limitations	Provider BA: So I would say you’re a good candidate for it were your risk high enough based on these numbers. I don’t know what your risk is really. I know you’re negative for the gene and you're not 1.6 or higher on this scale. So based on the data I have available you don't really fall into the category where it’s appropriate. However, I also know that you're in a little bit of a grey zone. (Participant A) Provider BB: I will quote you exact numbers—let me get you the exact numbers. I think it was about 1.4, so it was a very low risk. (Participant N)
Discussion in counselling: prompted by participant comfort and provider ambiguity	Participant U: So those who got there [into the study] … with atypical [biopsy findings], most likely had a— Provider BC: 86% risk reduction. Participant U: Over the 5-year? Or lifetime? Provider BC: Both, both. So that means we would reduce this [by] 86%, the lifetime risk of 30% down to … Participant U: By 86%? Provider BC: Well, down to 4–5%. Participant U AND Provider BC in unison: Which is lower than an average person. Participant V: We both understood [the previous provider] to say 40 to 50% [lifetime risk range]. Participant companion: 40 to 60%! Has that changed in the last year or two? Provider BC: No, that hasn’t changed. Participant V: So this is new information, … you’re saying on the high side 30%? Provider BC: Sure. Participant V: This is new information to me. I just felt like a 50% chance [told to me by the previous provider]. —I’m kind of like a ticking bomb!
Adding information to risk scores	Participant A: Does this number obviously go up as I age? Provider BA: As well, as well. Participant A: So as this number goes up to one(percent 5-yr risk), this [lifetime risk] number goes up higher? Provider BA: So you’re [at] one percent(5-yr risk), 20.6 [percent lifetime risk]. So I’m a little bit on the fence about this. Participant A: I’m wondering if this is the right time to start it, or next year if we revisit it, is that the best time to say, ‘As you’re approaching forty, you can come back and start [taking tamoxifen]’?

Discussion in counselling: prompted by participant comfort and provider ambiguity

Participants who engaged in discussion were those who expressed the most comfort with or certainty about the risk score presented. Those who had been previously counselled and had discussed their risk score in the past, or who mentioned in the consultation or interview that they worked with statistics, appeared most comfortable in discussing risk numbers during the consultation. When providers communicated uncertainty about the risk score, undermined the score or presented risk levels surprising to the participant, discussion was also prompted (table 3).

Adding information to risk scores

Some participants tried to understand the risk values presented to them during counselling by providing additional information, which they thought might change their risk level (eg, see table 3).

After counselling: four primary reactions to risk score information

In contrast to the counselling sessions, in the interview participants had lively discussions on risk scores and risk levels. Participants engaged primarily with their risk score in four ways during the interviews: (a) being at ease with the risk score; (b) being anxious about the risk score; (c) accepting the risk score or (d) being non-conversant about the risk scores (table 4). These responses were not mutually exclusive and overlapped in the ways participants engaged with the risk score. Patterns of engagement were viewed according to participants’ risk and health background, the actual risk score, and the way the providers had delivered the information.

Table 4.

After counselling: four primary reactions to risk score Information

Category type Important variables	Important variables	Quote
Being at ease with risk score	Counselling visitation	Participant U: Well, it took me a few times, like I said, initially to understand that 5-year risk for myself… vs 5-year average risk for someone that is not… I deal with probability all the time, but I was looking at it from a different perspective.
Anxious about risk score	Finding lobular carcinoma in situ Counselling visitation	Participant W: It was scary. Who wants to hear that they’re, you know, at double the risk for breast cancer? Nobody wants to hear that. Plus, I’m still confused … Participant V: I feel confused about these statics as I revisit this… I understand she’s doing her best to get a model for me… but it’s hard to combine the two. That’s what I remember thinking too after we left there, that that was going through my mind.
Accepting of risk score	Signalled trust in provider Age range (65 years+)	Participant X: Twenty-six is not a small—I mean, I guess that’s a high number… I would probably say I wouldn’t think so, but now that I have had [an atypical biopsy finding] there’s always a possibility that I could develop breast cancer—that I know is there. Participant R: She reiterated some things to me: ‘You’ll be at four percent if you take it, your risk factor is like 4% rather than maybe like 20%’. Just her numbers and everything make me feel comfortable taking it[medication]. Participant N: I mean, I think I was not more worried. I knew I was at a higher risk, but I didn’t think the ratio was that much more substantial…
Non-conversant about risk score	Comorbidity	Participant J: Basically [what the provider shared with me was how based on] my age, my family history, the probability or possibility within my age group of getting cancer within the next five years and thereafter. And the percentages are mostly what she explained… It’s a big concern for me, the possibility of having a stroke and certainly the possibility of having cancer because—and that’s another problem and situation, you know. The only thing is, you know, it’s weighing one I guess against the other.

Being at ease with the risk score

Some participants discussed the statistical calculations of their risk score easily during the interviews. These participants described their own prior expertise in statistics and familiarity with their breast cancer risk score from previous counselling discussions or through their own work. Irrespective of their risk score, they asked further questions about values or candidly discussed in the interview the way in which they understood a statistical presentation. They strongly signalled that they felt comfortable discussing numbers. In these cases, providers and participants familiar with risk scores shared a common language for describing them during their consultation session (table 4).

Anxious about risk score

A number of participants expressed feeling confused and/or anxious regarding what the risk score meant for them personally (table 4). Unease after discussing the risk score was expressed in one of two principal ways: either through confusion or through anxiety about the information. Most often, when the information was new to them, participants discussed how it was difficult to understand and how they struggled to grapple with the meaning of the risk score for them as an individual. Other participants had a more anxious reaction to their risk score, especially when ambiguity around the risk score was introduced. Although participants from both groups mentioned they were interested in engaging with the numbers and statistics, they also signalled that they had difficulty doing so for lack of experience or because risk factors (such as age or an LCIS diagnosis) rendered the statistical information uncertain and unclear.

Accepting of risk score

During the counselling and thereafter, the majority of participants indicated acceptance of the risk score they received, although this manifested in different ways (table 4). Some participants now felt at risk of getting breast cancer. Others accepted their risk score, but it did not change their feeling of being at risk. Acceptance of the risk score was marked by an expression of trust in the provider.

Non-conversant about risk score

A few participants, for example, some who were older or who had competing health risks, did not discuss their risk score in interviews at all. Some highlighted how other health experiences were of greater importance, which appeared to help them define their understanding and perception of their own risk (table 4).

Discussion

During clinical counselling sessions, risk score values were presented and tailored by providers in ways that gave meaning to the individual patient. As with the use of comparative risk,^{36 37} presentation can affect the way that risk assessment information is absorbed. We found that providers emphasised greater risk reduction benefits or the likelihood of developing disease, they bolstered their recommendations with persuasive statistics and augmented uncertain statistical values with clinical experience. Because challenges exist in determining how each patient comprehends information,²⁷ risk communication research suggest patient-preferred formats, such as visual representations, absolute risk values or comparisons may facilitate communication between the patient and provider.²⁸ However, communication tools in our study appeared to be most useful when patients and providers share a common sense of risk numeracy and trust. In the counselling sessions observed for this study, exchanges between most study participants and providers were provider-led, unless risk numeracy and trust was already established. Without the context of the interview, little meaning could be taken from watching many of the consultations, where there was little reaction to the risk score outside of small affirmations of listening. When counselling introduced uncertainty, however, the active discussion signalled how participants tried to reach an understanding of their risk score and risk factors.

A particular strength of our study is that after the consultation, we were able to parse out in the interviews whether and by what means participants scrutinised their risk score. Most participants integrated risk score understanding into their overall sense of risk and accepted or even embraced the risk score. However, some were unable to reconcile the numbers with what they knew about themselves, which caused anxiety. When risk scores were unimportant to individuals’ risk narratives, other health factors took precedence.

These findings add to previous findings about counselling in the larger study group of the NSABP DMP-1.^{25 26} The category types that describe how the women in the study engaged with the risk score presented to them disentangle how risk assessment and risk information are conveyed in primary cancer prevention settings, including the interactions between provider and participant. For instance, formats that provide numerical outcomes using simple percentages and frequencies, describe changes over time, tailor estimates and convey uncertainty^{36 38–40} are helpful. Despite a tailored presentation, risk scores that cannot be presented in absolute numbers or contain some ambiguity, may still cause anxiety for some individuals. Our study underscores the idea that for risk assessment information to become a meaningful tool for making treatment decisions, patients’ illness and risk experiences must be considered.^{27 28}

The sparse existing research into how risk assessments and risk scores are used in other health contexts points to gaps and limitations in applying assessment tools in counselling sessions with patients about their risk, and echoes our findings about how providers pragmatically tailor risk score information by adding their own knowledge and experience.10 16 41 42 Providers have been found to be more positively inclined toward prescribing prevention interventions when they are more knowledgeable about an intervention,⁹ a finding that is exemplified in our findings from the NSABP DMP-1 study whereby the most experienced counselling providers recommended SERM most frequently.²⁶ This study of how risk assessments are counselled on highlights how conducted risk scores are used to give meaning to recommendations.

Our work has some important limitations. Risk assessments in our sample were routinely conducted as part of breast cancer risk counselling and may not be applicable to other cancer prevention counselling where a risk assessment is not routinely used or such extensive discussions of risk scores might not take place. The number of providers in our sample was small and all but one had extensive experience in counselling. This has an important effect on how risk scores are discussed. Future research would benefit from a validation of these patient types with a wider range of cancer prevention counsellors in non-urban settings.

The findings from this analysis have implications for risk counselling practice using risk assessment tools. Taking a comprehensive view, our analysis of this sample previously suggested that posing patient-centred questions during breast cancer risk counselling will help women assess their own priorities²⁷ and this engagement will enhance the trust in a patient–provider relationship.²⁸ At the same time, the recognition that patient beliefs and understandings are vital to the decision-making process²⁶ advances a strong argument for providers to ask specifically about a patient’s comfort level with numeracy. Risk counselling could be adapted in various ways, for instance by using absolute risk values or visual aids for individuals who are less numerically comfortable or anxious about this information.

Conclusions

As risk assessments become a more frequently used tool in primary cancer prevention counselling, there is an increasing need to understand how providers present these scores and how patients use this information for decisions about their health. For instance, the ways in which sociodemographic factors such as racial/ethnic, educational and income disparities affect how patients are counselled on risk assessment remain unexplored. Providers work to build relationships that ideally lead to candid discussion and shared decision-making about health, however, the exchanges we observed in our sample were provider driven and prompted few exchanges on what these values meant. If risk score information is to move beyond a simple transactional information exchange, understanding the ways that this information is processed during and after counselling is crucial. Having awareness of how risk scores are presented is important because this may influence how information on primary cancer risk is understood and interpreted. Knowing how to make this information meaningful will augment its capacity as decision-making tool for providers and patients.

Data availability statement

No data are available. Data for this study are not publicly available for reasons of ensuring anonymity to participants.

Ethics statements

Patient consent for publication

Consent obtained directly from patient(s).

Ethics approval

This study involves human participants and was approved by the Boston University Medical Campus FWA00000301/IRB00008404/study application number H-31403; The University of Texas MD Anderson Cancer Center FWA00000363/IRB4-IRB00005015/protocol name NSABP DMP-1. Participants gave informed consent to participate in the study before taking part.