Background
Breast cancer is the most common cancer among women in the United States, with an estimated 1 in 8 women developing invasive breast cancer during her lifetime [
1]. More than 252,700 new cases and 40,610 deaths due to breast cancer are expected to occur among U.S. women in 2017 [
2]. Women with a 5-year invasive breast cancer risk greater than 1.67% or a lifetime risk greater than 20% based upon the Gail model [
3] have the option of taking a chemopreventive medication. Chemoprevention with selective estrogen receptor modulators (SERMs) and aromatase inhibitors (AIs) has been shown to reduce invasive breast cancer risk by up to 50–65% among high-risk women in randomized controlled trials [
4‐
7]. Grounded in strong evidence, the U.S. Preventive Services Task Force (USPSTF), National Comprehensive Cancer Network (NCCN), American Society for Clinical Oncology (ASCO), and the National Institute for Health and Care Excellence (NICE) recommend that clinicians discuss preventive therapy with high-risk women [
8‐
11]. However, uptake of SERMs or AIs for the prevention of breast cancer is extremely low [
12]. In a meta-analysis of therapeutic agent uptake to prevent breast cancer among women at increased risk, Smith et al. found that uptake was 25.2% among women screened for clinical trials, but only 8.7% (95% CI, 6.8–10.9) in non-trial settings [
13].
Decision aids can help to improve communication between providers and patients, and can assist patients clarify how important the potential benefits and harms are to them. However, studies that examined the effect of decision aids on SERM use found that while knowledge about risks and benefits increased, decisions were rarely influenced.
Guide to Decide, a decision aid that informed high-risk postmenopausal women about potential chemoprevention benefits and side effects resulted in no tamoxifen uptake and only 0.5% uptake of raloxifene [
14]. The
Ready, Set, GO GAIL! study involved PCPs using the Gail model to screen more than 5700 women [
15]. Although 868 (15.2%) women were classified as high-risk for breast cancer, only 14.7% were referred for risk counseling, 6.4% attended the consultation, and 2% started chemoprevention. The
BreastCARE randomized controlled trial revealed that more women were referred for high-risk consultation in the intervention group compared to controls, however there was limited communication about chemoprevention documented in the medical record [
16].
The literature is surprisingly scant on decision aids targeting both patients and providers [
17], and most studies on chemoprevention decision-making have been based on hypothetical scenarios to evaluate levels of interest, which may ineffectively predict actual uptake [
12]. Studies confirm the importance of the primary care provider recommendation for the decision to take a SERM for breast cancer risk reduction [
18‐
20]. However, this recommendation was more likely to be followed when SERM use discussions assess patient attitudes toward medication and relate to those when discussing chemoprevention options to make the information relevant to the patients [
18]. Since breast cancer chemoprevention is not generally diffused in the primary care setting, more effective tools are needed to inform both providers and patients about the risks and benefits of SERMs and AIs, help them to identify available options and deliberate those options in light of patients values and preferences.
In this paper, we report results of a pilot study conducted to examine the efficacy of two decision support tools, the
RealRisks decision aid (DA) for patients and the Breast cancer risk NAVigation (
BNAV) decision support tool for primary care providers (PCPs) [
21]. Providers received
BNAV decision support at the time their patient completed
RealRisks, thus the tools when integrated into clinical workflow were intended to complement one another. We sought to determine whether our decision support tools increase patient’s accuracy of breast cancer risk perceptions, and breast cancer and chemoprevention knowledge. We also sought to identify referrals for consultation at a high-risk breast clinic, and chemoprevention uptake among women at high risk for breast cancer post intervention.
Discussion
Our intervention demonstrated a significant improvement in accuracy of breast cancer risk perception and an increase in chemoprevention knowledge. It was notable that accuracy of breast cancer risk perceptions significantly improved post RealRisks, and was sustained at 6 months. The finding that decision conflict significantly increased from post-intervention to 6 months or after clinical encounters with PCPs was unexpected. The literature is not clear on why the decision to accept or reject chemoprevention would become more difficult with the passage of time. Possible explanations may reside with the decision itself, the care pathway when the DA was introduced, or a combination of both.
Decision conflict is not simply recognition that advantages and disadvantages exist for any given option. It is also an undesirable state of discomfort and internal conflict experienced when facing a difficult decision [
38]. The decision to take chemoprevention can raise an undesirable state of discomfort given the perception that SERMs or AIs are “cancer drugs” and tradeoffs between the risks and benefits of these medications. It has been proposed that the side effect profile of chemoprevention medications, coupled with wide-ranging concerns about the emotional impact of taking a medication, leads to medication avoidance. Specifically, women might decide to avoid chemoprevention because of the affect-laden responses associated with the term “side effect” and their belief that these medications will increase, rather than decrease, their level of health-related stress [
39]. [
40‐
42] Concerns about possible side effects, such as uterine cancer, thromboembolism, and menopausal symptoms, are the primary reasons why women are reluctant to start breast cancer chemoprevention [
43‐
49]. This may also explain results found in previous studies that improved knowledge may result in women becoming more reluctant to take medications that are associated with potentially harmful health risks [
14,
50].
It is recognized that negative emotions experienced while making a choice involving difficult tradeoffs can potentially impact decision conflict over time. The judgment and decision-making literature suggests that people can react to emotionally laden decisions by altering the amount or content of thought about the decision (emotion-focused coping). This can result in avoidant behaviors, for example declining to make a decision [
51], allowing another make the decision for you, or exhibiting an increased tolerance for the status quo option [
52]. Studies suggest that decision makers are likely to face between attribute tradeoffs required by decision conflict when the attributes are relatively low in emotional tradeoff difficulty. Conversely, they tend to elude these tradeoffs when attributes are higher in emotional tradeoff difficulty. Thus, it is plausible that increased decision conflict from post intervention to 6 months was due to decision avoidance, given that the decision to take chemoprevention requires women to evaluate a number of emotionally laden tradeoffs, mainly between the potential benefits and perceived barriers to chemoprevention medication uptake [
53].
As previously asserted in the literature, it is mostly acceptable for decisional conflict to be high if measured shortly after options have been presented, however after the patient has been given the opportunity to incorporate their preferences into the presented options and make a decision, decisional conflict should be low [
54]. An equally plausible assertion is that for a patient who prudently weighs competing options decision conflict will be high and even with the passage of time, a rational patient would state that the decision was difficult [
55,
56]. Few randomized controlled trials have investigated timed measurements of decision conflict, particularly in underserved populations. Our conclusions are limited due to study design and the lack of a control group in our pilot study, however this area warrants further investigation.
With respect to care pathway, although we tried to integrate the web-based patient and provider decision support tools into clinic workflow, use of the RealRisks DA was not closely linked with the PCP clinical encounter. Many of the patient-reported outcomes, such as accurate breast cancer risk perceptions and chemoprevention knowledge, diminished from immediately post-intervention to 6 months or after the clinical encounter. The increase in decision conflict may also reflect a diminished patient outcome since significantly fewer women were in the decision implementation phase according to the DCS. Nearly a quarter of women expressed interest in chemoprevention uptake after completing RealRisks, yet few women were referred for high-risk consultations and none of the participants had initiated chemoprevention at 6 months. Perhaps linking use of the DA more proximally to the clinical encounter will improve our patient reported outcomes. In future work, we have incorporated alerts and other cues to remind the patient to complete the RealRisks DA within a week prior to the clinical encounter.
Similar to previous studies, our intervention demonstrated that decision aids improve knowledge in those who use them. However, also like previous studies, increased knowledge does not lead to increased chemoprevention uptake for the purpose of reducing breast cancer risk. Improved accuracy of risk perceptions is an important measure of the quality of a decision aid [
57], and while this is a prerequisite to informed decision making, it is difficult to know if this results in clinically useful decision making. Even when breast cancer risk seems to be understood, willingness to take chemoprevention medication remains low among women who are identified as eligible based on their Gail risk score [
14,
58]. Previous research has demonstrated that health decision making may be based on heuristics and feelings, rather than on an accurate understanding of risk information [
59,
60]. As such, individuals may not always process and act upon the risk information presented to them in the ways that healthcare providers intend. In addition to precise probabilistic risk information, lived experiences and particularly individual experiences with cancer have been shown to influence chemoprevention decisions [
58,
61].
Several studies have demonstrated that recommendations from physicians and effective communication greatly affect patients’ decision making in chemoprevention uptake [
43]. Based upon data from key informant interviews, we found that PCPs reveal unfamiliarity with breast cancer risk assessment tools such as the Gail Model and a lack of confidence in prescribing chemoprevention. PCPs also reveal a preference to refer their patients to specialists for consultation about breast cancer risk reduction options, which may imply that the they were not well-informed about breast cancer preventive strategies available to patients [
25]. Although all high-risk women and PCPs were given information on the CUMC breast clinic for high-risk consultations, it appears this was insufficient to alter practice patterns for most PCPs.
Trials of decision support tools designed to increase uptake of breast cancer chemoprevention targeting both patients and providers have been limited. Uptake remains low [
14], [
15] and as demonstrated in a randomized controlled trial of the
BreastCARE intervention discussions about chemoprevention were still limited [
16]. This prior literature suggests that just targeting high-risk women or PCPs alone is ineffective. Additional work remains to better understand the impact of decision aids targeting both patients and providers.
Limitations
Limitations of our study include the lack of a concurrent control arm, the relatively small sample size, and conducting the study in an urban academic center with access to a high-risk clinic, all of which limit the generalizability of our findings. In addition, although we found no significant differences in age, race, and racial distribution between the pilot sample and KYRAS high-risk patients, our sample was self-selected from the larger KYRAS screening study. The study sample was older than what we expected and may not be totally representative of a higher risk younger population who are likely more appropriate for chemoprevention and the RealRisks DA. Additionally, we had higher than anticipated loss to follow-up of about 20%. Our short-term follow-up of 6 months may have been insufficient to assess clinical outcomes such as high-risk clinic referrals and actual chemoprevention uptake.
Conclusions
We developed decision support tools for both patients and their PCPs, which include personalized risk reports and education about breast cancer risk and chemoprevention. Our study population was racially and ethnically diverse and our patient-centered DA, which is available in English and Spanish, was rigorously tested in women of multiple ethnicities with varying levels of health literacy and numeracy. In addition to using validated outcome measures, we were able to assess referrals to the high risk breast clinic and actual chemoprevention uptake using electronic health records.
The results of our initial pilot study have informed the design and conduct of a larger randomized controlled trial of 300 high-risk women assigned to standard educational materials alone or in combination with
RealRisks and
BNAV (NCT03069742). We will target younger, healthier women with higher breast cancer risk, including those with high-risk benign breast lesions such as atypical hyperplasia and lobular carcinoma in situ. These women are likely to derive a greater benefit from breast cancer risk reduction and a lower risk of serious side effects. To reinforce use of the patient-centered
RealRisks DA, we will set up automated reminders to revisit the tool prior to their next PCP clinical encounter, next screening mammography visit, and birthday (as breast cancer risk increases with advancing age). We will enhance provider engagement with an enhanced
BNAV tool, which will offer continuing medical education (CME) credit and additional modules on breast cancer screening and other topics relevant to PCPs. We have already developed modules on risk communication and shared decision making. We will elicit patient preferences for chemoprevention, specifically factors that are most important and least important to chemoprevention decisions, which we will summarize for providers prior to the clinical encounter. Even if we do not observe a significant increase in chemoprevention uptake with the addition of
RealRisks and
BNAV compared to standard educational materials, our goal is to also increase informed choice, decrease decision conflict, and facilitate SDM during the clinical encounter. Facilitating discussions about breast cancer chemoprevention between clinicians and high-risk women is in accordance with recommendations from the U.S. Preventive Services Task Force (USPSTF), American Society for Clinical Oncology (ASCO), National Comprehensive Cancer Network (NCCN), and the National Institute for Health and Care Excellence (NICE) [
8‐
11].