Making Informed Choices On Incorporating Chemoprevention into carE (MiCHOICE, SWOG 1904): Design and methods of a cluster randomized controlled trial

Abstract

Introduction:

Women with atypical hyperplasia (AH) or lobular carcinoma in situ (LCIS) have a significantly increased risk of breast cancer, which can be substantially reduced with antiestrogen therapy for chemoprevention. However, antiestrogen therapy for breast cancer risk reduction remains underutilized. Improving knowledge about breast cancer risk and chemoprevention among high-risk patients and their healthcare providers may enhance informed decision-making about this critical breast cancer risk reduction strategy.

Methods/design:

We are conducting a cluster randomized controlled trial to evaluate the effectiveness and implementation of patient and provider decision support tools to improve informed choice about chemoprevention among women with AH or LCIS. We have cluster randomized 26 sites across the U.S. through the SWOG Cancer Research Network. A total of 415 patients and 200 healthcare providers are being recruited. They are assigned to standard educational materials alone or combined with the web-based decision support tools. Patient-reported and clinical outcomes are assessed at baseline, after a follow-up visit at 6 months, and yearly for 5 years. The primary outcome is chemoprevention informed choice after the follow-up visit. Secondary endpoints include other patient-reported outcomes, such as chemoprevention knowledge, decision conflict and regret, and self-reported chemoprevention usage. Barriers and facilitators to implementing decision support into clinic workflow are assessed through patient and provider interviews at baseline and mid-implementation.

Results/discussion:

With this hybrid effectiveness/implementation study, we seek to evaluate if a multi-level intervention effectively promotes informed decision-making about chemoprevention and provide valuable insights on how the intervention is implemented in U.S. clinical settings.

Trial Registration:

NCT04496739

Introduction

Breast cancer is the most common malignancy among women in the U.S.¹ Based upon results from several randomized controlled trials (RCTs), chemoprevention or risk-reducing medications with antiestrogen therapy, such as selective estrogen receptor modulators (SERMs) and aromatase inhibitors (AIs), have been shown to reduce breast cancer incidence by up to 50–65% among high-risk women.^2–6 Due to the strength of this evidence, 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) recommend clinicians discuss chemoprevention with high-risk women.^7–10 Women with atypical hyperplasia (AH) or lobular carcinoma in situ (LCIS) have a 4- to 10-fold increased risk of breast cancer compared to women with non-proliferative breast disease.¹¹ They derive a greater benefit from antiestrogen therapy compared to other high-risk populations, such as women with a strong family history.¹² For example, the relative breast cancer risk reduction from SERMS and AIs among the subgroup of women with AH ranged from 41–79%.^2,3,5,6,13 Therefore, this is an important high-risk population to target for use of antiestrogen therapy for primary prevention.¹⁴ Reasons for low uptake include insufficient patient and clinician knowledge about SERMs and AIs and concerns about side effects.^15,16 Further research is needed to determine how best to improve knowledge about breast cancer risk and chemoprevention among high-risk women and their healthcare providers.

Intervention trials of clinical decision support (CDS) tools that are designed to increase uptake of breast cancer chemoprevention have been met with limited success.^17–20 Our study intervention differs from the prior literature since we are targeting both patients and clinicians and our patient-facing decision aid has been rigorously tested among racially/ethnically diverse women.^21–26 We developed web-based decision support tools, RealRisks and Breast cancer risk NAVigation (BNAV), for patients and healthcare providers, respectively. An iterative design process was used to equitably maximize acceptability, appropriateness, and usability.^21–24 We completed a RCT of standard educational materials alone or combined with RealRisks and BNAV among 300 high-risk women eligible for chemoprevention screened in the primary care setting and their healthcare providers.^25–28 Comparing the intervention and control arms at 1 month, there was a significant increase in informed choice, defined as adequate chemoprevention knowledge and attitudes congruent with decision about chemoprevention (41% vs. 23%, p=0.003), and decreased mean decision conflict or uncertainty about decision-making (34.0 vs. 47.0, p<0.001), but no difference in chemoprevention uptake.²⁶

Since chemoprevention is not widely adopted by high-risk women, more effective approaches are needed to educate both patients and healthcare providers about the risks and benefits of SERMs and AIs for primary prevention. Studies involving specialized risk counseling at a breast clinic report chemoprevention uptake rates ranging from 11% to 58%.^29–35 Therefore, higher uptake may be achieved with health professionals and patients who have sufficient knowledge about breast cancer risk and risk-reducing options. Given the preference-sensitive nature of chemoprevention decision-making, our objective is to evaluate the effectiveness of CDS tools designed to promote informed shared decision-making about breast cancer chemoprevention.

Methods

Trial design

Our study design (Figure 1) is a multicenter cluster-RCT with 26 “recruitment centers” randomly assigned (1:1) to either the intervention or control arms. Recruitment centers are defined as an outpatient clinic or group of clinics belonging to the same National Clinical Trials Network (NCTN), NCI Community Oncology Research Program (NCORP), or Minority/Underserved (MU)-NCORP. Recruitment centers are randomized with stratification by patient volume (≤100 vs. >100 patients with AH or LCIS seen per year) and recruitment center type (MU-NCORP vs. non-MU-NCORP or NCTN). A total of 200 healthcare providers and 415 patients with AH or LCIS are enrolled and assigned to the control or intervention arms of standard educational materials alone or combined with RealRisks and BNAV, based upon randomization assignment at the site level. Online questionnaires are completed by healthcare providers at baseline and after the follow-up clinical encounter for each enrolled patient. Patient participants complete questionnaires at baseline, 6 months (after the follow-up study visit) and 12 months, then yearly follow-up for clinical outcomes. The primary outcome is chemoprevention informed choice, given that professional organizations recommend that clinicians discuss chemoprevention with high-risk women and the preference-sensitive nature of decision-making.

Figure 1.

Study Schema

Study setting

The trial is coordinated by the SWOG Cancer Research Network. Columbia University Irving Medical Center (CUIMC) is responsible for screening and selecting recruitment centers for randomization, recruited healthcare providers, centrally administering the study intervention, and conducting the virtual interviews of patients and providers. The SWOG Statistics and Data Management Center in Seattle, WA is responsible for providing the statistical design, performing cluster randomization of the selected recruitment centers, managing the registration of patient participants and collection of patient-reported and clinical outcomes, responding to recruitment center questions on patient-participant eligibility, study procedures and data submission, and reviewing patient-reported and clinical outcome data. The SWOG Network Operations Center in San Antonio, TX, manages all protocol-related updates, National Cancer Institute (NCI) Division of Cancer Prevention and NCI Central Institutional Review Board (CIRB) regulatory filings and communications to Clinical Trials Support Unit (CTSU) for document posting.

Ethical oversight

The protocol and its amendments have been approved by the NCI CIRB and registered at ClinicalTrials.gov NCT04496739. All patient and provider participants provided informed consent.

Eligibility criteria

Eligibility criteria for recruitment centers include the following: 1) NCTN, NCORP or MU-NCORP institutions. If the recruitment center has multiple clinics, they are required to be part of the same NCTN, NCORP, MU-NCORP, or subcomponent site; 2) have an active electronic health record (EHR) and patient portal used in the outpatient clinics; 3) see at least 50 patients with AH and/or LCIS patients per year; 4) identify a lead site principal investigator (PI) to facilitate recruitment and retention of patients and healthcare providers and participate in quarterly stakeholder conference calls; 5) willing to register a minimum of 5 healthcare providers and 8 patients to the study.

Eligibility criteria for provider participants include: 1) regularly seeing patients with AH or LCIS at a participating recruitment center, including physicians and advanced practice providers; 2) being willing to provide informed consent, complete an online baseline questionnaire and follow-up questionnaire after the 6-month study visits with enrolled patients; and 3) have an active registration in the Cancer Therapy Evaluation Program (CTEP).

Eligibility criteria for patient participants include: 5) women, age 35–74 years at registration (since the Breast Cancer Surveillance Consortium [BCSC] risk calculator is valid in this age range); 2) histologically-confirmed AH or LCIS documented by breast pathology report; 3) no history of invasive breast cancer or ductal carcinoma in situ (DCIS); 4) no prior or current use of SERMs or AIs; 5) no current use of hormone therapy; 6) no history of bilateral mastectomies or breast implants (since the BCSC risk calculator is not applicable in these women); 7) not pregnant or lactating; 8) premenopausal patients must not have a history of thromboembolism (since tamoxifen is contraindicated in these patients); 9) able to read and write in English or Spanish; and 10) able to access the internet and receive email or text messages. Only providers enrolled in S1904 are allowed to register their patients.

Patient and provider participants are eligible for the interview component of the study if they are in the intervention arm and agree to be re-contacted for future research when consenting to the parent trial.

Study procedures

The CUIMC research team screened recruitment center applications for eligibility. After a recruitment center is approved and randomized to the intervention or control arms, recruitment of healthcare providers from that center began. Our target accrual is 200 providers. The CUIMC research team works with the recruitment center’s lead PI to identify healthcare providers who regularly see patients with AH or LCIS and may be interested in participating. The CUIMC research team sends these providers recruitment emails. After providing an online informed consent, providers complete a baseline online questionnaire on their demographics and practice characteristics. If a provider practices at a recruitment center randomized to the intervention arm, s/he is given login credentials to access BNAV.

Following enrollment at recruitment centers, healthcare providers allow study personnel to approach their high-risk patients for participation. Patient participants are registered under an enrolled provider to create a study dyad. The study will enroll 415 patient participants, or an average of about 16 patients per recruitment center. Upon providing informed consent, patient participants complete a baseline questionnaire and schedule a follow-up visit (either virtual or in-person in the clinic) with their study provider at 6 months after registration. After completing the baseline questionnaire, patient participants are provided login credentials to access RealRisks. Upon logging into their accounts, patients from recruitment centers randomized to the intervention arm are able to access the full RealRisks decision aid and standard educational PDFs created by Susan G. Komen^®. Patients from recruitment centers randomized into the control arm are only able to access the standard educational PDFs (Figure 2). After the accounts are sent, CUIMC research staff send patients in the intervention arm three weekly emails or text messages reminding them to complete RealRisks. To link usage of RealRisks with the clinical encounter, email or text message reminders to revisit RealRisks are sent in the 2 weeks prior to the 6-month follow-up visit. The window for the follow-up study visit was expanded to 3–7 months to allow for earlier follow-up. After the follow-up visit, patient participants complete a follow-up questionnaire and providers complete a follow-up questionnaire for each of their enrolled patients to assess their shared decision-making during the clinical encounter. Patient participants complete a final questionnaire at 12 months after registration and are followed yearly for up to 5 years for clinical outcomes (Tables 1–2).

Figure 2.

Screenshots of RealRisks and BNAV decision support tools: (A) RealRisks modules for patients in the intervention arm; (B) Standard educational materials for patients in the control arm; (C) BNAV modules for healthcare providers in the intervention arm.

Table 1.

Schedule of Study Evaluations for Patients

Patient Measures	Description	Baseline	3–6 months1	12 months1	24 months1	36 months1	48 months1	60 months1
Baseline characteristics	Demographics and breast cancer risk factors	X
Chemoprevention Knowledge44	8-item scale assessing knowledge about the risks of hormonal symptoms, cataracts, broken bones, and breast cancer with the use of SERMs. Adequate knowledge is defined as answering at least 50% of the items correctly. We have added 10 additional items to assess knowledge on the risk of uterine cancer and blood clots with each SERM and the risks and benefits of AIs. Knowledge for the primary endpoint are scored based on the 8 core items; secondarily, we assess chemoprevention knowledge based upon the full 18 items.	X	X	X
Attitudes37	1-item five-point Likert-type scale about how good of a choice taking chemoprevention would be, with responses ranging from “For me it is not a good choice at all” (1) to “For me it is an extremely good choice” (5). Scores of 1 or 2 indicate a “negative attitude”, a score of 3 indicates a “neutral attitude”, and scores of 4 or 5 indicate a “positive attitude.”	X	X	X
Chemoprevention Decision	A 1-item multiple choice question assessing a patient’s decision about whether to take chemoprevention with responses of: no decision yet; decided to take tamoxifen, raloxifene, anastrozole or exemestane; or decided to take no chemoprevention.		X	X
Chemoprevention Informed Choice37	This measure combines three component measures: a chemoprevention knowledge scale, a 1-item attitudes scale, and the participant’s chemoprevention decision. Informed choice is defined as any one of the following: 1) having adequate knowledge per the knowledge scale, having a positive attitude towards chemoprevention, and deciding to take chemoprevention; 2) having adequate knowledge per the knowledge scale, having a negative attitude towards chemoprevention, and deciding not to take chemoprevention; and 3) having adequate knowledge per the knowledge scale, having a neutral attitude towards chemoprevention, and not yet making a decision on chemoprevention.		X2	X
Perceived Breast Cancer Risk45	1) Absolute estimate: “Very Low” to “Very High”; 2) Comparative risk: assessed on a 7-point Likert scale from “Much lower” (1) to “Much Higher” (7); and 3) Absolute numeric risk: on a visual analog scale from 0% to 100%. Actual absolute 5-year and 10-year invasive breast cancer risk estimates according to the BCSC risk calculator46 are recorded.	X	X	X
Accuracy of Risk Perception46	The difference between numeric perceived 10-year invasive breast cancer risk (subjective) and actual 10-year risk according to the BCSC risk calculator (objective). These are categorized as accurate if the difference between subjective and objective risk estimates is ≤10% in either direction, an underestimate if subjective risk is >10% below objective risk, and an overestimate if subjective risk is >10% above objective risk.45,47,48	X	X	X
Breast Cancer Worry49,50	2 Likert-style items with responses ranging from “Not at all” (1) to “All of the time” (7).	X	X	X
Decision Conflict51	10-item scale with 3 response categories (Yes/Unsure/No) to assess the level of conflict women feel about their decisions regarding chemoprevention. Scores range from 0 (no decisional conflict) to 100 (extremely high decisional conflict).	X	X	X
Shared Decision-Making for Patients52	9-item scale with Likert-style responses with scores ranging from 0 to 100		X
Decision Regret53	5-item scale composed of 5-point Likert-style items. Scores range from 0 (no regret) to 100 (high regret).			X
Chemoprevention Usage17	(1) report having started taking or currently taking a SERM or AI for primary prevention, as assessed by self-report; and (2) taking at least one dose of a SERM or AI for primary prevention, based on EHR logs. Reasons for never starting chemoprevention are also recorded.			X	X	X	X	X
Chemoprevention Adherence54,55	3-item scale only for patients currently taking a SERM or AI. Responses of 1 through 5 with 1 corresponding to perfect adherence. The three items are averaged if at least 2 of the 3 items have responses. Non-adherence is defined as an average score of greater than 1.			X	X	X	X	X
Reasons for Chemoprevention Discontinuation17	Include side effects, costs, other			X	X	X	X	X
Patient Mid-Implementation interviews			X3

¹Window for study evaluations is +/− 90 days

²Primary endpoint

³Within 90 days after the follow-up visit

Table 2.

Schedule of Study Evaluations for Providers

Provider Measures	Description	Baseline	Patient’s 6-month visit1	>6 months after enrollment
Personal and Practice Characteristics	Demographics and practice patterns with respect to high-risk women and chemoprevention	X
Shared Decision-Making for Providers56	9-item scale with Likert-style responses with scores ranging from 0 to 100.		X2
Provider Implementation Interviews		X		X

¹Window for study evaluations is +/− 90 days

²Conducted for each of provider’s enrolled patients

Additionally, we explore barriers and facilitators to chemoprevention use and assess the implementation of RealRisks and BNAV into clinical workflow in diverse practice settings. We conduct telephone/video-conference interviews of healthcare providers (baseline and mid-implementation) and high-risk women with AH or LCIS (after the follow-up study visit at 6 months) assigned to the active intervention.

Study interventions

RealRisks is designed to improve: 1) accuracy of breast cancer risk perceptions; 2) self-efficacy in engaging in dialogue about breast cancer risk and chemoprevention; and 3) chemoprevention informed choice. Figure 2 provides screenshots of the RealRisks modules for the intervention and control arms. To develop RealRisks, we conducted multiple design sessions, including participatory workshops²¹, usability studies²² and pilot testing²³, to arrive at guiding principles. These focused on: 1) options for providing information on breast cancer risk, 2) “interactive games” to communicate breast cancer risk, and 3) patient preference elicitation to weigh the risks and benefits of chemoprevention. We designed RealRisks to first provide general education in graphic novel format (narrative light) or slide presentations (narrative dense). The decision aid, which includes audio and Spanish translations, is organized into modules: 1) Breast Cancer Risk (breast cancer risk factors, calculation of personal breast cancer risk, interactive games on risk communication); 2) Family History and Genetic Testing; 3) Chemoprevention; and 4) Lifestyle Behaviors. Through RealRisks, we collect information on breast cancer risk factors (age, race/ethnicity, first-degree family history of breast cancer, breast biopsy results, and mammographic density) to calculate 5- and 10-year invasive breast cancer risk according to the BCSC risk calculator. Upon completion of RealRisks, an action plan is generated summarizing the patient’s personalized breast cancer risk profile and preference elicitation for chemoprevention.

The BNAV provider tool (Figure 2C) includes educational modules on breast cancer risk, genetic testing, screening, chemoprevention, and patient-centered care, and a patient dashboard with a summary of their patient’s breast cancer risk profile based upon her interactions with RealRisks.³⁶ Each educational module consists of: 1) self-paced lectures with slide presentations and audio; and 2) case-based learning modules with quizzes. Both RealRisks and BNAV are designed to provide self-directed learning outside of the clinical encounter. Each module takes about 10–15 minutes to complete and can be viewed during multiple sittings. Log-in times and usage logs are monitored for both tools. Both patients and providers in the control arm may be given access to RealRisks and BNAV, respectively, at the conclusion of the trial.

Outcome measures

Tables 1 and 2 summarize the primary and secondary outcomes, as well as the schedule of study evaluations for patients and providers, respectively. The primary endpoint is chemoprevention informed choice at 6 months from individual patient registration. This measure combines three component measures (Table 3): a chemoprevention knowledge scale, a 1-item attitudes scale, and the participant’s chemoprevention decision. Informed choice is defined as any one of the following: 1) having adequate knowledge per the knowledge scale, having a positive attitude towards chemoprevention, and deciding to take chemoprevention; 2) having adequate knowledge per the knowledge scale, having a negative attitude towards chemoprevention, and deciding not to take chemoprevention; and 3) having adequate knowledge per the knowledge scale, having a neutral attitude towards chemoprevention, and not yet making a decision on chemoprevention.³⁷

Table 3.

Outcome key for informed choice

Outcome Components			Outcome
Knowledge	Attitude	Decision (Intention)	Informed Choice
Adequate	Positive	Take chemoprevention	Yes
Adequate	Negative	Do not take chemoprevention	Yes
Adequate	Neutral	No decision yet	Yes
Any other combination (e.g., inadequate knowledge or discordance between attitude and decision)			No
If one or more components are missing			Missing

Statistical considerations

The primary endpoint is chemoprevention informed choice at 6 months from individual patient registration. In a RCT of RealRisks and BNAV conducted at CUIMC³⁸, we observed an absolute difference in informed choice of 20% between the intervention and control arms. Acknowledging potential anti-conservatism in this estimate, particularly when implemented across many sites, we have designed this trial to have adequate power (90%) to detect a 15% increase in the frequency of informed choice with a 1-sided 0.025 level test, with application of a continuity correction for binomial data. These calculations incorporate an intraclass correlation (ICC) parameter. The ICC adjusts for the added variability induced by clustering. There is minimal information to guide the choice of ICC in this setting, but assuming an ICC of 0.02, roughly equal accrual at each recruitment center, and 374 evaluable patients, we have 90% power to detect a 15% absolute difference in informed choice as long as the event rate in the control arm is ≤10%. Sensitivity to some of these parameters can be seen in Table 4.

Table 4.

Required sample size of eligible patients to achieve 90% power based on a one-sided 0.025 level test, assuming a mean of 15 eligible patients per cluster (N=26 total clusters).

Intraclass Correlation Coefficient	Control Group Probability of Informed Choice
		Intervention Group Probability of Informed Choice
		.2	.25	.3
0	.05	228	155	110
	.1	572	292	184
	.15	2504	708	348
0.01	.05	260	171	125
	.1	652	333	210
	.15	2855	807	397
0.02	.05	292	192	141
	.1	732	374	236
	.15	3205	906	445
0.03	.05	324	213	156
	.1	812	415	261
	.15	3556	1005	494
0.05	.05	388	255	187
	.1	972	496	313
	.15	4257	1204	592

These power calculations are based on a mean cluster size of 15 evaluable patients resulting in a requirement of 25 recruitment centers. The number of centers has been increased by one to provide balance between arms, for a total of 26 recruitment centers randomized 1:1 to either Group 1 (control) or Group 2 (intervention). Recognizing the potential losses to ineligibility, follow-up, and missing outcomes data (estimated 10%), the total accrual goal is 415. Accordingly, we require an average of 16 patients (415/26) registered per recruitment center.

If a recruitment center with any eligible patient accrual drops out, they are retained for follow-up and analysis. If a recruitment center drops out or is terminated without any patient accrual, we identify a replacement recruitment center for participation in the trial from a reserve list of eligible recruitment centers. We randomize the order of these replacement recruitment centers within stratification factors (if possible) to limit potential bias.

All enrolled patients who provide outcomes data at the 6-month timepoint are included in the primary analysis under the intention-to-treat (ITT) principle. The primary analysis of the intervention effect on informed choice at 6 months is based on a logistic regression model using a generalized estimating equation (GEE) approach to account for clustering. Additional supportive analyses adjust for both patient- and clinic-level factors. If there are substantial missing data for the components of the primary endpoint, a sensitivity analysis is conducted imputing a response of no informed decision-making for those women with missing information.

The analyses of secondary endpoints is following the ITT approach used for the primary endpoint to the extent possible. Initial comparisons between active and control groups are conducted using visual displays and Chi-square tests for categorical variables (i.e., accurate breast cancer risk perceptions) and Student’s t-tests for continuous variables (i.e., decision conflict). Depending on the scale of each outcome variable, the impact of the intervention on these outcome variables is assessed with linear or logistic regression models, fit under a GEE approach to account for the ICC assumed in this cluster randomized design.

Preliminary results

The trial was activated on September 1, 2020. Twenty-six recruitment centers were randomized to the intervention or control arms. However, six sites, four in the control arm and two in the intervention arm, dropped out prior to enrolling any patients. Five recruitment centers were randomized as replacement sites. Currently, there are fourteen sites randomized to the intervention arm (including 2 MU-NCORP and 5 high-volume sites) and eleven sites to the control arm (including 2 MU-NCORP and 3 high-volume sites).

As of April 1, 2024, 210 healthcare providers have been enrolled (including 10 from recruitment centers that dropped out prior to enrolling any patients) and 389 patients (94%) out of a total target accrual of 415 have been enrolled.

Conclusions

Chemoprevention uptake remains low among women at high-risk for breast cancer.¹⁴ Intervention trials of CDS tools designed to increase uptake of breast cancer chemoprevention have met with limited success. A web-based decision aid called Guide to Decide, which informed high-risk postmenopausal women about the risks and benefits of SERMs yielded only 0.5% uptake of raloxifene and no uptake of tamoxifen.¹⁷ The Ready, Set, GO GAIL! project involved systematically screening over 5,700 women by primary care providers (PCPs) using the Gail model.¹⁸ The study identified 868 women who met high-risk criteria, with 128 (14.7%) getting referred for specialized risk counseling, 60 (6.4%) who completed the consultation, and only 17 (2%) who initiated a SERM. The BreastCARE intervention used a tablet-based patient intake tool in the primary care setting that generated individualized breast cancer risk reports for patients and their physicians.¹⁹ In a RCT of the BreastCARE intervention among 1,235 women (age 40–74 years), more high-risk women were referred for specialized risk counseling in the intervention compared to the control arm (18.8% vs. 4.1%). However, few of these women had discussions about chemoprevention (1% vs. 0%). Orlando et al. reported on the MeTree family history screener, which generated tailored risk reports for patients and healthcare providers.²⁰ None of the 26 women eligible for chemoprevention initiated antiestrogen therapy for breast cancer risk reduction.

Women with high-risk benign breast lesions, such as AH and LCIS, have a 4–10 fold higher breast cancer risk compared to women with non-proliferative breast disease.³⁹ Because the majority of breast cancers that develop in these women are estrogen receptor-positive¹², these women derive a significant breast cancer risk reduction from SERMs and AIs. We assessed uptake of SERMs and AIs among 1,719 patients diagnosed with AH, LCIS, or DCIS.⁴⁰ About a third of these women were referred to a medical oncologist, which was associated with a greater than 5-fold increase in chemoprevention uptake compared to those who were not referred. This suggests that higher uptake may be achieved with providers who are knowledgeable about chemoprevention.

Prior literature suggests it is insufficient to target high-risk women or healthcare providers individually to increase chemoprevention uptake.^17–20 For this reason, we designed this study to operate on multiple levels of the social-ecological model.⁴¹ RealRisks and BNAV target high-risk women and healthcare providers, respectively. By empowering patients with self-directed learning, which has been demonstrated to be a successful knowledge translation strategy^42,43, and providing them and their providers with individualized breast cancer risk estimates and preference elicitation, this prepares both parties for shared decision-making. Through qualitative interviews of both patients and providers, we seek to better understand the implementation in diverse clinical settings and provide valuable data on successes and limitations.

The sites participating in the MiCHOICE trial include academic centers and community NCORP sites throughout the U.S. They include four MU-NCORP sites which will enhance the diversity of the study population and generalizability of our findings. To enhance accrual of Hispanic women, RealRisks is available in both English and Spanish and has been tested in minority groups with varying health literacy and numeracy.^21,22,24 The RealRisks and BNAV tools may be particularly useful at community sites and rural settings without access to high-risk breast clinics that provide specialized risk counseling.

Compared to prior research, this study offers the following advantages: 1) targeted recruitment of high-risk women with AH or LCIS, who are more likely to benefit from SERMs and AIs for primary prevention; 2) decision support tools designed for both patients and healthcare providers; 3) a patient-centered decision aid, which is available in English and Spanish and has been evaluated in ethnically diverse high-risk women with various educational backgrounds and health literacy; and 4) implementation of the study intervention in diverse geographic and clinical settings. Through this hybrid effectiveness/implementation study design, we seek to: 1) evaluate if our multi-level intervention is effective at increasing informed decision-making about chemoprevention; and 2) provide valuable insights on how CDS tools are effectively implemented in diverse settings.