Return of polygenic risk scores in research: Stakeholders’ views on the eMERGE-IV study

Summary

Research on polygenic risk scores (PRSs) for common, genetically complex chronic diseases aims to improve health-related predictions, tailor risk-reducing interventions, and improve health outcomes. Yet, the study and use of PRSs in clinical settings raise equity, clinical, and regulatory challenges that can be greater for individuals from historically marginalized racial, ethnic, and other minoritized communities. As part of the National Human Genome Research Institute-funded Electronic Medical Records and Genomics IV Network, we conducted online focus groups with patients/community members, clinicians, and members of institutional review boards to explore their views on key issues, including PRS research, return of PRS results, clinical translation, and barriers and facilitators to health behavioral changes in response to PRS results. Across stakeholder groups, our findings indicate support for PRS development and a strong interest in having PRS results returned to research participants. However, we also found multi-level barriers and significant differences in stakeholders’ views about what is needed and possible for successful implementation. These include researcher-participant interaction formats, health and genomic literacy, and a range of structural barriers, such as financial instability, insurance coverage, and the absence of health-supporting infrastructure and affordable healthy food options in poorer neighborhoods. Our findings highlight the need to revisit and implement measures in PRS studies (e.g., incentives and resources for follow-up care), as well as system-level policies to promote equity in genomic research and health outcomes.

A focus groups study with patients, community members, clinicians, and members of IRBs found support for the development of PRSs and interest in having PRS results returned to research participants but also multi-level barriers and significant differences in stakeholders’ views about what is needed and possible for successful implementation.

Introduction

Common chronic diseases have a major impact on individual wellbeing and are leading causes of death and disability. Studies indicate that 6 in 10 adults in the United States have a chronic disease and that 4 in 10 have ≥2 chronic diseases.^1,2 Chronic diseases also have significant costs for individuals and society. Eighty-six percent of U.S. annual healthcare costs cover healthcare services to patients with common chronic diseases,³ with additional indirect costs due to work absences, lost wages, and, more generally, lost economic productivity.⁴ Efforts to decrease the prevalence and outcomes of common chronic diseases are ongoing across social, medical, and scientific fields.

Polygenic risk scores (PRSs) are emerging as a promising tool to improve clinical outcomes for a range of common, genetically complex chronic diseases. Unlike genetic tests for single genetic variants of large effect, PRSs provide a summary risk score of the often individually minuscule effects of thousands of genetic variants. Today, PRSs are available to assess genetic susceptibility to outcomes such as breast and prostate cancers,^5,6,7 and they have demonstrated reliable but modest predictive value for other complex conditions such as diabetes and coronary artery disease^8,9,10 (we note that different approaches exist to calculate the predictive PRS scores, and/or to determine the threshold for “high-risk” PRS results. For further discussion on this issue, see Lewis et al.¹¹). Efforts to improve and validate PRS are central to the preventive goals of genomic medicine.^12,13 More specifically, hopes have been raised that health-related predictions based on PRSs will facilitate early screening to detect emerging health conditions with available risk-reducing interventions,¹⁴ and that genomic-based risk scores will increase motivation for behavioral changes among individuals at high risk for many common chronic diseases (e.g., dietary and exercise interventions for type 2 diabetes).¹⁴

Although studies that assess the long-term impact of genetic information on behavioral changes are limited across diverse participants, some studies are promising.^15,16 However, the use of PRSs in clinical settings raises equity, clinical, and research ethics concerns. The equity concerns arise, in part, because of the so-called portability or transferability problem. As existing genetic datasets comprise mostly European-ancestry participants, PRS results are likely to be less accurate for individuals of other ancestry groups that have been historically marginalized.^17,18 PRS results may also have variable accuracy for individuals within genetic ancestry groups due to differences in socioeconomic status, age, sex, and cultural variables.^19,20 Although large and diverse cohorts alongside advanced analytic techniques can improve the portability problem,^21,22 the current state of scientific knowledge raises questions about whether PRS results from research should be returned to research participants and what the risks and benefits of this practice are for research participants.

A second set of challenges relates to the implementation of medical recommendations in response to PRS results that indicate high risk for common chronic conditions (“translational genomics”). Even when PRS results can inform medical and public health interventions—a difficult and complicated task²⁰—their clinical utility hinges on both clinicians’ views on PRSs and patients’ willingness—and actual ability—to adopt risk-reducing behavioral changes, such as following a healthy diet and exercising regularly. Studies on the return of results of single genes associated with common disease risks suggest little impact on the adoption and maintenance of risk-reduction approaches that require lifestyle changes,²³ although genetic findings can increase the adoption of screening, medication, and prophylactic surgery for cancer risk reduction.²⁴ Whether and how these findings may apply to PRS results is understudied. Exploring how clinicians view the clinical utility of PRS findings, whether the processes for returning PRS results are responsive to the needs of patients/research participants, and what the barriers—and possible facilitators—are for behavioral changes in response to PRS results can shed light on the potential success of translational genomics efforts.

In addition, the possibility of returning PRS results raises issues related to research ethics that need to be considered. Since PRS testing is not currently considered standard clinical care, efforts to translate PRS knowledge into improved health outcomes center on research settings and necessitate a review of proposed protocols by institutional review boards (IRBs). However, little is known about the views of IRB members regarding the human subject protections that may be needed for PRS testing, return of results, and risk-reduction efforts in research studies—even as these views can impact study design, implementation, and the protective measures that researchers may be required to provide.

These concerns are compounded in PRS studies involving historically marginalized racial and ethnic communities. The prevalence of chronic diseases is associated with structural discrimination and more adverse social determinants of health and is disproportionately high among these marginalized communities.^25,26 Participation of individuals from these communities in PRS-related studies is essential for increasing the accuracy of PRSs across racial and ethnic groups and is in line with National Institutes of Health policy and the ethical principle of distributive justice that call for diverse enrollment into research studies.^27,28 However, participants from these communities are also at disproportionately higher risk for receiving inaccurate PRS results due to limitations in existing data. In the absence of resources to correct risk estimates over time, return of PRS results to these participants may result in the over- or under-estimation of actual risks and, thus, potential harm. Concurrently, the portability problem and concerns about challenges related to acting on the results²⁹ may impact views about a PRS study, its promise for translational efforts, and potential risks and benefits.

Our study aimed to explore the views and concerns related to PRS studies among key stakeholders: patients/community members (largely from historically marginalized racial and ethnic communities), clinicians, and IRB members. The study was conducted at Columbia University Irving Medical Center (CUIMC), a site of the National Human Genome Research Institute-funded Electronic Medical Records and Genomics (eMERGE) IV Network. The Network comprises 10 sites in the United States; it is working to validate PRSs for 10 common chronic diseases and to recruit 25,000 adult and pediatric patients, a majority from historically under-represented racial and ethnic communities. The eMERGE-IV study will return a Genomic Informed Risk Assessment (GIRA) report to research participants, including PRS results for the selected chronic conditions. Individuals receiving high-risk GIRA will be informed about their results in a one-on-one meeting with a genetic expert (genetic counselor, physician, or PharmD); those with results that do not meet study thresholds for high risk will be informed via mail or electronically (further information about the GIRA is available in Linder et al., 2023³⁰). The reports will be incorporated into patients’ electronic health records, and their clinical utility and psychosocial impacts on research participants and clinicians will be assessed. Our study compared views on key issues in PRS-related research on chronic diseases across stakeholder groups, including perspectives on the eMERGE study and research participation, return of results to diverse populations, and clinical utility, as well as barriers and facilitators to making changes in response to the PRS results.

Material and methods

Study design and instrument

We conducted eight focus groups with stakeholders: four focus groups with patients/community members, two focus groups with clinicians, and two focus groups with IRB members. This method was selected as it allows for interaction and exchange of ideas among participants with varying levels of health literacy—a strength in a study on a relatively new topic.^31,32 Focus groups were conducted online (time for data collection: February 2021 to April 2022) and in English; no other languages were offered as participants from diverse backgrounds were included in each focus group. Each focus group comprised four to eight participants and lasted 70–100 min (sample sizes varied due to challenges in recruiting clinicians and IRB members). Participants were provided with a consent form and requested to complete a brief demographic survey prior to their participation. Verbal consent was obtained at the beginning of each group.

The focus groups included a 10-min presentation on eMERGE-IV, adjusted to each stakeholder group and delivered by a member of the research team (M.S. or P.S.A.), including information about factors affecting risk for common diseases (environments, genetics, family history), an explanation of PRSs, potential benefits and challenges with PRS results (e.g., the goal of educating participants about their genetic risk and possible medical and behavioral interventions to reduce it; lower accuracy for marginalized racial and ethnic populations, respectively), and study goals and planned processes (see supplemental materials 1–3). The presentation was followed by a semi-structured discussion following a focus group guide that the research team developed based on existing literature on return of monogenic results,^23,24 and issues relevant to the implementation of the eMERGE-IV’s objectives, and the professional experience of research team members (M.S., W.K.C., and P.S.A.). The questions explored participants’ views on participation in eMERGE-IV, return of PRS results, including less accurate PRS results for groups under-represented in existing databases, preferred means of communicating results, and barriers and facilitators to making behavioral changes (such as following a healthy diet and regular exercise regimen) in response to PRS results. Participants’ yes/no agreement on some questions was collected (either as part of the focus group guide or as follow-up questions) and are included below in the results, as relevant (see supplemental materials 4–6). Focus group guides were largely similar, with some variation for each stakeholder group. For example, focus groups of patients/community members gauged their interest in receiving both high- and low-risk PRS results (the exact threshold for high-risk findings varied by condition); focus groups with clinicians asked about the potential barriers to communicating PRS results to patients/research participants, and the IRB member group included ethical and regulatory questions, such as determination of risk level for a study involving the return of PRS results and measures required to assure appropriate human subject protections. The focus groups were moderated by a professional moderator with 25 years’ experience; team members (M.S., P.S.A.) attended all the focus groups and had the opportunity to respond to questions about the study and follow-up on participants’ responses, as needed, at the end. Participants were offered a $50 gift card for their time. The IRB at CUIMC approved the study protocol.

Participants

Patient/community members were recruited by phone through the Community Engagement Core Resource of Columbia’s Clinical and Translational Science Awards program, which has accumulated a list of adult community members who agreed to be recontacted for research participation. Recruitment of patient/community members was done based on first-come, first-serve basis, aiming for 70% of participants from under-represented minority groups and gender balance, per the requests for applications (RFA) for the general eMERGE study. Clinicians (physicians and nurses) were recruited by email outreach to relevant medical divisions and departments in the New York City area. IRB members were recruited nationally by posting an invitation in the newsletter of Public Responsibility in Medicine and Research (PRIM&R) and circulating it through the PRIM&R IRB forum and other IRB listservs.

Analysis

Focus groups were recorded and professionally transcribed verbatim. Data were analyzed using qualitative, descriptive content analysis that allows for low-inference description from the data.³³ Analysis comprised a deductive phase of coding based on the exploratory questions of the focus group guide (rather than an explicit theoretical framework) and inductive coding to capture and categorize responses in the transcripts that were independently raised by participants without a direct prompt.³⁴ The research team created an initial codebook based on study aims and focus group guide questions and refined it after reading sample transcripts from each stakeholder group to add codes that emerged from the discussions. The codebook followed key questions from the guidebook, each with sub-codes based on responses to questions during the discussion. The first author applied the coding scheme to the transcripts with patient/community members and re-reviewed the questions and codes of transcripts with clinicians and IRB members, along with the research assistants; two research assistants then used NVivo qualitative analysis software to code each stakeholder group’s transcripts following the refined codebook. Codes were adjusted and/or added as they emerged from the material, resulting in a total of 47 codes. Memos were used to retain the context, and differences in coding were discussed and resolved by consensus. We organized and compared coded responses by key questions and stakeholder type.

Results

Study population

Overall, 50 individuals participated in the focus groups: 25 patient/community members, 12 clinicians, and 13 IRB members. There were differences in the racial and ethnic composition of the stakeholder groups, as well as education and income levels (Table 1).

Table 1.

Participants’ demographic characteristics

	Patients/community members N = 25 n(%)	Clinicians N = 12 n(%)	IRB members N = 13 n(%)
Age, years
18-29	4 (16)	0	0
30-49	12 (48)	7 (58)	5 (38)
50-69	8 (32)	5 (42)	7 (54)
≥70	1 (4)	0	1 (8)
Gender
Female	17 (68)	8 (67)	9 (69)
Male	7 (28)	4 (33)	4 (31)
Other/unknown	1 (4)	0	0
Ethnicity
Hispanic/Latinx	19 (76)	3 (25)	1 (8)
Non-Hispanic/Latinx	6 (24)	9 (75)	12 (92)
Race
Asian	0	3 (25)	0
Black/African American	3 (12)	1 (8)	1 (8)
White	2 (8)	7 (58)	11 (84)
Multiracial	0	1 (8)	1 (8)
Prefer not to answer	20 (80)a	0	0
Education level
≤HS, GED or vocational training	2 (8)	0	0
Some college	4 (16)	0	2 (15)
BA	8 (32)	0	2 (15)
Master’s degree	10 (40)	3 (25)	6 (46)
Doctoral/professional degree	0	9 (75)	3 (23)
Prefer not to answer	0	0	0
Unknown	1 (4)		0
Employment (all that apply)
Full time	13 (52)	9 (75)	10 (77)
Part time	5 (20)	1 (8)	1 (8)
Not employed	2 (8)	0	1 (8)
Stay-at-home parent	1 (4)	0	0
Retired	4 (16)	0	2 (15)
Volunteer	0	0	2 (15)
Prefer not to answer	0	0	1 (8)
Unknown	1 (4)	2 (17)	0
Household annual income
<$20,000	3 (12)	0	0
$20,000–$39,000	3 (12)	0	0
$40,000–$59,000	5 (20)	0	0
$60,000–$79,000	1 (4)	0	0
$80,000–$99,000	2 (8)	0	3 (23)
>$100,000	4 (16)	9 (75)	7 (54)
Prefer not to answer	6 (24)	1 (8)	3 (23)
Unknown	1 (4)	2 (17)	0

^aOf the 20 participants who responded they prefer not to answer the question on racial group, 17 identified as Hispanic.

General views on eMERGE

Participants in all stakeholder groups expressed general support for the eMERGE-IV study as a research endeavor and emphasized the key role of trust for study enrollment (e.g., learning about the study from a trusted source). Patient/community members focused on the potential benefits from research participation, whereas clinicians and IRB members also expressed concerns about the study.

Perceived benefits and concerns

Most patient/community member participants (80%) expressed interest in participating in the eMERGE-IV study. Key rationales included potential medical utility for themselves, their families, and their community, altruism, and broader interest in promoting scientific knowledge. These perceived benefits were intertwined and often future-looking—e.g., the multiple benefits of growing scientific knowledge in the long run. As one patient participant stated:

I would participate just to help myself, and to help my children, to help the community, just to help. See where this research can go. (Patient/community member #11)

Several clinicians reflected on their research endeavors and highlighted the importance of research to increase knowledge “in populations where the data are just not sufficient” (Clinician #4). They also raised concerns about the study’s plans to return PRS results and the implications for study enrollment. As one clinician stated:

[T]his [plan to return PRS results] should not be something that’s offered to people we don’t know it works in. … I’m not even sure if I would approach someone [to join the study] unless the data was validated in that subgroup, because it’s just going to lead to confusion. (Clinician #4)

IRB members indicated support for research in general (“I am an active research advocate”) and the eMERGE-IV study specifically (“knowing the importance, you know, of research […] I think that […] the, the goal is always to potentially get things done” [IRB member #13]). Most IRB members (10/13) stated that they would view the eMERGE-IV study as greater than minimal risk given the potential psychosocial impacts of PRS results (“It could be shocking. It could be alarming” [IRB member #4]), that use of PRSs is not yet part of routine clinical care, and the possibility that the PRS results may impact medical decisions.

I think what’s more than minimal risk is thinking about a subject’s psychological state. Something minor [i.e., a small risk increment] somebody could take as “I’m going to die.” Then, on the other side, it could be really serious, and they don’t understand the gravity of it. If you have somebody high strung with a lot of anxiety, the smallest little thing could just set them off. I think that would be one of the main things that would push it into greater than minimal risk. (IRB member #7)

Although IRB members viewed these concerns as applicable to the entire study, several emphasized that the concerns are far greater with regard to children. They questioned the rationale for pediatric enrollment (the study will recruit children 3 years and older), the process for securing assent, plans for getting re-consent from children reaching majority, and the impacts of returning results, including incorporation of results in the electronic health record (EHR). Three IRB members further indicated that they would have difficulty approving the protocol to allow pediatric enrollment.

I have a problem with the autonomy that’s not given to pediatric participants. So it’ll depend on how they’re assented, how much choice they have in it and what can be done with those results when they reach age of majority, if they can’t be expunged from their EHR, then I think there’s a problem. (IRB member #6)

However, IRB members emphasized the need to balance risks with the possible benefits of research and indicated that clarity and transparency in the consent process, including in particular the consent form, could address the concerns and secure protocol approval.

As long as its [all study information] disclosed, I can live with a lot of things that as long as they’re disclosed and it’s really understood, not just disclosed in some sort of a namby-pamby, you know, wording, it’s very clear what it means and they understand. (IRB member #12)

Enrollment approaches

Stakeholders suggested using a mix of platforms for informing and recruiting diverse participants into the study, including flyers, community events and centers, newspaper outlets, information in clinicians’ offices, and circulating study material through relevant listservs and via social media. Participants across stakeholder groups emphasized the role of trust in recruitment efforts, though their preferred recruitment methods and emphases on how to conduct recruitment differed.

Patient/community members highlighted the need for information about the study to come from a trusted source with whom they have had a positive experience (“Someone who has a sense of their integrity and their honesty and a trusted source, ideally related to healthcare” [Patient/community member #8]), though some also indicated concerns about clinicians having this role (“Sometimes the doctor, they don’t take the time enough to explain to the patients how’s it work … They don't take enough time to explain [to] all patients.” [Patient/community member #24]).

Clinicians believed that the rapport patients have with their primary care provider and office staff is critical for successful enrollment. Concurrently, they indicated concern about time limitations for discussing research opportunities with patients and suggested the use of remote engagement approaches. These include mailing, avatars, “record[ing] a webinar or some kind of filmed presentation that the patient can watch either at the physician’s office or at their own leisure if physicians tag them to be qualified for the trial” (Clinician #2), and contact through the EHR portal commonly used in U.S. hospitals to access and share electronic medical records.

IRB members emphasized the interconnectedness of trust, cultural competency, and recruitment efforts. Although IRB members viewed recruitment through clinicians as “the easiest access point” (several expressed preferences for such enrollment), they also raised concerns about clinicians “cherry-picking” participants, the risk for undue influence (“doctor-patient relationship … can be a little bit hierarchical maybe” (IRB member #8)), and the need for researchers to demonstrate how such risk would be minimized, as well as a detailed process for recruitment that assures consistency and separation between clinical care and research.

Perspectives on the return of PRS results

All participants were asked whether PRS results from the study should be returned, regardless of the portability or transferability problem, and to whom (i.e., only research participants or simultaneously also to the treating clinicians). Other questions were adjusted to each stakeholder group (see focus group guides).

Should PRS results be returned to all participants?

Across stakeholder groups, there was high support for the plan to return high-risk PRS results to participants. Most patient/community members expressed interest in receiving the results (21/25), including less accurate results for racial and ethnic minoritized groups (20/25; “something is better than nothing” [patient/community member #8]). Similarly, most clinicians (7/12) and all IRB members (13/13) thought that high-risk PRS results should be returned to participants from racial and ethnic minoritized communities.

Participants who did not support returning less accurate results to individuals from minoritized communities expressed concerns that it would create “worries about what to do” (Patient/community member #21) and questioned the usefulness of such information (“if it wasn’t accurate, I mean, then I wouldn’t know anything really” (Patient/community member #18); “at what point [the] issue become[s] worthless essentially because the accuracy is, is so bad” [IRB member #11]). A few clinicians highlighted that returning inaccurate results may evoke psychosocial harms and have negative implications for clinical care. As one clinician stated:

I’m on the fence leaning toward no [return of less accurate results], just putting my research hat on for a second and try to combine it with [a] clinical one. I’m hesitant to say, provide information to a patient that is known to be inaccurate at baseline and is probably known to not even be inaccurate by a certain factor, like a factor of one or two or three. And so I’m hesitant to provide that information because it could be dramatically different than we know, and this is a research study. (Clinician #6)

In contrast, participants raised several arguments in favor of returning results, including less accurate results. Patient/community members highlighted the medical benefit for them and future generations, the personal and societal value of being included in research, and the potential incentives for future participation of under-represented communities.

So I just think that diversity is good and it just creates knowledge base there. So, if there’s inaccuracy, it just kind of leads the way to further research it. And also, you feel like you’re included, the whole population is included, even if it’s minimal information. But at least there’s something there. (Patient/community member #12)

Several clinicians raised concerns about inclusion to explain why returning results across racial and ethnic groups is merited.

In my eyes, you can’t enroll someone and then not give them the results back. Or you don’t give the results back to anyone. But you can’t enroll some people and then not return the results back to some people and say, “Oh, you’re discriminated against because we don’t have enough information about your population.” So, if you choose to have a mechanism of returning results back, it needs to be universal. (Clinician #1)

IRB members highlighted the importance of information and equity considerations to explain the preference for returning results, regardless of accuracy.

I think it’s an equity issue. I think that if you have information that will definitely benefit one group but could potentially benefit another, and then you leave that out of the group it could potentially benefit, especially considering they are historically neglected populations, that’s a massive equity issue, one, and two, the appearance is just terrible. I mean, it harkens back to Tuskegee. I feel very strongly that all groups should be included. You just give the information that this might not be accurate because you are a member of a historically neglected population. (IRB member #8)

Participants across stakeholder groups emphasized that the researchers must be transparent about the level of, and reasons for, inaccuracy of results. However, the ensuing discussion indicated that the basis for this responsibility differed among stakeholder groups. Patient/community members emphasized experiences of marginalization, while clinicians considered this question from the perspective of clinical and genomic care, and IRB members referenced principles of research ethics.

We already mentioned that because of the issues of disparity, the numbers are going to be inaccurate. So we kind of know going into it that it's going to be somewhat not as accurate or correct. So [I prefer getting the inaccurate results and being told why they are inaccurate]. (Patient/community member #20)

We’re already there with monogenic genetic testing and variants of uncertain significance where a third of the results are that. So, I think you can have an option where you can tell the patient what the shortcomings are of the study and [find out] whether they don’t want to know results or whether they want to know all. (Clinician #3)

I like the spirit of transparency. You, you know, things may not always turn out the way you expect them to be, but at least you’re being honest. (IRB member #11)

To whom should PRS results be returned?

Participants’ views varied on whether results should be shared automatically with the treating clinicians. Several patient/community members stated a preference for a “medical team” approach and that returning results directly to clinicians, in addition to patients, would enable them (as patients) to better understand their condition and improve their healthcare outcomes. As one participant stated:

I believe in [a] team approach. So being that both [patient/research participant and clinician] are involved is great. Once the doctor receives the results, maybe they should have a focus visit on that, and then they could plan moving forward. That way nothing gets lost in between. (Patient/community member #12)

Other patient/community members, however, strongly believed that PRS results should only be returned to them and that they should have the choice whether and with whom to share these results. The rationales for this preference included privacy considerations, potential negative impacts of disclosure of genetic results and a desire to have control over their data and healthcare planning.

If I want to give it to the doctor, I’ll do that myself. Once genetic information is released, then it can be disclosed under … There’s some limited circumstances where it cannot be disclosed. And then somewhere it must be disclosed. And so I wouldn’t want my genetic information out there. It can impact certain things. And providers are not allowed to discriminate on genetic information in certain circumstances, but they are allowed under other circumstances. I don’t want my genetic information out there unnecessarily. I want to be in control of it. (Patient/community member #5)

Clinicians overwhelmingly preferred to receive the results at the same time as, and ideally before, the patient/research participant. Key rationales included having “no surprises,” ensuring time to prepare for questions about the results and mitigating potential psychosocial harms.

While a few indicated the need to ensure this process would be conveyed in the consent form, clinicians worried that patients receiving the results before them will “freak out” and that such a scenario “creates a lot of confusion and a lot of overwhelm for the patients.” (Clinician #10) As one clinician stated:

I think the clinician should get it first so that … the clinician has knowledge of clinical and social history of the patient so they can develop sort of a personalized response to whatever questions the patient may have. (Clinician #9)

IRB members largely believed that research participants should have a choice about whether to share results with their clinicians. They raised privacy and discrimination concerns regarding both the automatic sharing of PRS results with clinicians and incorporation of results into patients’ medical records, as well as worries that research participants would not understand the implications. As an IRB member stated:

I would consider offering an option for the participant to not have it be included in the medical chart, if there’s something that they may cause problems down the road with insurance or discrimination, then I think maybe they should have an option to not have it automatically be entered into the chart. (IRB member #10)

Several IRB members underscored the importance of communication among researchers, clinicians, and patients/research participants in the process of return of PRS results. They expressed concerns about the possible negative impacts of a staged process on clinical care, and conversely, the potential for clinician bias if they receive the results before the patient/research participant.

Hopefully the physician would be presenting his findings without bias. That was, that was my only concern. If you give it to the physician without the patient [at] the same time, maybe the physician’s gonna be biased and how he reviews it with the patient. (IRB member #11)

Re-interpretation of PRS results

The possibility of later re-interpretation of PRS results was not directly probed but was raised by some patient/community members and especially IRB members in discussions on the return of less accurate PRS results to participants from marginalized racial communities.

IRB members queried whether the study plans to re-interpret findings over time, as knowledge increases, return updated results, and cover the costs of genetic counseling. One participant indicated that “if the effect size [i.e., change following reinterpretation] is large enough, that it is a meaningful effect, it could be clinically meaningful. Then you inform somebody (IRB member #8).” Another IRB member stated that “It goes back to the principle of beneficence. And if the researchers know something that could potentially help the person, then they have an obligation that could communicate that” (IRB member #2). Ultimately, all IRB members (13/13) indicated that researchers have an ethical obligation to report to participants (or at least make a “reasonable effort” to communicate with them) about their reinterpreted PRS results—during the study and after its completion—and laid the associated costs on the researchers, research institutions, and funders.

IRB members further indicated that the ethical obligation to reinterpret PRS results and return them to research participants is heightened for pediatric participants:

I think I have more of a concern for children in this case [of re-interpretation], too, for the minors that are being enrolled. I mean, if you have their genetic data and you find a change in risk, you get their data when they’re five and you find a change in genetic risk when they’re 30 and you still have the identifiable data. I do have concerns about not informing, if you’re going to inform subjects, you are at risk for A, B and C, then I feel that there is an ethical obligation, if that risk changes, to reinform subjects: “Your risk for A, B, and C have now changed.’” And I can’t explain it any better other than there is that ethical obligation, because you told them once. (IRB member #8)

However, most IRB members did not view the absence of a plan to return re-interpreted results as a sufficient reason to disapprove a protocol.

So I’m going to say that none of these [i.e., not returning re-interpreted results and not offering related supports, such as genetic counseling] would be a hard “no” for me, but I would have a lot of questions and I would want a lot of clarification on the protocol and most importantly, in the consent form, if you’re not going to return the changes in results to people, changes in the PRS scores, then that needs to be very clear on the consent form and in the information that they’re given when they’re given the PRS scores, that this captures a moment in time, especially for racial and ethnic minority folks, these are subject to change. (IRB member #8)

Translational efforts, barriers, and facilitators

With regard to the impact of PRS-based health-related predictions on behavioral changes and risk-reducing interventions, participants underscored the need for the person returning PRS results to provide clear guidance about what patients/research participants can due to reduce their risk. As a patient/community member stated:

I think it [the PRS result] should go to someone who’s going to be able to translate it to you. Someone who’s going to be, break it down for you thoroughly. Whether it’s your personal doctor or whoever, whatever clinic you went to […] get the testing done. Someone should be able to break it down for you and explain the dos and don’ts from here on out. What you should have, what you shouldn’t be doing and just everything. (Patient/community member #7)

Participants also identified several barriers for translational efforts. Across stakeholder groups, participants pointed to personal challenges such as motivation, time management, and juggling family-work-personal health needs, highlighting the individual’s responsibility for making behavioral changes. Other identified barriers were low health education in historically marginalized communities (raised in all stakeholder groups), providers’ low genomic literacy (clinicians and IRB members), and structural challenges (raised in all stakeholder groups). Below we expand on the latter three barriers and on participants’ views on incentives for behavioral changes in response to PRS results.

Community health education

Several patient/research participants noted that limited health education in their community would preclude behavioral changes in response to high-risk PRS results and indicated that the study team should educate participants in this regard. As one participant stated:

Education. Educate your participant. If you don’t know how to do something, then that’s going to be an obstacle in itself. (Patient/community member #11)

Only a few clinicians raised this concern; those who did connected health literacy with “fear of the unknown” and a misunderstanding of genomics:

One [barrier] is health literacy. And that relates to fear of DNA is what I call it. People don’t like to use the word genetics because it sort of unleashes this nightmare of destiny. (Clinician #3)

IRB members raised community low health literacy in various contexts, including enrollment, pediatric participants, and translational efforts, highlighting the risk for psychosocial and medical harms. As one participant stated:

We’re expected to review these consent forms and have them be at an eighth-grade literacy level. Now, my question is, do eighth graders really understand, what is a Z score? What are polygenic risk scores? That understanding isn’t there. And that’s why I feel so strongly about the need for genetic counselors because that’s their job. That’s what they’re trained to do. And I just see this whole host of people freaking out and, and taking this on their own and taking drastic measures, going in for unnecessary testing, unnecessary follow-up because they don’t understand what’s going on. (IRB member #8)

Clinicians’ genomic literacy

While several patient/community member participants referenced their doctors as the most likely or expected individuals to return PRS results to them, both clinicians and IRB members raised concerns about clinicians’ genomic literacy. Clinicians noted that primary care providers may lack genetic expertise and would be uncomfortable returning PRS results not directly within their clinical specialty. Only 3 of 12 clinicians agreed they would feel comfortable answering questions about PRS and what the results mean; 4 clinicians stated they would need more educational support before answering their patients. As one clinician stated:

[P]atients expect the physicians to be prepared and the truth of the matter is there’s so much uncertainty about interpretation of these results and where they fall on the polygenic risk score. I probably would have quite a few questions as to the result pertaining in how we interpret that result for the patient. (Clinician #3)

Clinicians subsequently emphasized the responsibility of the study team to ensure there is a “system” for returning results, including a genetic counseling team and/or trained clinicians who may be asked to return results.

Some IRB members also raised the concern of clinicians’ low genomic literacy. They highlighted that “people can really be in a state of shock or depression or whatever, when they get this information” and that clinicians may be ill-equipped to deal with these emotional responses or answer relevant questions. As one IRB member stated:

[It would be important to have a genetic counselor because] you’re being given information that doesn’t have immediate quantifiable value, right? You’re being told that based on the PRS, you might have an inclination for this. You might have inherited something from somebody in your family. You might be passing it down to someone in your family. These are emotional issues. And I don’t think that most physicians are prepared to have that conversation to the depth or add a level of support that is needed. (IRB member #6)

IRB members strongly believed that a geneticist or genetic counselor is most suited to explain the results to research participants without overwhelming them with information, and all agreed that such an expert should be freely available to participants. Although they did not think they could require a study to have a geneticist/genetic counselor, they indicated that the presence (or absence) thereof will impact their determination of study risks. As an IRB member stated:

I think that the way to reduce the risk to being not greater than minimal risk, as I alluded to before, is to have genetic counselors be present. Again, my experience, most physicians don’t have a basic understanding in genetics and certainly not in medical genetics, in genetic epidemiology. I think it’s absolutely critical to have genetic counselors be present when these results are returned. I think that way you can reduce it to not greater than minimal risk. (IRB member #8)

Structural barriers

Structural issues that would preclude the implementation of translational efforts were raised in all stakeholder groups, though they were paramount in focus groups with patient/community members. Identified barriers included financial instability (raised in all stakeholder groups), insurability and limited access to specialists (patient/community members and IRB members), life-work imbalance (patient/community members, and clinicians), clinicians’ workload (clinicians), and the absence of health-supporting infrastructure and affordable healthy food options in poorer neighborhoods (raised in all stakeholder groups). Additional example quotes are included in supplemental material 7 (Table 1).

And so, a lot of people don’t have, maybe can’t afford groceries. … And some neighborhoods are food deserts. And so if the only thing available … If somebody can’t go a mile away to a supermarket, maybe the little corner stores, maybe just have junk food and maybe somebody has no choice. (Patient/community member #5)

Healthy eating costs more than cheeseburgers at McDonald’s and a good gym costs a lot to be a member of or even Planet Fitness costs $300 a year to be a member of. (Patient/community member #19)

So, often times patients I’ve had in the past have a hard time being able to buy fresh food and cook at home and end up buying lots of fast food and things like that because they’re working a lot of jobs and are unable to cook for their families at home. (Clinician #10)

Who we’re talking about as subjects might they not have the finances, insurance, whatever it would take to do some of those things. Say, for example, it was regular screening for something, testing for diabetes, would they be able to comply? (IRB member #5)

Incentives and resources

All stakeholders agreed that incentives may facilitate behavioral changes in response to PRS results. However, they differed in what they saw as necessary resources to address structural barriers, and whether researchers should provide them (additional example quotes are included in supplemental material 7 (Table 1)).

Patient/community members indicated high need for resources and ongoing interaction with the clinical or research team to support behavioral changes. Although a few only highlighted informational pamphlets and recommendations, most patient/community members suggested other initiatives, including creating support groups (19/25 expressed interest), regular text or phone reminders from staff members who are personally known to them, and resources to encourage behavioral changes, such as availability of relevant professionals on site (e.g., a nutritionist or a fitness coach), free medication and gym membership, provision of healthy food, and a gift card for achieving milestones. As patient/community members stated:

You have someone assigned to that person who’s struggling with that, and they call them up, let’s say twice a week, to see how they’re doing with the process. That will help a lot. (Patient/community member #7)

Let’s say going to the gym is a healthy activity, which we may prove to save you, put the years on your life or something. And then we provide you with like a membership to a gym for a year, or if it’s about a statin that may be the study, we’ll provide the cost of the statin for a year, for given time. That would be an incentive not only to participate in this study but to engage in the healthy behavior. (Patient/community member #8)

Clinicians were split about the usefulness of support groups that are not related to a specific diagnosis but to PRS results, though most indicated that referring research participants to support groups to motivate behavioral changes would be appropriate. Clinicians also raised questions about reminders. Although some clinicians thought that reminders could be helpful to keep patients/research participants on track, all clinicians strongly preferred for reminders to be automated through the EHR, avatars that send texts, or other online portals. Two clinicians noted that different approaches may be more suitable for older patients/research participants, but all underscored that clinicians should not—and due to workload, cannot—be responsible for these reminders and that the burden is on the patients/research participants to follow-up and make behavioral changes.

It should be automated. At that level, if we’re talking about it being [from the EHR] … with these pop ups and stuff, it should be a direct-to-the-patient reminder. (Clinician #3)

IRB members regularly mentioned principles of ethical research, including equity and justice, regarding study implementation. However, some IRB members questioned the process and usefulness of both support groups and reminders—especially automated ones—for promoting behavioral changes (“[T]hese text messages, and they’re just annoying. They’re not helpful. They make you feel bad about yourself if you’re not doing the things that they remind you to do. I don’t see it as particularly motivating” [IRB member #3]). In addition, IRB members emphasized the need for the study to provide clear information in the IRB protocol about the translational plans for underserved communities, including participation in support group and/or the possibility of receiving reminders.

When asked whether the study has an obligation to address structural barriers to increase the likelihood that participants will benefit from the study, two IRB members responded positively. These participants underscored the disparities that low-income participants may experience, which may result in biased findings (“You’re going to get SES [socioeconomic status] effects” [IRB member #8]; “We just make the assumption that the person is just resistant, but maybe that’s not what it is. They just need some support to do something”)). and principles in research ethics (“[Without some effort by the study] … the potential for benefit is not equitable, right? So it doesn’t meet that justice pillar from the Belmont Report, right, if everybody doesn’t benefit?” [IRB member #8]).

Further discussion highlighted that “addressing the barriers” meant the need for study teams to assess the impacts of such barriers on the equitable implementation of translational efforts (“at the very least you should be having some way of assessing whether folks are running up against barriers” [IRB member #5]) and provide research participants with information about available resources. An IRB member explained that, “They [researchers] don’t have to go to great lengths, but [have] something in place” (IRB member #8) and detailed in the protocol; and IRB members rejected the need for the study to provide financial support for participants to overcome the barriers. As an IRB member stated:

Um, so I would agree that it’s … They do need to address those barriers, but I wouldn’t say that they need to financially create something in order to overcome them. … I completely agree in helping them identify those resources is key. Not necessarily buying them a gym membership. (IRB member #9)

Discussion

We conducted focus groups to examine the views of patient/community members, clinicians, and IRB members on the return of PRS for chronic conditions, as planned by the eMERGE-IV Network, and the implications for patients/research participants, especially those from historically marginalized racial and ethnic communities. The study is embedded in a broader discussion on the growing uses and potential risks and benefits of PRS in research and clinical settings,³⁵ and adds a comparative perspective on key issues.

Our findings indicate support for the eMERGE-IV study and its goals, especially regarding the enrollment of adult participants, and belief in the interconnectedness between research and clinical settings for study recruitment and implementation. Patient/community members highlighted the potential medical benefits for research participants, their families, and respective communities, as well as the social benefits of research participation: being recognized as valued members of society. Across stakeholder groups, participants stressed the invaluable role of trust for research participation, especially to increase participation of marginalized racial and ethnic communities.

Across stakeholder groups, our findings indicate a strong interest in having PRS results returned to research participants. Some participants (mostly clinicians and IRB members) raised concerns about the psychosocial impacts on research participants and the limited clinical utility of results, especially if inaccurate. In contrast, similar to other studies on the return of PRS results,³⁶ patient/community members tended to view information as an absolute good: better to learn about inaccurate results than have no results at all. The reasons for this latter preference are seemingly based on expectations for longer term health benefits to individuals and communities from generating the data to improve risk predictions, an outcome that could be further explored in the future. IRB members further highlighted an ethical responsibility on the researchers to reinterpret the data over time and return updated PRS results, especially to pediatric participants and those from marginalized racial and ethnic communities. All stakeholder groups emphasized that the return of PRS results must come with researchers’ responsibility to be transparent about both the limited accuracy of the PRS results and the reasons for it—i.e., that the results are based on existing knowledge and limited cohort diversity.

Whether PRS results will lead to behavioral changes and improved health outcomes is debatable. Even with accurate PRS results, participants in our study identified multi-level barriers for making changes in response to PRS results. As in other studies,³⁶ these include access disparities based on socioeconomic factors (e.g., access to testing and post-test care, healthy food options), financial barriers to care, and health insurance coverage. Our findings highlight also structural barriers, including limited genomic knowledge and system-level efforts to increase health education,³⁷ work-life imbalance and/or work overload among both patient/community members and clinicians, and the absence of structures that promote healthier living in poorer neighborhoods (e.g., affordable gyms). In addition, acting on results with limited accuracy raises the risk of unneeded interventions or patients forgoing important ones (depending on the direction of inaccuracy). While there are ongoing studies to increase the accuracy of PRS results across populations and improve the consequent clinical recommendations, it will be important for the possible limited accuracy of the results and their implications to be conveyed to research participants in clear and understandable ways.

Moreover, stakeholders in our study indicated different visions about what is needed and can potentially be provided to yield successful translational efforts. Patient/community members expressed a strong preference for a “personal touch” throughout the study, including hands-on reminders about medical recommendations after the return of the PRS results—a preference that is likely to be impossible for researchers and clinicians to fulfill. Whether the use of EHRs and non-personal interactions (e.g., informative videos about the study) as preferred by clinicians will result in high interest and impact on health-promoting behaviors among patient/community members is unknown. Patient/community members expressed reliance on clinicians’ expertise in providing information about the meaning of PRS results and follow-up recommendations—yet clinicians indicated a significant need for genomic education (especially if not in their immediate area of expertise). The presence of genetic counselors can alleviate this challenge, as strongly recommended by IRB members. Expectations for the reinterpretation of PRS results over time are similarly likely to face funding challenges, especially once the study cycle is completed³⁸—although such a process and additional opportunity to assess long-term behavioral changes may be informative to translational genomic efforts.

All stakeholder groups agreed that structural barriers are likely to impede participants’ ability to make behavioral changes, especially participants from marginalized racial and ethnic communities. However, they held significantly different views about how to address them and who should bear the responsibility. Patient/community members suggested that researchers should proactively address structural barriers by providing a range of resources and incentives to encourage behavioral changes, whereas clinicians viewed behavioral changes as the primary responsibility of patients and their own role as only supportive—sending automated reminders, providing informational resources and making referrals; IRB members did not think it within their power to require that researchers act on structural barriers and were reluctant to require active investments as part of a study. Notwithstanding the concerns that IRB members raised about the study and its ramifications for the principles of equity and justice, IRB members overwhelmingly indicated they would approve the protocol—so long as the consent process is clear and the formalities of the informed consent forms are met.

Removing the barriers to increased screening for early detection of health conditions and to making behavioral changes following clinically confirmed high-risk PRS results will be critical if genomic research and improved health outcomes are to be achieved. To this end, the eMERGE-IV Network has worked extensively to improve the accuracy of PRS results across racial and ethnic groups³⁹ and develop accessible educational materials to explain relevant genomic information to those who are involved in the study.⁴⁰ In addition, the network is committed to returning high-risk results to research participants via qualified study staff (e.g., genetic counselor, physicians, PharmDs), communicating these findings to the treating clinician, and covering the cost of some clinical follow-up after the return of high-risk PRS results. Beyond this study, there is a need for research and system-level policies to ensure that research participation and access to care are equitable. IRBs can also be powerful actors in implementing justice in genomic research. They are often involved in the development of institutional policies related to research with human subjects^41,42 and in their role as oversight bodies, they can re-interpret key principles in research ethics—such as “distributive justice” and fairness—to require researchers and research institutions to consider, and take steps to address, the contexts of social injustice and marginalization in genomic research.⁴³

Limitations of the study

Our study has several limitations, including small samples, adult-only participants, and limited gender, racial, and ethnic diversity among the clinician and IRB participant groups. In addition, all focus groups were conducted in English and may be less generalizable to non-English-speaking individuals. As in all studies, it is also possible that individuals who opted to participate in our study are more likely to support research in general (we did not assess this). In addition, their views on eMERGE, as well as the design and analysis of this study, may have been impacted by the fact that eMERGE investigators developed the study material, presented information about the study, and were present (albeit off-screen during the discussions) in the focus groups. However, our samples represent the demographic characteristics of the respective stakeholder groups in our area and our findings contribute important insight into the challenges to translational genomic efforts—uniquely, looking at the issues from stakeholders’ perspectives. Future studies can explore possible clinical and socioeconomic interventions, their acceptability to stakeholders and efficacy in promoting translational genomic efforts.

Conclusions

The increasing knowledge related to the genetic underpinnings of chronic conditions is promising for improved health outcomes, especially among communities that are disproportionately burdened with such conditions. Participant diversity is a first step to increase the accuracy and validity of PRS across ancestry groups and it is strongly supported by participants in our study. The subsequent steps of returning PRS results and the ability of patients to act on them confront ethical, personal, financial, and structural challenges, and different visions among stakeholders about how best to implement translational efforts. Our findings highlight the need for study-specific measures and system-level policies that promote equity in genomic research and outcomes.

Data and code availability

The authors will make relevant study material and the anonymized dataset available to collaborators external to our team upon e-mail request, assuming that we can make appropriate arrangements to safeguard any sensitive information participants might have revealed in the course of the study and community concerns.