Uptake of Genetic Research Results and Patient-Reported Outcomes With Return of Results Incorporating Web-Based Predisclosure Education

Web-based predisclosure education for genetic research results is a viable alternative

Abstract

PURPOSE

We developed a web-based education intervention as an alternative to predisclosure education with a genetic counselor (GC) to reduce participant burden and provider costs with return of genetic research results.

METHODS

Women at three sites who participated in 11 gene discovery research studies were contacted to consider receiving cancer genetic research results. Participants could complete predisclosure education through web education or with a GC. Outcomes included uptake of research results, factors associated with uptake, and patient-reported outcomes.

RESULTS

Of 819 participants, 178 actively (21.7%) and 167 passively (20.4%) declined return of results; 474 (57.9%) were enrolled. Most (60.3%) received results although this was lower than the 70% uptake we hypothesized. Passive and active decliners were more likely to be Black, to have less education, and to have not received phone follow-up after the invitation letter. Most participants selected web education (88.5%) as an alternative to speaking with a GC, but some did not complete or receive results. Knowledge increased significantly from baseline to other time points with no significant differences between those who received web versus GC education. There were no significant increases in distress between web and GC education.

CONCLUSION

Interest in web-based predisclosure education for return of genetic research results was high although it did not increase uptake of results. We found no negative patient-reported outcomes with web education, suggesting that it is a viable alternative delivery model for reducing burdens and costs of returning genetic research results. Attention to attrition and lower uptake of results among Black participants and those with less formal education are important areas for future research.

INTRODUCTION

Genetic sequencing studies involving biobanked DNA raise questions about the obligation to share individual research results with participants. ^1-7 Arguments supporting return appeal to the principles of beneficence, autonomy, reciprocity, respect for persons, and clinical utility. ^4,7-9 Arguments opposing return raise concerns about the distinction between research and clinical care, the actionability of results, the right not to know, and costs. ^5,7,9-13 Although debates are ongoing, the consensus favors returning results that could be relevant to participants' health. ^2,4,5,14-16

CONTEXT

• Key Objective

• Is web-based education a viable alternative delivery model for predisclosure education for return of individual genetic research results?

• Knowledge Generated

• Uptake of web-based predisclosure education for return of individual genetic research results was high among enrolled participants. Among those who completed predisclosure education, most received their research results, which did not differ by web-based education versus education with a genetic counselor. No negative patient-reported outcomes with web education were found.

• Relevance (S.B. Wheeler)

• As return of genetic test results to patients continues to increase with clinically relevant genetic information becoming more widely available, appropriate and timely education to support interpretation is key. This study offers a potentially impactful and efficient approach to providing that education via web-based platforms.*

• *Relevance section written by JCO Associate Editor Stephanie B. Wheeler, PhD, MPH.

Research participants have reported high interest in receiving research results. ^16-25 However, lower uptake has been reported, ²⁶ particularly in studies involving biobanks for which return of results was not emphasized during enrollment. ^27-34 There are limited patient-reported outcomes (PROs) in the research setting ^27-29,35 although some genomic implementation studies have reported no psychological harms with return of genetic findings. ^36,37 In RESPECT1, we found favorable cognitive (eg, knowledge) and affective (eg, distress and uncertainty) responses with return of results among patients with breast cancer, but low uptake of results. ³⁸ In addition, we found that one third of participants reported that a self-directed web platform would be an acceptable alternative to speaking with a genetic counselor (GC) and could reduce steps in receiving results. ³⁹

In this multicenter, observational return of results study (RESPECT2), we developed a web-based predisclosure education intervention (web education) as an alternative to speaking with a GC to reduce participant burdens and steps in receiving genetic research results. Our primary aim was to evaluate uptake of genetic research results among research participants who provided a biospecimen for genetic research when using an alternative delivery model incorporating web education. We hypothesized that 70% would receive their results after predisclosure education. Secondary aims included understanding participant factors associated with uptake of results, changes in PROs, and whether outcomes differed by the method of predisclosure education (GC v web education).

METHODS

Participants were English- or Spanish-speaking adult women who had provided a biospecimen for genetic research (11 studies) at the University of Pennsylvania (UPenn), University of Chicago (UChicago), or Columbia University (Columbia) and had not previously had clinical multigene panel testing. Original consents stated that results that could affect participant health would be returned (n = 7) or would not be returned (n = 4). Individuals in the latter group were contacted so as not to assume that they would not want results.

Sequencing included 25 high- and moderate-penetrance genes (APC, BRCA1, BRCA2, CDH1, CDKN2A, PMS2, PTEN, MLH1, MSH2, MUTYH-homozygous, MSH6, STK11, TP53, ATM, BAP1, BARD1, BMPR1A, BRIP1, CHEK2, MRE11A, MUTYH-heterozygous, NBN, PALB2, RAD50, RAD51C, RAD51D) that had potential clinical relevance when the study was conducted. Sequencing was performed in institutional research laboratories. Clinical confirmation testing was recommended to participants for results that could potentially affect medical care (see below).

Institutional Review Board approval was obtained at all sites. Study invitation letters (English and Spanish versions at Columbia) explained that research testing had been completed and that participants could enroll to learn their research results. In eight studies (n = 1,583), the research team could follow up with participants by phone (three to five calls per site standards). In three studies (n = 379), original consents stated that results would not be returned and participants had to call or mail back a response card to be contacted.

Predisclosure Education by Web Intervention or Genetic Provider

On the basis of participant feedback in RESPECT1, we developed a self-directed web-based alternative for predisclosure education. ³⁹ The intervention was developed to cover the same content as predisclosure genetic counseling. Informed by the tiered-binned model, GCs reached consensus on indispensable tier 1 information that should be presented to all participants and additional or optional tier 2 information that could be provided to support variable information needs. ⁴⁰ The intervention and genetic counseling checklists included the same tier 1 content. The intervention consisted of seven modules and optional videos and was available in both English and Spanish (Appendix Table A1, online only).

Predisclosure, participants were offered access to web education or a GC session (conducted by seven GCs via phone or in person). ^38,40 GCs were blinded to participants' research results at predisclosure and completed counseling checklists to ensure fidelity to tier 1 content. ^41,42

Disclosure of Genetic Research Results

Participants received results by phone or in person with a GC. The result disclosure visit was separate from the education visit. Fidelity was assessed in 20% of counseling sessions, with a mean fidelity to checklists of 89% (predisclosure sessions) and 88% (disclosure sessions). On the basis of institutional policies, participants at Columbia with a positive result were told that there was a genetic finding but not the specific gene. Participants with a variant of uncertain significance (VUS) or negative result were told that there were no findings that could affect their health. UPenn and UChicago allowed participants with positive or VUS results to learn the specific gene and informed them that they should not change medical care until after confirmation testing.

Confirmation Testing in a Clinical Laboratory Improvement Amendments-Certified Laboratory and Clinical Follow-Up

Participants with a pathogenic or likely pathogenic research result in any gene were recommended to have clinical confirmation testing. We also recommended confirmation for VUS results in high-penetrance genes since a reclassification to pathogenic or likely pathogenic could affect medical care. This approach provided confirmation of the result along with an additional opinion regarding variant calling as clinical laboratories have access to additional data. It also allowed clinical laboratories to be responsible for updates to VUS results. The need for confirmation testing and the importance of not altering medical management until after confirmation was shared in predisclosure education and at disclosure. For discordant results, recommendations for medical management were based on the participant's clinical result and personal and family history, consistent with clinical care. GCs and research staff provided reminders and support to complete confirmation testing. Research funds covered all costs of confirmation testing at UChicago and costs not covered by insurance at UPenn and Columbia. Confirmation testing was completed by mailed saliva kits or phlebotomy during a clinical visit. Confirmation testing results were shared by phone or in person. All participants were recommended to return for follow-up care.

PROs

As previously described, the selection of relevant outcomes after the receipt of genetic research results was informed by our conceptual model, ^38,40,41 which is grounded in the Self-Regulation Theory of Health Behavior. ^43,44 Our model proposes that uptake of genetic research results and response to (eg, psychosocial adjustment) and use of (eg, performance of health behaviors) genetic information are products of an individual's understanding (eg, knowledge of genetic disease) and perception of disease threat (eg, cancer risk). ^41,43 Participants completed surveys at baseline (T0), after predisclosure education (T1), after disclosure counseling (T2), and at 6 months (T3).

Knowledge of genetic disease was evaluated (T0-T3) using an adapted version of the Cancer Genetics Knowledge Scale and ClinSeq knowledge scale ^45-47 and included knowledge of inheritance and test interpretation (nine items), benefits (three items) and limitations (six items) of multigene testing, and differences between research and clinical testing (five items; Cronbach's α = .79-.83). ³⁸

Perceived risk of cancer (T0) was measured on a Likert scale in relation to the average woman (much higher, higher, same, lower, much lower) and, in a second item, as a numerical lifetime risk (0%-100%) of getting breast cancer (or breast cancer again).

Psychosocial adjustment included the following: (1) state anxiety (T0-T2), measured with the 20-item State Inventory ^48,49 (Cronbach's α = .94-.95); (2) general anxiety and depression (T0-T3), assessed with the Hospital Anxiety and Depression Scale ⁵⁰ (Cronbach's α = .83-.85 and .81-.83); and (3) cancer-specific distress (T0-T3), evaluated with the 15-item Impact of Events Scale ⁵¹ (Cronbach's α = .86-.89).

Satisfaction with genetic services was measured (T2) with a 13-item scale used in related research ^38,42,52,53 (Cronbach's α = .81-.83).

Uncertainty was assessed (T0-T3) using a three-item scale adapted from the Multi-Dimensional Impact of Cancer Risk Assessment ⁵⁴ (Cronbach's α = .84).

Perceived utility was assessed (T0-T3) with a novel scale developed to evaluate patient perceptions of the utility of genetic results, including two 12-item subscales evaluating medical and personal utility, both now and in the future (Cronbach's α = .96-.97). ^38,55

Statistical Analyses

We characterized the samples using means, standard deviations, and proportions. We used logistic regressions to examine characteristics associated with responding to enrollment status groups and completion of the intervention. For psychosocial outcomes, we examined linear regressions of change scores between baseline and follow-up times (ie, follow-up minus baseline scores). In regressions, we controlled for variables that were found or hypothesized to be associated with longitudinal follow-up. These included site, Hispanic ethnicity, education, number of children, age, and number of relatives with previously diagnosed cancer. We used multiple imputation methods with 100 multiply imputed data sets to account for missing data in the regression analyses. ⁵⁶ Analyses were conducted in STATA (StataCorp, College Station, TX) and SAS (SAS Institute, Cary, NC).

We hypothesized that among 1,200 participants, 70% could be reached, 50% would enroll, and 70% would receive results. With 50% enrollment, we would be able to estimate the true uptake with at least a precision of approximately 4.9%. For secondary analyses, we aimed for a sample size of 420 to have 85% power to detect a standardized odds ratio of 1.5 in logistic regressions. We set the type I error rate and P value to 1% (two-sided) to partially account for multiple hypothesis testing. Additional power calculations are provided in our Protocol (online only).

RESULTS

Participants

Across three sites, 1,962 potential participants were contacted (April 2015-October 2019; Fig 1). Some letters were returned (6.6%), and 94 participants were found to be deceased or ineligible, leaving 1,739 eligible mailings. We were not able to communicate with 920 research participants to discuss the option to receive research results (ie, nonresponders; 52.9%). Of 819 participants who could be reached, 178 (21.7%) actively declined receipt, 167 (20.4%) passively declined (ie, expressed interest but were lost to follow-up), and 474 (57.8%) enrolled (Table 1).

FIG 1.

CONSORT diagram. ^aTwelve patients already had clinical MGPT, 155 had missing contact information, 103 research samples were not available, 38 were already known to be clinically BRCA+, and 81 were living abroad. ^bThirty-four patients were found to be deceased, 47 already had clinical MGPT, two were previously disclosed, three had language barriers, two did not have decision making capacity, and six did not have valid consent. ^cReasons for declining included privacy concerns, concerns about time burdens, not being interested in genetic information, only wanting actionable results, preferring clinical testing, and concerns about uncertainty or distress. ^dThe time from baseline survey to web education link being sent to participants was a median of 1.0 days and a mean of 4.98 days (SD, 16.06). ^eTwenty-one were completed with a GC. ^fFourteen pathogenic or likely pathogenic variants (positive result) were returned, including five high-penetrance genes (BRCA1 [2], BRCA2 [2], MSH6) and nine moderate-penetrance genes (ATM [2], BARD1, CHEK2 [4], NBN, PALB2, RAD51D). There were 23 results with at least one VUS (14 variants in moderate-penetrance genes and nine in high-penetrance genes where confirmation testing was recommended). There were two additional VUS results as a second finding. There were 250 with no findings. One individual received results without completing visit 1. ^gSixteen withdrew from the study. ^hTwo hundred twenty-nine of 256 who ultimately completed by web received results, and 56 of 67 who ultimately completed by GC received results. MGPT, multigene panel testing; ROR, return of results; T0, baseline; T1, after predisclosure counseling; T2, after disclosure counseling; VUS, variant of uncertain significance.

TABLE 1.

Characteristics of All Potential Research Participants Contacted

Uptake of Web-Based Predisclosure Education and Genetic Research Results

Most participants selected web education (88.5%) as an alternative to speaking with a GC for predisclosure education (Fig 1); 310 logged on, and 82.5% completed web education (Appendix Table A2). Research staff contacted participants who did not log on or complete to offer assistance or the option to speak with a GC. Among participants who selected to speak with a GC, 46 (90.2%) completed.

Among participants who consented to RESPECT2 to learn about receiving research results, 286 (60.3%) received results (Table 2). Most participants (88.2%) who completed predisclosure education received results, which did not differ by method (83.6% with a GC v 89.5% by web education; P = .2). Fourteen (4.9%) participants received a positive result, 23 received a VUS (8.0%), and 250 (87.4%) had no findings (Fig 1).

TABLE 2.

Uptake of Research Results Hypothesized and Achieved

Factors Associated With Uptake of Research Results

Research participants who could not be reached were more likely to be non-White, have lower education, have no history of cancer, have not allowed phone follow-up, and be at Columbia or UPenn. Research participants who actively or passively declined results were more likely to be Black, have lower education, not allow phone follow-up, and be at Columbia or UChicago (Appendix Table A3).

Enrolled participants who selected web education were more likely to have college education, be at UChicago or UPenn, not have children, have higher baseline knowledge, and have lower depression (Table 3). Completing predisclosure education was associated with selecting a GC, having college education, having higher baseline knowledge, having more relatives with cancer, and being at UPenn (Table 3). Completing predisclosure web education was associated with having higher education, having more relatives with cancer, and being at UPenn. Overall, uptake of research results among those who could be reached was associated with being White, allowing phone follow-up after letter notification, and being at UPenn or Columbia (Table 3).

TABLE 3.

Factors Associated With Interest in and Uptake of Individual Genetic Research Results in Multivariable Models

PROs

Knowledge increased significantly from baseline to all other time points (Appendix Table A4) and did not differ significantly by web versus GC education (Fig 2A). Similarly, there were no significant increases in distress for those who completed education by web versus GC. There was significantly greater reduction in short-term anxiety and uncertainty (T0-T1) and anxiety and depression (T0-T2) among those who completed by web versus GC although there were no significant differences in the long term (Figs 2A and 2B; Appendix Table A5).

FIG 2.

Change in patient-reported outcomes for web-based predisclosure education users relative to predisclosure education with a genetic counselor after predisclosure education and disclosure of results. The x-axis labels denote study wave, with T0 as baseline time point. The y-scale represents the absolute means of the scores. The range differs for each measure, which allowed us to present the results in the same figure to reduce space. The range and means at each time period are reported in Appendix Table A2. There were no significant differences in the percentage of participants who met cutoffs for clinically significant distress between groups. GC, genetic counselor; T0, baseline; T1, after predisclosure counseling; T2, after disclosure counseling; T3, at 6 months.

Among participants with a positive result, knowledge increased and depression declined over time (Appendix Table A6). Among those with a VUS or negative result, there was a significant increase in knowledge after the receipt of results. There was also a significant reduction in general anxiety, perceived utility of results (VUS and negative), state anxiety, cancer-specific distress, and depression (negative only) after the receipt of results.

Clinical Confirmation Testing

Most participants (82.6%) underwent recommended clinical confirmation testing (12 of 14 positive results, seven of nine VUS results in high-penetrance genes). Five of 14 (35.7%) participants with a VUS in a low- to moderate-penetrance genes had confirmation testing (testing optional). Among 17 participants for whom testing was submitted to insurance, three (18%) were not covered or had out-of-pocket costs and were covered by research funds.

Five of 24 samples (20.8%) sent for confirmation testing were discordant with research testing. These discordant results included a BRCA2 variant, which was classified as pathogenic by the research laboratory and as a VUS by the clinical laboratory (although this variant has been classified as pathogenic in ClinVar), and a common CHEK2 variant, which was not detected by the clinical laboratory because of sample mix-up, or analytic error. Others included MSH6, MUTYH, and PALB2 VUSs, which were not reported by the clinical laboratory. For these, we do not know if discordance was due to a difference in interpretation, sample mix-up or analytic error.

DISCUSSION

To our knowledge, this is the first return of genetic research results study reporting uptake and PROs with a web-based alternative to predisclosure genetic counseling. Although we met our minimum hypothesized uptake of results, as in our previous return of results study, many participants could not be reached. In addition, 88% of participants chose web education although some did not complete it or receive their results. Although uptake of results was not higher among those who requested web education, there was no evidence of misunderstanding or greater distress among those who chose web education as compared with a GC. This suggests that web education is a viable delivery alternative as long as completion rates can be addressed.

Interest in web education was higher than expected on the basis of stakeholder interviews, ³⁹ but was still consistent with other studies of patients with cancer undergoing genetic testing. ^57-60 Although most who selected web education completed the intervention, 26% did not and 18% never logged on. This finding is consistent with our stakeholder interviews in which some participants reported that they might be more likely to not complete a web alternative because it is not a scheduled appointment and therefore easier to forget to complete. ³⁹ Participants who never logged on were more likely to have lower education and lower genetic knowledge, raising the concern that eHealth alternatives could increase health disparities. It is possible that staff resources for reminding participants to log on varied and that automated reminders, follow-up, or additional assistance (eg, digital navigators) could be beneficial. In addition, our data suggest that retaining the option to speak with a GC is important for some participants.

Although we expected that web education might decrease barriers and increase uptake of research results, receipt of results among eligible contacted participants was no higher than that in RESPECT1. ³⁸ Other studies have similarly reported difficulty in recontacting research participants, ^38,61 with passive or active decline of research results as high as 40%. ^{27,32-34,38,62} Site differences in reaching research participants and in uptake of research results suggest that the level of engagement with the research cohort may also be important. Lower uptake among non-White participants and those with lower education suggests that disparities continue to exist with respect to interest in receiving genetic information ^32,38 and that providing the opportunity to decline genetic research results remains important.

Among those who completed predisclosure education, the majority (88%) chose to receive their research results, which did not differ by web versus GC education. Knowledge did not significantly differ by education method, and there were no increases in negative affective outcomes for web education. In addition, how participants used the intervention (eg, number of times accessed, content accessed) could be informative. More extensive secondary analyses addressing these questions are ongoing.

An additional challenge for research programs that are not sequencing in Clinical Laboratory Improvement Amendments-approved laboratories is the need to confirm research results. Four participants did not complete clinical confirmation testing even with significant support and coverage of costs, highlighting the importance of understanding barriers to confirmation testing. ^27,28,38 The need for clinical confirmation was further underscored by the fact that we had at least one pathogenic result that was not confirmed and several results with discordant interpretations. Discordance in our study is likely partially related to the limited data and standards for variant calling at the time the study was conducted. Nevertheless, discordance is not uncommon, even in clinical testing, and provides additional rationale for confirmation testing, including for VUSs. ⁶³

We acknowledge several limitations. This was an observational study, and many research participants could not be reached, creating a potential nonresponse bias. Reaching participants when research results are not immediately available remains a real-world challenge, making understanding barriers to recontact especially important. Because the study was not randomly assigned, there may be unmeasured differences between groups on the basis of participant self-selection. In addition, the study only included women and was focused on those with a personal or family history of breast cancer. As we had few individuals with positive or VUS results, it is important to confirm findings for these subgroups. Although we used standardized counseling checklists, other differences in counseling could affect outcomes. Differences in levels of engagement with the research cohort and site policies regarding the number of follow-up calls may also be relevant. Finally, outcomes for web-based education could be affected by individual use of the intervention.

In conclusion, we found high interest in a web-based alternative for predisclosure education for return of genetic research results although some patients did not complete web education or receive results. Notably, while uptake of results was not higher among those who requested web education, there was no evidence of misunderstanding or greater distress among those who chose web education. Attention to attrition and lower uptake of results among Black participants and those with less formal education are important areas for future research.

APPENDIX 1

TABLE A1.

The Multimodality RESPECT2 Web-Based Predisclosure Education Intervention

TABLE A2.

Completion of Predisclosure Education by Web Versus Genetic Counseling (n = 427)

TABLE A3.

Factors Associated With Nonresponse and Declining Research Results

TABLE A4.

PROs With Receipt of Genetic Research Results (n = 286)

TABLE A5.

Change in Patient-Reported Outcomes for Web-Based Predisclosure Education Users Relative to Predisclosure Education With a GC After Predisclosure Education and Disclosure

TABLE A6.

Patient-Reported Outcomes With Receipt of Genetic Research Results by Test Result Among All Participants Regardless of the Pretest Education Method

AUTHOR CONTRIBUTIONS

Conception and design: Brian L. Egleston, Linda Fleisher, Jill Bennett Gaieski, Susan M. Domchek, Katherine L. Nathanson, Angela R. Bradbury

Financial support: Katherine L. Nathanson

Administrative support: Wendy K. Chung, Aileen Espinal, Monica Palese

Provision of study materials or patients: Wendy K. Chung, Olufunmilayo Olopade, Payal Shah, Mary Beth Terry, Katherine L. Nathanson

Collection and assembly of data: Wendy K. Chung, Olufunmilayo Olopade, Kara N. Maxwell, Payal Shah, Jane E. Churpek, Mary Beth Terry, Dominique Fetzer, Jill Bennett Gaieski, Jessica Bulafka, Aileen Espinal, Kelsey Karpink, Sarah Walser, Davone Singleton, Monica Palese, Ilona Siljander, Amanda Brandt, Dana Clark, Carrie Koval, Julia Wynn, Jessica M. Long, Danielle McKenna, Jacquelyn Powers, Sarah Nielsen, Susan M. Domchek, Katherine L. Nathanson, Angela R. Bradbury

Data analysis and interpretation: Madison Kilbride, Brian L. Egleston, Wendy K. Chung, Kara N. Maxwell, Linda Fleisher, Mary Beth Terry, Dominique Fetzer, Jill Bennett Gaieski, Kelsey Karpink, Davone Singleton, Susan M. Domchek, Katherine L. Nathanson, Angela R. Bradbury

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Uptake of Genetic Research Results and Patient-Reported Outcomes With Return of Results Incorporating Web-Based Predisclosure Education

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).