The ENGAGE study: a 3-arm randomized hybrid type 1 effectiveness and implementation study of an in-home, collaborative PCP model of remote telegenetic services to increase uptake of cancer genetic services in childhood cancer survivors

Remotely providing genetic services by phone or videoconference as an alternative to in-person delivery could address access barriers. Randomized studies established that patient-reported outcomes (e.g. knowledge, distress) with remote phone services were no worse than in-person services for cancer genetic testing, although uptake of testing in some studies was lower in the remote phone arm than in-person services [27‐30]. Remote videoconference services have been utilized as an alternative to in-person services and have increased during the COVID pandemic [31‐36]. In cancer genetics, remote videoconference services have demonstrated feasibility and high patient satisfaction, but published studies are limited, heterogeneous in setting and delivery, nonrandomized, small and have limited patient-reported outcomes [37‐42]. Given lower uptake of genetic testing in two studies utilizing phone services, and preliminary data suggesting potential benefits in knowledge with videoconference in community-based populations, further evaluation of the relative benefits of videoconference over phone are needed. Additionally, moderator analyses can help identify who benefits less or more from access to videoconference over phone services [43].

In a randomized study of remote telegenetic services delivered on-site at community adult oncology practices, 80% of patients who met NCCN criteria had genetic services, as compared to 16% (OR 30.5, p < 0.001) in the usual care arm (in all cases with their local PCP and not a genetic provider) [44]. While remote genetic services can significantly increase uptake of genetic testing in patients in community practices, the on-site model requires multi-level support within practices and still requires proximity to select community sites, which limits scalability for a nationally distributed population (Fig. 1). Thus, a model where any patient or doctor can access services, where services are provided in the home (eliminating any remaining travel burdens) and are not tied to a participating center, could provide even greater scalability, access to genetic services and uptake of genetic testing, which will be critical to realizing the potential of precision medicine.

To address the gap in access to genetic services, we propose to evaluate the effectiveness of an in-home, collaborative PCP model of remote delivery of genetic services to increase the uptake of recommended genetic assessment and testing in childhood cancer survivors. A PCP Advisory Board, including 8 Family Physicians, General Internists, and an OB/Gyn that practice in diverse community practices (urban, rural, suburban, Northeast, Southeast, Midwest, South and West Coast), informed the development and procedures of the in-home collaborative PCP model and confirm that it is a feasible and acceptable approach (Fig. 1). Advisory board members confirmed the value of access to genetic services, as many PCPs don’t have sufficient expertise to address genetic testing guidelines. They felt many PCPs would be willing to participate if the process was simple and did not require significant provider time. PCPs confirmed that rare and high-risk conditions (e.g. childhood cancer survivors), having patients approach them with requests for services or guideline based recommendations is appropriate (“right place at the right time”). They reported that older PCPs, and those in practices without an electronic medical record or with limited medical and administrative support, might experience barriers to uptake. Recommendations to increase participation included making clear the value of testing, streamlining and simplifying PCP/practice steps, including reminders to minimize burden on the practices, providing access to education materials and answers to frequently asked patient questions and access to the genetic counselor for specific clinical and screening questions even after testing.

Our secondary aims are to evaluate the effectiveness of remote videoconferencing to provide greater increase in knowledge and decrease in distress and depression compared with remote phone services (Aim 2a), to examine the moderators of patient outcomes with remote telegenetic services, to understand who benefits less or more from remote services as compared to usual care, and videoconference as compared to phone counseling (Aim 2b), and to estimate intervention costs and incremental cost-effectiveness of the three study arms (Aim 2c).

Eligibility criteria include individuals enrolled in CCSS, who are 18 years or older, able to understand and communicate in English, reside in the United States, have a history of a CNS tumor or sarcoma (excluding Ewing sarcoma), one or more SMN or have a family history of cancer qualifying them for genetic testing according to the NCCN guidelines [1]. Exclusion criteria include uncorrected or uncompensated speech defects that would lead to the participant being unable to communicate effectively with a medical provider, uncontrolled psychiatric/mental condition or severe physical, neurological, or cognitive deficits rendering the individual unable to understand study goals or tasks. Participants who have already received clinically appropriate multi-gene panel genetic testing are excluded. We elected not to exclude individuals without a usual source of care (e.g. primary care provider) or those without health care insurance. For individuals in these categories, we provide educational materials and resources for obtaining care or insurance.

All potential participants are identified through the CCSS Coordinating Center. The CCSS Coordinating Center uses a combination of electronic (email and text message), mailed, and phone-based recruitment methods to contact all potential participants. The introductory letter includes five videos reviewing the value of genetic testing for childhood cancer survivors (brief animated video, physician explaining the benefit of genetic testing, GC explaining the steps of genetic testing, GC explaining costs and addressing genetic discrimination and GC discussing overcoming apprehension regarding genetic testing), including a study invitation letter by email with five reminders and follow up calls.

The ENGAGE study utilizes a participant portal called myLTFU, a HIPAA-compliant interface that allows direct and interactive messaging and web portal-based data collection. This portal allows participants to consent, view educational content, complete surveys, track study progress outcomes, upload and download documents and pictures, and securely interact with the study team. Communication with participants takes place through the DatStat Connect Platform, which allows for pre-programmed messages and workflows to be activated by situational triggers. Recruitment invitations started in August, 2021.

As outlined below, our enrollment goal for the primary aim is 120 per arm or 360 participants total. We elected to contact childhood cancer survivors of osteosarcoma, soft-tissue sarcoma, and multiple primary cancers, as no additional information is needed to confirm their eligibility for genetic testing. If necessary, we would next contact individuals with a first-degree relative with ovary, male breast or pancreatic cancer as they meet NCCN criteria with this history alone. Next, we would contact those with a family history of colon, uterine or breast cancer, although family history details would need to be assessed to determine if they meet NCCN criteria for genetic testing. As shown in Table 1, we expect to have sufficient CCSS participants to meet our enrollment goal of 360, although if needed we could extend to all childhood cancer survivors and assess family history for eligibility (N = 19,154). Importantly, to increase the generalizability of our findings, we seek to enroll 30% of participants from under-represented minorities.

Participants who contact Penn Telegenetics in Arm A will complete their pre-test counseling session by videoconference in the home or their selected personal location with a Penn Telegenetic Counselor (GC) utilizing communication protocols from related studies [44, 52]. The Telegenetics team will contact the participant up to 6 times (3 calls and 3 emails) to schedule their pre-test counseling session. If a scheduled participant does not show for their appointment, they are contacted 3 times to reschedule. Participants are provided links to download secure videoconferencing software on their home computer or device. The Penn Telegenetics team utilizes a HIPAA-compliant technology platform (Blue Jeans) for videoconference visits. In our communication protocols, if videoconference technology fails, GCs will convert the session to phone. Based on our experience at study start (> 800 telegenetic sessions across studies and clinical contracts), failures occur only 4% of the time, although this may be higher when adapting our model for in-home videoconferencing.

Consistent with standard clinical practice, GCs will review personal and family history (FH), the risks, benefits and limitations of genetic testing, testing options based on their personal and family history (e.g. TP53 testing or a panel of cancer susceptibility genes) and the costs associated with genetic testing. The Penn GC works with the patient to determine appropriate testing and which lab to utilize for testing, provides instructions for completing the test kit and reviews information on testing costs. GCs order the testing through clinical labs (e.g. Invitae or Ambry Genetics) and the labs send test kits to the participant’s home. Through the ENGAGE portal, participants get 3 email and 3 text reminders to return their test kits.

After results are available, the Penn GC shares results with the patient by videoconference at a scheduled visit, and provides the PCP with the chart note, summarizing the results, implications, cancer risk estimates, and standard risk-reducing or screening strategies and implications for relatives. This chart note is also provided to patients, consistent with routine clinical care. Penn GCs are available to answer PCP questions and facilitate referral to regional centers for genetic carriers as indicated. Patients are recommended to follow-up with their PCP to implement any screening and medical (including risk reduction) recommendations.

Participants who contact Penn Telegenetics in Arm B complete their pre-test counseling session by phone in the home or their selected personal location with a Penn Telegenetic Counselor (GC) utilizing communication protocols from related studies [30, 44, 52]. The remainder of procedures for pre-test counseling, genetic testing and post-test counseling sessions are as described above for Arm A.

As above, participants in the usual care arm will receive print materials reviewing ways to obtain genetic services. These include speaking with their doctor about testing or accessing the National Society of Genetic Counselors, “find a counselor” function on their website. At 6 months participants will be contacted to evaluate if they had genetic testing (see 6-month status survey in Effectiveness Outcomes below). All Arm C participants who have not had genetic testing at 6 months will be offered remote telegenetic services in a waitlist design (Fig. 2). If they accept remote services, they are re-randomized to Arm A or Arm B. This future potential for Telegenetic services (e.g. waitlist design) is not shared with them at the time of enrollment. Rather, they are told that after they complete the 6-month status survey.

Genetic testing in all arms will be clinical cancer genetic testing through standard clinical commercial labs, consistent with real-world practice and as indicated based on their personal and family history. Genetic testing is intentionally not covered by the study to reflect “real-world” practice. For the subset of participants who are not able to get testing covered by insurance, commercial labs offer financial assistance programs and out-of-pocket testing is often no more than $250. Based on prior experience in populations meeting NCCN criteria for testing, we anticipate this will be < 5% of participants [44].

If patients elect to proceed with genetic testing, they will be required to provide their PCP’s contact information, who will be the ordering provider as a licensed physician is required to order genetic testing in most states. Consistent with the Penn Telegenetics Patient Access Program procedures, the Penn Telegenetics Team contacts the PCP office, provides information on Penn Telegenetics Program and requests that the provider register with our program. Registration includes providing their credentials and contact information, so that the GC can collaborate with the PCP to facilitate testing. PCP practice staff will be permitted to complete registration on behalf of the PCP. As above, these procedures have been successful to date in a pilot study and were informed by a PCP Advisory Board.

Outcomes in the ENGAGE study have been informed by our conceptual model grounded in The Self-Regulation Theory of Health Behavior (SRTHB) to evaluate innovations in the delivery of genetic services [49, 52‐54]. This model proposes that the reaction to, and use of health (genetic) information is the product of an individual’s understanding, knowledge, and perception of the disease threat and risk reduction behaviors [53, 55‐61]. It emphasizes “common-sense” representations rather than medical or scientific definitions, and incorporates individual cognitive, emotional, familial, and cultural experiences that might contribute to individual variability in understanding. Thus, it has been proposed that the SRTHB is an ideal framework for considering the outcomes of genetic testing [54, 59, 62]. Understanding the moderators of outcomes with delivery innovations are critical to understanding who benefits more, or less from delivery innovations. The literature supports the hypothesis that cognitive, psychological, and behavioral factors will be moderated by biological test results, [63‐66] cancer history, [64, 67‐69] sociodemographic factors (e.g. education, race/ethnicity) [64, 69‐71] and cognitive and emotional factors (e.g. health literacy [72‐74] and self-efficacy [75, 76]).

Our conceptual model is also guided by the Consolidated Framework for Implementation Research (CFIR), which provides and overarching theoretical framework to evaluate barriers, enhancements, and adaptations to increase successful implementation [77]. Even after health-related interventions have proven efficacy and effectiveness, many fail to translate into clinical settings [78, 79]. Thus, there is increasing recognition of the importance of evaluating and addressing barriers to implementation across diverse health care settings. The Consolidated Framework for Implementation Research (CFIR) provides an overarching theoretical framework to evaluate barriers, enhancements and adaptations to increase successful implementation [77]. Based on our stakeholder interviews and implementation science experts, we have included selected constructs from each of the 5 major CFIR domains (intervention characteristics, outer setting, inner setting, characteristics of individuals and process) as they relate to our intervention (Fig. 5) which will be evaluated in our multi-stakeholder mixed-methods process evaluation (Aim 3). As in other studies, [80‐82] evaluation of CFIR components is expected to identify patient, provider and system barriers and enablers of our remote delivery interventions and increase future dissemination to increase access and adoption of genetic services.

The primary outcome of Aim 1 is a composite variable of whether the participant had genetic testing or counseling by six months, and uptake of genetic counseling and identification of genetic carriers are considered secondary outcomes. Uptake of counseling, testing and identification of carriers at 6 months (Aim 1) will be obtained through study records for Arms A and B (intervention arms) and through the 6-month status survey in Arm C (usual care) and participants in Arms A and B who did not complete remote services. This time frame was selected as it allows for a range of delays in completion of counseling and testing and was found to be a reasonable time frame in our prior study [44]. The 6-month status survey includes two closed ended questions for the primary outcome (“Have you completed genetic counseling/testing in the past six months?)”. Secondary closed and open-ended items evaluate how they obtained usual care genetic services (provider, setting, method, and costs), their experience (e.g. what was easy/hard? do you have questions about your results?). For those who did not complete genetic services, secondary items explore barriers to services and what might change their interest in genetic services. The 6-month status survey is designed to be interviewer administered, but if necessary can be completed self-administered. Participants who complete the 6-month outcome survey receive a $75 gift card code.

Outcomes to evaluate the effectiveness of remote videoconference as compared to phone services are shown in Table 2. Upon completion of the T0 survey, T1 survey, and T2 survey, participants receive a $25 gift card code. Participants who complete the T3 survey and T4 survey receive a $30 gift card code.

Implementation outcomes are determined by identifying key CFIR-informed constructs across the five domains to evaluate factors related to uptake of remote telegenetic services, implementation facilitators, and barriers and adaptations (Fig. 5, Table 3). Key informant interviews will be conducted with patients, PCPs and practice staff as part of our Aim 3 implementation evaluation. Key informants will include:

We seek to continue recruitment to the above enrollment goals or until data saturation occurs across the theoretical domain (e.g. no new themes introduced within the constructs included in key informant interviews) [106]. Interviews will be conducted by phone by research staff, audio recorded and transcribed.

We do not expect substantial missing data, although there may be loss to follow-up over time. For the primary outcomes, we will assume that those lost to follow-up did not have pre-test counseling or testing (i.e. failures under the intention-to-treat paradigm). Data from our community practice stakeholder interviews suggests this is reasonable since many patients do not get genetic services even when referred. In secondary analyses, we will account for missing data using the multiple imputation technique of Raghunathan with 25 imputed datasets [107]. We will contrast the results obtained through imputation and those obtained from complete case analyses to study if missing data bias could be substantially affecting our inferences.

We chose our sample size to have sufficient power for both Aims 1 and 2. For Aim 1, the sample size was selected to detect average differences in the primary endpoint (a composite variable indicating either genetic testing or genetic counseling) between the primary randomization arms of interest. The primary comparison groups will be the usual care/waitlist group versus the combined remote telegenetics groups (phone or video conference as one group); the primary analysis will consider the groups prior to the waitlist rerandomization. Preliminary data demonstrated that uptake of genetic testing from an ongoing registered randomized trial of remote telegenetic services vs. usual care for cancer genetic testing in community practices was at least 53% (29/55 with uptake) in the intervention arm, but only 17% (4/24 with uptake) in the control arm. With 120 patients in the usual care/waitlist group and 240 patients in the combined telegenetics groups with complete data (after accounting for loss to follow up), we will have > 99% power to detect a similar difference in uptake of genetic testing. This assumes a 1% Type I error rate (2-sided) and the use of Fisher's Exact Test. We set the Type I error rate to a conservative 1% to partially account for multiple hypothesis testing when including secondary outcomes and potential moderators in Aim 3. Given the magnitude of the pilot data differences, we anticipate excellent power for the moderator/subgroup analyses. The comparison of the usual care/waitlist group to the combined telegenetics group will be the comparison used to determine if the study accomplishes its primary objective. The uptake of pre-test counseling (Aim 1 secondary objective) and comparison between the phone and videoconference groups (see Aim 2a) are secondary objectives.

We hypothesize that remote telegenetic services by videoconferencing will be associated with greater decreases in cancer related distress and depression and increases in knowledge when compared to phone services. We will compare the usual care at baseline group to the combined arms receiving remote services at baseline. For comparisons between the remote videoconferencing versus phone groups, we will assign participants to the remote group assigned at baseline, or after rerandomization for the waitlist group. Our primary change scores for the first three variables will be change between baseline and immediately post genetic testing. For the waitlist group, there will be a second baseline measurement after the usual care period has ended. For the 3 primary analyses in Aim 2, we will use T-tests. We will assess balance of potential confounders (e.g. age, race, study site, waitlist assignment) between arms using T-tests or Fisher’s exact tests as appropriate. Waitlist group will be considered a confounder in these secondary regression analyses. We will use multiple linear regression models to control for potential confounders inadequately balanced. We will include in the models the confounding variables and indicator (binary 0/1 variables) to indicate randomization arms (leaving one group out as the reference). For longitudinal analyses, we will examine time trajectories using regressions estimated by Generalized Estimating Equations (GEE) to account for within subject temporal correlation. Panel time will be included via indicator variables. We will also include interaction terms between randomization arm and time indicators to investigate temporal effects [108]. We will repeat analyses, but assign participants to per-protocol (i.e. restricting sample to those who properly used the assigned method of disclosure) or as-treated groups (e.g. assigning all participants to a group based on the method of disclosure used, regardless of assignment). For Aim 2 secondary outcomes, we will use multiple linear regressions for continuous outcomes and multiple logistic regressions for binary outcomes using GEE-estimation as described.

We further chose our sample size such that we would have sufficient power for Aim 2 comparisons of the videoconference versus phone arms. After re-randomization of the waitlist group, we anticipate potentially having 180 people/arm randomized to the videoconference or phone arms. Based on prior experience, we anticipate that loss to follow-up will be less than 28% in each arm, leaving us with 130 evaluable participants with follow-up data in each group (180–180*28% = approximately 130). Some of the loss to follow-up will occur prior to the waitlist group being rerandomized. We used our pilot data to determine power for this aim (see Table 2 for estimates). We assumed 85% power and a 1.67% Type I error rate. We used a 1.67% Type 1 error rate (2-sided) by applying a Bonferroni correction for the three outcomes of interest to the typical 5% Type I error rate (5%/3 = 1.67%). We see in Table 4 that we have excellent power for all three arms with at least 95/arm, which is well below our expected 130/arm anticipated sample size. Table 4 also demonstrates that we have sufficient power for subgroup analyses, with a sample size of just 22/arm needed for knowledge comparisons between arms.

To evaluate moderators of patient cognitive, affective and behavioral outcomes (Aim 2b), we will use logistic regressions in which we include variables of interest as covariates. For moderation (i.e. effect modifier analyses), we will include indicator variables for randomization arms in the GEE-estimated multiple linear (for continuous outcomes) or logistic regressions (for uptake). We will also include panel time indicators, the potential moderator variables, and all two-way and three-way interactions among the arm/time/moderator variables. Interactions are created by multiplying two variables together [108]. We will examine non-time moderators separately, and will examine multiple regression models with potential confounding variables.

We will repeat the analyses but assign participants to per-protocol (i.e. restricting sample to those who properly used the method of disclosure as assigned) or as-treated groups (e.g. assigning all participants to a group based on the method of disclosure used, regardless of group assignment).

The economic impact of remote telegenetic services (Aim 2c) will be assessed by performing: 1) cost analysis and 2) cost-effectiveness analysis. The goal of the cost analysis is to estimate the cost of delivering remote telegenetic services, by phone and videoconference, compared to usual care. As above, the cost analysis will take a societal perspective, including the costs of all resources consumed for the implementation and delivery of remote genetic services. Separately, we will examine costs from the health system perspective that can inform decisions about the provision and reimbursement of services, and ultimately will influence the effective dissemination and implementation of remote telegenetic services beyond the study. Cost-effectiveness will be estimated as the incremental cost per (1) additional survivor who receives genetic testing, and (2) additional mutations detected. The numerator of the incremental cost-effectiveness ratio (ICER) will include intervention costs and patient time and travel costs, estimated as described above. In the first analysis, the denominator will be defined by the primary trial endpoint of genetic testing received within 6 months. Because receipt of genetic testing is the measure of effectiveness in this case (i.e. the denominator of the ICER), neither the cost of testing nor the cost of related services provided after genetic testing will be included in the numerator. For the second analysis, the numerator of the ICER will include both genetic test costs and any related health care service costs (additional tests, visits, counseling) associated with genetic testing at 6 months. Incremental cost effectiveness will be estimated using standard methods, and sensitivity analysis will be used to assess the impact of assumptions and uncertainty on results and conclusions [109, 110].

We hypothesize that our CFIR-informed multi-stakeholder mixed-methods process evaluation will identify barriers to uptake and inform recommendations for future adaptation and sustainability. We will quantitatively characterize factors associated with uptake using T-tests (continuous predictors), chi- squared tests (categorical predictors), and multiple logistic regressions (multivariable models). We will control for randomization arm in multivariable models. Qualitative data will be analyzed using a deductive content analysis approach and CFIR as the coding framework [77, 82, 111]. The qualitative data will include patient surveys and key informant interviews, notes from the PCP Advisory Board and notes and debriefing from the telegenetics service and research team meetings. Guided by consensual qualitative methods [112, 113], two independent coders review notes and assign responses to CFIR constructs. Finally, using a convergent mixed-methods approach [114], we will merge the quantitative and qualitative data, organized by the 5 CFIR domains to evaluate which constructs were associated with, or more prevalent among: a) patients who had testing and those who did not and b) PCPs who registered and their patient completed testing, PCPs who registered and their patient did not complete testing, PCPs who never registered (were contacted by the patient but did not support testing through our model of remote telegenetic services). As in other studies [80‐82], evaluation of CFIR components is expected to identify patient, provider and system barriers and enablers of our remote delivery interventions and increase future dissemination to increase access and adoption of genetic services.