Authors: Kathleen F. Mittendorf (1.Division of Hematology/Oncology, Department of Medicine, Vanderbilt University Medical Center, Nashville, TN), Harris T. Bland (2.Department of Biomedical Informatics, Department of Medicine, Vanderbilt University Medical Center, Nashville, TN), Justin Andujar (3.Division of Genetic Medicine, Department of Medicine, Vanderbilt-Ingram Cancer Center, Vanderbilt University Medical Center, Nashville, TN), Natasha Celaya-Cobbs (3.Division of Genetic Medicine, Department of Medicine, Vanderbilt-Ingram Cancer Center, Vanderbilt University Medical Center, Nashville, TN), Clasherrol Edwards (4.Department of Microbiology, Immunology and Physiology, Meharry Medical College, Nashville, TN, 37208), Meredith Gerhart (1.Division of Hematology/Oncology, Department of Medicine, Vanderbilt University Medical Center, Nashville, TN), Gillian Hooker (3.Division of Genetic Medicine, Department of Medicine, Vanderbilt-Ingram Cancer Center, Vanderbilt University Medical Center, Nashville, TN; 5.Concert Genetics, Nashville, TN), Mryia Hubert (1.Division of Hematology/Oncology, Department of Medicine, Vanderbilt University Medical Center, Nashville, TN), Sarah H. Jones (6.Institute for Medicine and Public Health, Vanderbilt University Medical Center, Nashville, TN), Dana R. Marshall (7.Department of Pathology, Anatomy and Cell Biology, Meharry Medical College, Nashville TN), Rachel A Myers (8.Department of Medicine Clinical Research Unit, Duke University School of Medicine, Durham, NC, 27705), Siddharth Pratap (4.Department of Microbiology, Immunology and Physiology, Meharry Medical College, Nashville, TN, 37208), S. Trent Rosenbloom (2.Department of Biomedical Informatics, Department of Medicine, Vanderbilt University Medical Center, Nashville, TN), Azita Sadeghpour (9.Duke Precision Medicine Program, Department of Medicine, Duke University, Durham, NC), R. Ryanne Wu (9.Duke Precision Medicine Program, Department of Medicine, Duke University, Durham, NC; 10.23andMe, Sunnyvale, CA), Lori A. Orlando (9.Duke Precision Medicine Program, Department of Medicine, Duke University, Durham, NC), Georgia L. Wiesner (3.Division of Genetic Medicine, Department of Medicine, Vanderbilt-Ingram Cancer Center, Vanderbilt University Medical Center, Nashville, TN)
Categories: Article, Hereditary cancer, risk assessment, genetics, digital health, service delivery models, SMART on FHIR
Source: Contemporary clinical trials
Authors: Kathleen F. Mittendorf, Harris T. Bland, Justin Andujar, Natasha Celaya-Cobbs, Clasherrol Edwards, Meredith Gerhart, Gillian Hooker, Mryia Hubert, Sarah H. Jones, Dana R. Marshall, Rachel A Myers, Siddharth Pratap, S. Trent Rosenbloom, Azita Sadeghpour, R. Ryanne Wu, Lori A. Orlando, Georgia L. Wiesner
Hereditary cancer syndromes cause a high lifetime risk of early, aggressive cancers. Early recognition of individuals at risk can allow risk-reducing interventions that improve morbidity and mortality. Family health history applications that gather data directly from patients could alleviate barriers to risk assessment in the clinical appointment, such as lack of provider knowledge of genetics guidelines and limited time in the clinical appointment. New approaches allow linking these applications to patient health portals and their electronic health records (EHRs), offering an end-to-end solution for patient-input family history information and risk result clinical decision support for their provider.
We describe the design of the first large-scale evaluation of an EHR-integrable, patient-facing family history software platform based on the Substitutable Medical Applications and Reusable Technologies on Fast Healthcare Interoperability Resources (SMART on FHIR) standard. In our study, we leverage an established implementation science framework to evaluate the success of our model to facilitate scalable, systematic risk assessment for hereditary cancers in diverse clinical environments in a large pragmatic study at two sites. We will also evaluate the success of the approach to improve the efficiency of downstream genetic counseling resulting from pre-counseling pedigree generation.
Our research study will provide evidence regarding a new care delivery model that is scalable and sustainable for a variety of medical centers and clinics.
This study was registered on ClinicalTrials.gov under NCT05079334 on 15 October 2021.
Hereditary cancer syndromes affect approximately 1–2 in 200 people and cause a high lifetime risk of multiple cancer types, which often onset earlier and are more aggressive.^1–3^ Early identification of patients at risk for hereditary cancers allows cancer prevention and early detection interventions that greatly reduce morbidity and mortality.^2, 4–10^ To improve outcomes, at-risk individuals must be recognized, offered genetic counseling and testing, and referred for subsequent care, ideally before cancer develops.
Despite guideline recommendations to assess family health history (FHH) in primary care, not all clinicians are aware of or implement these guidelines, which are complex and frequently revised.^11–18^ Increasing demands on primary care clinicians leave little time for family risk assessment. Lack of genetics knowledge and comfort managing patients at risk for hereditary syndromes exacerbates this barrier.^16–18^ When family history is collected, it is often detail-limited, making it difficult to meaningfully utilize.^11, 13^ As a result, many eligible individuals are not offered genetic counseling and testing, a care gap that more drastically impacts medically marginalized populations.^14, 18–23^
These barriers contribute to inequities inherent in the traditional genetics services delivery model, which is labor intensive and strains an overburdened primary care and genetic counseling workforce (Figure 1A). An efficient care delivery model is needed to improve recognition of patients at risk for hereditary cancer syndromes while not overburdening the genetics workforce.^24^ The Family History and Cancer Risk Study (FOREST) study, funded by the National Cancer Institute (NCI) as part of the Beau Biden Cancer Moonshot^SM^ Initiative, aims to address these systems-level barriers. We leverage 1) a patient portal recruitment strategy; 2) a validated, patient-facing electronic FHH application that can be integrated with patient portals and their electronic health records (EHRs) using the SMART-on-FHIR standard; and 3) clinician-targeted clinical decision support (CDS) to provide guideline-concordant recommendations on follow-up care for at-risk patients (Figure 1B). To our knowledge, this is the first large-scale evaluation of a SMART on FHIR-based FHH application. This approach allows vendor-agnostic integration of FHH with patient portals and EHRs, immediate availability of risk results to patients and clinicians, and EHR-integrated clinical decision support for downstream management. We present the clinical trial design (NCT05079334) for evaluating this unique strategy.
The FOREST study is deploying a care delivery model to facilitate systematic hereditary cancer risk assessment in diverse clinical environments; we hypothesize that our model will improve identification of individuals eligible for genetic counseling. We are using MeTree, a validated patient-facing, electronic FHH ascertainment and clinical decision support platform.^25–27^ MeTree uses patient-reported FHH data to generate a pedigree and create a personalized report as well as provider-directed clinical decision support. We are evaluating whether this model can facilitate scalable, systematic risk assessment for hereditary cancers in diverse clinical environments in a large pragmatic implementation science study at two sites.
We are using the Reach Effectiveness Adoption Implementation Maintenance (RE-AIM) model^28, 29^ in the pre-implementation, implementation, and maintenance phases of the study to guide site-specific implementation strategies. This framework ensures external validity and sustainability in “real-world” settings. Because each site requires tailored strategies, the RE-AIM outcomes drive adjustments during pre-implementation, implementation, and maintenance to ensure “real-world” success. Our primary outcome concerns Reach but our secondary outcomes explore other aspects of the RE-AIM framework (Table 1).
The VUMC Institutional Review Board (Approval #201202) approved this study. Other participating sites’ IRBs ceded to the VUMC IRB. The study is registered on clinicaltrials.gov (NCT05079334).
We are recruiting adult participants from two health systems in Nashville, TN, USA: Vanderbilt University Medical Center (VUMC) and Meharry Medical College (MMC). VUMC is a comprehensive academic healthcare facility and home to Vanderbilt-Ingram Cancer Center (VICC). MMC is one of the nation’s oldest historically Black academic health science institutions. It partners with Meharry Medical Group and is home to several inner-city clinics that predominantly serve medically marginalized populations.
The VUMC population is sourced from English-speaking adult users of VUMC’s patient portal, My Health at Vanderbilt (MHAV). As of quarter three conclusion 2022, over 770,000 unique English-speaking adult patients have an MHAV account. English-speaking MHAV users are approximately 63% female (legal sex) and 72% non-Hispanic White. Based on primary race/ethnicity, 9% are Black and 3% have one or more Hispanic/Latino-related ethnicities. Approximately 94% of English-speaking patients whose primary medical home is VUMC (those who had encounters in primary care, medicine subspecialties, or in internal medicine in the prior fiscal year) have an MHAV account.
We are recruiting at two Meharry 1) Family Medicine, which is part of the Meharry Medical Group (MMG), and 2) Internal Medicine, which is part of Nashville General Hospital (NGH). Over 70,000 patients are cared for at MMC, and over 10,000 patients are seen annually in the Internal Medicine, Family Medicine, and other affiliated clinics. Internal Medicine is the largest Meharry clinic, with ~7,500 patient encounters annually. About 53% are female, 33% are white, and 56% are Black/African American. For patient-portal-based recruitment, the MMG portal has ~8,000 registered users. In the Family Medicine Clinic, ~2,500 patients met inclusion criteria, were web-enabled on the portal, and had encounters in 2023–2024.
Because we hypothesize that FHH collection and pedigree creation by MeTree can make the genetic counseling process more efficient, we will analyze MeTree’s influence on the length of genetic counseling appointments at the VUMC Hereditary Cancer Clinic (HCC). The HCC serves patients in the Nashville, TN, area and has approximately 1,400 patient visits annually. A review of 5,680 unique new HCC patient visits showed that patients were mostly female (86%); over half (53.4%) had a diagnosis of cancer, and 98% had a family history of cancer; 67.5% were non-Hispanic White, 5.8% were African American, 2.4% were Hispanic, and 3.1% belonged to other racial/ethnic groups.
The FOREST study has inclusion and exclusion criteria related to ability to utilize MeTree (Table 2).
During pre-implementation we assessed existing barriers, resources, and current workflow based on input from key clinic staff/managers at VUMC and MMC. This assessment informed adaptations to implementation, including site-specific MeTree deployment plans, such as site-tailored technical, clinical, and patient portal elements. Pre-implementation work will be described in greater detail in another publication.
At VUMC the platform was integrated into proof-of-concept test systems to evaluate workflow and data sharing capabilities through SMART on FHIR from the patient and clinician standpoints. Community engagement studios of patients and community leaders were held using a standardized format via the Vanderbilt Institute for Clinical and Translational Research. The VUMC patient community engagement studio was focused on user experience. Suggested changes to MeTree were implemented wherever feasible. Studio participants also helped improve recruitment messaging approaches.
The MMC EHR had theoretical but not concrete support for SMART on FHIR; thus EHR integration was identified as a significant barrier to recruitment and enrollment. A manual recruitment process was predominantly used. Two community engagement studios were conducted pre-implementation to gather patient concerns and integrate input into the final recruitment workflow. The studios were most informative about the Implementation (Pauses and Barriers) aspects of the RE-AIM framework. During study course, technical adaptations to improve SMART on FHIR integration at low-resource sites are tracked to address the Maintenance (Resources to maintain) aspects of RE-AIM.
Following patient portal invitation (VUMC) or in-person recruitment (MMC), prospective participants complete an eligibility assessment (Figure 2, Table 2). Eligible participants are offered electronic consent, and consented individuals complete a baseline survey collecting measures of interest. After baseline survey completion, participants are directed to the MeTree family health history platform. Following MeTree completion, participants receive a risk summary and guidance suggesting genetic counseling for those meeting guideline criteria. Participants who complete MeTree receive a post-MeTree follow-up survey; participants who attend genetic counseling receive another follow-up survey.
We will recruit numbers sufficient to reach 500 completed MeTree pedigrees based on a power analysis for the primary outcome (see Evaluation of Study Outcomes section, below).
Recruitment at VUMC began in October 2021. We leveraged MHAV to identify and send mass research recruitment messages to adult patients with a VUMC visit in the last three years and an active MHAV account. Prior to November 2022, messages were sent via MHAV or text message, depending on research notifications consent type. If no notification consent was provided, study information appeared on their research studies page in MHAV; following implementation of an opt-out consent approach in November 2022, recruitment messages were sent to all VUMC MHAV account holders who did not opt-out of MHAV-based recruitment messages. The study is also listed on Research Match, a nonprofit program funded by the National Institutes of Health (NIH) for connecting people interested in research studies with researchers.^30^ Interested individuals can reach out to the FOREST study email to request an MHAV invitation.
Recruitment began at MMC in September 2022 in primary care clinics. A clinical research coordinator asks patients attending a primary care visit are asked about interest in study participation. Interested patients who meet eligibility requirements and have an MMC provider can enroll either on-site or on-line in partnership with the Nashville General Hospital Internal Medicine Clinic and Meharry Medical Group Family Medicine Clinic. Study participants may choose an enrollment process guided by study staff using a laptop at the time of enrollment, a scheduled appointment with the research coordinator, or an independent process completed on their own time and device. Recruitment was updated to include patient portal-based messaging similar to the VUMC recruitment methods in September 2024. Messaging was sent by either the patient portal platform or text to eligible patients who met inclusion criteria, had a clinic encounter in 2023–2024, and were text/web-enabled on the portal.
Participants enroll on a rolling basis using an eligibility survey to evaluate study inclusion/exclusion criteria (Table 1). Enrollment, electronic consent (eConsent), and baseline surveys are administered and tracked using REDCap (Research Electronic Data Capture).^31, 32^
Once a VUMC patient clicks on the notification link in the email/text or from the MHAV homepage, they are redirected to a research studies page where they can indicate interest in FOREST. REDCap receives participant interest messages, and either sends a follow-up email with more information about the study and a link to the eligibility survey (participants who choose “interested”) or registers the participant as not interested to prevent further study contact (participants who choose “not interested”). At VUMC, participants complete the study online, but contact information is provided for those needing assistance to complete any study-related surveys and/or MeTree..
At MMC, participant interest/non-interest is not recorded due to limitations with the in-person recruitment approach. Interested participants are offered the eligibility survey in the clinic on a laptop by a study team member. Time constrained individuals are offered the option to receive an email to study materials or to return to the clinic to complete eligibility, consent, baseline, and/or MeTree according to participant preference. Participants who intend to complete the entire process from eligibility through MeTree in clinic are told that this process may take up to two hours. Study navigation is a method that has been previously shown to be helpful in medically marginalized populations and has been most helpful in prior MMC studies.^33^
Participants receive a REDCap-hosted baseline survey to collect sociodemographic and other relevant measures of interest such as genetic knowledge and self-efficacy. All baseline measures were selected from validated or established survey instruments or measures (Table 3). Participants receive a $10 incentive for completing steps through the baseline survey.
Following baseline survey completion participants are immediately directed to MeTree via electronic link, which they also receive by email. Upon MeTree entry, participants can enter personal health, demographics, and FHH information for 143 conditions, including 20 hereditary cancer syndromes (Table 4). At least six family members (parents, maternal and paternal grandparents) need to be entered for MeTree to provide risk assessment. Because MeTree assesses risk for many conditions, not just those related to cancer, MeTree completion time varies based on family structure and medical history; prior work found an average time of 27 minutes (though a limited dataset is allowed to proceed with a cancer risk assessment in the FOREST study).^26^
Participants can save and return to MeTree to update or add new information for three weeks following their index MeTree access date, after which time the account locks to prevent additional modification and to finalize risk assessment. Participants who do not access MeTree within 30 days of consent are sent up to four weekly reminder emails. After completing MeTree, participants receive a personalized patient report indicating whether they meet criteria for cancer genetic counseling for hereditary cancer syndromes according to American College of Medical Genetics and Genomics (ACMG) and National Society of Genetic Counselors (NSGC) guidelines.^34, 35^ At risk patients are offered genetic counseling. Individuals receive a 40 (MMC) incentive for completing MeTree. Specific VUMC and MMC processes are outlined below. Four to six weeks following MeTree completion, participants are sent an email link to a REDCap-hosted post-MeTree survey (Table 3; see Surveys section, below), for which up to four weekly reminder emails are sent.
A link to MeTree is provided by email following completion of the baseline survey. Once activated, the link initiates a SMART authentication and authorization protocol that launches the MeTree interface using the SMART authentication token provided by the VUMC MHAV patient portal and executes five FHIR APIs supported by VUMCs EHR (patient, familymemberhistory, condition, observation, and procedure). These FHIR APIs pre-populate fields in MeTree with information from the medical record to minimize data entry burden and enhance accuracy. Participants then complete data entry for the remaining unpopulated fields and add additional relatives to their pedigree as needed. Following account lock, the MeTree-generated pedigree and risk report is imported into the study REDCap and uploaded to the EHR. VUMC providers are notified via secure email that their patient has a MeTree risk assessment and a clinical decision support report is available in the medical chart (accessible via a SMART link and as a PDF document in the media tab). If the patient is at risk for a hereditary cancer syndrome, they are informed that they qualify for genetic counseling, which is subject to health insurance coverage.
At MMC, a study team member manually creates participant MeTree accounts. Participants have the option to complete data entry in-person using a laptop, to have the team member assist the participant either in-person or virtually, or to complete data entry with their own device independently. Login credentials and a site-specific link to MeTree are sent via email for those completing MeTree independently. Study staff manually upload the report into the EHR and (for at risk patients) send an encrypted email alerting the MMC provider that a risk report is available. If the MMC participant is found to be at risk for hereditary cancer syndromes, the MMC study team contacts the participant by phone or email to advise that genetic counseling is available with the VUMC HCC. They are also encouraged to discuss their risk with their MMC doctors. The MMC study team assists participants in navigating insurance and, for uninsured participants, covers the cost of the genetic counseling visit.
Four to six weeks following MeTree completion, participants receive a brief follow-up post-intervention survey (Table 3; up to 35 items depending on branching logic), designed to assess participant experience using MeTree and the impact of MeTree on participant health behaviors and information sharing with their health provider(s) and families. Participants receive an additional 40 (VUMC) or $60 (MMC).
Participants who accept counseling attend a standard telemedicine or in-person clinical appointment in the VUMC HCC. The counselor provides genetic counseling, risk assessment, education, and genetic testing, as appropriate according to standard practice. Using REDCap, we track uptake and time spent in the appointment. Providers are notified of their patient’s results via usual practice. Post-test counseling is offered to participants who undergo genetic testing. When a likely pathogenic or pathogenic variant is found, the genetic counselor initiates risk management recommendations.
For participants who elect to see a genetic counselor, we distribute a post-genetic counseling survey (up to 40 items depending on branching logic and genetic testing decisions) designed to assess the impact of MeTree on the genetic counseling experience and participant health behaviors following genetic counseling. Participants receive an additional 50 (VUMC) or $70 (MMC) total study compensation.
We will report the number and proportion of participants who complete each step of the FOREST study. REDCap tracks all data necessary to evaluate completion of each step upstream of MeTree, as well as the post-MeTree and post-genetic counseling surveys. Back-end functionality in MeTree tracks which participants create an account, the amount of data they complete, their individual data inputs, and their computed results.
Because it is difficult to directly measure whether risk assessment occurs in the control condition of usual care, we will utilize a proxy measure for the ability of MeTree to improve systematic risk the number of patients identified as at risk of a hereditary cancer syndrome in the usual care (control) and intervention conditions. The control condition utilizes historical data from the participants prior to their enrollment date in FOREST to determine the number and proportion of participants who would be identifiable as at risk without the implementation of MeTree. EHR billing code data will be used to determine the proportion of enrolled participants who had indicators of risk ascertainable from the EHR prior to their participation in FOREST.
There is no single billing code that fully encompasses hereditary cancer risk; as such, we will define our codes indicating risk based on billing code data associated with an existing cohort of VUMC HCC patients that were documented prior to HCC referral, such as specific cancer bused as a proxy to evaluate what proportion of participants with an “at-risk” MeTree were identifiable from EHR data prior to engagement with the MeTree app. This evaluation will demonstrate whether MeTree has added value as a clinical screening tool. The RE-AIM effectiveness of MeTree will be assessed by comparing the incidence of at-risk identification before and after MeTree, using the McNemar test of marginal homogeneity. The McNemar test was selected for its suitability – in paired samples, such as pre- and post-MeTree exposure for a given participant. Significance will be assessed using a two-sided test with a p-value threshold of 0.05.
Sample size projections to determine effectiveness of MeTree at identifying at risk patients were assessed within the participants who completed MeTree. The required sample size was estimated assuming a McNemar test for detecting differences between MeTree and EHR review with the following the probability of not being identified as at risk based on the EHR and being at risk based on MeTree (p01) is 0.10 and the probability of being identified as at risk based on the EHR and not at risk by MeTree (p10) is 0.05. Under these assumptions, 500 participants completing MeTree are required to power the study at 80% with a significance level of 0.05.
After a visit occurs, study staff request the Vanderbilt Clinical Informatics Core (VCLIC) to extract timestamps for visit start and complete times by provider. VCLIC extracts data using Clarity by Epic. Appointment time data from the control cohort of patients (referred to the HCC through usual care) will be compared to the intervention cohort (referred to the HCC after completing MeTree). We hypothesize that there will be at least a 5-minute reduction in pre-test counseling appointments due to the extensive family history gathering and pedigree generation by MeTree. The clinical genetic counselor is also surveyed about the value of the MeTree-generated pedigree in the clinic session.
For at-risk patients who receive care recommendations from the genetic counselor and/or their MeTree report, EHR data is used to assess uptake of and adherence to appropriate risk management strategies. This data is used in concert with data from the baseline survey.
To ascertain whether participants share their at-risk results with family members, the study tracks how many times participant’s share their unique family resource web link, whether the link was accessed (clicked on), and the number of family members who enroll in the study. These data will be reported in descriptive analyses only.
FHH has been a critical component of identification of patients and family members who may be at risk for a hereditary cancer syndrome. However, FHH is under-collected in current clinical practice, meaning most at-risk patients go unidentified, a care gap that disproportionately impacts patients from medically marginalized groups.^11, 13, 14, 19, 36^ Additional leaks exist within this pipeline in terms of receipt of appropriate referrals and counseling, and these leaks disproportionately hemorrhage patients from underserved groups.^14, 19–21, 23, 36^ As such, the traditional model is not equitable, scalable, nor sustainable, and the resource-intensive nature of the traditional model is poorly suited to community health and low-resource settings, like that of MMC.
Informatics and telecommunication innovations can be used to develop a sustainable, scalable, replicable genomics care delivery model that can be deployed across a medical system to address leaks in the genetics services delivery pipeline. MeTree is a validated, flexible patient-facing FHH collection tool previously used as the backbone of the Implementing GeNomics In pracTicE (IGNITE) network’s family history clinical utility study,^37, 38^ where it showed clear improvements in the quality and quantity of FHH collected in 5 geographically diverse primary care practices.^39^ SMART on FHIR is EHR-vendor it is already supported by most major EHR vendors, and the 21st Century Cures Act and the Federal Interoperability and Patient Access Final Rule required SMART on FHIR support for all certified health IT by the end of 2022.^40, 41^ Our study will provide the first large-scale evaluation of an FHH platform that can be directly integrated with EHRs and associated patient portals in a vendor-agnostic fashion leveraging current and evolving informatics standards. It will also provide evidence that alternative approaches can be implemented in systems where technical support does not yet exist. Such programs can act as a stop-gap and create institutional readiness for technology adoption when it becomes available.
Our outreach strategy and the patient-driven risk assessment approach via MeTree is designed to address multiple structural barriers to identification and genetics evaluation of at-risk individuals that contribute to the leaky pipeline. The first roadblock MeTree removes is the time and knowledge required by the primary care clinician to assess FHH and calculate risk. FOREST aims to automate and systematize patient identification and FHH-based risk assessment, so that more patients have full access to appropriate FHH-based risk assessment. Further, by integrating guideline-concordant clinical decision support directly into the EHR, it removes the burden on the primary care provider to stay up-to-date on guidelines in the rapidly evolving specialty field of genetics. Systematic FHH collection should also streamline resultant genetic counseling appointments by reducing FHH collection time during the appointment, addressing another important structural barrier related to genetics workforce shortages and the traditional labor-intensive clinical model.^24, 42^ We expect that the length of time can be reduced up to 20 minutes based on a prior VUMC HCC quality improvement project.
By evaluating this intervention in two settings, one of which serves predominantly patients from medically marginalized populations, we will evaluate whether this approach has clinical utility and can improve ascertainment of patients at risk for hereditary cancer syndromes.