The screening pilot occurs in Vermont, United States. Vermont is among the few states making strides toward healthcare reform with emphasis on value-based care (Grembowski and Marcus-Smith, 2018; Kissam et al., 2019). The focus signals openness to investment in innovative health prevention activities. Vermont’s accountable care organization (ACO), OneCare Vermont, is facilitating the transition to value-based care models. Federal, state, and private health insurers contract with the ACO and enrolled providers for a risk-adjusted, quality-focused, single annual payment for healthcare services. Alignment with the ACO allows better visibility into the real-world health impacts of innovations in population health screening.

Research involving return of actionable genomic sequencing results to patients for clinical use (Duow and Marjanovic, 2016; Linderman et al., 2016; Sanderson et al., 2016; Suckiel et al., 2016; Ryan et al., 2017; Murray et al., 2018; Rego et al., 2018; Reuter et al., 2018; Sapp et al., 2018; Schwartz et al., 2018; David et al., 2019; Nussbaum et al., 2019; Williams, 2019; Zoltick et al., 2019; Walton et al., 2020; David et al., 2021; Kelly et al., 2021; Khoury and Dotson, 2021; Lemke et al., 2021; Miller et al., 2021), potential harms of proactive testing, quality of next generation sequencing technology, and implementation of genomic medicine (Weitzel et al., 2016; Ginsburg et al., 2019; Williams, 2019) strongly informed our design.

Primary care is not a traditional setting for genetic testing or screening. Primary care providers do order pre-conception and prenatal screens and sample for newborn screening. Genomic literacy and competency among primary care providers is limited outside those areas. Upon receiving a positive genetic screening result, primary care providers’ responsive actions may be limited to patient notification and referral to a relevant specialist, or to following scripts, such as those provided by newborn screening laboratories. In general, time is the most limited resource for primary care providers and their staff. At the same time, risk assessment and directing and managing preventive care, the main objectives of genomic population health, occurs principally in the primary care setting.

Professional guidelines and resources for actionability of results, including the ACMG secondary findings guidance (American College of Medical Genetics and Genomics, 2019; Directors of the American College of Medical Genetics and Genomics, 2019; Nussbaum et al., 2019), ClinGen expert assessments (Rivera-Muñoz et al., 2018), and locally sourced specialty specific guidance, served as anchors for the design. However, updated non-genetics specialty practice guidelines are scarce for many of the health risk genes or are based on data from patients screened because of affected family members, often after an affected member had a positive indication-based test result. Current breast cancer genetic testing guidelines fail to identify almost half of individuals with a breast cancer risk gene pathogenic or likely pathogenic variant (Beitsch et al., 2019).

Clinical genetics laboratories now have extensive experience classifying the pathogenicity of gene sequence variations according to standardized systems (Richards et al., 2015; Nykamp et al., 2017) and linking variants to peer-reviewed literature supporting clinical validity. New variant and clinical validity/utility information evolves and justifies re-classification of variants with a necessity to update clinical reports. Nonetheless, evidence is lacking to accurately classify much of the human genomic sequence variation as pathogenic, benign, or likely so. For these variants of unknown/uncertain clinical significance (VUSs), it is not currently known whether they impact health.

Anecdotal reports describe missed, inappropriate, and or unnecessary medical responses after genetic or health-risk testing. These have been used to warn against broad-based genomic screening at population scale (Murray et al., 2018). Restricted genetic competency among non-geneticists tasked with interpreting genetic test results may facilitate insufficient responses even when preventive opportunities exist. Genetic disease expertise clearly has a role in population genomic screening (Lemke et al., 2021).

The popularity of consumer-oriented genomic testing and concerted efforts to increase the genomic literacy of Americans has fostered growing public awareness of links between heritable genetic variation and disease. Programs that performed health-related genomic screening tests for physicians and health administrators have helped them personally identify with the potential for routine genomic risk screening and raised awareness and interest among non-genetic specialists and primary care leaders (Briggs, 2016; Masterson, 2016).

At the same time, widespread testing has raised concern regarding the privacy of genetic information, genetic discrimination, as well as the commoditization of genetic data. Many people are unaware that a genetic result obtained outside of a healthcare setting is not subject to HIPAA privacy law protections nor CLIA laboratory quality certification, and many lack clarity about the extent of protections against genetic discrimination provided by federal and state laws.

Information relevant to a patient’s health is recorded in the health record. Yet electronic health records (EHRs) generally lack robust, expansible, accessible, and readily implementable functions to store, annotate, retrieve, and update germline genetic information and annotations that may remain clinically relevant for many decades (Walton et al., 2020).

While large cohorts of research participants have received exome or genome sequencing results, fully clinical programs screening large numbers of health risk genes have until recently been offered only in clinics catering to self-pay clients. Research screening programs are being adapted to a clinical model.

The University of Vermont Medical Center (UVMMC) is a regional academic tertiary care center serving a largely rural population in Vermont and northern New York, where Northern-European ancestry and white race are claimed by most of the population. UVMMC is the academic teaching hospital of the UVM Health Network that includes five other rural hospitals, home health and hospice, a physician organization, and collaboration with a Federally Qualified Health Center. By the end of 2022, all will operate on UVMMC’s Epic Systems EHR instance. UVMMC and Network partner Porter Hospital have multiple community primary care clinics in Chittenden and Addison counties, VT.

Traditional models of genetic disease detection and prevention are practiced, including mandated newborn screening, variable documentation of family health history, genetic specialist evaluation, genetic counseling, and genetic testing of individuals and families at risk or manifesting genetic conditions. No DNA-based primary screening of people without risk factors occurs. Individuals at higher risk of genetic predisposition due to a diagnosis of colon or endometrial cancer are screened for Lynch Syndrome using immunohistochemistry. Individuals with a family history suggesting predisposition to cancer may be referred to the Familial Cancer Program of genetic oncologists and genetic counselors.

An on-site Genomic Medicine Laboratory, directed by molecular pathologists, two Ph.D. molecular biologists, and a clinical and laboratory geneticist, performs NGS sequencing of tumor DNA and RNA for precision oncology therapy. All clinical germline testing is sent to referral laboratories.

UVMMC has a robust Patient and Family Advisors (PFA) program (Celenza et al., 2017; Wahlberg et al., 2021). PFAs are volunteers invited to provide patient- and family-centered perspectives to implementation teams during project planning.

UVMMC health information technology (HIT) resources are extensive yet principally focused on business operations and dissemination of Epic Systems products throughout the health system.

The pilot was envisioned, designed and supported by the Chair of the Department of Pathology and Laboratory Medicine (DGBL), a molecular pathologist who founded the Genomic Medicine Program. The Genomic Population Screening Program implementation was led by an ABMG Clinical Geneticist (RSW) with both laboratory and patient care expertise. Both (DGBL and RSW) have been involved in national efforts to promote realization of the genomic medicine potential in health care. The geneticist has broad experience in genetic and genomic medicine, including solo genetics practice in rural and suburban areas, academic and non-academic settings, workforce training, education, policy, clinical molecular genetics, and rare disease research. The Chair of Family Medicine (TCP) and the Family Medicine champion provider (AWR) both had professional experiences arising from the Illumina “Understanding Your Genome” (UYG) program performed locally in 2017 that informed their participation and commitment (Briggs, 2016; Masterson, 2016). Both Chairs are leaders at the UVMMC and the UVM Health Network with access to health system leaders.

An experienced clinical and laboratory certified genetic counselor (CAG) helped plan and execute the pilot, she provided the genetic counselor’s perspective and performed genetic counseling. A second genetic counselor provided temporary, part-time support.

Three retired non-medical professionals from the community who volunteer as PFAs provide input during both planning and execution phases. Ten PFA volunteers contributed as a group to develop a new written clinical consent form, as well as a brief animated video providing a patient-oriented overview of the program.

Participating PCPs largely belong to two multi-site family medicine practice groups of the UVM Health Network Medical Group. Most were recruited informally by the PCP champion and other participating PCPs, while a few were approached by knowledgeable patients. Most are physicians, but nurse practitioners and physician assistants also participate. None received participation incentives.

Patients offered the test must meet these eligibility criteria: at least 18 years old, they and their partner are not pregnant, their PCP participates in the pilot and received program training, and the patient is attributed to Vermont’s ACO. There are no restrictions based on health status, family history, or other health risk factors.

Testing began 1 November 2019, in one Family Medicine practice with one PCP champion. Additional PCPs and practices joined as roll-out issues were resolved and as clinic workloads permitted. By March 2020, four additional PCPs and one additional clinic site were offering testing. Nearly all patients were tested by a Family Medicine PCP. The remainder were Internal Medicine patients who heard about and requested the test. Since patients are offered testing in the context of primary care visits, they reflect the demographics and health status of individuals frequenting primary care offices.

Two years after testing began, twenty different PCPs had ordered at least one test. One quarter of the providers ordered three quarters of the tests. 186 patients between 18 and 92 years old had been tested. Median age was 58. Thirteen percent of tests had no reportable variants. The rest reported one or more dominant or recessive likely pathogenic or pathogenic variants.

Figure 2 shows monthly tests and the sources and timing of unanticipated inner setting demands on primary care and Genomic Medicine. Operational disruptions from the COVID-19 pandemic interrupted testing for about 2 months. Staffing issues quieted hoped-for expansion of the perceived optional activity to more primary care providers. We built a public web page where patients engaged through telemedicine visits can view the animated educational video, and download test information, educational resource documents, and the consent form, including contact information for the GMRC (Wildin, 2020). A home saliva sampling kit option was also added.

In response to laboratory wide needs, an HIT systems architect was engaged. This experienced professional performs a critical adaptor function to the HIT operations and prevailing culture of our setting.

To assess early patients’ perceptions of the program’s implementation effectiveness and to focus quality improvement efforts, in June 2020 we mailed a two-page survey to the first 61 patients tested along with a postage-paid return envelope. After two reminder letters, 19 surveys were returned. One was blank and excluded from tabulation. The Supplementary Material shows 18 tabulated responses in the survey instrument format. Aside from logistical challenges like receiving printed results in the mail, which we worked to improve, the survey indicated general satisfaction or enthusiasm about the testing design and process, and for the value proposition. Of note, patients strongly endorsed that the PCP’s office is the right place to offer this testing.

Table 2 describes events captured by our quality surveillance processes and how we responded. Most resemble those occasionally encountered in health care and none affected patient health. While data about the rate of patients choosing to test when offered testing, and why patients declined testing, are potentially informative, their collection was not practical during this pilot implementation.

The key goals of this pilot implementation of clinical genomic population health screening of any-health-status adults were accomplished (Table 1). This demonstrates the feasibility of translating lessons from prior population sequencing and return of results research into clinical practice, which was the primary goal. Key differentiators of our implementation include placing primary care at the center, using a large, pre-defined clinically relevant target gene panel performed in a clinical laboratory, offering testing as part of usual preventive care at no cost, providing a written action plan with the test reports, and not being a research protocol.

The implementation we describe here leveraged all the opportunities and overcame most of the challenges cited for “non-traditional genetic testing” in the American College of Medical Genetics and Genomics’ “Points to Consider” analyses, including the important roles primary care providers contribute (Bean et al., 2021; Murray et al., 2021). Strengths included leadership engagement with tools like a personal genomics test that occurred years prior to beginning the pilot, getting formal buy in from medical center administration with a non-financial business plan, involving diverse stakeholders in the design and implementation process and making it worth their time, and leveraging existing workflows wherever possible.

We contracted for existing validated tests and primary reporting with a commercial laboratory. This allowed us to move forward sooner and with less expense than if we had to implement germline testing, variant interpretation, and reporting ourselves.

Indirect measures of success include that new PCPs continue volunteering, most PCPs involved have continued to offer the test, and patients continue to get tested. Notably, no participation incentives are provided to the PCPs. Recruiting new PCPs was actively limited due to unrelated staff shortages and suspended during the COVID-19 public health crisis, redemonstrating the susceptibility of new prevention-oriented programs to externally imposed prioritization.

Patient complaints are few, are related to process and communication, and are easily addressed. Unanticipated resource demands have not surfaced, and no critical element of the complex multi-disciplinary design has failed or had to be withdrawn. Our patient quality survey is a direct measure addressing some of the same data types as the survey by Orlando et al. (2018). The results are generally positive and support the assertion that the process is sufficiently patient-centric.

Barriers to scaling up are common in new interventions. We underestimated the need for leadership engagement in HIT and the relative priority for planned system-wide HIT transformation, where tension for change was far higher. HIT resources were unavailable to build consent, order, and resulting experiences familiar to the PCPs. The EHR-plus-paper order process we used instead burdens clinic staff and dissuaded some PCPs from participating. This adaptation is also the principal source of tracked process errors. EHR-based improvements will be prioritized once the system-wide Epic implementation is completed in April 2022. A separate, secure data system was built internally to track the multiple process steps. The solution allows oversight but is neither interfaced nor scalable. The criterion that tested patients are attributable in Vermont’s ACO was similarly challenging because ACO status is not reliably reflected in our EHR. It requires a manual inter-institutional lookup process.

The strong knowledge and experience of the principal implementers and of the primary care and other key partners, and the continued involvement of the PFAs, all contributed to resilience in the face of disruptive shifts in the setting that eluded anticipation, such as the COVID-19 pandemic and a UVMMC cyberattack.

The implementation of genomic population health screening in primary care at our institution benefited from elements in the outer setting, like the ACO, and in the inner setting, like engaged leaders who embrace innovation, champion providers, a highly collaborative team with broad expertise and capabilities, and availability of non-research funding of the pilot. While not unique, these advantages are not universal. Our guiding design principles may not be shared in every instance, and situations calling for pragmatism may also diverge. Presenting our implementation openly and in a recognized framework may help others identify their unique paths to success.

While not a goal of the pilot, we recognize that the patients’ clinical results combined with their personal and family health histories represent data types underlying a key phase of learning healthcare systems (LHS) (Schwartz et al., 2018; Williams et al., 2018). Having met our goal of demonstrating feasibility, we anticipate building a real-world LHS with related implementation, outcomes, return on investment, personal, educational, and health system research that can be combined or compared with similar data from other genomic population health screening programs.

We wish to increase testing for younger and healthy adults, who visit their PCP less often, by engaging them through EHR patient portal messages (Christensen et al., 2021) and by expanding testing to women’s health clinics. To accomplish enhanced risk assessment for genetic disease risks, family health history and genomic population health risk information should be co-analyzed (Wildin et al., 2021). This adds complexity but could propel family member (“cascade”) testing, an important added value for genomic population health screening.

Finally, since our pilot’s funding is finite, there is a need for both stable and scalable investment in this and similar programs that support the enhanced prevention focus of value-based care. We envision genomic population screening as a future benefit in value-based care payment contracts, supporting the preservation of a healthy state in both individuals and populations.