According to the Precision Medicine Initiative, precision medicine is “an approach for disease treatment and prevention that takes into account individual variability in genes, environment and lifestyles.”
Precision medicine and personalized medicine often are used interchangeably, but have slightly different connotations – with the former focused more on the clinical realm of genomics and the latter taking a more expansive view of social and behavioral health.
Both hold huge potential for better health outcomes – but also require complex and challenging technology deployments, changes to clinical workflow, and education for physicians and patients alike.
“It is important that the provider CIO help to lead their organization into this new world by considering how existing technologies can be optimized and how new, disruptive technologies can be anticipated over multiple years of capital budget investments,” said Dan Kinsella, managing director, healthcare and life science, at consulting giant Deloitte.
“Of paramount importance to the typical provider CIO is how to operationalize precision medicine at the point of care. There is not a one-size-fits-all solution for healthcare providers, but there are leading practices to consider whether you are an academic medical center, an integrated delivery network or a community hospital.”
In this special report, seven precision medicine technology experts – from Accenture, CereCore, Chilmark Research, Deloitte and Orion Health – offer healthcare provider organization CIOs and other health IT leaders best practices for optimizing this technology.
7-step implementation process
Some optimization techniques for precision medicine technologies can take place during system implementation. Implementing precision medicine technology is no different from any other IT implementation project, said Ian McCrae, CEO of Orion Health, a healthcare technology company delivering interoperability, population health and precision medicine systems.
Healthcare CIOs and other health IT leaders must get the basics of change management right by following seven steps, McCrae advised.
“First, know what problem you are trying to solve,” he said. “Have this clearly defined from the outset. Don’t make the mistake of trying to implement the tech if you haven’t identified what you will be using it for. Second, ensure the solution makes life easier and delivers a better outcome. If the project fails in either of these areas, then it will fail overall. If the precision medicine tech doesn’t make life easier for clinicians, or deliver a better outcome for patients, then why are you implementing it?”
Third, have clear roles and responsibilities, including data stewardship, governance and ethics, he suggested. The principles of data governance and stewardship are critical, and must not be overlooked if a project is to be successful, he said.
“What are your guidelines for governing the data you will extract?” he asked. “These guidelines should be clearly aligned with your organization’s strategic vision and values. Ethics of data use is another critical area: informed patient consent, the right to withdraw, confidentiality, objectivity … the list is long.”
A recipe for frustration
Fourth, CIOs need to connect the dots with precision medicine technologies, McCrae advised.
“Providing a better prediction without a means to act on it will be a recipe for frustration,” he said. “Once you have the technology to enable improved predictions, will you also have the resources to apply the learnings? If you can’t deliver a better outcome for patients, then it’s likely your project will fail. Fifth, remember accuracy isn’t necessarily the most important thing.
“We often compare solutions by how often they get the answer right, without understanding what people want to do with the answer,” he added. “Knowing that someone is 61.3% likely to get cancer versus 59.8% isn’t as important as how quickly you can know it, and what you can do when you find out.”
Sixth, stick to the plan and do not get distracted by failures along the way, he said.
“We find it hard to continue the development of something when the first stage isn’t as successful as we had hoped,” he noted. “If we are aiming to make precision medicine the ‘gold standard’ across different fields but the first application isn’t successful, that doesn’t mean you should throw out the goal.
“And seventh, start with specialties where the application is clear,” said McCrae. “Rather than aiming to implement the tech into a multitude of areas, select one or two specialties where the value of precision medicine is clear. Learn from those before expanding into new areas.”
Establish foundational infrastructure
Dr. Charles Bell, chief medical officer at CereCore, a health IT consulting firm, advised that getting the foundational infrastructure established before precision medicine can be applied via the EHR is one best practice for optimizing the use of the technology.
“Precision medicine relies on genomics – genomics, including pharmacogenomics, has created a vast amount of data, whereas the advent of the EHR has established an enormous data repository,” he said. “The success of advancing the technology is dependent on the genomic data residing in a repository that the EHR can readily provide access to. Therefore, there is a foundational infrastructure that must be established before precision medicine can be applied leveraging the EHR platforms.”
Genomic medicine is currently informing clinical care. Notable examples are in the treatment of some cancer types, cystic fibrosis and heart disease.
“The integration of the EHR, the data repository and the genomics medicine platform becomes essential to translate relevant and crucial data to drive precision medicine care,” Bell said. “A streamlined workflow must be established that allows clinicians to provide appropriate care from within the EHR using genomics and precision medicine.”
Enterprise solution architecture
Precision medicine requires capturing and analyzing complex data so that it is actionable at the point of care. Evolution of clinician workflow to support precision medicine use cases – even those that are relatively simple, such as pharmacogenomics – requires multidisciplinary change-management efforts and thoughtful systems integration, said Kinsella of Deloitte.
“Furthermore, the challenges of leveraging next-gen sequencing data in clinical decision support exceeds the capability of current EHR systems, except in certain use-cases such as pharmacogenomics,” said Kinsella’s colleague Connor O’Brien, manager at Deloitte Consulting.
“This requires external decision support analysis, which often is a manual process, such as the outputs of diagnostic review boards, although we are seeing many attempts at automation being applied, such as the decision-support platforms being deployed by GenomOncology, 2bPrecise, Syapse and others.”
When it comes to oncology and other service line roadmaps, health IT leaders should work with their service-line leaders to understand any gaps they have in the technology required to enable excellence in care delivery, Kinsella suggested.
“With oncology specifically, ensure that genomic requirements are understood as the capital investments may require multiple fiscal years,” he said. “Refine your technology roadmap for tumor boards as the future state is likely to include a variety of external contributors such as leading academic medical centers and drug and biotech companies.”
Social determinants of health
Then there are social determinants of health (SDoH). Precision requires understanding of variability in environment and lifestyle in addition to genetics. While most provider organizations are oriented to patients, expansion to the notion of “member” as an individual who may or may not have a medical record is required, Kinsella insisted.
“Value-based contracts with payers define specific cohorts (members) for whom the provider has assumed a level of accountability,” he explained. “Background and lifestyle questions not typically the focus of most EHR-centric workflows are crucial to the personalization of the care we deliver.”
With precision medicine come institutional alliance relationships, said Kinsella’s colleague Kate Liebelt, a manager with the Precision Medicine Community of Practice at Deloitte Consulting.
“In addition to having the logo on your website, what is the essence of your relationships with your external partners?” she asked. “Are you sending your data out to a registry without distilling the value of that information for care of your own patients? Increasingly, providers are licensing proprietary data to industry partners. For example, Cancer Commons is a not-for-profit network focused on connecting patients, physicians and providers to access cutting-edge personalized treatments beyond the traditional standard of care, through data sharing.”
Entities like the Texas Medical Center Accelerator harness innovation and talent from area healthcare organizations and generate start-up companies with regional, local and international reach, she added.
“Real-world evidence is driving innovation in value-based contracting and reimbursement strategies as demonstrated by the CMS Oncology Care Model – a new payment and delivery model designed to improve the effectiveness and efficiency of specialty care,” she explained. “Enablement of precision medicine helps AMCs continue to meet their tripartite mission of education, care delivery and research.”
And on a related note, interoperability. Sending and receiving data from across the evolving ecosystem requires that one be at the top of one's game regarding interoperability – and, importantly, cybersecurity and compliance – from FTTP, to HL7, to FHIR API and beyond, O’Brien said.
“Don’t leave out your CISO or legal and compliance teams,” he said. “Current architectures integrate insights from external clinical-decision-support systems, with the EHR serving as the transactional system of record: insights derived from external decision support FHIR API-based integrations that trigger EHR transactions such as pre-populated order sets, modifications to problem lists, and incorporation of CLIA test reports into clinical documentation modules in EHRs.”
Integrating genomic data across systems
Jody Ranck, senior analyst at Chilmark Research, a healthcare IT research and consulting firm, advised that integration of genomic data across different EHR systems and across different laboratory and precision medicine platforms is key and challenging for most organizations.
“Genetic test results tend to be large files that are difficult to integrate into an EHR,” he said. “Therefore, having a road map for your precision medicine approach is essential to think ahead several years and analyze which clinical areas will be impacted by the precision medicine program first. Oncology tends to be the most well-developed area, but in our COVID-19 moment, we may see the need for adjustments as significant caseloads of patients are those recovering from treatment with long-term challenges and new knowledge of the virus expands.”
The impact of the pandemic on precision medicine may have some long-term consequences for best practices.
“There will be a distributional shift of baseline health characteristics at the population level for the datasets that machine learning algorithms were trained on and new features to these populations that may interact with specific precision medicine initiatives,” Ranck said.
“The pandemic also has highlighted how poorly prepared the health IT infrastructure was for a public health crisis. Future federal funding, if funded wisely, will have significant funding to enhance precision public health initiatives, particularly those that bring social determinants into the picture. CIOs will face growing pressure to find effective ways to leverage and enhance SDoH efforts through more precise allocation knowledge and financial resources to address the sequelae of the pandemic.”
Integration standards for inpatient and outpatient
One best practice for optimizing precision medicine technology is to create integration standards that support treatment across ambulatory and inpatient settings, said Bell of CereCore.
“The large amount of data that has been generated in both the ambulatory and inpatient settings creates a challenge for integration of the information,” he said.
“Standards need to be established and refined to aid in the adoption of the technology that will support precision medicine. Clinical-decision-support capabilities must be integrated within the EHR. The evolution of the use of genomics to support precision medicine is dependent on collaborative development by multiple stakeholders.”
The list of requirements includes, but is not limited to, genomics specifications, clinical decision support, systems capable of handling genomic information, and resources to bridge the gaps between the data and its use clinically, he added.
“An example of the use of pharmacogenetics is that of Warfarin dosing,” he said. “For a decade now, recommendations for Warfarin dose requirements have been influenced by gene studies. Though there continue to be questions of the effect on specific genotypes in some patient populations, there still has been an improvement in treatment of identified patients with warfarin therapy. The result is that information is gained for a more effective treatment plan and a decreased risk of potentially harmful side effects.”
The more specific needs of varied patient populations can be addressed with further use of genetic data that is standardized across the patient’s settings, he added.
“Most EHRs offer a genomics solution to address providers’ workflow,” Bell noted. “An order is entered into the system and a pathway provides information to enhance clinical decision-making. It takes into account clinical decision support as well as alternatives if genomic results do not exist or are not accessed within the system. For all vendors, including Meditech, Cerner and Epic, storage and access to genomic repositories needs to be resolved.”
eMerge and ClinGen are examples of organizations, along with other resources and efforts, that are developing approaches to integrate genomic information into precise clinical care, he added.
Analytics strategy refresh
To enable precision medicine, leading provider organizations are refreshing their existing analytics strategies, and hardening core data-management capabilities, said Kinsella of Deloitte. Note that analytics includes descriptive (reports on what happened yesterday), predictive (what might happen in the future) and prescriptive (for example, precision medicine leading practices), he explained.
“Regarding reference architecture, use what you have, buy what you need and build what you must,” Kinsella said. “Explore the capabilities of your core enterprise applications including EHR, ERP and cost accounting, and adjust known levers – for example, clinical-decision-support capabilities, lab-management systems, and billing and coding management – to operationalize a precision medicine program. Focus on the tools you may require to ensure collection, curation, calculation and consumption of data to generate analytic insights.”
On a related front, there are edge technologies and big data. By leveraging open source and edge solutions, providers can augment legacy analytics and data management capacity, Deloitte’s O’Brien said.
“For example, providers increasingly are commissioning data lakes to collect and curate data from a variety of internal and external sources,” he noted. “The velocity of data, including streaming, enables monitoring (for example, sepsis data), disease management and population health surveillance (for example, SDoH), and remote patient-monitoring, tapping into the tsunami of data generated from wearables and IoT.”
The need for analysis provenance and traceability of results becomes amplified when dealing with molecular-level data, due to the dynamic nature of scientific discovery, he added.
“Genomic variants that are classified as ‘variants of unknown significance’ today can become clinically significant as scientific knowledge progresses,” he said. “These requirements will become even more critical as more dynamic types of omics data become clinically significant, such as being realized in the case of metabolomic and proteomic data. Put simply, today’s information exhaust may become tomorrow’s rocket fuel.”
Data governance and data excellence
In the continuous pursuit of data excellence, CIOs should collaborate with CMIOs, CNIOs and clinical informatics to ensure that key data elements are understood, configured to be captured by the enterprise applications, and, most important, align the workflow so that data is collected predictably, Kinsella said.
“Registries, often a standard feature of enterprise EHRs, represent untapped potential,” he noted. “Typical features include definition of inclusion rules and calculation instructions for specific cohorts of patients. When, for example, does a diabetic patient get tagged as a diabetic patient in the diabetes registry?”
Threaded throughout the emerging theme of precision medicine enablement is education around analytics: training in data science, and the application of descriptive, predictive and prescriptive analytics, he added. Increasingly, provider organizations are hiring in-house analytics experts and partnering with entities on their data strategies and capabilities, he said.
“Review your organization strategy and align your data sharing approach accordingly,” added Deloitte’s Liebelt. “Are you motivated by social good? Academic pursuit of new science? Are you open to earning revenue by sharing de-identified data by building bandwidth to drive robust real-world evidence programs and innovative industry partnerships?”
Patient registries and patient-reported outcomes-measurement are a significant means of value creation for provider organizations, particularly in the areas of oncology, rare and orphan disease, and chronic disease management, she said.
“Theoretically, providers can predict and validate a patient’s predisposition to diabetes and track and measure their progress on various treatment regiments through the systematic collection of patient data, for example, population-level data, lab results, patient-reported outcomes, etc.,” she explained.
“As providers continue to make their real-world data available in open, closed or hybrid networks, there is an emergence of innovative partnership opportunities with other provider organizations, pharmaceutical/biotechnology/medical device companies, health insurance companies, and publicly and privately funded research institutions.”
Reducing the burden for physicians
On another front, precision medicine is a significant mind-shift for both patients and providers, and the integration of genomic data, or more importantly, knowledge, is a significant challenge, said Ranck of Chilmark Research.
“The process of obtaining genetic information is not always as straightforward, and interpreting these results for a patient can be difficult,” he said. “Most diseases are not a one gene equals ‘X’ disease type of phenomenon.”
Physicians will need more time to digest precision medicine data and render this into actionable information for the patient, he said.
“In the context of standard clinical workflows, this is a challenge,” he observed. “However, there are platforms that can reduce the burden for physicians, but rigorous evaluation of these solutions and the underlying science needs to be done by physicians and scientists with sufficient knowledge of statistics, machine learning and genetics.
“Genetic counselors will be essential and may not be in adequate supply as precision medicine matures,” he added. “Precision medicine is not solely a technological issue and needs to be understood as socio-technical in nature.”
Precision medicine for cancer and drug prescribing
Dr. Kaveh Safavi, senior managing director at Accenture Health, offers two best practices when trying to optimize precision medicine technology.
“Good clinical practice today needs therapy to be tailored to the genetics of the tumor and the patient’s immune system for many types of cancer,” he explained.
“From a CIO perspective, precision medicine achievements mean building a new environment for data acquisition, analysis and decision support in near real time. Oncology decision-support platforms will require managing genetic information of the patient, the patient’s tumor and other phenotypic data that may not be part of the typical electronic health record.”
Since much of oncology care is provided in an ambulatory setting, it also will require seamless data sharing across care settings that may cross boundaries of a clinical enterprise but be essential to treating a patient’s condition in the most appropriate way possible, Safavi said.
“And on another note, there is a growing body of knowledge that combines pharmacology and genomics to develop effective and safe medications and doses tailored to a patient’s genetic makeup,” he said. “A delicate part of a CIO’s responsibility is selecting and investing in an informatics strategy to support this highly dynamic aspect of clinical care.”
An informed drug-prescribing platform requires the ability to gather biological information found in genomes, microbiomes, proteomes, metabolomes, phenotypes and endotypes, he concluded, and applying them to drug-prescribing decision-support platforms used by prescribers should take into account looking for technology architectures with the greatest flexibility to predictably handle large data volumes and data types.