Abstract
Electronic health records (EHRs) – systems that control all hospital data – have been around since the early 1970s. One of the first, the Health Evaluation thru Logical Processing (HELP) system, was developed by medical informatics pioneer Homer Warner at LDS Hospital, now Intermountain Healthcare, in Salt Lake City, Utah.
GrafVishenka / Getty Images
Warner's system used medical logic to evaluate patient data and provide clinical decision support (CDS). With time it was refined to include more information, such as clinical laboratory and pharmacy data and cardiology and pulmonary function test results, and it is still in use today.
EHRs were initially developed and used by academic institutions, but in the 1990s, when hardware became more affordable and the internet led to the use of web-based systems, commercial EHR vendors gradually took over. Although these commercial systems now have huge capabilities, they currently struggle to support the increasing use of genomic data in medicine.
Identifying the problems with genomic data
The issue is widely recognized, so much so that the Association for Molecular Pathology (AMP) set up the EHR Interoperability for Clinical Genomics Data Working Group to assess the challenges and look for solutions to the current problems associated with incorporating genomic data into EHRs.
The working group's co-chair Somak Roy, associate professor and director of molecular and genomic pathology services at Cincinnati Children's Hospital Medical Center, said they are also working with molecular professionals, laboratory directors, clinicians, genetic counselors, clinical informaticists, EHR vendors, application developers, and other professional societies, to identify and develop solutions for addressing the gaps in EHR interoperability of genomic data.
Somak Roy, Associate Professor, Cincinnati Children's Hospital
The group hopes to develop standards for clinical genomic data interoperability in the EHR that can be broadly adopted by healthcare systems and have recently published a list of 13 key challenges in this area in the Journal of Molecular Diagnostics1. “The most important and fundamental challenge that current EHR systems pose is the inability to appropriately receive, store, manage, and display discrete genomic data elements,” said Roy. “Storage of discrete clinical genomic data elements in EHR allows the implementation of additional solutions to enhance the use and visualization of this data.”
More specifically, the Working Group found that EHRs are not ready to send accurate, coded, clinical and family history data to laboratories and often lack sufficient information to allow test ordering for specific variants. Once produced, the genomic reports vary in structure and content between laboratories, there are no standardized genomic variant data structures between the laboratory information system and the EHR, and standards created to improve interoperability, such as HL7 version 2, use syntax with limited hierarchy and inadequate coding systems.
There are also no set standards for displaying variant data collected over time and between sample sources, tests, and laboratories that keep clinical context and associated interpretation intact, no consensus on how to request and reclassify variants, and no national or international standard for CDS rules. In addition, the Working Group says that genomic reports are difficult for medical professionals and patients to understand. This is particularly important given that in 2020 the U.S. Department of Health and Human Services (HHS) ruled2 that patients should have the right to access their medical records, including genomic test reports.
“This HHS ruling has important implications on how genomic data will be presented and viewed by the patients,” said Roy. He explained that EHRs have been traditionally designed for use by medical professionals, but it is now important that clinical laboratories are mindful of how the content of their reports may be interpreted by patients.
“Most genetic test results are complex and require review by specialized healthcare professionals for a patient to understand the meaning of the test results comprehensively and accurately, Roy remarked. He added that “continued work is required by clinical laboratories and providers to fully understand the implications and address any unexpected issues that may arise in the future.”
Finally, the Working Group reports that “current coding and interoperability standards are not adequate for genomic data,” and that technology lacks the functionality needed to make sure that any genomic data released for research studies or other reasons receives informed consent from the patient and/or institutional review board.
A number of organizations and initiatives, such as DIGITizE, eMERGE, and ClinGen, are also working toward defining standards in the transfer and display of genomic data and tackling additional problems such as how to store such large volumes of data. Yet, in most cases genomic test results are still displayed in text or PDF files that make it difficult to cross-check variants between different reports for the same patient and receive alerts for therapy that may be contraindicated on the basis of the genetic results.
Moving away from PDF reports
“If it's in a PDF, it's not computable. You can't do anything with it, but if it's in there, as structured data, you can say, you know, based on X variant, we want to make these recommendations,” said Nephi Walton, clinical geneticist and associate medical director of Intermountain Precision Genomics in Utah.
Nephi Walton, Associate Medical Director, Intermountain Precision Genomics
Walton previously implemented CDS and the use of structured genomic data at Geisinger Health in Pennsylvania and now oversees the HerediGene Population Study at Intermountain, along with director of genetic counseling Brent Hafen. He therefore has a greater interest than many in the need for EHRs to fully support genomic data; the HerediGene study has already sequenced around 100,000 people and hopes to ultimately increase number to 500,000.
For the study, the researchers will be looking at pharmacogenomic data, as well as mendelian disease risk genes, CDC Tier 1 genes (those involved in hereditary breast and ovarian cancer, Lynch syndrome, and familial hypercholesterolemia), and the American College of Medical Genetics' reportable as well as their own set of genes.
To manage all the data they are generating, Walton and colleagues have built their own IT framework because the Cerner EHR used at Intermountain does not yet support structured genomic information.
“I've been talking to them and they're working on it,” Walton remarked.
The recent $28.3 billion acquisition of Cerner by the software company Oracle may help to boost progress. In a statement on the Oracle Cerner website the company says it is committed to open APIs [application programming interfaces] to ensure any authorized user can consume health data and insights.” Another priority will be the development of develop cloud-enabled capabilities that allow providers to access information when and where they need it.
Although the there is no mention of a strategy for integrating genomic data into their systems, one would hope that this will also be a post-investment priority.
Data standards
Walton agrees with Roy that one of the biggest challenges to integrating genomic information into the EHR is the standardization of data. “If you get reports from seven different labs, they're probably going to have seven different ways that they describe it, which isn't a problem if you're a health care provider that's reading it, but if you are a machine that's trying to say I want to create decision support based on this then that is a problem,” he said.
Recently, HL7 developed Fast Healthcare Interoperability Resources (FHIR)3, a next-generation standards framework designed to modernize data transfer and thus improve interoperability.
However, as the EHR Interoperability for Clinical Genomics Data Working Group points out, the only U.S. federal requirement to use FHIR is for communication between EHRs and third-party applications that access EHR systems using APIs. For communications between laboratories and EHRs, the law mandates that HL7 version 2.5.1 is used. Unfortunately, two interfaces are not compatible, which means significant investment will be needed by vendors and laboratories to switch to FHIR for interfaces between laboratory instruments, laboratory information systems, and EHRs.
Speaking of investment, Walton says this is another obstacle faced by multiple institutions that is preventing precision medicine from moving ahead. “Healthcare systems are not really investing in IT needed to deploy it,” he remarked.
Genetic sequencing is becoming cheaper and Walton believes that ultimately everybody in a healthcare system will be sequenced. Therefore, “one of the biggest things that will be in the EHR for the patient's is their genomic data, but I don't see CFOs [chief financial officers] and hospital leaders realizing this and understanding the importance of IT,” he said.
Even with investment and the ability to adequately handle the genomic data, the EHRs need to be structured in a way that avoids overwhelming physicians who may not be overly familiar with the information they are being given. Walton said it is important to provide concise information and CDS to primary care providers so that they can manage the patients in the short time they are given for an appointment.
What are EHR vendors doing to address the problems?
Epic is one of the largest EHR vendors in the United States. They also have a worldwide presence, storing the medical records of more than 250 million people across 15 countries in their systems.
Headquarters of the electronic health systems vendor Epic in Verona, Wisconsin.
They have taken a number of steps to allow the integration of genomic data into their software.
“Epic software integrates ordering and resulting genetic tests into provider workflows, and Epic's framework can display genomic results, translate them for a variety of users, and make them actionable at the point of care,” said Catherine Procknow, a software developer on the genomics team at the company.
Catherine Procknow, Genomics Product Lead & Software Developer, Epic
She added: “Because genetic testing results are more complicated than typical lab results, we've built specialized data structures for storage and display. Storing these results - including genomic variants - in a discrete and accessible format is an essential first step in enabling their use throughout the system. Clinical decision support tools and content can translate those discrete results into genomically informed recommendations, screening plans, warnings, and education, which can be made available to different users based on their needs.”
The Epic CDS feature can be tailored to each company's needs and has the ability to include information from genomic knowledgebases such as ClinVar, the Clinical Pharmacogenetics Implementation Consortium, and the Dutch Pharmacogenetics Working Group (DPWG), as well as experts at the organization, and other sources of genomic knowledge.
Procknow noted that the company are also working closely with key stakeholders such as the HL7 Clinical Genomics Working Group and Global Alliance for Genomics and Health, genetic testing labs, and their customers to refine industry standards and develop more user-friendly formats for genetic data.
EPIC have not yet adopted the FHIR framework but they are aiming to future-proof their systems by storing sequence data in the Cloud. “In the future, we'll connect this sequence data to Epic's clinical and research tools to help users make new discoveries and inform care with up-to-date knowledge,” Procknow said.
In addition, the company is looking at ways to expand the use of genetic risk factors to inform preventative care and screening decisions and to improve access to genetic testing for all patients who are likely to benefit.
Future-proofing is also something that Intermountain have incorporated into their EHRs. Walton explained that they have a genomic surveillance system that allows them to continually reassess variants, and when they are actionable, pull them to the front and put them into clinical action.
Another major player in the EHR field is MEDITECH. They have 2,250 customers in 23 countries worldwide and have recently introduced MEDITECH Expanse Genomics, which integrates discrete genomic data directly into the patient's chart and clinician's workflows. This “enables physicians to order tests, identify genetic markers, and access clinical decision support within the workflow of their EHR.” explained Jen Ford, product manager, genomics and laboratoryi nformation systems at MEDITECH. The system is designed with standard interfaces that reduce the amount of time spent ordering and interpreting genomic tests, which was one of the challenges previously identified by MEDITECH users. “Prior to Expanse Genomics, all ordering was miscellaneous, not integrated, and could take up to 25 minutes per order on a genomics reference laboratory website due to lack of integration. Each reference lab used its own proprietary interface. Additionally, genomic data was scanned in PDFs and difficult to locate in a patient's chart,” said Ford.
Jen Ford, Product Manager, Genomics and Laboratory Information Systems, MEDITECH
She also noted that MEDITECH are now working on how they can globally integrate genomic data so a clinician can access it regardless of where it is stored.
Process chart for an electronic health record system that incorporates MEDITECH's Expanse Genomics software.
“For example, if a medication order is placed in MEDITECH and the patient does not have any pharmacogenomic results in the system, data can be retrieved from other databases for use such as large consumer facing laboratories that perform testing. In addition, if a patient receives genetic testing in one EHR and receives treatment once moved, the data can be communicated to the new EHR,” Ford remarked.
The future of EHRs
In terms of the future of EHRs, Walton thinks the best way to manage genomic data will be to use a large system similar to the Picture Archiving and Communication Systems that are being successfully used to improve the access, storage, and transfer of digital images. These systems are outside of the EHR and send the information to the EHR when it is needed.
“I don't see electronic health records really doing really well with management in such a dynamic field, with such dynamic data. I think ultimately it needs to be a separate system,” he said, adding that that is why Intermountain have already built their separate framework.
However, Roy thinks that through “active engagement of the EHR vendors and relevant stakeholders from clinical and laboratory practices, it is expected EHRs will be able to accept and store granular genomic data and effectively visualize and use it for functionalities such as clinical decision support in the foreseeable future.”