CHICAGO (GenomeWeb) – Count Cerner among those electronic health records vendors that want clinicians to see genomic data as just another part of a patient's overall picture.
"We believe that genomic data and the genomic aspects of precision medicine should be incorporated into the routine clinical workflow in as seamless and transparent a fashion as possible," said David McCallie Jr., Cerner's senior vice president of medical informatics.
"It's not like a brand-new thing that requires a whole new system. It should just fit right in," McCallie added.
North Kansas City, Missouri-based Cerner and rival Epic Systems together account for half of the US hospital market for EHRs, according to the Department of Health and Human Services.
To handle genomes, Cerner has had to "extend" its core Millennium-branded EHR with some new data fields, according to McCallie. However, the EHR stores just genetic test summaries rather than raw genomic information, just like other laboratory tests.
"We think that the essence or the summary of the genetic data and genomic data should be in the EHR," in the form of a human-readable document, he said. Clinicians should be able to access deeper information in outside databases with a click from within the EHR, but Cerner does not plan on storing whole-exome or whole-genome sequences on its servers.
"I think in the early days, people got all excited and said, 'Oh my goodness, there's terabytes of genetic data. Where are they going to put it?' It actually turns out that labs know how to do that," McCallie said. "They can handle that."
In other words, genomics is just like medical imaging, from the perspective of Cerner. "Nobody stores the raw data of an MRI scan in the EHR. They store the report and then they store a link out to a viewer, and a clinician who needs to see the actual image can launch the viewer and browse it," McCallie said. "We think genomic data will be delivered in a very similar fashion."
Instead, McCallie envisions a future based on SMART on FHIR, a Harvard-developed framework officially called Substitutable Medical Apps, Reusable Technology (SMART) that follows the Health Level 7 International specification known as Fast Healthcare Interoperability Resources, or FHIR.
Though still not fully developed, SMART on FHIR provides a means for third parties to build apps into core healthcare information systems.
"If you have a patient who had complete exome sequencing or even whole-genome sequencing, you're not going to pull all that data into the EHR, but you might pull a link to a SMART app into the EHR so that if the clinician needs to dive deep into the results, he can launch a genome browser and do it right there from within the EHR," McCallie said. Though the data is managed by the testing lab, it would feel to the end user like the results are right there in the EHR.
"We think that the EHR of the future will have lots of little apps or components that deliver some of the really deep and complicated clinical knowledge that the EHR vendor [itself] is not going to try to manage," McCallie continued. "We're not going to build a genome browser, but we would be happy to put a SMART app into the workflow that makes it easy for someone else's genome browser to be accessible to the clinician" through a browser-agnostic interface.
McCallie said that Cerner is "very committed to the SMART app approach," both within and outside genomics. "We have several dozen deployed commercial SMART apps in our client base already and we expect to see a number of those apps focused on genomics."
He reported that one outside SMART app, a version of YouScript personalized prescribing software, is now going through Cerner's validation process for pharmacogenomics. "It pulls in the patient's medication profile and gives the clinician a detailed analysis of the potential risks, not only looking at genomic variants that might be relevant, but also [at] polypharmacy, when you have a patient who's taking several drugs that all share the same metabolic pathway," McCallie said.
YouScript has also been integrated with Epic, though not via the SMART framework.
"YouScript, they've got a very nice package that does a deep analysis that's more than just simple pharmacogenomics, and we'll make that available as a SMART app to our clients who want to contract with YouScript to deploy that," McCallie said. "I think there will be others like that."
Indeed, Cerner's genomics journey initially is focusing on integrating pharmacogenomics into the clinical workflow. PGx is also where fellow EHR vendor Meditech started when it comes to genomics.
PGx is "both low-hanging fruit and very valuable fruit" that is relevant to a large number of patients, McCallie said.
Eventually, Cerner would like the PGx and other genomics reports to arrive from labs in machine-readable format to inform clinical decision support systems, which is what standards such as FHIR can enable. However, there does not appear to be much customer demand for that to happen quickly.
"Everyone asks us about our precision medicine strategy. They all want to know where it's headed, but I think it's a relatively small number who have pushed hard to get, for example, the pharmacogenomic integration to move forward," McCallie said. "The uptake has been a little bit slower than I might have predicted."
Much of that is due to the clinical utility of PDF lab reports. "If a physician orders a pharmacogenomic screening panel and they get back a report that they can read, that's 95 percent of the problem," McCallie said. We can certainly make it more automatic, but it's not like they can't [use] it today. They can do it just fine. It's just not as automatic."
What Cerner would like to do soon is add PGx alerts. While clinicians are quite comfortable working with human-readable documents like PDFs or even unstructured text, it's easy for a doctor to forget about a potential pharmacogenomics issue at the point of care, McCallie noted.
To prepare for the days when machine-readable reports come in, Cerner is building PGx rules from the Clinical Pharmacogenomics Implementation Consortium into its Millennium EHR. This, McCallie said, will help clinicians quickly identify the "highest-value combinations" of drug and genetic variants.
The vendor also is in the process of adding a genomics component to its "model system," a reference build for clients to use when upgrading their EHRs that includes many of the most common interfaces and standardizations.
"We pre-build a lot of the complexity into the reference build so that they don't have to do it themselves. As you can imagine with genomics, there's literally thousands of variants that you have to have very precise naming conventions for or it won't work right," McCallie said.
This work started in pediatrics wards and hospitals — notably, St. Jude's Children's Research Hospital in Memphis, Tennessee — which tend to be among the most advanced Cerner users when it comes to adopting clinical pharmacogenomics. "We took their input and are putting this model system together, feeding it back for them to test it and try it and see if they agree," McCallie said.
"When we started asking around about phamacogenomics, we had several pediatric clients that were doing advanced work … so we could pull it together and coordinate it," McCallie explained.
In time, Cerner expects to add in new HL7 standards, including the FHIR Genomics offshoot, to help move structured data around. "That's not quite ready for prime time yet," according to McCallie.
"Today, [genomic reports] come back in a variety of incompatible and nonstandard formats that often, I'm sorry to say, results in some custom coding having to be written to pull the structure out of the report," McCallie said. "I really hope that all switches to FHIR in the future."