Drug company researchers would prefer to get high-quality data straight from the EHR in real time, rather than from claims data or dumps of EHR data that they have to mine, said Rebecca Kush, CEO of CDISC, a not-for-profit that promotes consensus around the use of data standards by research organizations.
During a July 7 webinar, Ron Fitzmartin, senior adviser in the office of strategic programs at the FDA's Center for Drug Evaluation and Research, practically begged healthcare providers, researchers and technology vendors to help the FDA end swivel-chair interoperability. Within five years, the FDA wants to close the so-called air gap between EHRs and EDC systems.
“We know that EDC systems are fantastic, but there's always this intervening paper document where source data (from EHRs) are collected” and clinical trial data are written down on paper and entered manually into the EDC systems, Fitzmartin said. “We want to promote and encourage the elimination of that duplication of data.”
Fitzmartin co-hosted the FDA webinar to explain the agency's formal request, published in June in the Federal Register, for volunteers to partner with it in developing pilot projects to connect EHRs and EDC systems.
The interoperability idea seems simple enough: Pluck pertinent patient data from an EHR for a clinical trial and send it to an EDC system. But the devil has been in the details—for decades.
EDC systems and EHRs can collect similar types of information, but the data are often asked for and stored in different ways. Patient-identifying information is common to both types of systems and might transfer smoothly. But data about smoking cessation, for instance, may not, and that can create a disconnect for a clinical trial.
That's because clinicians and researchers elicit information from patients in different ways, said Dr. Charles Jaffe, CEO of Health Level Seven International, a standards development organization. “ 'Do you smoke? No. Did you smoke?' That's the way that doctors and patients talk. The FDA calls that dirty data. The question the FDA asks is, 'Have you smoked in the last five years?' For a trial, it's either yes or no, zero or one.”
In addition, case report forms and EDC systems for clinical trials collect patient data specific to the trial, such as absorption of a drug or how long it remains in the body. But such data are not normally collected by a physician and stored in the EHR. Only about half the data elements needed for the trial may be captured by the EHR, according to drug researchers.
The industry's failure to link the two types of systems hasn't been for lack of trying, CDISC's Kush said. “I have slides showing our vision of trying to do this since 1997, the year we started CDISC.”
And the failure hasn't been for a lack of standards or implementation specifications.
CDISC-supported standards are widely used by EDC system vendors. Health Level Seven has its own boatload of standards used by EHR developers. A third not-for-profit organization promoting interoperability, Integrating the Healthcare Enterprise, also has worked for years on an EHR/EDC system connection and standards implementation.
The rub has been harmonizing the standards that already exist and obtaining buy-in from all industry participants and the government. “If you have a group of people who believe they have the right approach, they say, 'If everybody would just use my system, everything would be perfect,' ” said veteran medical informaticist Dr. Doug Fridsma. “The reality is, it's not going to be one size fits all.”
In the 2000s, the American National Standards Institute, under HHS contract, convened the Healthcare Information Technology Standards Panel, largely to seek consensus on the use of existing healthcare data standards to solve specific interoperability problems. Among its final acts before being disbanded in 2010, the panel released a clinical research interoperability specification to link EHRs and EDC systems. The effort was orphaned when the federal health IT coordinator's office shifted focus to the EHR incentive payment program.
To say EDC systems and EHRs have never talked to each other would be an exaggeration. A successful pilot called ASTER linked an EHR to an electronic report-writing system, which is like an EDC system. It plucked data from the EHR at Brigham and Women's Hospital and helped physicians send electronic adverse drug reaction reports to the FDA. But ASTER technology was used for only a few months in the pilot project, said Michael Ibara, a pharmacist who helped organize the trial.
Glen de Vries, co-founder and president of Medidata Solutions, an EDC system software vendor, said his firm's system interfaces with the homegrown EHR of a research clinic that specializes in running small Phase I drug trials.
The federal health IT coordinator's office is looking at a different approach—Structured Data Capture, which tags and grabs data elements from an EHR and shares them with other IT systems, databases and mobile devices. It's leaning toward an HL7 initiative called Fast Healthcare Interoperability Resources, or FHIR, which is based on Internet-like interoperability principles. FHIR is seen by some major EHR vendors as having good potential for extracting discrete data elements from EHRs and sharing them with other IT systems and devices.
Several leading EHR vendors, including Athenahealth, Cerner Corp., Epic Systems Corp. and McKesson Corp., have formed a group called the Argonaut Project to explore FHIR's potential, including for EHR/EDC system interoperability in drug trials.
Cerner has participated in many attempts to standardize data flow from EHRs to EDC systems, said Dr. David McCallie, the company's vice president of medical informatics. “However, we have found those standards to be inadequate and are now focused on the development of newer approaches that leverage HL7's emerging FHIR standard. While these new approaches are still developing, Cerner intends to submit a proposed project to the FDA based on the FHIR standard.”
Jaffe, HL7's CEO, said FHIR could shorten the EHR/EDC system interoperability timeline to two years.
But other experts foresee a much longer road to EHR/EDC system interoperability, and some wonder whether the effort will ever produce commensurate results. “There are just unrealistic expectations about what people are gathering in regular clinical care and the level of specificity and very narrow definitions that are used in standard research protocols,” said Dr. Michael Kahn, co-director of the Colorado Clinical and Translational Sciences Institute at the University of Colorado.
Kahn's research found that a standard ambulatory EHR was capable of providing only between 30% and 50% of the data needed by researchers. And the average trial now requires even more data that cannot be obtained from a standard EHR. “We really need to re-think the way we're doing clinical trials,” he said.
But Medidata's de Vries sees the interoperability gap between EDC systems and EHRs closing in five years in the U.S. He predicts the FDA will modify trial designs to accommodate more rapid data-sharing that computer technology enables. “It doesn't even have to be 100% of the data,” he said. “Even if it's 20%, it's going to reduce the friction.”
Meanwhile, clinical research coordinators in their swivel chairs will continue to fill the gap. Bain said that's an embarrassment to healthcare organizations running trials. His group did an analysis for the Centers for Disease Control and Prevention and asked officials there how to get data into their survey forms. “Guess what? It's done with swivel chair,” he said.