How mCODE is Driving EHR Interoperability for Cancer Research

Historically, restricted EHR interoperability has hindered the cancer research community's ability to harness big data.

With siloed data across the healthcare continuum, cancer research has centered on clinical trials. However, these studies only include 6.3 percent of cancer patients nationwide, leaving vast amounts of treatment data unstudied across health systems.

"We're not learning everything we can from the patients that are being treated, and if you ask my colleagues why, it's because the data are hard to get to," Travis Osterman, DO, MS, FAMIA, FASCO, associate vice president for research informatics at Vanderbilt University Medical Center (VUMC) and director of cancer clinical informatics in the Vanderbilt-Ingram Cancer Center, told EHRIntelligence in an interview.

"There are many data standards out there, and they all seem insufficient to actually answer meaningful questions in oncology," said Osterman, who holds dual faculty appointments at VUMC in the Department of Biomedical Informatics and the Department of Medicine in the Division of Hematology and Oncology.

However, a data standards project led by the American Society of Clinical Oncology (ASCO) is looking to change that.

The Minimal Common Oncology Data Elements (mCODE) initiative aims to facilitate the interoperability of research-quality cancer data.

ASCO convened a workgroup of oncologists, informaticians, experts in terminologies and standards, and researchers, including Osterman, to create the standard in 2018.

"One of the biggest goals for mCODE is if you're being treated, we should learn everything we can from how your treatment went so that everyone that comes after you can benefit," said Osterman.

He acknowledged that many members of the health IT community were hesitant at the mention of a new data standard. However, mCODE filled a gap in the data exchange ecosystem.

"I think anyone in the standards community will always look at a proposal for a new data standard and say, 'Why do we need another data standard?'" Osterman noted. "I think many of us were skeptical that this would be a good project to move forward with, but I think we all started to really see the vision as we got deeper into the work."

He explained that while most data standards go deep and with a narrow view, the mCODE workgroup sought to create a standard that was broad but not necessarily deep. The goal was to create the smallest dataset that describes a cancer patient's journey to help researchers answer meaningful questions in oncology.

The specification includes six high-level domains (disease, genomics, outcome, treatment, patient, and assessment), each with an accompanying set of data elements.

The workgroup leveraged existing data standards for the specification, such as HL7 Fast Healthcare Interoperability Resources (FHIR), to encourage adoption.

"We didn't want to reinvent the wheel," Osterman emphasized. "FHIR provides the abstraction, and then mCODE provides the data elements that get bundled together."

After first announcing the standard at the ASCO 2019 annual meeting, the mCODE team has worked with HL7 to drive adoption. Currently, the standard has over 70 implementations, including large EHR vendors such as Epic.

"The project is only five years old, and for a data standards project to gain such widespread adoption, we're incredibly pleased with the success," said Osterman, who serves as chair of the mCODE executive committee.

One of mCODE's main use cases is supporting the exchange of electronic consent forms and case report forms for clinical trials.

"mCODE provides a standard language that's an abstraction above the EHR, so it doesn't matter whether you're on Epic or Cerner or AllScripts; mCODE can live on top of that and then can connect to the clinical trial sponsors," he said.

mCODE also helps optimize data exchange in the radiation oncology space. Radiation oncology providers typically leverage systems from smaller EHR vendors, which has posed challenges for interoperability with other providers, Osterman noted.

"Upwards of 90 percent of radiation oncology health record systems support mCODE natively, which is incredible and just makes it much easier than for us to pass data back and forth between health record systems," he said.

Osterman said that the mCODE closed-loop feedback ecosystem has been a big contributor to the project's success.

For instance, CodeX is a member-driven HL7 FHIR accelerator that focuses largely on supporting mCODE implementers. CodeX members work with the mCODE executive committee's technical review group to give feedback on the standard.

"Almost all the time, implementers will find gaps where mCODE doesn't quite get them all the way to solve their problems," Osterman said.

For instance, mCODE implementers brought up gaps in terms of radiation oncology data elements.

The executive committee then worked with the technical review group and HL7 to update the standard.

"Radiation oncologists and the physicists wanted even more granularity than what we included, so they have a supplemental implementation guide that's specific then to their subdomain," Osterman explained.

"I think we will see this model more and more for groups using mCODE," he said. "mCODE will provide 90 percent of the data, and then for the 10 percent of data elements that we don't cover, you may need to implement a specialty implementation guide."

An additional mCODE implementation includes CC Direct, a White House initiative focused on pediatric malignancies.

"We have not had a project in the pediatric malignancy space, so we are working closely with those implementers across governmental agencies to understand the spots where mCODE might need to be adjusted," Osterman said. "Then we'll make those considerations and potentially make updates to future versions of mCODE to reflect that feedback."

Ultimately, enhanced EHR interoperability through mCODE adoption is set to advance cancer care, allowing investigators to access data more easily for clinical research.