What does cancer have in common with opioid addiction, heart disease, and diabetes? Unfortunately, numbers—very large numbers.

According to the National Cancer Institute, approximately 38.5% of men and women will be diagnosed with cancer at some point during their lifetime. In 2014, an estimated 14.7M people were living with cancer in the United States.

While numbers for cancer and other major diseases are staggering in terms of human suffering, such large figures also provide potential solutions. Here’s why: odds are that most of those 14.7M people have EHRs of some kind. And while each patient’s record contains a wealth of information about symptoms, treatments, and results that are important to him or her personally, that data also holds potentially valuable information that could help researchers identify better treatments for other people, as well.

Essentially, each patient’s real-world record may provide clues to help solve real-world problems for many.

There are several barriers, however, to achieving the promise of using vast quantities of real-world data to achieve better patient outcomes. One is incompatible data across the many EHR systems. There are more than 1,500 different electronic medical and health record IT system products in use, many of which have been configured to their users’ needs. This increases the complexity and incompatibility across different systems.

Another barrier is that clinicians capture too much information as free text in ways unique to the clinician and thus difficult to collect across systems.

Both barriers are part of the fundamental problem: today’s health IT systems contain semantically incompatible information. Because of the great variety of the data models of EHRs, transferring information from one health IT system to another frequently results in the distortion or loss of information, blocking of critical details, or introduction of erroneous data. This may explain why, even after health IT has been almost universally adopted, clinicians still routinely share information using old-school methods—such as fax machines.

With only about 3 percent of cancer patients participating in clinical trials, researchers are limited to a small pool of patient data to evaluate the safety and efficacy of cancer treatments. Yet, clinical trials are currently the gold standard for gathering the complete and precise data critical to cancer research. Clinical trials are expensive, time-consuming to set up, and restricted by limited guidelines. For instance, while a researcher may learn how a specific sequence of treatments affects men under the age of 30, how will it affect women over 60? And what if the sequence of treatments changes? How will that affect the outcome?

With clinical trials, changing parameters requires the expense and time of recruiting a new group of patients. In contrast, within the tremendous volume of EHR data, myriad combinations and comparisons of parameters are possible. This data has the promise to provide substantial insights into the understanding of cancer and the design of future clinical trials—if we can overcome the barriers of poor quality and incomplete records due to incompatibility.

In 2017, MITRE launched an effort to address these challenges, which we call the Standard Health Record (SHR) Collaborative. The Collaborative’s central focus is to establish standards for the structure and content of health record information.

Creating an SHR does not mean eliminating the beneficial competition of having multiple EHR systems. It means these systems will share a consistent set of structures and content. The promise of SHR is that no matter where patients go—including among different healthcare systems—they and their providers will have access to a complete and accurate set of the same data. The resulting SHR addresses the dynamic data needs of providers, patients, and caregivers by including data specifications for many areas related to social determinants of health. In fact, the standardization of the data will give us a greater understanding of the variation in patients’ stories and experiences.

Each EHR system may present the information in different ways but the underlying data will be readily accessible to patients and to those providers given permission to share it. This will ease the burden on providers, greatly reduce the “starting all over again” process for patients whenever they see new specialists, and empower researchers with relevant, more easily analyzed data.

SHR will also accelerate secondary uses in public health, such as disease “early warning” surveillance, post-approval monitoring, and patient-centered outcomes research.

While SHR has the potential to improve outcomes for any disease or healthcare problem, we believe cancer research is the best place to start. MITRE and its many partners are working on a data standardization project, known as the Minimal Common Oncology Data Elements (mCODE) initiative. The goal of mCODE is to define a foundational set of critical data elements to enable clinical care and research via the EHR. The mCODE initiative is in line with the broader SHR vision and aims to demonstrate the SHR approach for cancer.

mCODE is supported by MITRE, in collaboration with the American Society of Clinical Oncology (ASCO) championed by Dr. Monica M. Bertagnolli, the Oncology Center of Excellence of the U.S. Food and Drug Administration (FDA), the Alliance for Clinical Trials in Oncology Foundation (Alliance), and the Clinical Information Interoperability Council of Health Level Seven International (HL7).  Institutional collaborators include Intermountain Healthcare, Salt Lake City, and Partners Healthcare and the Dana Farber Cancer Institute (DFCI), Boston.

The mCODE data standards are being developed through a use case approach.  Collaborators bring forward proposals for clinical research to the mCODE Development Team, which works with collaborating organizations to identify the data elements necessary for analyzing treatment across different EHRs and clinical care settings.  The resulting mCODE-based research or care coordination project produces high quality data from the clinical care setting that does not require extensive curation and can be shared with other mCODE-based datasets without reformatting.

Each new use case contributes to building a more versatile and effective mCODE.

Today, mCODE is being evaluated through point-of-care pilots to assess the suitability of its representation to support target use cases. The initial use case for mCODE is in support of comparative effectiveness. ASCO, CancerLinQ, MITRE, and Intermountain are piloting and evaluating this initial version of mCODE. The partners will use data collected via mCODE to create a “patients like this” comparative effectiveness tool, designed to be used by clinicians and patients to improve shared decision-making.

Through this partnership, ASCO and CancerLinQ are providing clinical expertise and leadership in developing, piloting, and endorsing mCODE. MITRE is developing Compass™, an mCODE-based FHIR implementation and Substitutable Medical Applications and Reusable Technologies (SMART)-on-FHIR application. Compass will extract mCODE data in computable formats and deliver reports to providers and patients—empowering shared decision making.

These comparative effectiveness reports will use the mCODE data for each patient to construct, analyze, and surface a population of similar patients stratified by treatment and outcome. Intermountain is operating test sites for capturing mCODE data and leveraging these reports.

In another example, MITRE is working with the Alliance, the FDA, DFCI, and Partners Healthcare on a project known as ICAREdata, which stands for Integrating Clinical Trials and Real-World Endpoints data.  This project involves patients who are part of a prospective, randomized clinical trial that obtains study data using conventional electronic case report forms.  For each study participant, the team is also collecting treatment response and toxicity data directly from the EHR using the mCODE format. The goal is to validate the clinical endpoint data collected via the EHR by showing that it matches the results obtained using standard study-specific electronic case report forms.

The ICAREdata project team is developing and testing new treatment response data elements for capture via the EHR.  The approach has been to configure the EHR to prompt clinicians to answer a limited number of structured questions. The answers to these questions are automatically placed within the clinical note as a tagged phrase, which can be easily extracted from data submitted via an API. The clinicians involved in this project are clinical trialists who have a vested interest in developing more efficient methods of gathering study data. 

The goal for the ICAREdata project is to use the EHR as a tool to conduct clinical trials that develop new treatments, particularly for situations not currently served by clinical trials—such as for patients with rare tumors, or to assess overall survival in patients who receive multiple lines of treatment.

The partners believe mCODE will enable clinicians and researchers to provide better treatments and, ultimately, cures for cancer patients by using the invaluable information contained in cancer patients’ EHRs. We will be able to explore the data from millions of patients and study the myriad combinations and comparisons of treatment parameters to provide substantial insights.

This information could improve patient care, empower shared decision making, drive innovation, and set the foundation for a national cancer health learning system.

Collaboration among a broad range of stakeholders is critical to the acceptance and scalability of mCODE. Many of these stakeholders are involved in mCODE Workgroups. Participants include Dr. John Halamka, a professor at Harvard Medical School and CIO of Beth Israel Deaconess Medical Center. (Dr. Halamka is leading a HIMSS presentation on Precision Medicine, which will touch on mCODE.)

By September 2019, a core group of elements for mCODE will have been defined. MITRE will have led the effort in developing corresponding SMART-on-FHIR applications (known as Compass), FHIR profiles, and FHIR Implementation Guides. And Intermountain will have piloted and proven the model.

Note: MITRE team members will discuss mCODE and the roll out of the corresponding SMART-on-FHIR application—known as Compass—at several HIMSS events.

The mCODE partners have committed the necessary expertise and resources to take this project from concept to implementation in two years (we started in the fall of 2017). Each partner contributes its own resources and expertise as part of this collaboration.

We’re looking for more partners with the same passion and dedication to bring mCODE to its full potential.

For more information, contact Mary Jo Fitton Sullivan at mCODE@MITRE.org.