At the Children’s Hospital of Philadelphia, work on genomic clinical decision support has started with a focus on primary care. CHOP researchers are seeking to understand the potential benefits and challenges of genomic clinical decision support in pediatric care.
“We are focused on solutions, not hype,” said Jeff Pennington, the Department of Biomedical Informatics’ senior director of translational informatics, in a recent presentation. At CHOP, he leads a team of data integration analysts, programmers, and technology developers.
In 2015 CHOP clinicians started ordering clinical exome tests for many conditions and diseases. So how do clinicians get the results?
“The current state of art is a seven- or eight-page PDF. This information-rich document ends up in the EHR and has everything a clinician might need to know, but ends up being not that accessible,” Pennington said. “We saw an opportunity to make this data more accessible to clinicians.”
CHOP took an empirical approach and is conducting a three-phase study. First they did a survey of providers to study the need, and then worked to study provider workflow for real clinical cases. Phase 3 involves developing a prototype tool to put into practice in Epic.
“We started with primary care because they are often involved in care coordination,” Pennington said. “They are a lifespan caregiver and can get involved in family care, which genomic testing has implications for.”
In Phase 1, the team did open-ended interviews with six physicians. The team sought to assess their experience caring for patients with complex and rare genetic findings and hear their opinions about how the EHR currently supports this role.
The physicians said the notification of results is a big deal to them, but the reports can be confusing. They expressed that they need to know more about genomics. “They said they felt confident that they understood basic principles and thought they should be involved in the process, yet they were underprepared to engage with patients on genetic conditions that had some consequence,” Pennington said. “They are at a bit of a loss as to how to weave this into their practice.”
In Phase 2, the team then created a mockup of a clinical decision support tool. The researchers took this mockup to 26 providers in CHOP’s Practice-Based Research Network and did a walkthrough of the tool and gave clinicians a survey.
The mockup involved a reality-based case study. The patient had a genetic variant that conferred risk of adverse reaction to anesthetics. The tool gave them a view of the result that would allow doctors to receive education, deliver education to patients, and do documentation. They get inbox notification of the results, with access to the underlying test report. “They thought it was useful and that it would be useful and enhance work flow,” Pennington said. “Primary care providers are ready for help in this area. They think they have a role to play. We now have an evidence-based model for how we could provide decision support.”
In Phase 3 they plan to build a prototype to put into practice. “We are incredibly fortunate. No other institution that has the ability to customize Epic at the lowest level that CHOP does. We can unlock the guts of the system and pretty much do anything. We can highly customize our implementation. That is good for us.”
One big issue, he added, is how to address knowledge management. Someone has to curate a knowledge base — from gene variance to clinical guidelines to educational material, with recommendations and rules for gene variants. “We have to take into account the fact that the literature changes over time. How do we re-analyze these data in the context of new information?” It requires an expert curated knowledge base. There aren’t many third-party options, but does CHOP want to build such a knowledge base? “I can’t overstate how important this is to doing decision support beyond a pilot research study.”
Pennington said health systems would have to work through these issues soon. “This is here and happening now,” he said. Exome tests are being ordered. How do they get integrated into care? These patients have siblings. How are they notified? How does research fit? “How do we want the information to be adopted?” Pennington asked. “A cardiologist ordered a test, and a medical interpretation is done by a genetics center. There is a primary care doctor out there on the other side of that who needs some help.”