Clinical decision support (CDS) is designed to improve care by presenting providers with evidence-based guidelines at the point of care. The lack of simple processes for translating guidelines into code understood by EHRs, however, has prevented widespread adoption. Recognizing the potential of CDS, the Agency for Healthcare Research and Quality (AHRQ) and the Office of the National Coordinator for Health IT (ONC) are working together to learn what works and how it can be widely implemented.
“Without CDS, EHRs may just be another means of recording information,” according to Eta S. Berner, EdD, professor of health informatics at the University of Alabama at Birmingham and a member of AHRQ’s CDS Initiative technical expert panel. However, the challenges of building, implementing and optimizing CDS systems are significant, she acknowledges.
“When clinical guidelines are written by professional associations, they write a text document in plain English, not in computer code,” says Jon White, MD, director of the AHRQ health IT portfolio. Translating prosaic clinical knowledge into code that any EHR system can understand is difficult. However, with Meaningful Use (MU) requiring providers to use CDS, AHRQ and ONC are seeking to scale CDS.
The organizations work together, but operate in separate, distinct roles: AHRQ identifies and funds research projects aimed at translating narrative recommendations into coded formats, and ONC works with stakeholders to determine best practices for making the resulting, computer-executable CDS rules available and compatible with different EHR systems.
From silos to standards
Providers have long used checklists and evidence-based guidelines to aid their decision-making, but digitizing a vast library of clinical knowledge makes the right information available at their fingertips when needed. To make information more readily available, AHRQ is funding two demonstrations with $2.5 million grants: the Guidelines into Decision Support (GLIDES) project at Yale University School of Medicine in New Haven, Conn., and the CDS Consortium at Partners HealthCare in Boston. The goal of these programs is to turn the siloed approach to CDS development into one that is standards-based, according to White.
When vague, clinical guidelines are nearly impossible to transform into tools that actually influence provider behaviors. So says an article published in January 2012 in the Journal of the American Medical Informatics Association (JAMIA), by Richard N. Shiffman, MD, director of GLIDES, and his colleagues. To facilitate the authoring of guidelines that deliver clear, actionable information that can be codified in XML format understood by EHRs, Shiffman et al developed a software application to help authors choose appropriate language.
Similarly, BRIDGE-Wiz (Building Recommendations In a Developer’s Guideline Editor) operates to the common “wizard” applications that guide personal computer users through software installations and system configurations. A series of 15 prompts walk authors through guideline development. The first prompt requires users to select an action from a drop-down list, the second prompt requires users to select a verb based on that action, the third prompt requires users to define an object for the verb and so on. Five guideline panels that tested BRIDGE-Wiz agreed it was useful for guideline development, and it has since served as a resource for the ONC and its partners on CDS standardization, says White.
Whereas AHRQ operates upstream from clinical practice, ONC operates “much closer to where the rubber meets the road,” says the agency’s chief medical officer, Jacob Reider, MD. ONC’s Health eDecisions initiative is currently engaged in consensus building to determine the best processes for publishing clinical knowledge and building it into EHR systems.
Shortly after ONC was established in 2004, the agency began contributing to a roadmap for national action on CDS that was published by the American Medical Informatics Association in June 2006. The stated goal of the roadmap is to “ensure that optimal, usable and effective CDS is widely available to providers, patients and individuals where and when they need it to make healthcare decisions.”
When the roadmap was published, it was difficult to imagine EHR vendors cooperating with one another, but they have since “set aside historical differences to focus on a shared vision of what this could really be,” according to Reider. “We’ve got these contrasting factions to agree that there can and should be a common goal.”
Today, various stakeholders are working together to develop a knowledge-sharing framework comprising a central knowledge repository, structured knowledge resources supportive of MU criteria, a web portal for accessing these knowledge resources and executable CDS tools using these knowledge resources.
That knowledge-sharing framework is being tested at Partners, where several vendors have partnered with the CDS Consortium, a group aimed at storing and making accessible a wide body of CDS tools. The results have been promising so far, according to an Aug. 4, 2012, JAMIA article. Vendors are now integrating third-party CDS applications into their EHRs, interfacing systems with the CDS Consortium’s online service and converting structured knowledge resources into their own knowledge representation formats.
For now, CDS progress is incremental. The CDS Consortium demonstrates the feasibility and potential of a central knowledge repository, which would allow providers to access a library of up-to-date CDS, but plug-and-play functionality is still out of reach. Providers who want CDS must rely on what their vendors give them or build their own. However, achievements made by GLIDES and the CDS Consortium mark important landmarks on the CDS roadmap. “Perfection can be the enemy in health IT,” Reider says. “It’s about what’s going to work well enough.”
MU has helped the federal government get better at measuring healthcare quality, says Reider. Now providers need the tools to help drive quality improvement. Currently focused on the development of a simple, non-disruptive CDS to remind providers that a patient is due for blood glucose testing, th e work is paving the way for more advanced applications using emerging natural language processing and data analytics tools. “With MU regulations pushing the use of CDS, we will see the fruits of the research in the near future,” Berner said.