Better clinical decision support depends on 'population, protocol and patient'
Clinical decision support is designed to deliver the most relevant patient data to the physician at the time it is most needed – namely, when a critical choice about care has to be made. It is not a new concept and the healthcare industry certainly has the technology available to make it work at an optimal level.
Still, there is room for improvement on both the provider and vendor ends, say specialists in the CDS field.
"The traditional definition of CDS is what you can do within the electronic health record to support better decisions, but across the industry the state of decision-making is really bad," said Dale Sanders, senior vice president of Salt Lake City, Utah-based Health Catalyst.
"You have spots of innovation in some areas, but as an industry CDS at the EHR level is really bad. On a scale of one to 10, I'd give it a three or four. But there is great movement around improving it, so we're optimistic."
The keys to optimizing clinical decision support are three levels that Sanders calls the "three p's – population, protocol and patient." Each level has its own self-contained purpose, but together they coalesce into an effective program.
"When you're making decisions and putting data in front of patients, they are as important to CDS as doctor is – both parties have to be involved," Sanders said. "The decisions you make about clinical care and strategy at the population level is a different skill set, different strategy and different method than at the patient level.
"The next level down is the protocol level, where you narrow the number of patients affected," he added. "Within the population, it is about developing specifics for clinical protocols of a certain type – the temporal dimension of decision making, measured in months and weeks. The final tip is delivering to the personalized level for the patient."
In grading the industry based on his "Three Ps" benchmark, Sanders says efforts at the protocol and population levels get "passing grades" due to increased emphasis on making them better. However, he imposes "a failing grade" at the patient level due to "a lack of vision, priority and leadership" around the topic.
Extracting actionable data
The machinery is in place for deploying CDS, but the industry has faced various obstacles in getting it up to speed, says Foad Dabiri, chief technology officer at San Francisco-based WANDA. Principally, he says, the challenge has been with extracting actionable information from the various system silos.
"It is getting actionable information – what you can collect and record," Dabiri said. "The question is, what information is the physician looking for and how can it be seen?"
Having instant access to actionable data is paramount when making decisions to keep chronic disease patients from costly hospital readmissions, he said. Diagnosing the symptoms of a homebound patient with congestive heart failure, for instance, requires enough granular information so that the physician can determine if an episode such as fluid retention can be handled with a remote intervention or if the situation is more urgent in nature.
"A patient can show signs of something wrong, but that is very different than a tangible outcome," he said. "What we do is combine various symptoms and vital signs, and through analysis correlate the historical value and combine it into a single decision."
Technology has advanced to the point where CDS should be readily utilized, but Dabiri maintains that many provider organizations are still overly reliant on proprietary legacy systems that prevent mobile access.
"The industry recognizes the need to upgrade – providers need to move from an in-network system to a cloud-based system," he said. "That would provide more scalability, reliability and access to electronic records."
For physicians contemplating implementing a CDS system in their practices, there are some considerations they need to make, such as determining what type of CDS is most suitable as well as adhering to HITECH Act requirements.
Allan Ridings, senior risk management and patient safety specialist with the Cooperative of American Physicians concedes that there have been "trust issues" with CDS in the physician community, which is why the sector is not as proficient in its utilization as it should be. Moreover, he says EHRs evolved in a backwards fashion, starting with the claims and financial data and moving to clinical diagnostics instead of the other way around.
In looking at CDS systems, physicians will find two kinds – a knowledge-based system and a data mining system. It is essential they know the difference, Ridings said.
A knowledge-based system obtains patient data from a result engine and "reveals all discoveries based upon the data being enquired upon," he said. "This type of system is also known as ITTT –'if this-then that' and could be used for determining drug interactions."
Data mining is based on algorithms, artificial intelligence and machine learning from previous entries. "They might be used to examine a patient's medical history in conjunction with reliable clinical research," Ridings said.
Whichever system they decide upon, physicians need to make reducing risk and maximizing patient safety their highest priority, he said.
Replacing the 'gut'
As a physician himself, David Delaney, MD, chief medical officer for the SAP Public Services and Health Care Industries team in Newtown Square, Pennsylvania, understands the traditional medical process of "using your gut, intuition and experience" in decision making. And while some old-school docs might still prefer that method, Delaney realizes the tremendous clinical advantages of CDS.
"The ability to leverage organizational knowledge to bring better decisions has been lacking in the industry," he said. "The data also includes claims and financial data that are needed to understand the 'value' in value-based care. I believe we are ready to pivot into an era where if there is information available that can impact a decision, it must be brought to bear."