A couple of months ago, I spoke at a conference about the patient-centered medical home. The attendees were really engaged, and they were asking lots of questions, especially about predictive analytics and how you incorporate those into the clinical workflow.
At one point, a gentleman from a health system—a system whose name you’d recognize, one that has managed population risk for over a decade—stood up and shared an informal analysis that his informatics team had run.
His team was searching for the best predictors of which patients would be the highest cost to the health system in the next year. The team had all the data they wanted at their fingertips—claims data, clinical data, and many other risk factors.
But what they found was that none of their data was as good as human intuition. The most accurate predictor of health care utilization, far and above any number they could pull from their data sets, was to ask physicians one simple but elegant question:
"Which patients would not surprise you if they died within the next year?"
Why is this question brilliant? It doesn't ask doctors to classify which patients are "highest-risk," an exercise that often yields patients who are simply challenging to manage, along with patients who have real clinical problems.
And it doesn't just identify patients who have medical issues. It also flags patients with social or other non-clinical needs: for example, the elderly patient whose wife died last month and now has no support at home for managing his dozen daily medications.
“Physician input into the creation and fine-tuning of risk models is critical but to continually ask providers to assess potential patient risk could be perceived as an additional burden in an already busy day.” —Advisory Board member
Do you agree? Scroll down to see what your peers are saying and log in to tell us what you think.
No need to wait for IT perfection
Could a simple question really outperform sophisticated data analysis? Surprising, but I can’t say I was completely shocked.
So many health systems looking to manage populations are searching for the perfect care management solution, the one that’s able to crunch every last number and identify the high-risk and highest-cost patients with flawless accuracy.
But there’s no need to wait for IT perfection to start down the path of care transformation. If you talk to the best population health managers, they’re still refining their technology approach, just as they’re continuing to develop and improve their care management processes.
Incremental and piecemeal: Not always dirty words
Actually, some research suggests that the health systems that are spending the most money upfront on IT, in an effort to set up a comprehensive backbone for care management, actually tend to be less successful at improving population health than peers that take a more incremental IT approach.
Yes, IT systems are critically important components of care transformation infrastructure. The only way to measure population health at scale, let alone benchmark opportunity, predict performance and prompt patient-level interventions, is through information technology.
In particular, analytics are the key to finding "rising-risk" patients, the patients who ought to be managed more closely in models such as the patient-centered medical home.
But to find your highest-cost patients, rather than searching in numbers, try looking for them in your sites of care, by asking your ED and physicians that "one simple question." No guarantees—this is just one health system’s analysis, after all—but I’d love to know what you find out.
One patient, one ED, one year. Over 200 visits?
Up to 28% of ED visits come from high-utilizer populations. See how Community Regional Medical Center achieved impressive results with a frequent user program that provides outreach, engagement, and case management services. Watch now.
Like this post? Read more like it in the Care Transformation Center Blog and subscribe to get blog alerts as soon as new content is posted.