Developing an Alzheimer’s Risk Score as an Epic Widget
As researchers investigate therapies that could eventually alleviate or eliminate the symptoms of Alzheimer’s disease, many clinical neurologists are trying to get patients to be more proactive in thinking about brain health in much the same way they think about heart health, with a focus on diet, sleep and exercise. The Center for Brain Health at NorthShore University HealthSystem in the Chicago area is relying on the health system’s informatics infrastructure to help neurologists identify patients at higher risk for developing Alzheimer’s. They also are integrating mobile applications to help track patient lifestyle changes.
In a recent interview, Demetrius Maraganore, M.D., chairman of NorthShore’s Department of Neurology and medical director of its Neurological Institute, said many research studies have shown that if you exercise daily and eat a diet rich in fruits and vegetables, and don’t smoke or abuse alcohol, your likelihood of getting Alzheimer’s disease is much lower than if you didn’t do any of those things. “We know from observations made in tens of thousands of subjects over decades that how you live and behave is strongly associated with your likelihood of getting Alzheimer’s disease,” he said.
Of course, of the 20 factors that can double or halve a person’s likelihood of getting Alzheimer’s, the king is genetic type and the variants of the APOE gene. If you carry a variant that increases your risk of developing Alzheimer’s, that should motivate you to work harder on brain health activities and take it more seriously, Maraganore said. “Genetic type can help us decide which people need to make the most effort to work on brain health,” he added.
The Center for Brain Health has created some direct-to-consumer tools, including a brain health quiz that also links up to the family health history tool called Health Heritage available though the NorthShore portal.
The organization also is taking a “big data” approach. “We are creating a risk prediction model built into Epic using routinely captured data and generating alerts and lists that facilitate population health management,” Maraganore said, adding that he was unaware of any other health system that has deployed a system-wide Alzheimer’s risk score.
Within its Epic system, NorthShore captures dozens of discrete data elements about an individual during every visit, and that data gets archived in the enterprise data warehouse. Researchers are starting to look back at the subset of patients who developed Alzheimer’s disease over the past five years. “We are going back to 2009 and saying, what were the discrete data elements that were captured in that historical cohort in 2009, and do any of these data elements predict accurately who got Alzheimer’s in 2014 and who didn’t?” Maraganore explained.
“We are using the data routinely captured by Epic to determine someone’s likelihood to be free of Alzheimer’s or their hazard of developing Alzheimer’s in the next five years,” he explained. “And we can assign a risk score to that person and decide what risk score sets off bells and whistles so that we should inform this person and their doctor that this person is at increased risk.”
Within Epic, NorthShore is building a widget that will allow physicians to access a patient’s Alzheimer’s risk score. If they consider it to be actionable, they can refer the patient to the Center for Brain Health. “We can take it a step further and rather than having the physician look at the widget and access the patient’s risk score, we can provide a report on which patients in their practice have crossed a risk threshold that they consider actionable,” Maraganore said. “We can generate those reports from the EMR and send messages to physicians these are the people in your practice with an increased risk for Alzheimer’s, the same way we would use the EMR to identify people in the health system who may have a diagnosis of hypertension.”
Maraganore said that between October 2015 and September 2016 he anticipates that all NorthShore primary-care physicians will be getting lists of their patients who have a high risk for Alzheimer’s.
He said the elements that make up the risk score would be very discovery-driven. “We may be surprised by some of the things we find. We may find things totally unexpected like gallstones. It could be that we have made an important discovery about a risk factor for dementia that was heretofore unknown,” he noted. “What we will do is publish results and challenge the scientific community to replicate or refute those findings.”
But Maraganore said NorthShore’s predictive model would undoubtedly reproduce some prior prediction models used in research studies that identified things such as hypertension, smoking, head injury, obesity, heart disease and stroke as factors. “I think our model will have validity because it will include factors that are known to be associated with Alzheimer’s, but we will also make new discoveries, and have a way of stratifying our patient population at large into higher risk and lower risk, so we can more actively manage the higher risk population.”
NorthShore also is prescribing multi-modal interventions involving diet, exercise and sleep. “Because we know it is hard to change behavior without motivation, we are incorporating digital health tools commercially available today,” Maraganore said. “If you are sedentary, we can set a goal of 10,000 steps a day and if you reach that goal fewer than four days a week we can alert you through a HIPAA-compliant e-mail or text message or via phone. We are keen to use these available technologies and integrate them through NorthShore’s digital health initiative.”
Maraganore has received funding from the federal Agency for Healthcare Research and Quality to do a study called “Quality Improvement and Practice-Based Research in Neurology Using the Electronic Medical Record.” It is helping to create a neurology practice-based research network so that tools such as the one NorthShore has built in Epic that discretely captures information about neurological disorders can be disseminated to other health systems at no cost in exchange for sharing de-identified data. Participants are federating the database in support of quality improvement. “The prediction model we have built in Epic can be shared with other neurology departments across the country, and they can stratify their patient populations if they wish,” he added.