Deploying Big Data Into Better Diabetes Care

Lonny Reisman, MD, CEO of HealthReveal, told attendees at Patient-Centered Diabetes Care that harnessing multiple disparate data sources can help physicians deliver better solutions at the point of care.

The burden of diabetes is large: more than 29 million Americans, more than 422 million people worldwide. The cost of care is high and growing; in 2014, spending for diabetes therapy reached $97.68 per member per year, leading all other diseases, for commercially insured patients covered by Express Scripts.

The good news, according to Lonny Reisman, MD, formerly with Aetna and now the CEO of HealthReveal, is that diabetes has more therapeutic options than ever. The bad news? Many of these drugs are expensive, and with the flood of information on diabetes available, how is the average practitioner—especially those in primary care—to keep up?

Data can hold the key, said Reisman, who offered the keynote address, “The State of Big Data in Diabetes,” which opened Friday’s session at Patient-Centered Diabetes Care, presented by The American Journal of Managed Care in Teaneck, New Jersey.

Even with advances and more effective treatments for diabetes, “There are huge issues in terms of awareness and education,” Reisman said, especially in addressing patient behaviors that affect the condition. As a cardiologist, Reisman sees the effects in his field: 57% of the costs of diabetes come from strokes and coronary disease.

And the costs are large: a well-known study by the American Diabetes Association found that of the $176 billion in annual direct medical costs, 43% came from inpatient hospital stays. Medical spending is typically 2.3 times higher if a patient has diabetes, Reisman said.

“Despite the advent of programs, new drugs, and diagnostic techniques, we actually do a pretty poor job,” he said. If healthcare seeks to deliver better quality care, improve population health, and lower costs in diabetes care, the system is falling short.

The question, Reisman asked, is “Why aren’t we in a better place?”

A huge challenge, he said, is bridging the gap between new evidence and what’s happening in clinical practice. Even if physicians have time to digest new studies, it is unlikely they will recall information on subgroups and exclusions for the patient sitting in front of them.

At the same time, huge amounts of data are being gathered on patients each day and collected in electronic health records (EHR), so that “machine learning” could be ongoing, he said.

To improve care at the community level, “We really need a better solution,” Reisman said. “We really need to take advantage of what’s available to us, to exploit all these data sources.”

How can this be done? Reisman showed how today’s medical information exists in silos: claims data, clinical data, data from research by pharmaceutical companies, and new, rapidly accumulating amounts of behavioral data being collected through social media. The promise of “Big Data,” he said, is the hope that these data sources can be integrated and harnessed for the betterment of patients at the point of care.

“Certainly, if we exploited the information available, we can do some pretty exciting things,” Reisman said. The challenge is creating “a bridge to the patient on the street.”

Harnessing Big Data means figuring out how to deal with the “4 Vs”: the volume of data, the variety of data sources, the velocity or constant stream of data, and the veracity or accuracy of data. If information from EHRs is mixed with non-traditional sources, steps must be taken when data are captured and interpreted to ensure data are useful for decision-making.

Harnessing Big Data is based on several principles:

· Prevention is better than waiting for disease to happen

· Advanced analytics can predict costs and adverse outcomes

· Care can be tailored to the person

· Large amounts of data can track population health trends

· Technology can be used to engage patients

Challenges involving patient privacy, standards, and fragmentation must be overcome. Some sources of data have good incentives to share it, and some do not.

But when data are used well, they can reveal or confirm risk factors that can be deployed into clinical practice. Reisman pointed to a study that had just been presented at the recent meeting of the American College of Cardiology that showed how waist circumference is a more important risk factor than overall weight or BMI for persons with diabetes.

“Not everything can be done through a clinical trial,” Reisman said. The study, by Intermountain Health and Johns Hopkins, is a good example of the “machine learning” that extracts patient information to gain new insights.

He pointed to the well-publicized results for empagliflozin (Jardiance), from a large, very expensive randomized clinical trial that showed a CV benefit for the diabetes drug. But Reisman asked whether extrapolation of data could be used to show benefits or risks in the future for similar drugs.

The real promise, Reisman said, will come when implantable devices or wearables can be integrated with evidence-based standards for at-risk patients, and warn patients or physicians when that patient’s physiological indicators show something is going awry. These are the solutions Reisman is pursuing at HealthReveal.

“To the extent there is clear deviation to the literature, we would have to the ability to extend to the treating physician,” Reisman said, who said the physician might recommend a new medication, or something more intense depending on what the indicator suggests.

“This is relatively new, but the benefits are huge,” he said.

Related Videos
Related Content
© 2023 MJH Life Sciences
All rights reserved.