Healthcare Analytics, Population Health Management, Healthcare Big Data

Population Health News

Prioritizing Preventive Value-Based Care with Big Data Analytics

Success in the value-based care environment requires a familiarity with big data analytics that can fuel population health management efforts and preventive care.

By Jennifer Bresnick

- The old adage that “an ounce of prevention is worth a pound of cure” is nowhere more applicable than in the primary care setting, where healthcare organizations moving to value-based care arrangements are eagerly seeking data-driven strategies to make them more nimble, efficient, and responsive to patient needs.

Value-based care and big data population health management

As the health system’s financial mechanisms start to incentivize providers to engage in population health management techniques like risk stratification, chronic disease screenings, and health coaching, organizations are becoming more interested in adopting big data analytics tools to deliver the insights they need to keep their patients healthier for longer.

At the Fall 2016 Value-Based Care Summit on November 15, panelists urged organizations to prioritize prevention and proactivity by understanding what big data really is and how, when, and why to harness it.

“We now have algorithms that allow us to take more data in different forms and do things with it that were impossible five years ago,” said Bharat Rao, Principal in KPMG’s Advisory Services. 

“That’s because of the increased digitalization of data and the increased computing power through the cloud.  That technology has transformed many industries, and it’s inevitable that it will transform healthcare through population health and value-based care.”

READ MORE: Big Data Analytics, Population Health Top Executive Priority List

But the term “big data” can be misleading, he added, and produce some fundamental misunderstandings of how to collect, manage, and leverage health information. 

“When you think about ‘big data,’ you immediately think about volume – lots and lots of data,” he said.  Volume is, in fact, one of the “three V’s” of big data – volume, variety, and velocity – but the “big” part of “big data” is more about the data’s complexity than how much of it there is.

Organizations create “big data” when they combine two or more disparate sources of information into a single dataset that can produce novel insights.  These datasets can be large or tiny – it’s the act of bringing multiple types of data together that makes it “big.”

In the healthcare industry, this can mean integrating claims data with electronic health records to figure out the cost burden of certain comorbidities, comparing public transportation schedules with missed appointment records, or examining local air quality data with the rate of ED visits for asthma to pinpoint opportunities to teach patients about using their inhalers. 

By combining these very different types of information, providers can learn something new about their patient populations, their behaviors, and their outcomes.  These richer insights are crucial for designing and executing preventive care strategies, including risk stratification, predictive analytics around patient safety issues like sepsis, and targeted patient outreach or education.

READ MORE: Claims Data Analytics Show Diabetes Has Third Largest Health Impact

“At Atrius Health, when we think about the uses of data and analytics in the value-based environment, the theme is trying to swim upstream to get to prevention,” said Joe Kimura, MD, MPH, Chief Medical Officer of the Massachusetts-based health system.

“That makes care less costly and more valuable, not only to the health system, but to the patients, society, and the community.  The question we need to ask ourselves is how we can go upstream faster, more efficiently, and more effectively.  We have to think about efficiency, effectiveness, and reliability.” 

The Difference Between Big Data and Smart Data in Healthcare

Providers aren’t just financially accountable for prevention under value-based care models.  They also have a responsibility to their patients to deliver responsive, holistic care that will reduce the likelihood of preventable injuries or illnesses.

“As a primary care internal medicine physician, we’re really trying to help people not break their hip,” Kimura said.  “It’s not only about figuring out what happens after they’ve fallen and broken the hip, but how to prevent falls that might result in that injury.”

READ MORE: Using Risk Scores, Stratification for Population Health Management

In order to accomplish that goal, providers need a variety of data sources, he continued.

“You absolutely need structured and unstructured EHR data at a minimum,” he stated.  “In the value-based environment, you’re also going to use claims and administrative data.  You need to be able to wrap that together to see what’s happening.”

“And on top of that, you need patient-reported outcomes; you need information from the Internet of Things – that’s going to be real-time data coming in that will allow you to pick up signals and allow a delivery system to start figuring out what they can do in specific situations that will prevent something bad from happening tomorrow, a week from now, a month from now, or six months from now.”

Collecting and synthesizing all these data sources, many of which are unstandardized, unstructured, and hard to verify, is a very difficult proposition, Kimura acknowledged. 

“Big data is classically about volume, variety, and velocity, but there’s also a fourth vector, and that’s veracity.”

Patient-reported outcomes (PROs) and Internet of Things (IoT) data are notoriously short on integrity and can be difficult to interpret and rely upon in a regimented clinical environment. 

Recent surveys have found that only 20 percent of hospitals are routinely using PROs to inform their care decisions, while forty-two percent of organizations suffer from IoT “data overload,” and have not yet figured out how to integrate wearable devices, sensors, and remote monitors into their overall data strategies.

The Role of Healthcare Data Governance in Big Data Analytics

While devices like FitBits and Apple Watches are gaining in popularity among tech-savvy consumers, and value-based reimbursements are increasing providers’ consideration of the patient’s voice in his or her care, few health IT developers have really cracked the conundrum of how to present this information in a meaningful, actionable way.

The challenge of utilizing big data for care improvements gets even more complicated when providers start thinking about the difference between process measures and outcomes measures, added Rebecca Williams, RN, Care Coordination Management at St. Joseph Hospital.

“When it comes to population health management, there are some quality measures that are more about if the patient had something done, not necessarily what the value of it was,” she explained.   “Diabetics, for example, are measured on if they had their A1c tested, not whether or not the number was within an optimal range.”

Providers looking to use this type of data for preventive care must carefully consider how to create equilibrium between processes and outcomes to highlight trouble spots that may develop for healthy patients down the road.

“It’s very important to look at outcomes, but by the time we’re looking at outcomes for chronic disease, it’s almost too late,” Williams asserted.  “We want to keep people healthy.  It’s very important to care for patients who are already sick, of course, but what if we could dial that back a little bit and start looking at the 30-year-old whose blood sugars are just starting to tip over the edge?”

“You want to start looking at risk stratification, predictive analytics, and being able to proactively address the future instead of reacting to the past, but you have to do that in a way that doesn’t sacrifice the care of any patient population.”

To strike the right balance, Williams suggests that providers develop a roadmap for population health that targets high-priority patient groups most likely to benefit from interventions, including low-cost strategies like telephone follow-up from a care coordinator.

“When you’re thinking about data, you need to stop and decide what type of population you’re trying to impact.  Are you going to go after that top five percent of the population that incurs 70 percent of the costs, or are you going to go after the middle tier of patients that could be on their way to moving into that high-utilization bracket ten years from now?”

“Every organization has to pick an area of focus if they want to get the most out of their data and their investment,” she said.  “That may mean that the higher utilizers get formal chronic care management, and your medium-risk patients get a health coach.  Not everyone requires the same intensity of services, so that is a major consideration for designing a workable strategic plan.”

Continue to site...