Harnessing Big Data to Enhance Population Health Management

August 13, 2020 - As the industry continues its quest to provide holistic, comprehensive, and cost-effective care to patients, the term population health management has emerged as a crucial task for organizations to undertake.

Understanding the needs of particular patient groups, targeting resources to those who need them most, and measuring outcomes are all part of advanced quality care delivery. But in order to successfully manage patient populations, institutions must have a solid grip on something the industry contains in multitudes: Data.

“Healthcare is typically slow to take things up, but the industry was always big data even before there was big data,” Brian Coffey, senior vice president and chief data insight and innovation officer at Southwestern Health Resources (SWHR), told HealthITAnalytics.com.

While clinical, social, and claims information is in no short supply across the healthcare landscape, drawing actionable conclusions from this big data is another challenge entirely. With 29 hospital locations and more than 650 outpatient clinics, SWHR is especially well-versed in the trials of big data. 

“You can imagine how much information we have with the 700,000-plus lives that we manage, and using that information is paramount,” Coffey said. “We’re a data-driven organization and everything we do depends on the availability, accuracy, and actionability of the information that we produce.”

READ MORE: Sharing SDOH Data for Improved Population Health Management

Despite the numerous challenges that can come with so much data, Coffey noted that when it comes to proactive, innovative care, less is certainly not more.

“Claims data is great to have, but it is not always complete data. It only gives a partial picture of an individual member. The EHR is really where this information lives. It lives in the notes, it lives in the documentation,” he explained.

“For example, using EHR data, we can monitor patients admitted to the emergency department. We can track their length of stay throughout that encounter, and then ensure that the patient has a follow-up with their PCP. We can do all of this even before the claims data drops, because we can see it in the EHR and the registry data.”

For institutions looking to leverage big data and improve their population health management strategies, it helps to have a plan, Coffey added.

“Organizations need to look at how they're going to use that data to model their populations,” he said.

READ MORE: How Artificial Intelligence, Big Data Can Determine COVID-19 Severity

“For instance, you could develop a model for complex seniors to be able to predict the risk of fall. What is the risk of fall in 75- to 85-year-old men or women in the population of interest? That is where EHR data becomes powerful, because you're able to look at the drugs they’re taking and whether they have a history of falls, as well as social factors like their living conditions and transportation options. These things can help you understand your population.”

To gain further insights into population health and patient needs, entities will need to leverage big data analytics tools that can help providers anticipate future trends.

“At SWHR, we do the traditional retrospective reporting that most organizations in an ACO world do, but we’re also currently implementing machine learning models to look at EHR data on a larger scale,” Coffey said.

“We’re using these tools to take a deep dive into physician notes that may lead to recognition of a condition that wasn't coded as such. This will hint at what direction we should take with our patient populations.”

When building machine learning and predictive analytics models, socioeconomic data is just as essential as clinical information, Coffey stated. This data can offer new insights that providers may have otherwise overlooked, leading to more accurate risk predictions and treatment plans.

“Social determinants aren't just out there for case workers to use, they're there for the analyst to offer information to the provider to help them properly manage their population,” he said.

“With social determinants data, for example, I've seen fall risk models work with 90 to 95 percent accuracy. It’s astounding what kind of interventions can be done once the providers understand their populations.”

Linking patients’ social determinants of health with their clinical information can also facilitate more proactive care, particularly in the context of COVID-19.

“We’ve started looking at correlations between social determinants and not only claims data, but also EHR data,” Coffey said.

“Especially now during the pandemic, we're leveraging this information to conduct patient outreach and try to get them in for screenings – colorectal screenings, breast cancer screenings, cervical cancer screenings, and so on. We're mining that information and working with their physicians or their PCPs to help the physicians connect with their patients and get them in.”

Additionally, socioeconomic data and analytics technologies can help physicians recognize and reduce health disparities, leading to better outcomes for patients.

“We're in the process of working with our physicians on quality care gaps. This is an area where we’re using a lot of machine learning. When members do come in to see their PCP, those gaps can be covered properly and the patients can get the right care at the right time,” Coffey said.

“In this case, the value is in actually working with our physician partners to show where their population trends are emerging as far as quality gaps and conditions becoming more prevalent.”

Going forward, extracting meaningful insights from big data will likely require the use of advanced analytics tools – a trend that will contribute to better management of patient needs.

“The translation from research to operations is what’s driving adoption of analytics models. Researchers have time to vet these tools and determine where they can be used, whether it be at the bedside, in the clinic, or in operations for case management. Implementation of predictive technologies and other analytics tools will only increase in healthcare,” Coffey concluded.

“Machine learning, predictive analytics, and AI tools are gaining a foothold, especially now during the pandemic, because these technologies can help providers connect with their population. These tools are able to go in and mine the EHR so that the best data is right in your lap and you can take action. The technology is just going to keep moving on up in healthcare.”