Rethinking How to Engage with At-Risk Populations

By Matt Phillion

Population health risk stratification combined with social determinants of health (SDOH) data is reshaping care delivery, enabling providers to predict outcomes, reduce costs, and address health disparities.

Proactive, equitable care has led to reduced admissions, significant cost savings, and stronger patient engagement, explains Mike Hoxter, CTO of Lightbeam Health Solutions. For example, a large integrated delivery network in the Southwest, operating a health plan for the state’s Medicaid program, deployed a 30-day avoidable admission model to proactively identify high-risk members and reduce avoidable admissions among its Medicaid population. The intervention resulted in a 5.3% reduction in admission rate, leading to a $637,000 reduction in medical costs over a 12-month period—and an estimated 65 prevented admissions.

“There’s a lot of risk stratification methods that exist, for a lot of reasons. There’ s an absolute ton of data available out there,” says Hoxter. “There’s a lot of systems that are built to work with the lowest common denominator of what’s available to them.”

A lot of this data is coming from insurance companies, or from Medicare or Medicaid, based on claims and bills, and as a result this is the basis for most basic risk stratification.

“That doesn’t make it what’s best,” says Hoxter.

Looking at social determinants of health, there’s a wealth of data that opens up information at the census block level, Hoxter notes.

“Looking at this data, I know if the patient lives in a food desert. I know about the walkability of the area. The percentage of the population in the area that has a car: I won’t necessarily know if the patient has a car, but if I know only 30% of the population around them has a car, there’s a 70% chance they don’t,” says Hoxter. “And then I know I can’t order something as part of your care if you don’t have a way to get there.”

All of this information is incredibly useful to understand a patient but isn’t used in payer-based risk scores.

“You need to keep the lights on, and you have to get paid, but the risk score for how you get paid versus who actually needs help. This Venn diagram between the two is not a perfect circle; there are a lot of patients who are just not accounted for,” says Hoxter. “This information is readily available as social determinants of health data points. It’s just doing the proper data engineering and cleanup to make it available and tying it all together so you can start to see the root causes of condition prevalence. While the data’s out there, it is widely underused.”

Revising the focus of care

These social determinants of health should be looked at in the same way a care gap is, Hoxter explains.

“If the patient lives in a food desert, that’s a problem. Maybe they have it under control, maybe they don’t, but the next step is addressing that care gap,” he says.

Healthcare is at a point in its evolution where this isn’t optional long-term, Hoxter notes.

“Statistically, as a nation, the amount of our GDP we spend on healthcare does not leave us in a good spot,” he says. “We’re getting older, our average age is increasing, our life expectancies are increasing. We’ve got to take better care of our patients. How do you do an efficient job with the resources you have? It’s not a question of what happens if we don’t address these issues, because we’re largely seeing what happens if we don’t do enough about it now. Our costs have got to come down, and there’s a lot of really preventable care—so how do we get out in front of that?”

Hoxter does believe healthcare is headed in the right direction with rising percentages of the population included in value-based care models.

“People tend to do what they are incentivized to do. So, when you start to have a financial incentive associated with these behaviors that has nothing to do with billing another CPT code, if you’re getting paid because you took good care of the patient, preventive care works,” he says.

Getting there requires a change in company and industry culture.

“Realistically, this is not the way it’s been done. It’s not the status quo,” says Hoxter. “We need a cultural adoption on the operational side.”

And it requires time and resources. You can provide an organization with the data demonstrating a list of patients they can do something about, but if that becomes the 10^th to-do item and they’re struggling to get through their top three, it’s not incentivizing change. Freeing up the time to make those cultural changes by alleviating things like administrative burden can help organizations get there, though.

“The more you know about these patients, the more you can do to help them,” says Hoxter. “One way is to semi-automate communication. This is a population that needs help, but it isn’t clear until they’ve bubbled up to a person assigned to that patient.”

To be as preventive as possible requires tools that help these populations see if they need intervention or don’t, whether that’s enabling them to get to their own doctor visit, fill and pick up their prescriptions, or addressing whatever issue they are encountering and streamlining that communication so providers and organizations know what that person needs.

“There are a lot of companies that provide automated patient outreach. You’d like for them to have a care manager, but you only have so many care managers, and you can’t cover the entire population you wish you could manage,” says Hoxter. “How can we automate that? Using AI or automation tools to streamline communications can help make sure those patients are getting what they need. Most of your patients are probably fine, but the ones who aren’t, require you to take action, and automation can help flag those patients.”

It’s imperative from the quality-of-care perspective and the financial perspective to get ahead of these cases, Hoxter notes.

“A lot of times you find out the patient needs help when they show up in the emergency room,” he says. “That’s a really expensive way to find out they need help, and it’s bad for everyone. The patient doesn’t want to end up there, the provider doesn’t want it, the insurance company doesn’t want it.”

It will take both high-level influence and on-the-ground mental shifts to make this happen, Hoxter notes.

“It would be great to see some level of government adoption of this,” says Hoxter. “Better reimbursement for automated care and better support for telemedicine. That’s gotten better, but there’s room to grow.”

As noted earlier, people will do what they are incentivized to do, but if it has a positive effect, they will be even more likely to do what’s asked of them, Hoxter says.

“That starts at the government level and trickles down,” he adds. “I feel like there’s an attitude or mental shift that has to happen alongside this shift from a fee-for-service mentality to a more value-based care view of the patient. Our responsibility does not end when the patient walks out the door, and I think a lot of companies have viewed their patient populations that way.”

With a culture shift in the right direction, Hoxter says, the industry can find that spot where quality of care and value connect.

“Where technology and AI will be in five years is anybody’s guess, but we are continually increasing the average age in our patient population. We’re going to have the largest Medicare population we’ve ever had,” says Hoxter. “We’re going to need to do a better job making sure we have the ability to grow with those needs. We’ve got to educate people that this is the way medicine is going to work. We need to take care of these patients when they are outside our offices, when they’re at home, when you’re not billing them. They are just as valuable then. We need to get everybody to view patients as important all the time.”

Matt Phillion is a freelance writer covering healthcare, cybersecurity, and more. He can be reached at matthew.phillion@gmail.com.