It's one thing to build a population health management infrastructure if your patient panel is composed of people with stable employment and housing, consistent insurance coverage, and no trouble getting access to a nutritious diet.

It's quite another to get a grip on achieving better outcomes if some of your patients live at or below the poverty level, are perhaps homeless, and wondering where their next meal will come from.

"The big problem in healthcare is that all our systems and data sources are focused on the medical diagnoses, and all the risk scoring tools revolve around the medical diagnoses, and in our population that's only about 10 to 15 percent of the story," says Nancy Garrett, chief analytics officer of Hennepin County Medical Center in Minneapolis.

The medical center is one of the founding partners of Hennepin Health, a Medicaid ACO that launched in January 2012 for single adults aged 21 to 64 without dependents. The organization integrates medical, behavioral health and social services, offering shared financial incentives for collaboration. For the plan to succeed, collecting enough contextual data well beyond the walls of a hospital or doctor's office is critical.

Early results, according to data published by the Agency for Healthcare Research and Quality, are promising:

*Through November of 2013, the rate of outpatient visits increased 3.6 percent, suggesting that efforts to engage members with primary care are having an effect.

*The rate of ED visits per 1,000 Hennepin Health members was reduced by 8.4 percent.

*The rate of inpatient hospital admissions decreased by 2.5 percent.

*Intense care coordination and management have led to reductions of 40 percent to 95 percent in the costs of caring for previously high-using enrollees. A 90-day review indicated that more intense medication management for these enrollees has led to a 50-percent reduction in their medication costs.

Data sharing is obviously an integral ingredient to this coordination. To that end, the system gives clinicians and social workers involved in care access to its Epic EHR system. A centralized data warehouse pools clinical and utilization data and lists the patient’s social services case worker in a single patient record. Because enrollee-specific details on social services cannot legally be imported, providers can call the case worker to discuss individual cases and share information as needed. And lastly, the data warehouse organizes the patient’s health and utilization information into a patient-specific summary known as a “radar report" within the Epic platform, which Garrett says is used as a work flow guide for the program's care coordinators.

"They log into Epic and can see things like 'These are the 10 of my patients who came into the ED in the last 24 hours, and these 10 were admitted,' she says. "Then they actually go visit them and make sure they are coordinating their care."

Perfecting the way to get the large percentage of data outside the clinical record will be critical to delivering better individual care, Garrett says, and the organization has instituted a lifestyle overview survey to establish a baseline. The overview, she says, is helpful in truly understanding patients' needs and goals on an individual basis. For example, a diabetic patient who is also homeless has a myriad of obstacles in adopting a lifestyle of assuring a nutritious diet and a stable medication regimen, factors that obviously affect their health.

"If the goal is to get into stable housing, we can get the social workers on it," Garrett says. "Then we can work on dealing with the diabetes after they have a place to cook a meal and store their insulin."

Indeed, she says, the overview process is not only an example of data collection, but also "an intervention in itself."

For example, according to Garrett, the system interviewer doesn't stop at asking a patient if he or she is homeless--instead, the questioning goes deeper, into asking about any concerns the patient has about potential trouble ahead.

"We are trying to get at the issue of housing stability," she says. "It's really a continuum. So that series of questions is documented in the medical record and even just having the conversation is helpful in terms of understanding what that patient's needs are."

Garrett also says the system's data architects created an indicator called unstable housing. If a patient's answer matches the location of any of the homeless shelters in the area, or if they said they get their mail at a general post office box--which often denotes an unstable home address--the system can use that indicator in conjunction with the medical risks to try to better understand the population.

"That's just kind of a baby step, but I feel it's a huge emerging area of where we need to do a lot more research," she says. "How do we collect the data on those social factors and then how do we use it in our risk models?"

Demographic Factors a Nationwide Concern

The questions surrounding how best to collect and make use of risk factors outside the clinical setting go far beyond Hennepin Health, of course. Garrett was a member of a recent National Quality Forum panel that investigated these very factors. Ignoring socioeconomic status, and failing to create new care matrices that take factors like homelessness, food insecurity, and inconsistent medication regimens into account, she says, could lead to troubling disparities in the coming age of incentives and disincentives for readmissions and "redundant care."

The panel received more than 650 comments, and Garrett says the majority of the provider community that commented was in favor of including socioeconomic factors in risk adjustment formulas. Ultimately, she says, the NQF did not adopt the panel's full recommendation--that every time the Forum endorses a measure, whether it is appropriate conceptually and empirically to do that additional risk adjustment for the social and demographic factors should be considered, and if it is, it should be done. Instead, the Forum recommended a "robust trial period" of evaluating the effects of including such data. While the Forum's stance does not officially change the policy, Garrett says "it's good, it pushes the envelope."

Such advances in data integration will have to become de rigueur to fully and efficiently serve the vulnerable populations in such Medicaid ACOs, she says.

"To be able to do those population health measures and to do the program evaluation to understand what interventions are working and what aren't, you need the cost data from the health plan combined with the clinical data," Garrett says. "Then, bringing in the social services data is one of the most innovative parts of what we're doing. We're just scratching the surface of how that's going to be useful."

Register or login for access to this item and much more

All Health Data Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access