Identify high-risk populations early to cut costs, improve care
Identify high-risk populations early to cut costs, improve care
Get a jump on patients’ needs with sophisticated screening tools
With 36,000 members enrolled in its Medicare service line and 600 more coming on board every month, chronic care managers at Portland, OR-based Providence Health Systems knew they needed to develop a system capable of quickly identifying patients at high risk of using health services.
By conducting routine initial screenings designed to assess new patients’ risk of utilization, staff could more effectively target patients for early interventions before costs got out of control.
"We really started with a population management approach," says Cindy Klug, MST, CHES, director of Providence’s regional chronic care program. "So our risk screen was developed as a case finding tool. We wanted to target those people that case management should be aware of, look at the resource needs, and be better able to coordinate care for those high-risk people."
Such an approach has become crucial in the managed care arena, particularly among older populations, says Chad Boult, MD, MPH, associate professor at the University of Minnesota Medical School in Minneapolis. "Ten percent of an older population typically accounts for about 70% of the health care costs of that population," Boult says. "So, if you’re part of a health plan and especially if you’ve entered into a Medicare risk contract where you’re responsible for paying for all of the members of the population, it’s certainly to your benefit to know in advance which 10% are going to cost you 70% of your payments."
Most approaches to risk identification begin with a screening instrument, typically a questionnaire given to new enrollees either by mail or by phone. Most of these instruments use questions that have been determined to be good predictors of an enrollee’s future use of health care services.
Based on the initial screening tool, enrollees typically are separated into low, medium, and high-risk categories. "There would usually be an algorithm or triggers already embedded in the scoring process," says Deborah L. Paone, vice president for member services at the National Chronic Care Consortium (NCCC), based in Bloomington, MN. "So that if someone has had three previous hospitalizations in the last 12 months, has three chronic conditions, and takes seven different medications, all these risk factors would pop up. And the combination of all those things equals high risk."
Initial screening alone won’t do
Paone cautions, however, that while a quick initial screen can raise red flags about a particular patient, by itself it can’t provide enough information on which to plan an effective intervention. What’s needed is a more in-depth follow-up assessment capable of giving a thorough breakdown of the patient’s medical needs. (See risk identification table, p. 108.) At this point, patients are often stratified according to their chronic conditions, Paone adds.
In developing their own risk identification plan, NCCC administrators pooled information from member organizations, collected information from the literature, examined various examples of risk-identification tools, and compared notes with physicians, says Paone. "The literature, as well as physicians’ own tools and experience, pointed to the fact that there are certain key pieces of information that seem to predict individuals who are at higher risk for adverse outcomes," says Paone. (See flowchart, p. 107.)
Interestingly, the best predictor of high risk is the enrollee’s own assessment of his or her health status, says Paone.
"If you ask a group of people [to assess their health], and then ask their doctors, who have just done a history and a physical on them, to predict the likelihood that the person will survive the next five years, the person’s own self-rated health is a better predictor than an informed physician’s estimate of mortality," says Boult. "We’ve found that it’s also a good predictor of use of health services."
A disease management approach to risk identification typically involves risk stratification or the separating out of enrollees by chronic condition, says Paone. "It’s a similar kind of thought process, but the tools would be very different," she says. For example, the terms low, medium, and high risk take on a somewhat different meaning when applied to a population of congestive heart failure (CHF) patients than when applied to a general elderly population.
"Low risk in that instance might refer to people who are exercising, have their weight under control, and are properly monitoring their symptoms," Paone says. "Whereas there are others who, despite their best attempts, remain unstable and have high utilization of services. So there’s still an opportunity for primary, secondary, and tertiary prevention within a disease management strategy."
The problem with the risk stratification approach, Paone contends, is the likelihood of comorbid conditions. "What if a patient has CHF, diabetes, and arthritis?" she asks. "How do you layer and intermingle the various tools, the approaches, the care teams? Can they coordinate all these much more in-depth assessments and disease-specific interventions?"
Klug concedes she’s still looking for answers to those questions at Providence. "Comorbidities represents the biggest single issue I’m dealing with right now," she says. We’re being inundated with all these disease state management companies, and I ask every single one of them the same thing: How do you deal with comorbidities?’ Because, basically, I’m seeing programs where a patient could end up getting four or five different approaches, types of information, and care managers."
The approach Klug takes at Providence is to incorporate disease management into the standard risk identification process. "For example, from our risk screen, we could print out a list of all the people enrolling in the plan who are diabetics," Klug says. "So you could do a disease state management program for them from that listing or hook them into your education or whatever types of interventions you have available. What we’re hoping to do is some cross-coordination of the programs so that we can better target diabetics with heart disease or diabetics who smoke and develop more focused interventions for them because you have both sets of information."
However high-risk patients are identified, Paone stresses the importance of having in place an information system capable of quickly disseminating screening information to the proper care providers. (See Risk Identification Components and Goals, at left.) For example, by using a networked database, information from the questionnaire can be added to medical records electronically. "That way, the next morning, after the information’s been added, the primary care office can turn on its computer, and up pops risk-assessment information for a particular physician’s panel of patients," Paone says.
But deciding how to organize and implement an effective risk identification system can be a complex process, Paone adds. Questions to ask include:
• What system or software will be used?
• Is there a computer scanner to scan returned questionnaires?
• Who should do the follow-up?
• Who should conduct the risk screen?
• Under whose control is the whole risk identification process?
"You want to be sure you have the technology, the information system, and the scanner," says Paone. "You also want to be careful about creating a stand-alone system that isn’t linked to the core database, especially if you already have some electronic information transfer over multiple sites. For everyone to get the information they need out of the system in a timely fashion, obviously, there’s a lot of hardwiring and linkages that need to happen."
Computer reads questionnaires
At Providence, new enrollees return their risk screening questionnaires by mail to the plan’s outcomes research department that uses a computer scanner to enter the information into a system-wide database. In addition to recording the information, the computer scores the questionnaires according to a pre-established formula (for example, a poor self-assessment might be given more "weight" than another factor). According to the resulting scores, the computer assigns each enrollee to a risk category.
Finally, staff in the research department use the database to generate patient reports that include such information as health care utilization in the past year, overall health status, and social support. Reports on high-risk patients are then disseminated to the appropriate primary care physicians and case managers. "That way, when the case managers and physicians hold their weekly meetings, they can staff the people who have been identified in the risk report and put a plan of care in place for them," Klug says.
Be sure info is up to date
Paone stresses, however, that such reports, however valuable, only represent "a snapshot in time." Accordingly, it must be obvious to the reader where the information came from and when it was collected. "If someone categorized as low risk has been in the ER or experienced two hospitalizations since the questionnaire was done three months ago, how useful is it at this point?" she says. "It’s important to be capable of having some red flags come up on the screen that say Warning, this low-risk person has been hospitalized twice. A re-screening may be necessary.’"
As risk identification becomes more common, Klug anticipates other problems as well particularly the prospect of caregivers becoming flooded with patient reports from a variety of insurance plans. "I think there’s a possibility that primary care physicians could be getting five different types of reports back because they’re working with five different carriers," she says. "So I think the next issue to tackle will be working among the health plans to standardize reporting methods. That’s one of the biggest emerging issues: As more people do this, how do you coordinate it from the physician’s perspective?"
[For more information on risk identification, contact:
Chad Boult, MD, associate professor, University of Minnesota Medical School, Minneapolis. Telephone: (612) 627-4686.
Cindy Klug, MST, CHES, director, regional chronic care program, Providence Health Systems, Portland, OR. Telephone: (503) 215-3729.
Deborah L. Paone, vice president for member services, National Chronic Care Consortium, 8100 26th Ave. South, Suite 120, Bloomington, MN 55425. Telephone: (612) 858-8999.]
Subscribe Now for Access
You have reached your article limit for the month. We hope you found our articles both enjoyable and insightful. For information on new subscriptions, product trials, alternative billing arrangements or group and site discounts please call 800-688-2421. We look forward to having you as a long-term member of the Relias Media community.