By Brian Schilling
Three years ago, when Deanna Scott, vice president of human resources and corporate operations for the Scooter Store was busy establishing the organization's wellness program, a report came across her desk which, in a disturbing way, confirmed the "ordinariness" of her firm. It showed that 82 of the company's 2,700 employees accounted for about half the organization's overall annual claims.
Scott wasn't surprised. Nationally, the most expensive 5 percent of patients generally account for about 50 percent of all health care costs, but seeing the report still set off alarm bells.1
"Even with blinded claims data, when you're looking at a summary of the high-cost cases for your own firm, you can't escape thinking that there are real people behind those dollar figures who need help that they may not be getting," says Scott.
The Scooter Store prides itself on being a great place to work— the manufacturer of scooters for the mobility impaired was twice ranked by Fortune as one of the nation's 100 best workplaces. Scott's first reaction upon seeing the data was to commit herself to making sure the organization's wellness program would make a real difference to employees.
Claims Data Paint a Picture
The report that Scott received, which showed a decidedly lopsided distribution of health care costs among her employees, came from WellNet, one of a growing number of vendors that offer claims data analysis and predictive modeling services. WellNet President Keith Lemer explains, "If companies have access to claims data—and every company really should—data analysis and predictive modeling can tell them what their top health care issues and costs are going to be down the road."
A 2011 New Yorker article showed the potential of using those data to cut costs. The article profiled the experiences of an initiative in Camden, New Jersey that used outreach to very sick patients to cut the hospital bills of 36 so-called "super-utilizers" by 56 percent. A similar effort in Atlantic City, New Jersey cut medical costs among casino and hotel workers by an estimated 25 percent while markedly improving the health of the more than 1,200 patients involved in the effort.
But getting claims data can be hard and expensive. When asked about the difficulty in getting health plans to share data or provide real-time access, benefits consultant Bill Lavis of the Oakland-based firm Sitzman, Morris and Lavis, Inc. said, "It's a huge headache. You'd be surprised how many firms sign contracts saying that they don't actually own the data for their own employees. Plans really want to hold onto it in many cases. They don't necessarily want you to be able to see the difference between claims costs and your premiums." He went on to elaborate that self-insured employers are likely to have an easier time getting their own data, but that they should expect to pay for it.
Claims data are blinded, of course, so reports from vendors don't ever say which employee has what problem, but they can still be fairly specific. "We can tell you you've got X people with diabetes, Y with serious heart conditions, and Z with psychosocial issues," said Lemer.
Lemer says the most important data are pharmacy data. "Pharmacy data are available immediately, they're unambiguous, and they're highly accurate," he says. "For medical data, you might wait 60 days before you see a claim and by that time the subsequent cost you might want to try to prevent has already occurred."
Finding Candidates for Outreach
WellNet uses the Scooter Store's pharmacy data to look for good candidates for personalized outreach. For instance, if a patient has a claim for hypertension, then fills a prescription for diabetes medication, and later seeks reimbursement for an inhaler, that patient gets flagged as having a high probability of experiencing serious medical conditions in the future, such as a heart attack or stroke. "If you can prevent that person from having a heart attack by helping them up front," says Scott, "you've done some real good and you're also saving a ton of money."
Many serious health conditions can be recognized by patterns in the claims data archive. Wellnet and other firms use software to recognize these patterns and engage members at the earliest possible point.
Other claims patterns are also apparent—a diabetic not filling her prescription routinely, a patient with a heart condition who hasn't had a stress test in two years, a recently discharged patient with substance abuse issues who has missed all follow-up visits. Such patterns allow WellNet and similar companies to fairly accurately predict the health issues that are likely to crop up in the future. "With enough data," says Lemer, "we can provide you with a very good forecast of what health problems your employees will be facing over the next two to five years."
In a frank moment, Lemer acknowledges that such reports are not useful on their own. "Many analytics firms can do predictive modeling and do it well," he says. "It's what you do with the reports that makes a difference."
The Scooter Store's Approach
With the support of management, Scott mapped out an aggressive wellness program with many different components, all working together under the firm's "Live, Work, Be Well" initiative. The program is designed with all Scooter Store employees in mind, not just the most expensive "high users." Highlights of the program include:
Scott says that what makes the program work is not simply that good care is available, but that the Scooter Store markets it aggressively.
"We don't pull any punches when it comes to getting people to come in for the care they need," says Scott. "We're comfortable being very aggressive in that regard. I'm not afraid to call someone who hasn't responded to repeated contacts from our vendor to say, 'Hey, I don't know what you're dealing with, but we pay those vendors to make sure you get the care you need. They're on your side. Please call them back."
The response, she says, has been overwhelmingly positive.
The art of encouraging beneficiaries to participate in available disease management or other programs must be practiced in a way that complies with the Health Insurance Portability and Accountability Act of 1996, which makes it illegal for employers to receive identifiable health information about employees (See Related Article.). But the law doesn't bar third parties from doing essentially the same thing.
In the case of the Scooter Store, that third party is WellNet. "The value that we sell is highly targeted member engagement," said Lemer. The engagement rate at the Scooter Store is around 35 percent, which is quite high by industry standards."
To achieve that level of participation, WellNet relies on everything from kiosks at the company health fair to newsletter articles to text messaging to phone calls to mailings. "There's no one approach that will work for a given population. You can't reach a 25-year-old the same way you reach a 55-year-old. The messages are custom, the medium is unique, even the timing is different."
For the Scooter Store, the key to effectively promoting the wellness program has been to work with WellNet to segment its employee population across several different strata, including age, income, gender, retired/working, etc. Each distinct segment gets its own unique pitch with messages, incentives, and penalties designed with them in mind.
"Blanket approaches to outreach across an entire population aren't as effective," explains Lemer. "Highly paid executives won't respond to a $150 incentive the same way a younger (and less highly compensated) employee will. But give them a $150 penalty for not filling out a health risk assessment and the response rate might hit 70 percent."
Scott recalls feeling somewhat anxious about pushing the program too hard when it was first launched. "We were a little fearful of employees reacting from a Big Brother standpoint: 'You're invading my life. You're going to know everything about my medical condition, and that's way too personal," she says. That was never the case, of course. Scott has never seen personally identifiable health information about her employees and doesn't want to.
Two years in to the effort, Scott is no longer anxious about pushing the program too aggressively. "At this point I think what we really need to do is just communicate more often and in different ways and do a better job of promoting it at health fairs and things like that, which we're now doing."
Churn at the Top
It is a little-known fact that there is a high degree of "churn" among high-cost individuals. That is, people who are high utilizers of health care one year may not be high utilizers the next year. A 2003 study found that of all people ranked in the top 5 percent in terms of health care expenditures, only 34 percent were again in the top 5 percent the following year. In other words, roughly two-thirds of the highest-cost individuals for any given year were low-to-moderate users of health care the previous year.3
For employers hoping to prevent health issues from becoming more serious (and more expensive) in the future, the implication is clear. You need to monitor claims data on close to a daily basis or you'll constantly be playing catch up. High-cost employees are not a static group.
The Scooter Store does monitor claims daily, albeit through a vendor. "If someone fills prescriptions for a beta blocker, cholesterol medication, and a diuretic, it is a good indication that they would benefit from some communication directing them to the resources of our wellness program. If our vendor learns about that the next day, they can really do something with that information," says Scott. "But if they learn about it at the end of the year, they've missed the window."
Because the ongoing data monitoring is HIPAA-compliant, Scott is out of the loop on a day-to-day basis. "We have WellNet coordinate with our large case management vendor and they are empowered to work together without our intervention. We just want to see that people who need to be managed are managed, and that's clearly happening."
For the Scooter Store, the savings have been striking. Year-over-year claims costs from 2009 to 2010 went from $6,600 per employee per year to $5,400, a 22 percent reduction that translates to $2.3 million in savings.
But Scott says those aren't the results that matter most. "One of my managers who has struggled with weight issues for a long time sent me an e-mail recently saying that he started taking walks in the evening as a result of all the outreach we did. He ended up losing 103 pounds over the next year."
At the Scooter Store, Scott believes the best is yet to come. "It's hard to say which piece of the puzzle saved what amount of money. But in the aggregate, we're off to a great start."
1 A. C. Monheit, "Persistence in Health Expenditures in the Short Run: Prevalence and Consequences," Medical Care, July 2003 41(7):53–64.
3 Ibid. The same study explains that while care for chronic conditions, such as diabetes and asthma, is indeed fairly expensive, such care alone will not put individuals with those conditions at the very top of health care spenders. Acute episodes or complications, by contrast, certainly will.