In 2002, a young primary care doctor in Camden, New Jersey, began working with the city’s police department after witnessing a shooting near his home. Using the department’s data, he found there were “hot spots” in the city that were responsible for a large portion of its crime.
With healthcare costs rising, brokers and employers want to identify their largest sources of spending.
A Colorado study found the highest-cost patients were those with terminal cancer and those receiving emergency dialysis.
Some experts believe predictive analytics can help determine future high-cost users; others say such projections are more difficult to achieve.
Read the Sidebars
After he unsuccessfully attempted to change Camden’s policing around these locations, the doctor, Jeffrey Brenner, transferred this knowledge to his own field. Brenner aggregated data from area hospitals and found a similar phenomenon. People from just two of the city’s neighborhoods, one with a large nursing home and another with a low-income housing complex, accounted for more than 4,000 hospital visits and $200 million in healthcare costs.
Brenner had found his own hot spots. Just a small group of patients with unmet healthcare needs were accounting for an enormous percentage of Camden’s healthcare costs.
Out of his work was borne the Camden Coalition of Healthcare Providers, which began identifying patients who had been to the hospital two or more times in the past six months and were battling social complexities that might be impacting their health. They sent out care teams to meet people at their hospital bedsides while in crisis. They determined their needs, then provided short-term, wraparound services such as housing assistance, transportation to doctor’s appointments, and substance use assistance to help them better manage their health. For his work, Brenner was awarded a MacArthur Fellowship—the so-called Genius Grant—in 2013.
Like Brenner’s work, much of the research surrounding these superusers has been focused on lower-income people or Medicaid recipients. But it’s not just these populations that have a top tier of patients responsible for well more than their share of costs.
According to the Agency for Healthcare Research and Quality, 5% of patients in the United States account for 59% of all healthcare costs. Dr. Ronald Leopold, chief medical officer at Lockton Companies, says among its 1.67 million covered client employees, 3% account for 56.5% of healthcare costs. And 1.4% of that population spends more than $50,000 a year.
“We are increasingly seeing more and more of our clients’ overall dollar spend jam-packed into a very small number of members, and that trend is growing,” Leopold says. “If you are looking for cost mitigation, the real runaway costs are in this population.”
With healthcare costs consistently on the rise, being able to identify where a large portion of the spending is going is a valuable skill for any broker or employer. But it’s not an easy thing to do. It takes access to gads of data, followed by careful analysis, and then a range of solutions to meet the needs of the population. It may not be simple, but some groups, such as Lockton and the Camden Coalition, have found ways to improve care for these superusers while realizing healthcare savings.
A group out of Denver Health and the University of Colorado School of Medicine performed a comprehensive study of this medically complex, high-cost population in 2015. They analyzed records of more than 4,500 publicly insured or uninsured superusers at an urban safety-net system over a two-year period. They found 3% of adults over that time met superuser criteria, accounting for 30% of the adult healthcare costs in the system (costing more than $113,000 per capita).
They also found the top 3% tended to have the same demographics, health status, payer source and spending. A vast majority had multiple chronic conditions, and nearly half had a serious mental health diagnosis.
The highest-cost patients were those with terminal cancer and those receiving emergency dialysis.
What was even more interesting about the discussion, however, was that the superusers weren’t the same people from year to year. Fewer than half were still superusers just seven months after being identified as one, and even fewer were superusers a year later.
“What we found was dramatic,” says Tracy Johnson, director of healthcare reform initiatives at Denver Health and the study’s lead author. “The population nearly turns over in a two-year period.”
And though this was a group of Medicaid patients, private insurers are likely to find similar results. It can be called regressing to the mean—or what essentially amounts to the ebb and flow of an individual’s health.
Dr. Alan Glaseroff, an adjunct professor of medicine at Stanford, says it’s normal for about two thirds of superusers in any population to reduce their healthcare spending the year after they are in that group. It makes sense, he says, that someone who has uncontrolled diabetes and needs to have surgery or another costly intervention won’t have those same charges again immediately.
“The person doesn’t change,” he says. “They will just have periods that cost more or less.”
Identification Is Critical
The revolving character of this population is partially why identifying them can be like hitting a moving target. Each organization’s complex, high-cost group is going to look slightly different, so it’s important to choose a segment and tease out who among them may be able to receive better, low-cost care.
Lockton breaks groups into three sections: people whose annual healthcare costs run between $25,000 and $50,000, those who fall between $50,000 and $100,000, and the $100,000-plus spenders. While Lockton tries to reduce spending in the upper echelon, Leopold says the company focuses mostly on the $50,000-$100,000 group because of the higher likelihood of improving preventable conditions, such as diabetes, and thus lowering costs.
One in three people in the $50,000 group can trace their healthcare spending to hospitalizations and attached complications. Among the rest, 23% have chronic conditions such as diabetes and congestive heart failure, 20% have experienced trauma and musculoskeletal issues, and 16% are being treated for cancer.
Dr. Eric Bricker, chief medical officer at Compass Professional Health Services, an Alight Company, says among its 1,700 clients, about half of the top spenders have unmanaged chronic conditions, mental health and substance use disorders, or some combination of both. The chronic conditions among the fully insured tend to be related to musculoskeletal problems, cancer and cardiovascular disease. At businesses with a younger workforce, maternity and its complications also rank high in spending.
Prediction Versus Real Time
There is typically one of two tracks used when it comes to identifying high-cost patients. Both tracks are performed with the help of some sort of proprietary algorithm or intensive data crunching.
The first is trying to determine risk in advance. For example, Glaseroff’s group employed predictive analytics to help determine future high-cost users. The group used Milliman Advanced Risk Adjusters, which Glaseroff says could predict with about 30% accuracy who was going to be a high-risk claimant next year. Milliman takes a person’s health information and analyzes factors like medications and claims to try to understand what spending might occur in areas like hospitalizations, ER visits and pharmacy.
We are increasingly seeing more and more of our clients’ overall dollar spend jam-packed into a very small number of members, and that trend is growing. If you are looking for cost mitigation, the real runaway costs are in this population.Tweet
In 2017, Aetna launched a new program called AetnaCare, in which the insurer works with accountable care organizations in New Jersey to identify and provide support to high-need, complex patients. Aetna uses its own algorithm to determine who might be high-risk.
Dr. Sunny Ramchandani, Aetna’s deputy chief medical officer, says the company runs data from a given population through its program to determine who has the highest risk scores. “We went to the ACOs and said, ‘We’ve found some high-cost folks in your market, and we can work with you to tackle them,’” Ramchandani says. “We’ve been able to lower healthcare costs for many of them.”
But some say predictive analytics in healthcare can be tricky. Mark Rosenberg, president of healthcare analytics benefits and HR consulting at Gallagher, concedes that healthcare is the toughest industry in which to use predictive modeling, because each patient is so very different.
“You can have five 50-year-old males brought up the same, living in the same place with the same condition, and they may all react differently to certain medications or treatments,” Rosenberg says.
That theory leads to a second way to stratify high-risk patients, which is to do it in real time, like the Camden Coalition. “Lots of insurers love working with predictive models and algorithms,” says Natasha Dravid, director for clinical redesign initiatives at the Camden Coalition. “We are real-time in our data. We are able to look at who was in the hospital yesterday and identify them when they are in those high-risk moments.”
The Camden Coalition uses hospital utilization as a proxy for high-risk patients—they search for people admitted to the hospital two or more times a month or those going to the ER more than six times in a few months. Dravid says this is a good option because they are able to connect with people at the moment their spending is high to improve care. And also because of the regression to the mean: high users today might not need as much care next year.
“We really think that looking at who is going to the hospital is where to start,” she says. “If they are there that much, something’s going wrong with their care. It’s about getting the right care for the right folks at the right time.”
Finding the Gaps
In addition to stratifying employees into cost groups, organizations also must determine the care gaps and other cost drivers.
There are always two pools in any high-cost group, says Jeff Hadden, partner and president at LHD Benefit Advisors. One consists of those with unpreventable conditions, such as the birth of a preemie. The second, which Hadden says accounts for more than one third of high-cost claims, is made up of those dealing with generally preventable conditions like diabetes or its complications. Catching these early is critical to reducing costs.
But even among the preventable conditions, identifying where to put resources can be a challenge. “Just because you have a group of diabetics, it doesn’t mean you can bring in a vendor and make them cost less,” Rosenberg says. “They may be high-cost but are already doing what they need to do—see their doctor, take medicine and track their glucose levels. You have to understand where the gaps are, and then you can bring in a solution.”
While Gallagher does work with clients’ higher-cost populations to make change, the brokerage really focus its efforts on the middle of the pack—those responsible for 30% to 35% of spend—because, Rosenberg says, they can have the most impact on that population.
The upper echelons can be difficult to change because their expenses may be incurred during an event like a premature birth or come from people with multiple complex conditions. For the latter group, Rosenberg says, Gallagher works to make sure employees are seeing physicians who are in-network (which can dramatically lower costs) and provides individualized case management to help them better navigate the system.
In the rare case there is a group of these employees with similar conditions, outside vendors like diabetes management groups may be called on to improve care.
Mary Delaney, president of Vital Incite, a population health consulting firm that works with employers, came from the healthcare side of things, where she learned that most employers didn’t know what kind of programs to use to improve their population’s health and that many advisors didn’t know if solutions they were putting in place were working.
But data like that which her organization crunches can create risk scores based on an individual’s disease burden. Using this can help identify which people are likely to become high-cost claimants and what resources they may need beforehand to keep that from happening.
“We can analyze every health plan to find where the waste is and needs are that, if they are met, could drive down future dependency on the healthcare plan,” she says.
According to one of the organization’s case studies, they were able to reduce emergency rooms visits and lower hospital costs per admission by almost 35% over a two-year period. Vital Incite did this by analyzing employee data, health plans and healthcare usage. To plug gaps in the system, it worked to increase use of an on-site clinic (where it completed new-hire physicals), connected employees with primary care providers, and implemented programs for diabetes and high blood pressure.
Some Simpler Solutions
There are a host of options for improving care and reducing cost among the top healthcare users. They begin with simple and free options, such as changing plan design to encourage employees to make better health choices. One possibility Rosenberg recommends is encouraging—and paying for—second opinions when a major diagnosis is given. He says this often leads to better decisions on potential treatments.
Or insurers could offer free prescriptions to people who manage their care well. Diabetes patients may get metformin at no expense if they get regular blood work and do a physical and eye and foot exams. These initiatives could be coupled with creative solutions like a communications campaign around a particular condition to encourage people to manage it better.
One of Rosenberg’s recent clients found the top three conditions for which people used the ER were headaches, sore throats and urinary tract infections. When Gallagher looked at the client’s plan, it realized patients paid only a $25 co-pay for emergency room visits. These conditions could easily be treated at a primary care provider, so the client changed the plan to deter employees from ER overuse. The co-pay was increased to $100, and the client saved $1.2 million in medical costs in one year.
“Members suddenly had some skin in the game and were thinking more about which provider to use even though it was just bumped to $100,” he says. “An answer could be as simple as that.”
Delaney favors requiring people to get annual physicals. First, she says, a physical can identify conditions earlier, preventing an employee from going into the high-risk group. They also can help people who are already diagnosed with complicated conditions from amassing significant costs down the road.
Hadden says just connecting people with a primary care provider can make a big difference in costs down the road. He says studies have shown that cancer tends to be detected earlier in people who have a good relationship with a primary care provider, mainly due to increased screening rates among these patients. (For more on the advantages of primary care, see the sidebar “Primary Care Benefits.”)
We went to the ACOs and said, ‘We’ve found some high-cost folks in your market, and we can work with you to tackle them. We’ve been able to lower healthcare costs for many of them.Tweet
Many of these solutions are good for employers who are apprehensive to try moving the needle even though they may want to reduce their costs. Typically, businesses with fewer than 500 employees will have a tougher time implementing programs that require much financial input on their part.
“It will be dependent upon the client and their appetite for change and ability to spend a little money to save a lot,” Rosenberg says. “They are spending a fixed amount of known money to hopefully reduce cost, but because it’s not guaranteed, they see it as a cost.”
Going All In
For larger organizations with more spending power, high-touch case-management options can be effective in better serving complex patients. Glaseroff was able to use the resources of Stanford to back his program, which he labeled an ambulatory intensive caring unit, or AICU.
The unit began by determining the top spenders in the workforce and interviewing them to create an individualized care plan. Among the top of the spending spectrum, the unit found high rates of diabetes and hypertension and some cancers and neurologic conditions, along with chronic obstructive pulmonary disease (COPD) and asthma.
Healthcare providers set goals for the patients and worked with them to ensure success. For example, Glaseroff says, if a patient says she needs to start exercising, the provider might prod her by asking what type of exercise she wants to do, when she would do it, with whom she plans to exercise, and how confident she is about following through. If walking is her goal, the provider might request she do a short walk the next day, just to get started. The day after that, the provider would call the patient to see how the walk went.
Glaseroff’s group averaged about one contact each week per patient via messaging, phone or in-person visits. They worked with pharmacy, physical therapy, diabetes educators, nurses and social workers to provide wraparound services. Providers were paid through capitation and received shared savings when a patient’s costs were reduced.
Ramchandani says Aetna’s program is similarly high-touch for a short duration. When superusers are identified, nurses located near the patients are dispatched to their homes. Seeing the patients face to face helps Aetna design a more personalized care plan, he says. This plan, or care map, offers health actions members can take to improve their chronic conditions, including treatment adherence, meeting social needs and improving lifestyle behaviors. The goal is to improve patients’ health while making sure they have the knowledge and skills to manage their condition after a two- or three-month intervention.
Aetna has measured its results on more than 100 patients, and Ramchandani says engagement rates are 65% (compared to the 20% to 30% industry average for this type of program), inpatient hospital use is down 70%, and per-member per-month cost is down 45%. Part of Aetna’s success, Ramchandani says, is because they “curate and integrate a host of ecosystem services.” This includes a medical estimator tool that helps plan recipients purchase more inexpensive medications and a partnership with a local grocery store where members can shop for groceries with a nutritionist.
Alcohol Under the Radar
Because of the tremendous costs of some of these top-tier individuals and the complexity of their conditions, Leopold says, he has seen a proliferation of vendors coming to the market offering programs that target specific high-cost populations. “There are a lot of third-party players and carve-out opportunities where employers can plug in cost-management solutions,” he says.
Annum Health is one of these. It was created by Michael Laskoff in 2008 when he realized there was one major population flying under the wellness radar: people with alcohol issues.
And there is a large unmet need here. According to Laskoff, one in four Americans binge drinks, and one in six does so more than four times a month. Two thirds of all people with a self-admitted drinking issue are employed, which he estimates costs the workforce $80 billion annually in lost productivity.
Most of the more than 16 million people who meet clinical guidelines for an alcohol-use disorder never get help for the condition, Laskoff says, even though there are thousands of addiction clinics across the country.
And even if they do, a vast majority relapse even with costly inpatient rehabilitation, which can run well over $30,000.
These grim rehabilitation statistics led Annum to focus on a modern treatment alternative for heavy drinkers. “The product has to be better,” Laskoff says. “Right now, it’s expensive, stigmatizing, inconvenient and generally ineffective. People would rather live with the problem than pay for the solution.”
Laskoff says a majority of people with drinking issues wish their employer would offer some sort of private, effective treatment option. But this can be a tricky proposition. To avoid HIPAA issues, Annum reaches out to all employees to see if they would like to take part in the service. This helps Annum and the employer avoid pointing fingers and allows employees who might be worried about seeking treatment to get the help they want.
People who enroll in the program are eligible for a year’s worth of treatment. During this time, the employee meets with a local social worker in six to eight live sessions. Employees can also meet via video with physicians who can prescribe medication or guide the employees to reducing or stopping their drinking. They also receive a year’s worth of support and coaching through an app and texts. And for those who want a social component, they can take part in private, online, moderated group sessions.
The goal is to track the amount of alcohol consumed and help employees drink less or stop altogether. Employees are monitored via the app, on which they answer basic questions each day: “Did you drink yesterday?” “If so, how much?” “Did you work yesterday?” “If so, how was your mood and productivity?”
Laskoff says the yearlong program is offered for about 75% less than the cost of inpatient treatment. Annum partners with the health plan to offer the program and gets paid only for patients treated. Laskoff couldn’t provide specific results at this point but says the program has shown to be two to three times more effective than traditional behavioral health interventions.
“Our results are terrific, and a lot of that is because we don’t dictate what success is,” Laskoff says. “They have a goal of zero unsafe drinking days per year…at the same time, they are reminded that lapsing is a normal part of recovery.”
Another vendor, Mymee, works with people who are severely ill with autoimmune diseases to see if they can stop taking expensive specialty medications.
About 24 million Americans have at least one autoimmune condition. Mymee doesn’t target the whole spectrum of specialty drugs for these conditions but instead focuses on the most costly ones, such as Humira and Enbrel, which treat a range of autoimmune disorders, including rheumatoid arthritis and Crohn’s disease. These are the two most expensive drugs on the market that treat these autoimmune conditions, costing $50,000 to $60,000 annually per person.
Mymee also works with people who are already on specialty medications. This enables them not only to improve patient care but also to track their results, as patients are able to reduce the amount of medications they take. The program is a 16-week digital therapeutic, which uses digital technology treatments including health monitoring, apps and virtual health coaching to treat health conditions.
Just because you have a group of diabetics, it doesn’t mean you can bring in a vendor and make them cost less. They may be high-cost but are already doing what they need to do—see their doctor, take medicine and track their glucose levels. You have to understand where the gaps are, and then you can bring in a solution.Tweet
Mymee is tailored to each participant and is a high-tech, high-touch, functional medicine approach to this population. They work to zero in on each person’s triggers—lifestyle, diet and environmental factors—so those can be avoided, helping reduce their symptoms. It begins with participants tracking their diet and activity for six weeks. Then they are onboarded through a call with a health coach to talk about how they are functioning. Each week they work with the coach to find ways to avoid problematic foods, activities or other triggers and, in this way, to make their body run better.
Mymee is a risk-based system for the organization. They work with self-insured employers and treat employees who cost $50,000 or more a year. Employers pay for the service only when employees begin to see results. They are planning to begin working with insurers in the coming months.
Mymee did some testing to prove its worth among patients with lupus. It focused in this space because there are not a lot of options for treatment among this population and treatments can sometimes be dramatic.
Patients can take steroids or immunosuppressants or have surgery to remove affected organs.
Their test was small, consisting of only 18 people, three of whom had experienced organ removal prior to working with Mymee. By the end of the study, eight of these patients were off some or all of their medications. All 18 also reported an improved quality of life with its use, says Mette Dyhrberg, Mymee’s CEO.
Mymee is currently working with people who have active, moderate to severe symptoms of lupus, rheumatoid arthritis, Crohn’s disease, inflammatory bowel disease and Sjögren’s syndrome. Mymee will also work with psoriatic arthritis and some less common autoimmune diseases, such as hidradenitis suppurativa, Mette says.
Delaney says the first lesson she learned moving from the healthcare space into consulting was that the results weren’t as good on almost every product out there as they expected them to be.
For this reason, Vital Incite sets very clear goals with vendors it works with—on-site clinics, diabetes management programs, wellness groups. Vital Incite reviews its specific expectations with the vendor (like a 1% reduction in A1c levels among uncontrolled diabetics) and determines whether it will help the employer meet its ROI. Then, Vital Incite measures the results and meets with vendors regularly to ensure those goals are being met. Another thing she learned, Delaney says, is that outcomes improve when performance guarantees are put in place.
Hadden cautions against jumping on board with each new product a vendor offers to improve the health of top-tier spenders. “There are a billion of them now, and they don’t all improve health outcomes,” he says.
“The problem is they are difficult to vet without putting them in place.”
Whatever option an employer picks to tackle the issue of complex, high-cost employees, Rosenberg recommends implementing a three- to five-year strategic plan on ways to address the issue. And be prepared for wins and losses.
“Different solutions can be short- or long-term, and clients all take different approaches,” Rosenberg says. “They just have to try to move the needle, because they are not going to be able to address the whole problem or even a majority of it. They just have to try to chip away at it. Look at better ways to deliver care and make sure people are getting it at the right place.”
Worth is a contributing writer. firstname.lastname@example.org