As the chief operating officer for TriHealth Corporate Health, Terri Hanlon oversees a team of health coaches who work with individuals to help them adopt — and maintain — healthy behaviors.
“As we all know, not everybody wants to engage in healthy behaviors on a regular basis consistently,” she says. “So this nagging question in my mind has always been: What really motivates folks who do what they do?”
She doesn’t have the total answer, but she is one step closer, courtesy of a consumer segmentation strategy that arms health coaches with new insight into the individuals they work with. TriHealth, based in Cincinnati, is a pioneer in the use of consumer segmentation to improve patient engagement in their health status and health care.
Another pioneer is Novant Health, a four-state network based in Winston-Salem, N.C. Its marketing team is developing a segmentation strategy that will inform the system’s interactions with its patients. On the one hand, segmentation is all about the patients; on the other, it is a key to Novant Health’s population health management program.
“If we better understand our segments so we know what their preferences are and how they want to be engaged, we can become more aligned with their needs,” says Jesse Cureton, chief consumer officer at Novant Health. “The other side of that coin is about the clinical care. Better engagement correlates with our ability to provide better health care and ultimately improve the health of the communities that we are in.”
Of course, health systems routinely segment patients for various purposes, such as insurance status, risk stratification or marketing. What’s new is combining previously untapped data and information from a wide range of sources, including patients themselves, and analyzing it to support more customized — and effective — care delivery.
This kind of consumer insight research is standard practice in other industries, says Dan Clarin, senior vice president at consulting firm Kaufman Hall & Associates. He previously worked at Walgreens, one of many retailers that are pushing rapidly into health care delivery and have deep knowledge about their customers’ preferences and buying habits. Traditional provider organizations will need to catch up.
“There’s almost an air of inevitability that segmentation is going to have to come to health care,” he says. “There are so many case examples in other industries that providers can look to and see that it has worked really well. So why aren’t we doing it?”
Coaching them up
TriHealth Corporate Health tiptoed into consumer segmentation with a three-month pilot program involving 210 individuals who work with health coaches to manage either diabetes or musculoskeletal disorders. Based on that experience, Hanlon is eager to expand into a second phase.
“We’re taking this to all of our health coaches, both in condition management and in wellness coaching,” she says. “This will reach about 30 individual coaches and probably about 3,000 individuals.”
Through this work, TriHealth Corporate Health classifies each individual into one of five psychographic segments, based on the participant’s responses to a 12-question survey. Members of each segment — direction-takers, balance-seekers, willful endurers, priority jugglers and self-achievers — share certain attitudes or motivations about health, wellness and interactions with health care providers.
That set of five segments was identified by a vendor, c2b Solutions, based on a 400-question survey of about 9,000 Americans from a range of demographic and socioeconomic groups. After the segments were established, c2b analyzed the survey data to develop 12 questions to assign individuals to the right segment.
TriHealth’s coaches are trained to use their knowledge about segmentation to tailor their communications. “If someone is a balance seeker, you have to give them choices, and you cannot tell them what to do,” Hanlon says. “But with direction-takers, you need to say, “I want you to do this,’ and they will take that and go with it.”
The pilot program demonstrated that the consumer-segmentation strategy improved goal attainment, which is how health coaches measure success. It also improved satisfaction for both participants and coaches. “Where coaches in the past had been frustrated and thinking, ‘Oh, what do I do for this person? I’ve tried everything,’ ” all of a sudden, those frustrations were being resolved and eliminated, and the interaction became a win-win,” Hanlon says. “The coach felt good about her intervention with the individual, and the individuals felt good because they were meeting their goals.”
When coaches learned about the segments, they quickly realized they were most comfortable working with individuals who were in their own segment. Their success improves when they learn to “speak in a different language,” as Hanlon puts it, to communicate with individuals who see the world differently than they do.
Another surprise finding: More of TriHealth’s participants fell into the “balance-seeker” and “self-achiever” segments than the general population, as identified by cb2, while the other three segments were underrepresented.
“That told us that we may not be attracting individuals who would really benefit from the program, because we are probably not marketing it in the language that they would react to or engage with,” Hanlon says.
Better care management
Carolinas HealthCare System, operating in more than 900 locations throughout North and South Carolina, is using an entirely different approach to segment its patients.
Avery Ashby, assistant vice president at the health system’s Dickson Advanced Analytics group, leads a population health analytics team of data scientists, statisticians, business intelligence specialists, epidemiologists and others. Using data from a three-year period, he and his colleagues have segmented CHS’ 2.2 million patients into seven segments to inform the health system’s care management strategy.
“We are getting ahead of the curve to make sure we have the infrastructure to be attractive to those commercial payers that may want additional care management services that for so long they have had to provide on their own,” Ashby says.
The team aggregated data from CHS’ electronic health record system and its billing system supplemented with behavior, consumer and geospatial data that identify the census tract in which each patient lives.
Instead of segmenting patients by cost or diagnosis, the analysts used a sophisticated hierarchical cluster analysis to “let the data talk for itself,” as Ashby puts it.
That means that the team was not looking for a certain number of segments or trying to sort patients by a set of preconceived assumptions. Rather, the analytical techniques sorted all patients into seven mutually exclusive groups. In six of those groups, patients “clustered” together because of traits they shared:
Advanced cancer: These patients, who account for 0.6 percent of CHS’ patient population, clustered together on the basis of their high billed charges (a proxy for resource utilization) and cancer diagnoses. (Patients with less serious or less costly cancer diagnoses may not fit into this group because they are more similar to the clinical characteristics in another segment.)
Complex chronic: These patients — typically older, low-income, covered by Medicare, widowed, and struggling with behavioral health problems and multiple chronic conditions — account for 6.6 percent of the patient population.
Aging rising risk: Younger than patients in the “complex chronic” segment, these people typically are married, commercially insured and have just one chronic condition, such as Type 2 diabetes, that is under control. They make up 16.7 percent of the patient population.
Mental health: A large percentage of patients have a mental health diagnosis, of course, but the 0.1 percent who clustered to create this segment share some specific commonalities. They are older teens and young adults, most of whom are male, who have serious mental health conditions that require a great deal of support.
Pregnancies and deliveries: These healthy women make up 2.6 percent of the system’s patients.
Newborns and toddlers: In general, these are healthy children under age 3 who account for 2.5 percent of CHS’ patients.
Together, the patients in the first four segments make up 24 percent of CHS’ patients — and 76 percent of billed charges. The seventh segment — dubbed “sparse information, acute and well” — accounts for 70.8 percent of patients and only 17.4 percent of charges.
That segmentation allows for care management protocols that target specific slices of the patient population. For example, CHS knows that around 145,000 patients fall into the “complex chronic” segment, which is the second most expensive segment. That’s an unmanageable number, Ashby says. But having so much data about those patients in an enterprise data warehouse lets the analysts identify subsets — for example, those complex chronic patients who have increasing year-over-year charges, have a certain combination of diagnoses, have had avoidable utilization and belong to certain demographics — that warrant specific attention.
“So, let’s come up with a plan for that microsegment within complex chronic because they’re all similar,” Ashby says. “These are the conversations that we have with clinical and operational leadership when we’re all looking at data together. We’re able to answer their questions about our patients in almost real time.”
Informing strategic decisions
Cureton came to Novant Health from the banking industry, where segmenting customers for a variety of purposes is standard practice.
“Historically, health care has been an industry that is considered a perpetual business because people will always need health care—and segmentation may not have been considered as a strategy to inform products,” he says. “But, with the advent of the Affordable Care Act and the increasing complexities within health care, the need to improve outcomes and the need to reduce cost, segmentation becomes very important.”
To support its population health strategy, Novant Health, which serves some 4 million patients in Virginia, North Carolina, South Carolina and Georgia, aggregates a wide range of information about its patients into a single database. Cureton ticks off a long list of details the system is collecting that goes far beyond clinical and demographic information.
“We know the number of prescriptions that they have. We have a sense of their income. And we also have a sense of their willingness to pay for quicker or easier access,” he says. “We know the level of education, the number of children in the household— just tons of information that deepens the level of knowledge that we have.”
Cureton foresees the day that asking patients about how they want to access care and engage with the health system will become routine.
“We’re not there today, but what I see is that, in addition totaking information about a patient’s clinical condition — how you’re feeling today, how often do you exercise, what stress do you have in your life — our physicians will ask questions about preferences,” he says. “That will be a way of continuing to increase our information.”
is already using its trove of information to segment patients into six distinct groups. The segmentation strategy is proprietary, but one of the organizing principles is patients varying levels of engagement with health care. One segment includes individuals, many of whom are caregivers for other patients, who are highly engaged with the health care system atients in another segment tend to avoid interacting with health care providers unless it is absolutely necessary.
Cureton is equally interested in both segments. Highly engaged patients are good customers to have — while individuals in the highly unengaged segment pose a challenge for population health management.
“That’s the segment of the population that very likely could have chronic illness that will become very costly for the health system,” Cureton says. “We want to get them engaged because we want to be proactive. Although this is a very small population, we have to understand who they are, where they are, and we have to begin to do things to get them thinking about our health care system and their health.”
Among other things, Novant Health will use its knowledge about patient segments to make decisions about access points that give patients what they want. Some patients want early-morning appointments at traditional medical clinics; others want round-the-clock access via telemedicine; and others want a convenient location for a walk-in visit on a weekend.
“As we become smarter about our patient population, all of that understanding begins to inform strategically about the kinds of products and offerings we make to grow our business,” he says.
Most health systems have more patient data in-house than their leaders may realize, Clarin says.
“We see systems that have a lot of information but are not necessarily sure how to use it to better understand consumers,” he says. “That’s the first place to start.”
For example, organizations may be able to identify which patients typically do or do not comply with physician recommendations or referrals, which may lead to insights about patterns of patient behavior.
Next comes segmentation, and Clarin says using one of the global market segmentation schemes is a good first step because it documents the fact that health care consumers are not homogeneous.
“It begins to build more of a consumer-centric focus for health care organizations and helps with the (internal) education that there’s not just one consumer,” he says. “There are different segments that have different needs and wants as they relate to health care decision-making.”
From there, consumer insights can be applied to specific challenges, such as planning an ambulatory strategy, overhauling primary care access or reworking any other offering that needs be more consumer-oriented than it has in the past. The goal is for the health system to change its ways — so consumers change theirs.
“We can use these insights to determine what types of incentives — financial and otherwise — are going to be meaningful for consumers based on their attitudes and their motivations and their fears,” Clarin says. “What types of incentives are going to get patients to actually think about changing their behaviors? That’s what is exciting here.”
Four Ways to Apply Segmented Consumer Data in Health Care
Using new kinds of data to segment consumers in new ways has many applications in health care, says Dan Clarin, senior vice president at consulting firm Kaufman Hall & Associates. Those include:
1. Engaging patients for purposes of population health management. A disease registry will provide a list of patients with diabetes, but those patients are hardly homogeneous. “Layering information about attitudes and motivations and fears on top of that clinical information to understand what is the best way to communicate with and engage with different types of patients is definitely an opportunity to improve their care,” Clarin says.
2. Messaging. Knowing a consumer’s behaviors and attitudes helps health systems to know what type of message likely will produce a response.
3. Service distribution strategy. “Understanding which consumer segments are more prevalent in certain neighborhoods could help you determine what types of providers and what types of services to place in different parts of your market,” Clarin says.
4. New service or product development. Recognizing that pregnant women are not all alike, MultiCare Health System in Tacoma, Wash., invites expectant moms to choose their own mode of care. The standard regimen of in-office prenatal visits and classes generally appeals to first-time mothers, while second-time and stay-at-home mothers may opt for some group visits led by a nurse practitioner. Working mothers, meanwhile, tend to like completely virtual visits with an NP, supplemented with standard ob-gyn visits when needed.