Next Generation Care Technology

In this episode, we explore the origins of care management, reactive versus proactive models of care, ways to coordinate care with real time data, practical uses for data and analytics for diagnosis and treatment, as well as the ideal care management dashboards for frontline healthcare workers, managers and directors of care management programs.

Host: Hi, Everyone; this is Episode 2, Next Generation Care Technology. I’m excited to welcome Dr. Chris Johnson to the show. Dr. Chris Johnson designs and implements health management solutions for government-sponsored health plans. Building on his academic training in medicine, public health, and medical informatics, he has pioneered new workflow methods for health management technologies. He co-founded Performance Clinical Systems to develop what is now Advantasure’s health management platform. He currently leads a growing team to continue innovating solutions for health payers and provider networks. Welcome, Dr. Chris.  

Dr. Chris: Thank you. I’m glad to be here.  

Host: You’re a physician, so you have unique insight into the provider side of health management, and now as a technology platform director and advisor for health plans in the government-sponsored health plan space. What do you see as a forefront problem in care management today?  

Dr. Chris: Let’s start by giving a historical overview; historically, care management was case management. That’s really the origins of where it began. In case management, we have a large member population as a health plan; we know that every year, a small percentage of these people are going to be high cost, high risk, and our aim is to help them manage their health risks better. Manage their diabetes so it doesn’t spiral out of control. Manage the fall risk in the home so they don’t fall and break a hip with an expensive replacement and then potentially having it result in pneumonia from being in the hospital. This is historically where it began—with case management starting with a small percentage of the population where you could intervene and help members avoid these catastrophic events. Over time, case management has evolved into care management as we, the industry, started asking how can we help not just that few percent at the margin rather how do we begin to influence the broader member population to help them have healthier lives, help them avoid costs, and avoid the associated suffering. If we can help prevent a member from getting high blood pressure that may eventually lead to a stroke, it’s good for resources, but it’s also good for the member and their family. That’s where the industry is going, towards care management. The challenge is that as you move into looking at the larger member population, you need better and better data. You need to be able to identify in the broader population who is likely to have an event and in what time frame. Going back historically, case management identified those who were already in the hospital and sought to manage that hospital transition. Or manage a person who has already had a ton of ED visits. So, it was a reactive model. Versus today, care management is proactive, and the only way to proactively identify and manage members is by leveraging data and analytics and developing predictive modeling. and the question we’re asking is, “How do we begin to point our teams— our nurses, pharmacists, social workers to the members before they get sick, before a catastrophic event.” This is where the industry is going, but we can only get there with strong technology support.  

Host: It’s such a simple concept of reaching people early on. We all know the extraordinary costs associated with preventable illness or something as simple as medication adherence. This proactive shift in healthcare can really do great things for healthcare costs and improving the quality of member’s lives. Let’s talk real-world examples. How are today’s care management teams working together as an integrated unit, and how is technology aiding this effort?  

Dr. Chris: Historically, RNs were the case managers, and still are today—as care managers. They were reliant on claims data to find the people to go and help. But the claims data from the health plan would finally arrive in the nurse’s hands months later. For example, they might notice a member has had 3 or 4 ED visits in the past few months, so they reach out to the member to connect them with a primary care doctor and help them build a relationship. But it was historical data, claims data alone, and from that, the case manager had to try to understand who to help in the member community.

Now, the industry is moving to care management and real-time data updates to the health plan so the health plan knows the minute one of its members hits the ED or is discharged from the hospital. Now, the health plan can intervene with the member while they’re in the hospital, scared and motivated to take action. That’s a lot more compelling than waiting for claims data 3 months later. That’s the power of real-time data.

Other valuable sources of new data streams come from EMRs, social determinants of health, and clinical lab data—utilization data. Care management teams are more effective when their interventions are driven by thorough data sets. “Oh look, I saw you were in the ED two days ago. I see from the EMR data that the member’s A1C values are rising, and social determinants of health data indicate the member is having difficulty getting transportation —it’s hard for them to get to the doctor or clinic. Now, the case manager has a better sense of what they can do to intervene. They can help the member get public transportation vouchers so they can get to their doctor. A1C is going in the wrong direction; let’s get this member plugged in with the right kind of clinic to see an endocrinologist. At the end of the day, the member is in the system, the A1C is better managed, and we’ve avoided future costs inpatient visits and ED visits. Most importantly, we’ve impacted the member’s life.  

To realize this vision and make it the new standard for care teams, we have to provide these care teams with integrated sources of data that are usable and in real time. The other thing we’re supporting is the unification of provider teams and health plans so they can work together.

Classically, a case manager had the claims data, and they were operating in a silo. They would call the member and say I’m here to help. The member would say I just talked to you guys yesterday. And the case manager would say, “Oh, I didn’t know.” Now, the mission is to bring those data streams together and ensure that everyone has a shared view of the member to provide the best care and the best member experience possible. So how do we do this? You must have a system that integrates care management, utilization management, risk analytics, care gaps, and social determinants of health into one system. So, it’s all happening in the same place. If care managers are integrated with utilization management in the same system when there’s a utilization management event like a wheelchair, the wheelchair is requested, and the wheelchair is approved, the care manager knows that immediately. They weren’t involved with that per se; it came from a provider authorization request. But now, when the care manager calls, they can say, “I see your wheelchair is on the way.” The member feels assured because they know everyone is on the same page.  

There are 3 points here to make and areas we’re continually improving on: how do you bring in more data from the outside? Clinical data, lab data, social determinants of health. How can we use analytics on that data to predict who will have challenges? And how do you fuse that data and ensure different members of the team—the nurses, pharmacists, authorization, and medical directors—everyone has a common view? This is what it takes to provide members with a better experience.  

Host: Let’s talk more about analytics; it’s a big area right now.  

Chris: Yes, it is. It’s also a topic that’s not well-defined right now. People say, “We’ve got analytics, or we’re adding analytics.” And that’s great. But it’s like saying the secret to being wealthy is to buy low and sell high. It’s true, but it’s not particularly helpful. The devil’s in the details. When someone says, analytics are the future. Ok, great. But tell me, how are you going to use analytics, which analytics, which business problems are you going to solve with those analytics? That’s where the real difference is made.  

How do you identify rising risk? And then what are you going to do about it? What’s the right intervention? I’m an internist originally, and I was trained that diagnosis and treatment are two different steps. First, get all the data, do the history, get the labs, and make your diagnosis. Stop. Separately, treatment. Given the diagnoses, how are we going to treat them? What are the steps? There’s a very sharp distinction. And I think a lot of this holds true for analytics. It’s an open and exciting frontier. You can have phenomenal analytics that do the diagnosis step to using those different data sources—clinical lab data, social determinants of health—and all kinds of nontraditional data sources. All of that’s great; you’ve got new machine learning AI models that can do predictive. But it’s still the diagnosis step—the first step. When I look at where the exciting opportunities are in analytics within this space, there’s a lot of room for innovation around the treatment aspect. All of this data comes in, the analytics streams, the models, and this tells me whom I should go after and what the problem is. But now, what do I do? What’s the most effective intervention? Who’s the most effective care manager to do that intervention? No one is doing this yet. But this is where the industry is going—analytics on the diagnosis side and analytics on the treatment side.  

Host: This is very exciting. Can you illustrate how this looks in a practical setting?

Dr Chris: Let’s say I have phenomenal analytics on the diagnosis side for care management, and I’m bringing in all of these different data sets. I’ve got modern machine learning AI models and phenomenally predictive algorithm scores that tell me exactly who to go after in my member population. Assume all of that. Now, there’s this vast open area where I can do analytics on the characteristics of my nurses. Who’s going to have the greatest rapport with one patient versus another? Within my care manager pool, how can I find the right characteristics to match with the patient so they click and have that rapport? So now, my excellent diagnostics tell me who to go after, my analytics on the treatment side tell me that Bob’s the right care manager to go talk to this person cause they’re going to click. Now, based on the analytics, I can determine if it’s best to start a care plan at this point. Is it best to refer them to a community? If they’re extroverts, they might do better in a support group where they can talk to others and share. If they’re introverts, maybe it’s better to give them reading materials so they can study about their diabetes. These are all examples of this huge open area for us to move into. Truly individualized healthcare fueled not by analytics but by actionable analytics.  

Host: That’s fascinating. You can’t find that on Google. So, building on this concept of actionable analytics, in your mind, what does the ideal care management dashboard look like for frontline workers, and then how does that differ from the dashboards for the managers and directors of care management programs?  

Dr. Chris: A frontline healthcare worker working with the person directly will need to know what the most pressing health problems are for this person, and they’re also going to need to know what the person’s perceived most pressing health problem is. The two can be different. If I’m going in and talking to somebody, I know all the data about this person, here are the key things we think we want to work on with them, but if I can also have any degree of analytics from questionnaires or social media posts—anything that helps me understand what they’re most worried about, that is the most powerful combination of data. I might say their blood pressure is the highest priority from a physiological standpoint. But, if they’re terrified of falling and breaking a hip, then we need to address that in tandem because that’s what they’re most motivated by. This is an ideal person-to-person care manager dashboard.

The ideal dashboard for managers and directors of the care management program is going to look different. This dashboard will answer the questions: within the population, what are the major disease categories, which are poorly managed, who has a health burden that is poorly managed, and who’s impactable? Impactability is a key variable; it’s an important distinction between simply knowing who has a health condition and who has a health condition that is poorly managed. Not everyone with a health condition is poorly managed, and if these analytics are to prove useful, they will have to point to who’s in need. Those already receiving care, great, let their care teams take the lead. Then there are those with health conditions that are poorly managed but not impactable. These are the folks who will never pick up the phone; they will hit unsubscribe—they won’t take your calls. Poorly managed and impactable—put those together, and that’s the ideal dashboard for me as a director. Now, I can see across my members who both need help and want help. That’s who I go after first.  

Host: It sounds like a well-rounded approach to care management—taking psychology, patient perception research, and motivational markers and integrating them into the physiological data and claims data sources.  

Dr. Chris: Yes, this is a well-researched and well-understood area using the Patient Activation Method, PAM, a survey that captures in a quantifiable score how well a person understands their condition and how ready they are to be helped. Right now, we depend on this survey to the degree that we can get this information from data—a pseudo-PAM if you will. If we can get this data organically, we can begin to be proactive even if they never fill out a survey.  

Host: So, is the survey being used like this today, or is this a vision for the future?

Dr. Chris: Although the PAM has been around for a while, I don’t think it’s being used in this way yet, as a proactive way to identify impactable members. It’s a good stewardship of resources, and it’s good for the member. This is where the PAM comes in handy. If you’re going to help people avoid outcomes, you want to know they’re engaged, ready, and want to be helped.  

Host: Earlier, you mentioned gathering this data in the absence of the survey. How can this be accomplished?  

Dr. Chris: This is an open frontier. Some examples of ways this is being explored are with NLP and Natural Language Processing, which is the ability to look through clinical notes and extract key phrases. Clinicians have always done this manually. They search through discharge notes and look for key phrases. AI and new technologies are enabling us to automate this. So you can feed a discharge summary and other kinds of notes from an EMR or other data sources. The NLP system can search for these key phrases that may indicate the patient is looking for help or if the patient is worried about what if this happens again. Those are phrases that can be pulled out and feed the algorithm.  

Another source is social media, and there’s a lot of work happening here; it’s where people live these days. People are reaching out and looking for answers online. You want to be sensitive to privacy, and that’s a balance to strike. But some data can be pulled that indicates a person is looking for information online about their health conditions, indicating they are impactable.  

These new resources that didn’t exist five years ago are becoming available in care management soon.  

Host: It sounds like the dream is to have all of the data integrated, where the health plan is feeding data, providers are feeding data, and members are driving data into the health management platform, so it’s one unified repository.  

Dr. Chris: Spot on. That’s exactly where we hope to go.  

Host: Well, Dr. Chris Johnson, thank you for joining us. I know the listeners appreciate your insight and vision to the industry. Thank you to our listeners. Like the episode and share it with your colleagues. Thank you so much for listening.