AxisPoint Health is a Westminster, Colo.-based company that provides care management services to a variety of entities, including state Medicaid programs and other insurance and medical providers. It focuses in particular on condition management—the careful tracking of the health status and condition of individuals who are members of large insured cohorts, and the alerting of providers to changes in condition or health status that suggest interventions.
Recently, senior leaders at AxisPoint Health conducted a study of a Medicaid population, to determine what kinds of differences the use of condition management might make in that population. Among those involved in the preparation and analysis of the study were Gregory Berg, Ph.D., associate vice president, research and outcomes, and Margaret Flaum, R.N., vice president of account management, for the company. How many plan members in this universe?
Berg, Flaum, and their colleagues looked at two populations for comparison. The broader group encompassed 25,000 Medicaid recipients in a western state who were continuously eligible for condition management by AxisPoint Health professionals; the trend, or control, group, consisted of just under 5,800 individuals who were not being condition-managed.
In analyzing data gathered for the study, the AxisPoint Health study leaders found that total expenditures for those in the trend, or control group—those not receiving condition management—saw an annual increase of 100 percent in cost inflation—though it should immediately be noted that those individuals had not been identified as having a chronic condition meriting inclusion in care management or condition management. Meanwhile, those in the group being condition-managed saw a flat trendline, implying that care or condition management did indeed make a significant difference in managing their costs.
In the first month of the study, June 2014, Medicaid plan members in the condition management program averaged per member per month (PMPM) costs of $1,018, while those in the control group (without identified chronic conditions) averaged $105 PMPM in costs to the program. Over the next 18 months, the average PMPM costs of those in the condition management program remained flat, while the PMPM costs of those in the control (general population) group rose by 400 percent, indicating that the condition management program was indeed helping to manage not only the clinical conditions of those plan members, but also their PMPM costs. Even more dramatically, the inpatient utilization rate of those in the condition management program rose between 10 and 15 percent within the 18 months of the study, while inpatient utilization rose 100 percent in the non-managed control group. Similarly, pharmacy costs remained essentially flat, with a possible slight increase, in the care-managed group, while they rose from around $20 PMPM to over $100 PMPM in the general population.
Meanwhile, emergency department (ED) utilization declined slightly in both groups; study leaders wrote this in their study’s summary: “In summary, within the constraints of unreconciled data, the key financial metrics of the program show generally positive trends when compared to a similar population. If the program did not have a positive effect on the members, one would suspect mirroring of, or even exaggeration of, the trend population. Continued effort will be applied to decrease Emergency Department utilization.”
Berg and Flaum spoke recently with HCI Editor-in-Chief Mark Hagland about the study and its implications for care management and population health management efforts in U.S. healthcare. Below are excerpts from that interview.
Overall, though one needs to be careful about over-generalizing, you found clear positive benefits from care management. That is the bottom-line result of the study, correct?
Margaret Flaum, R.N.: Yes. This program is a population-based care management program. And through analytics and communications to all the members and providers, we were able to identify through claims data, and risk-stratify, the population, based on individuals’ levels of claims risk. Then, using social workers, behavioral specialists, and community health workers, we were able to reach out to members, and do care coordination and care management in concert with their providers, to manage their risk factors. So ultimately, this program, which encompassed telephonic intervention, on-the-ground intervention, and the engagement of members, provides a good example of how programs like this that make use of proactive interventions, will be able to influence all of the key metrics we found in the study’s results.
Gregory Berg, Ph.D.: So, we first looked at the PMPM costs of the 25,000 plan members in the program, and found that their costs were relatively flat. We didn’t initially know whether that was a good or a bad thing, so we compared that level of cost increase [flat] to a control group of 5,800 plan members who were not in the care management group, in order to infer the impact of participation in the program, on PMPM costs. Comparing the flat level of cost increases among the care-managed plan members to the increases among the general plan population, we could infer that the care management program had had some effect.
Flaum: Looking at the PMPM costs of the two groups in June 2014, the costs for the general population, or control group, were $105, while those in the intervention, or care-managed group, were $1,018. Those in the care-managed group saw no cost increase over 18 months, while costs for the control group, while much lower, rose by 400 percent. Meanwhile, inpatient utilization in the care-managed group rose only 10 to 15 percent, while inpatient utilization among the general population rose 100 percent during the same period of time. Now, it is important to note that Medicaid costs overall haven’t risen significantly, but the majority of Medicaid health plan members are not sick, either.
Accepting the inference that care management/condition management does lower PMPM costs, based on your analysis, what elements of care/condition management are leading to those results?
Flaum: Once the analytics have been applied, and we risk-stratify individuals in the program. There are four levels of health risk: very, very high; severe; moderately severe; and low. And nurses and social workers and behavioral health specialists and community health workers, collaborate to perform a health risk assessment. They’ll also look at claims histories. And the health risk assessment will look in real time at barriers to care, social impacts, and so on. Do plan members have a primary care medical home? Do they understand they should contact their PCPs? Nurses do a full clinical assessment based on those interviews. And that prioritizes for the member and the provider, and clarifies for us different areas to focus on, to help that person be empowered to manage their condition.
In other words, it involves a lot of assessing them in real time, assuring that they have a care management team, and then really giving them the tools and knowledge to address their conditions early. Are there barriers to them getting their medication? Do we need to help them with that? Are asthmatics using their inhalers correctly? Do patients with CHF have scales to weigh themselves on? And having support systems, whether face-to-face, or going to their provider’s office? Just making sure they have those ongoing touchpoints to make sure that exacerbations don’t occur. And it takes time. You can start seeing impact within the first 18 months, and generally, the longer-term that these programs are in place, the more impact we see.
What are the keys in doing the analytics correctly?
Berg: The biggest key for us is the risk stratification, finding and targeting those individuals who need intervention.
Is there a right data or analytics process involved in performing the risk stratification?
Berg: One of the things is looking for impactability--how impactable a person is. They may be high-cost, but that cost may be unavoidable in the future, as opposed to someone who’s lower-cost, but whose trajectory we can impact in the future.
What should our audience understand from what you’ve learned?
Flaum: On the one hand, your audience may or may not know about doing analytics in this way, and they may want to know, how can they improve the data they’re providing. In other words, you have to have the best data you can get. If you’re the CIO of a hospital, it may come down to the accuracy of coding and billing. And the coding and billing people may be focused on billing for the hospital, but if they know it will improve the potential for care management, that is important.
And ideally, you want to identify these members before exacerbations—inpatient hospitalizations and ED visits—occur. We want to be able to do that before these events occur. And many of these patients can be hard to find. And if they do have an acute exacerbation, that’s an ideal time for care management programs. We have staff who will go in and meet with these patients at that time. So there’s tremendous opportunity with these programs to where, once you’re working with an individual, to really help break down some of the barriers to care, and get them much more stable and comfortable in accessing healthcare as they need it, long-term.
Is there anything you'd like to add to what’s already been said?
Flaum: I’ll just add that we’re very passionate about these programs. We believe in them. They need to be very comprehensive in order to be effective. And they really need to be tailored and customized to the geographic aspects, the cultural aspects, the importance of really working with their primary care providers—that can’t be emphasized enough. Our goal is to really empower those members, and help them feel more confident in ultimately managing their care.