Before COVID-19 entered the nation’s lexicon, pioneering health plans were seeking innovative ways to stratify their members to better target high-risk individuals for needed health and/or social services. Traditionally, much of this risk identification work relied on utilization, claims, and diagnostic data, but the pandemic and the existing health inequities it has highlighted — especially in communities of color — are leading some innovative health care players to expand their perspective on what types of data should be used to understand risk. The case for health plans to incorporate social health data (race, ethnicity, socioeconomic status, etc.) into their analytics and partnership work is strengthening as the pandemic continues. Health plans can use this data both to develop a more holistic perspective of patients’ medical needs and to support provider networks.
CareOregon, a Medicaid managed care plan participating in the Center for Health Care Strategies’ (CHCS) Complex Care Innovation Lab and its Rising Risk initiative, is a leader in using data to identify and coordinate care for its most complex members. CHCS spoke with Amit Shah, MD, CareOregon’s chief medical officer, about how the organization’s data analytics approach was impacted by COVID-19 and the implications this has for the health care system to improve the collection of social needs data.
Q: What impact has the COVID-19 pandemic had on your data analytics work?
COVID-19 galvanized our provider network’s interest in using our data to further identify and manage individuals who are at high risk.
A: Prior to COVID-19, our population health analytics team had already been identifying the levels of risk in our populations and organizing our teams and clinical strategies around high risk and rising risk groups, but there had been some challenges in engaging our provider network to use our data to support their patient management efforts. Then as COVID-19 started to spread, we had to rally around these tools and strategies to quickly figure out who was at high risk for COVID-19, how to get that data to providers, and how to manage the associated care. COVID-19 galvanized our provider network’s interest in using our data to further identify and manage individuals who are at high risk. When we produced a list of high-risk patients and disseminated to our clinics, they found it to be accurate and valuable.
Q: How have you focused your efforts amid the pandemic to identify rising-risk and high-risk patients?
A: Our first step was identifying groups of people who had clinical definitions of high risk, a more useful indicator of risk than relying solely on the predictive value of cost and utilization. According to CDC guidelines, there is a group of patients who have certain clinical conditions (COPD or diabetes) that you know if they get COVID-19, they are at risk for poor outcomes. This group is not particularly hard to find in the data, and when you share that with the provider network, they can confirm through electronic health records and conduct the proper outreach.
The second group of people we wanted to identify were those who are at risk for negative health outcomes because COVID-19 disrupted needed services. For example, patients using infusion therapy in the rural community could no longer receive care because the local clinics were closed. We used pharmacy data to identify these patients. We know without these types of outpatient treatments, these patients would be in trouble, so we needed to focus on them as well.
The third group are those who need behavioral health services and are at high-risk for a negative outcome without them. A relatively healthy 30-year old with a substance use disorder and can no longer receive daily Methadone treatment when the clinics are closed would be an example of this. CareOregon has recently taken over the behavioral health benefit in all of our regions. We now have these data readily available for our members, but are working with a new network of behavioral health providers, who are both not used to getting our data and who historically may have had only limited insight into the physical health risks of their patients. There are legal barriers to sharing certain behavioral health data across networks, and some of these members only receive behavioral health care and not primary care through our network. Now, we are figuring out how to navigate all of that and include behavioral health data as part of our risk stratification process. Making sure we have both behavioral and physical health data on these patients is crucial for fully understanding their risk.
Q: How have you leveraged your analytics work to support your provider network during the pandemic and the financial challenges it has precipitated for health care systems?
If these clinics were to fold, where would the most vulnerable members, our Black and Latinx communities who have higher infection rates of COVID-19, go?
A: CareOregon implemented a proactive provider network stabilization payment plan to help prevent the most fragile parts of our provider network from potentially reducing services and access drastically or even closing. Since utilization was dramatically decreased, in some cases to none, they were not receiving the necessary revenue to stay afloat. Instead of holding onto this money, we went ahead and paid these clinics based on their prospective utilization. We targeted specific provider network partners (including physical, behavioral, and dental providers) based on data highlighting which member populations were facing the greatest risks of COVID-19 and where they were being served. If these clinics were to fold, where would the most vulnerable members, our Black and Latinx communities who have higher infection rates of COVID-19, go?
Q: Are there other data points that you have considered using for your risk models, given how COVID-19 has affected these marginalized communities?
A: In general, the data around inequalities around race, ethnicity, poverty, and other social health data can be hard to find in traditional health plan datasets. But if anything, COVID-19 further showed us that the social health piece was critical from a data perspective. While we did not solely rely on utilization data for identifying who was at risk for COVID-19, examining specific utilization data actually showed us what interventions our community could benefit from. For example, Oregon has health-related services benefits as part of its Medicaid waiver, which permits health plans to cover additional services as a supplement to covered benefits. The network can submit requests for things such as temporary rent assistance, laptops to increase access to telehealth, even air conditioners. In the past few months, we have seen a massive utilization of this service. The need for housing, food, transportation, and other social needs demonstrated how health-related social needs often are the true safety net.
This has really highlighted an opportunity for CareOregon to grow its role in the health care system by supporting social services. It was clear that when you focus on the social health data, a bright light was shone on existing inequalities. You can really see how certain racial and ethnic minority populations have been disproportionately affected and how great the social needs are. We are getting better at this type of data collection as we go about developing our social health data platform. CareOregon is pursuing the use of Unite Us, an online platform that builds care coordination networks within local community-based organizations. We anticipate that this will start us on the right path to addressing these inequalities.
Q: What are the opportunities to build on the insights you have gained in recent months?
Social health data can help address social injustices by highlighting inequalities and letting us know where action is needed most.
A: Here at CareOregon, we will continue to build and develop the ability to do predictive analytics. While we did a really great job after COVID-19 struck, it should not really matter if a pandemic happens or not. You should have the ability and the data to predict any sort of situation, and be able to “slice and dice” the data in a way that shows whom the at-risk populations are. As soon as you see a trend increasing in real time, you should be able to implement proactive interventions way ahead of time in order to protect and support these individuals. We also need to be better at refining our data, so that when it is given to the provider network, it is curated in a way that is timely and actionable.
Another aspect health care systems need to work on that is timely and relevant to the current unrest in the nation is we have to be more transparent with data to demonstrate where inequalities exist. Social health data can help address social injustices by highlighting inequalities and letting us know where action is needed most. We have a responsibility to actually do something about inequality and pivoting our data use is one step in a long road to building a socially just health system all of us want.