1. What were your first steps when beginning this analysis?
When we kicked this project off, our analytics, research, and program teams came together to determine key measures to focus on. The analytics and research team helped us determine how to build meaningful and accurate analytics. There were many different aspects of end of life care that we could have looked into and we knew we couldn’t do them all at once. We researched and prioritized the measures, and then created a project plan with phases to be completed over the next couple years. The next big hurdle was to match people in the Colorado All Payer Claims Database (CO APCD) with people who had deceased using death data from the Colorado Department of Public Health and Environment (CDPHE).
2. Were there specific considerations you needed to keep in mind based on the data?
There are limitations to what we can do with this data. Palliative Care is one of the areas that we wanted to inform with this project and it is also something that is difficult to capture in claims because it is not always paid for and therefore doesn’t have an associated code. There is a code, however, that indicates that someone had a palliative care consult, which enabled us to have a broader dialogue with palliative care providers and advocates as to how we should move forward to inform their work. That led us to start looking at common items that are key to palliative care benefits that do exist. Those services, such as pain management, home health care, and medication management were explored to demonstrate the cost for services and the value that comes to the person when they receive those services. In terms of data, the population is included in the analysis does not represent all people who have deceased in the time period evaluated. With multiple inclusion and exclusion criteria, we decided to focus the analysis on people who did who had multiple chronic diseases and comorbidities and those who had significant health insurance claims before death.
3. What challenges did you encounter while performing the analysis? How did you overcome them?
This is a very complex analysis that relies on understanding the population, trends that we would expect, billing practices, and more. We relied heavily on internal expertise as well as leveraging our partners (Department of Health Care Policy and Financing, Center to Advance Palliative Care, Coalition to Transform Advanced Care, and others) to ensure we are capturing meaningful and accurate data. In our next phase, we’ll be exploring the differences in billing practices between Medicare, which pays based on a 30-day look back period, and Medicaid, which pays on for individual services. That has the potential to drastically skew the analysis if we don’t account for the difference. Moreover, there are codes for certain measure that providers are not incentivized to use and therefore are hard to capture in the claims data. This includes information for home health, advance care planning and hospice care. We are reaching out and consulting institutes and providers with expertise in these areas to better understand the billing and service process which will help in our future analyses. Also, people with chronic and comorbid diseases and conditions, tend to have multiple health insurance payers. As a result, assigning people and claims to a specific payer was an arduous task.
4. Without delving into results, did anything surprise you about this analysis or the process of executing it?
It was surprising that we were seeing high cost and utilization of services in the Medicare Advantage population. Over the last few years, the number of Medicare Advantage plans have drastically increased. As a patient, it is hard to compare all of the Medicare Advantage coverage options and pick the one that best fits your needs. It can also be difficult to determine how to access services and understand what services are covered under each plan. With such high cost and utilization of these types of plans, we want to dig further into claims related to Medicare Advantage plans to understand what types of services are being accessed by which providers, and whether there is a difference between the cost and utilization of services by geography.
5. What did you learn while performing this analysis?
End of life care is very complex and requires knowledge beyond strictly analyzing the data. Understanding the contextual elements, such as the Advance Care Planning (ACP) codes that were implemented in 2016 and the different billing practices for Medicaid versus Medicare, are important to ensure accurate reporting. We also need to take COVID-19 into consideration, and see how it impacted the data and health care needs across the state.
