Healthcare organizations across the United States are pursuing more equitable health outcomes for divers populations. Our team of health economists, data scientists, and researchers draw insights from data sources to help clients solve inequities. Data stratification empowers decision makers, clinicians and healthcare administrators to identify their patients, assess risks, measure success and implement innovative care models.
What is data stratification?
Data stratification is the process of sorting data into defined segments or groups and can make data analysis efforts more comprehensive and easier to evaluate. When applied to population health, data stratification can anchor initiatives that require data, such as care coordination, and inform population health management workflows.
In an interview recorded earlier this year, learn how data stratification can be used to combat health inequities, the motivating factors for instituting data stratification, new health equity accreditations, and more.
You can listen to an audio version of my interview here:
Studies show that improving equity impacts disease outcomes
How do you convince payers to allocate money and time to invest in something like data stratification? What is their economic incentive?
I would first say there's a moral imperative that we do something about disparities and outcomes across multiple indicators of health and wellbeing of our population. I think there's also mounting evidence in certain areas of care, for example behavioral health, that improving equity and access to behavioral health has a direct impact on outcomes for other kinds of diseases like diabetes, cardiovascular disease, and others.
I think we have a lot of opportunity to address the population health level outcomes, where we know we have longstanding disparities in care, and more importantly, in the health outcomes of populations that are not White. I would also say that health equity encompasses a lot of things beyond just race and ethnicity. Health equity encompasses equity and access for rural populations, for people who identify as LGBTQ, for people who have many different dimensions to their whole being that equity concerns.
Stratification is essential to understanding social determinants of health
When we talk about stratification, is the objective to improve outcomes and access, or is it to identify the underlying social determinants? Do you need to do one before you do the other? Talk a little bit about the dichotomy between the social determinants and existing inequities and how stratification can help address one or both?
Data stratification is the key to understanding the population that a payer or a provider is serving. It’s a way to really understand the specific determinants or drivers of health affecting a population that a provider is responsible for, under value-based contracts or attribution and, of course, at the payer level as well. The ability to stratify your population—understand all the dimensions of their social determinants of health—impacts your ability to target, in a smart way, where payers and providers spend their time and their resources. This improves focus for the maximum impact on outcomes and achieving health equity. It’s really hard to do that without understanding the underlying population’s characteristics.
Stratification involves both identification of race, ethnicity, and other sorts of demographic components, as well as specific health conditions, correct?
If you're trying to understand the social determinants of health, you should be gathering that information directly from the patient themselves, ideally, or from the member of a health plan themselves. And then you can combine that information with other information the health plan may already have about the healthcare utilization patterns of certain people and then you begin to understand or see more clearly. What are the specific social drivers of health that you may be seeing among certain subsets of patients. So it's a way to really get that more comprehensive picture of the patient population that you're serving to drive better population health programs, and also to evaluate their impact over time.
Plans and providers share the responsibility of stratifying data
I would have assumed that plans have traditionally done a pretty good job of stratifying their populations to better understand their members. Is that not the case, or are we just talking about taking it to different levels?
Plans traditionally have done this inconsistently and it’s better for some lines of business than others. For example, they may have more complete information for members in their Medicare advantage plans, but not in their commercial plans. And so part of what's happening now at the national level is an increasing focus on a move towards accountability for addressing health equity. And step one of that is to actually put in place accreditation requirements, for example, through NCQA or new requirements rule making from CMS that is turning up the level of requirement and focus on really gathering and using that social determinants of health data.
Is it the primary responsibility of the payer or the provider to collect the information and to stratify it? What is the vision of labor to ensure the back and forth necessary to collect as much relevant information as possible, both clinical and demographic?
I think the ability to collect that data, share it and then combine it with other data has been extremely challenging. I would say that the responsibility for gathering it is a shared responsibility. There's evidence of course, that members of health plans are more reluctant to share information about themselves, about their drivers of health, directly with their health plan. They may be much more comfortable in fact, sharing that information with provide. At the same time, providers worry that if they gain this fuller picture about their patients, but can't do anything to address say, transportation issues or food insecurity issues or referrals for other kinds of community-based services and supports that their individual patients may need, why collect it?
If I can't do anything about it, if it's completely outside of my control then why collect it? This has been part of the barriers and pushback in the past. I think what we're seeing now under value-based contracts in particular, is much more collaboration between payers and providers about, “All right how do we do this?" We both have this shared responsibility for better understanding the populations that we serve under our value-based arrangements with each other, so we're seeing more novel implementations of ways to gather and share that data. But I would say that it's certainly not terribly widespread or standardized. And I think that's really the frontier where we're going, is to get much better data.
There's also a lot of variation between states. And in some cases, you have MCOs that have much more stringent requirements around collecting information about social drivers of health. For example, in North Carolina, there's NC360, which has been implemented where providers use a standardized instrument that goes into a standardized statewide database. We're seeing those kinds of efforts not just in North Carolina, but in other states as well, so you're getting a more comprehensive picture. Then the question is around the data sharing and how do you link that data with other data that you might have that helps you to better address population health and equity.
Is it primarily the responsibility of the payer to take the lead in combining this information from various sources and then providing it to the provider as needed or vice versa or is that even clear yet?
I think that as CMS engages in upcoming rule making we’ll see greater clarity emerge. They're putting together sort of their health equity strategy. I have a feeling that that's going to result in changes that are going to affect providers, but we're also going to see things coming through on the payer side on regulations affecting the health plans and payers themselves.
Emerging health equity standards are changing the demand for data collection
Should payers and providers focus on developing a day strategy or stratification of existing data?
In some cases, the big need is figuring out how we are going to collect this data directly from patients so that we are in compliance with emerging requirements, for example, from NCQA. Beyond that how do we actually use that data as part of our population health management efforts? These are requirements from NCQA accreditation, which is what most health plans pursue here in the US.
Could you explain the five HEDIS measures that support health equity and how those fit into the broader context?
Around health equity there's a couple of things that are emerging and they're sort of complementary to each other. One is that they have developed a health equity accreditation and a health equity plus accreditation. So two new accreditation programs that they have pilot tested with a number of health plans and are implementing, that's a voluntary accreditation. Health plans are not required to pursue health equity accreditation status. Many will choose to, and it'll be interesting to see what the experiences are of health plans that pursue health equity or health equity plus accreditation. In addition to that, NCQA actually is opening that accreditation to other kinds of organizations. So you don't just have to be a payer or a provider, other kinds of organizations—community based organizations, healthcare disruptors of various kinds—can achieve that same accreditation.
The other thing is that the population health management standards themselves that are part of the NCQA accreditation that most health plans have are undergoing some revisions. I'm not exactly sure of the timeline of when and how those are going to change, but I believe that we will see more requirements around understanding your patient population. Then addressing health equity through population health management approaches and becoming part of those population health management standards in the future.
It used to be that NCQA would allow payers to use third party data to say, "This is what we estimate our racial, ethnic, and other sort of social determinants of health distributions in our population that we serve without actually knowing that that is their actual member bases sort of picture." They're using third party data and it was adequate to submit that third party data. As for these five measures, you can still do that this year, but going forward, NCQA is also looking for direct data collection from your members in addition to that third party data with an eventual switch to the majority of your understanding of your population social determinants of health coming from direct data gathered from your members. I think it's by 2025.
It’s a massive change. And health plans are working hard now and have been because the knowledge that this was coming has been around for a little while. So preparations on the part of health plans began right away as it became clear that this change was going to come. For some, I think it's a big lift and for others, it may be not quite as a big lift, but it's signaling a future in which the expectation is going to be. You're gathering this information, you are getting this information directly from patients members.
Improved data infrastructure will support evaluation of key external factors
What about the technical challenges of pulling together structured and unstructured information from a host of sources?
The data infrastructure is a significant challenge that a lot of payers, and also providers, are addressing through major IT initiatives. Even just the ability to pull in and aggregate data and really match data to individual members can be extremely challenging. It requires a lot of analytic work, as well as just the data exchange infrastructure. You might have data in a particular state from like health information exchanges. You may have APIs between providers and payers that might help with some of that data sharing. There’s still the need to aggregate and link up data. Yes, the data infrastructure and the data governance, as well, continue to be big issues that payers and providers are working out.
This sounds like an enormously complex and expensive undertaking. What is the motivation other than the common good? Why should a health plan invest time, energy and limited resources in this?
There’s going to be requirements, and the reason those requirements are going to exist is because there is evidence that when we address those social determinants of health, when we really understand the population that we're trying to serve, whose health outcomes we're trying to improve, those social drivers of health, and address them, it does lead to better outcomes. There may be times of course, where you might have some upfront increase in costs that leads to better downstream utilization patterns in the longer term.
Ultimately you want people to have the best care that leads to their highest level of wellbeing, far beyond what is just measured by HEDIS measures. And I think ultimately you can address the need of the population that you're trying to serve by recognizing that most people's health outcomes are far more determined by things that happen outside of clinical care than what happens within the clinical setting. That's the argument for really addressing those social determinants of health and achieving those better population health outcomes.
Think about behavioral healthcare, maternal and child outcomes. North Carolina for example, is a terrible state in terms of the disparities and outcomes for women who give birth. This is why understanding the specific outcomes relative to specific populations and specific places really has to drive what you do to achieve greater equity.