Artificial intelligence (AI) and machine learning (ML) are increasingly used in healthcare to combat unsustainable spending and produce better outcomes with limited resources, but healthcare organizations (HCOs) must take steps to ensure they are actively mitigating and avoiding algorithmic bias.

While AI/ML has the potential to identify and combat disparities, it also has the potential to inadvertently perpetuate and exacerbate health inequities—despite apparent objectivity. In fact, algorithmic bias in healthcare is already pervasive. The Chicago Booth Center for Applied Artificial Intelligence states, “algorithmic bias is everywhere…Biased algorithms are deployed throughout the healthcare system, influencing clinical care, operational workflows, and policy.”

What is Algorithmic Bias?

Algorithmic bias occurs when issues related to AI/ML model design, data, and sampling result in measurably different model performance for different subgroups. This has the potential to systematically produce results that are less favorable to individuals in a particular group compared to others, without any justification. In healthcare, this can lead to inequitable allocation or prioritization of limited resources.‍

High-profile cases of algorithmic bias have directly propagated racial health disparities. In 2019, Obermeyer et al., a seminal paper on algorithmic bias in healthcare, found evidence of racial bias in a widely used Optum algorithm that managed 70 million lives. Bias occurred because the algorithm inappropriately used health costs as a proxy for health needs. As Black people face more barriers to accessing care and systemic discrimination, less money is spent on caring for them compared to white people. As a result, Optum’s algorithm incorrectly learned that Black members are healthier than equally sick white members.

This systematically disadvantaged and produced worse outcomes for Black members while prioritizing white members for care and special programs, despite being less sick than Black members on average. Researchers found that “the algorithm’s bias effectively reduced the proportion of Black patients receiving extra help by more than half, from almost 50 percent to less than 20 percent. Those missing out on extra care potentially faced a greater chance of emergency room visits and hospital stays.”

What Causes Bias?

There are many causes of algorithmic bias, but the majority fall into two broad categories: subgroup invalidity and label choice bias.

Subgroup Invalidity

Subgroup invalidity occurs when AI/ML is predicting an appropriate outcome or measure, but the model does not perform well for particular subgroups. This may be due to poor calibration, a significant difference in the distribution of predicted and actual outcomes for certain subgroups. Generally, subgroup invalidity occurs when AI/ML models are trained on non-diverse populations or with data that underrepresents the subgroup or fails to include specific risk factors affecting them.

For example, a recent study of pulse oximeter algorithms, which were originally developed in populations that were not racially diverse, demonstrated subgroup invalidity bias. This measurement technology uses a cold light that shines through a person’s fingertip and makes it appear red. By analyzing the light that passes through the finger, an algorithm is capable of determining blood oxygen levels. However, the study found that “Black patients had nearly three times the frequency of occult hypoxemia that was not detected by pulse oximetry as white patients.” In this case, the algorithm performed poorly, likely because it was primarily trained on white people, potentially resulting in worse health outcomes for Black people.

Label Choice Bias

Label choice bias is more common and more difficult to detect than subgroup invalidity. It occurs when the algorithm’s predicted outcome is a proxy variable for the actual outcome it should be predicting. The source of racial bias Obermeyer et al. found in Optum’s algorithm (predicting healthcare utilization costs in an attempt to predict future health needs) is a quintessential example of label choice bias in healthcare. Using cost as a proxy to allocate extra resources or care demonstrates this form of bias because Black people are less likely to receive necessary care due to systematic discrimination, racial biases, and barriers to care. This results in lower healthcare costs and significantly biases cost as a proxy for future health needs.

Four Steps to Address Bias

Algorithmic bias in healthcare is not inevitable. Organizations are taking major steps to ensure AI/ML is unbiased, fair, and explainable. The University of Chicago Booth School of Business has developed a playbook to guide HCOs and policy leaders on defining, measuring, and mitigating bias.

While the playbook describes practical ways to mitigate bias in live AI/ML models, the steps below are also helpful to consider if you are in your initial considerations of AI/ML. It provides guidance for technical teams working with AI/ML daily, but also describes preventive oversight structures and ways to address bias from the outset.

Simply taking an inventory of AI/ML models your organization is currently using, developing, and considering implementing will enable you to begin assessing them for bias and establishing structural oversight. Your organization may not have a record of deployed models, but either way, talking to a diverse group of stakeholders and decision-makers across business units will help develop your overview and fill in key details about how AI/ML is being used. Gaining a comprehensive understanding of the kind of decisions your organization is making and the tools being used to support these processes is key to determine potential for bias.

A leader in your organization should also be responsible for overseeing algorithmic bias across all departments. Developing a comprehensive inventory that evolves over time will ideally be supported by active governance from someone with insight into high-level strategy and your organization’s goals. Members of your C-suite are great candidates for this role, and your organization should consider hiring a leader with experience in addressing algorithmic bias and advancing health equity if it lacks one.

Critically, this leader should not work in a silo, and should strive to collaborate with teams across the organization while considering diverse perspectives from internal and external stakeholders. They should work closely or form a team with people from diverse backgrounds to ensure appropriate representation and avoid inadvertently introducing their own biases into their analytical work.

Evaluate AI/ML models in your inventory for label choice bias and subgroup invalidity by carefully assessing the data used, potential discrepancies with the ideal predicted outcome, and variables used if possible.‍

You can begin by examining the data for problems related to lacking diversity and poor performance for underserved groups. Confirm that the data on which the models are trained is representative of the populations to which they will be applied. One-size-fits-all models that do not account for the unique demographics and qualities of the population they are applied to may exacerbate health disparities. You should also carefully consider how subgroups are defined and the fraction of each subgroup relative to the total population. If definitions differ from the training set to your population, the model may be biased.

Next, measure performance in each subgroup to ensure models are well calibrated and do not perform worse for any particular group. Good performance across subgroups does not guarantee a lack of bias, but it is a key step to avoiding it.

Once you’ve ruled out subgroup invalidity, you will also need to evaluate the model for label choice bias. Unfortunately, assessing label choice bias requires domain experience and cannot be automated. You should establish processes to systematically evaluate the potential for label choice bias with multiple stakeholders and also attempt to ensure the labels used in training data align with your organization’s intended use.

If your AI/ML model is predicting a proxy outcome, you will need to concretely define it and your ideal outcome in the data to compare calibration effectively. For example, “future health needs” is abstract, poorly-defined, and cannot be proxied with cost. Instead, your model will need a complex, measurable metric to better compare the ideal outcome and the outcome in use. Comparing the ideal outcome to a composite assessment of health needs, such as the Charlson Comorbidity Index, will enable you to better evaluate how the model actually performs across subgroups. In practice, proxy measures should be multifactorial, consider how model outputs will be used, the data you have available and input from a diverse group of stakeholders.

Finally, evaluate AI/ML models for feature bias, which occurs when the meaning of individual features (variables) in a model differ across subgroups. This can arise due to differences in access to care, how diagnoses are ascertained, and health data collection. As an example, a model making predictions related to respiratory disease may demonstrate feature bias if it uses air quality reports without considering that rural populations may have missing data, as smaller towns are not necessarily required to regularly report air quality to the EPA. Again, health care costs are a feature that can introduce significant bias. Using relevant and diverse features can help to achieve high accuracy and validity across different subgroups.

Biased algorithms directly affect health outcomes and perpetuate inequities, and ideally should not be deployed. If you’ve discovered bias in a deployed model, it must be retrained or discontinued.

To mitigate label choice bias, you may be able to retrain the model with the same variables you originally used to illustrate the presence of bias as your new composite outcome. The Chicago Booth School of Business team demonstrated this by replacing cost as the proxy variable for predicting future care needs and retraining “a new candidate model using active chronic conditions as the label, while leaving the rest of the pipeline intact. This simple change doubled the fraction of Black patients in the high-priority group: from 14% to 27%.” They note that this approach represents just one option, and that the best approach will depend on your specific circumstances.

Models will also need to be routinely retrained to avoid becoming biased over time. This occurs due to feature drift, in which the distribution of AI/ML variables in the target population begins to substantially differ from the distribution in the training population. Feature drift occurs naturally, and can be addressed by regularly monitoring deployed model performance and retraining when appropriate. Additionally, major updates to your population (e.g., a merger with another organization), changes to codesets, or shifts in chronic disease prevalence (e.g., respiratory disease rates during the COVID-19 pandemic) will all require models to be retrained to avoid bias and significant accuracy degradation.‍

Sometimes retraining a biased model to resolve label choice bias or subgroup invalidity may not be feasible. This is often the case if your organization is evaluating unexplainable “black-box” algorithms from third-party providers that are proprietary. In this case, your organization may be unable to retrain the model at all or evaluate its design in enough depth to realistically assess whether changes impactfully reduce bias.

If you cannot address and correct the causes of bias in a model, then you should not use it. Instead, you should consider using the first two steps above to begin the process anew and either produce or purchase a new model that avoids label choice bias, is well-calibrated across subgroups, and can be routinely evaluated and modified.

HCOs using AI/ML should establish structural processes for bias management that are directly overseen by an algorithmic bias steward and ideally a dedicated, diverse team formed to address bias. According to The Chicago Booth playbook, organizations should take the following steps:

• Create a system to report bias concerns. Enable everyone in your organization to formally report algorithmic bias concerns without worrying about any repercussions.
• Create standardized requirements for documenting algorithms. Detail all relevant information related to your models and expect clear, comprehensive documentation of any models you purchase from a third party.
• Establish a routine schedule for auditing AI/ML. Consistently monitor, manage, and govern any deployed AI/ML.
• Consider partnering with external oversight. Involving a third-party to audit your AI/ML can help to ensure accountability and provide guidance if bias is detected.
• Stay on the pulse of AI/ML developments. AI/ML is a rapidly growing field and suggested guidelines are constantly evolving.

WellSect’s Approach to Bias

At WellSect, we are committed to addressing algorithmic bias in healthcare and minimizing its impact on obstructing health equity. Our AI/ML positively impacts tens of millions of lives daily, and our data science platform is specifically designed to help HCOs identify and address bias.

In April of 2021, we won the Centers for Medicare and Medicaid Services (CMS) AI Health Outcomes Challenge — the largest healthcare-focused AI challenge in history. Demonstrating a comprehensive, effective approach to mitigating algorithmic bias was a key part of this $1.6 million challenge, which centered on creating explainable AI solutions to predict adverse health events. Our work during the challenge included feedback from Ziad Obermeyer and David Kent, two authors of seminal papers on bias in healthcare AI. To learn more about how we addressed bias and won the challenge, please watch this webinar led by our CTO, Dave DeCaprio.

Joseph Gartner, our director of data science and professional services, has written a series of posts to help HCOs identify and address algorithmic bias concerns that may widen equity gaps:

• Differentiating Bias and Fairness in Machine Learning
• Three Tangible Steps in Combating Prejudice in Data Science
• A New Metric for Quantifying Machine Learning Fairness in Healthcare

Each of his posts explain key concepts, provide best-practice suggestions for healthcare data practitioners, and break down why bias in healthcare must be approached differently compared to other industries. In particular, “A New Metric…” explains why disparate impact, the most common metric traditionally used to evaluate algorithmic fairness, is entirely unsuited to healthcare, and provides an alternative metric created by WellSect.

Other members of WellSect leadership have also written at length on addressing algorithmic bias. Carol McCall, our chief health analytics officer, recently published an article on the subject in STAT, and was featured on a MedCity News panel, How to incorporate Robust Bioethics in AI algorithms. In both the article and the panel, she emphasized the importance of explainability and argued that “black-box” models aren’t adequate for healthcare. She asserts that they must insist on “AI algorithms that are fully transparent, deeply explainable, completely traceable, and able to be audited. Anything less is unacceptable.”

Ultimately, AI/ML is not directly at fault for reinforcing systemic biases. Algorithms are designed by humans and told what to predict, what data to use, how to calculate predictions, and what population to make them on. As a result, they can inadvertently reflect and perpetuate the inequities and biases infused in the healthcare system. However, AI/ML is also an incredibly powerful tool for enacting positive change. Not only can we take steps to avoid and prevent  algorithmic bias, with AI/ML we can unearth inequities across healthcare and leverage the mountains of available data to produce better outcomes and advance health equity.

This post is the final part of our health equity series. If you’re interested in learning more about health equity and what can be done to achieve it, please check out our comprehensive overview post, Why Health Equity Matters in 2022, and our previous posts:

• Diving into CMMI’s New Health Equity Objective For 2022 and Beyond
• 3 Ways Healthcare Organizations Can Advance Health Equity by Addressing Social Determinants of Health
• How COVID-19 Exacerbated Health Disparities
• How You Can Develop and Launch a Strategy to Prioritize Health Equity

The post Four Steps to Measure and Mitigate Algorithmic Bias in Healthcare appeared first on WellSect.

Implementing a comprehensive strategy to advance health equity is a moral and financial imperative for healthcare organizations (HCOs).

Persistent health disparities create preventable suffering and excess costs, are fueled by social determinants of health (SDoH), and consistently disadvantage people of color. Recent studies of racial equity estimate that $135 billion could be saved annually if racial disparities in health were eliminated, including $93 billion in excess costs of care.

These issues are pervasive because they extend far beyond the traditional healthcare setting, and efforts to advance health equity must reflect a complete, organization-wide commitment. Many HCOs have voiced a commitment  to achieve health equity, but taking the next steps to develop a comprehensive strategic approach can be challenging.

Effective strategies hinge on effectively identifying specific community needs, establishing trust to overcome historic racism, and embedding health equity into business processes and objectives. As a result, every organization must develop a unique strategy tailored to their population’s most pressing issues, their own internal structure and processes, and their available resources.

Developing a Strategy to Prioritize Health Equity

Achieving health equity requires making it one of your organization’s key priorities and developing a robust strategy. However, cultivating a culture around health equity and executing on equity-centered initiatives is more than adding it to a list and holding a planning meeting.

Fortunately, the Institute for Healthcare Improvement (IHI) has outlined five universally applicable ways to make health equity a core strategy. We’ll explore these five key actions below.

Health equity should be a top-level focus that is aligned with business objectives and built into every major decision. Leadership is ultimately responsible for ensuring that health equity considerations are actually put into practice, and they must make it clear to employees that advancing equity is not a charitable side project but a key, organizational focus.‍

To ensure accountability, your organization may also consider incorporating measures of health equity advancement in executive compensation and departmental key results. As one executive said in an industry-wide survey by Deloitte, “When equity IS our culture, just the way we do business, and NOT a special set of circumstances—that is how to keep it as essential.”

Hiring a chief equity officer is one possible approach to creating a top-down focus on health equity. This position should be accountable and responsible for the development and execution of an organizational strategy—helping to maintain a focus on health equity in leadership/board meetings. Critically, hiring an equity-focused C-Suite role should not result in siloing equity efforts within their department. Instead, they should embed equity throughout the organization.

Advancing health equity requires dedicating organizational resources to establish new programs and initiatives. In turn, this necessitates creating a governance structure that oversees and manages these programs, ensuring that resource allocation is consistent, sufficient, and not mismanaged. The scope of these programs may depend heavily on organization size and capacity, but the teams that oversee them should be able to focus their undivided attention on health equity with repeatable processes, an appropriate budget, and well-defined goals.

Henry Ford Health System’s (HFHS) Healthcare Research Disparity Collaborative is an excellent example of structural support for health equity. HFHS founded the collaborative to create an arm of their business dedicated to researching and addressing racial and ethnic health disparities with other health systems and community-based organizations (CBOs). Since then, HFHS has maintained ongoing operational and financial support through the collaborative for a variety of health equity initiatives in Detroit, such as the Women Inspired Network, which works to reduce infant mortality for women of color.

To advance health equity, HCOs must identify the disparities that exist in their communities, determine the precise SDoH affecting their members, and develop initiatives designed to meet these needs. This can be challenging because the most relevant SDoH will vary by community and individual and will generally reflect factors outside the healthcare setting. Thus, HCOs will need to collect non-medical, demographic data to identify the most actionable areas for improvement and the population subsets at the highest risk. To conduct this analysis, HCOs can distribute social needs assessment surveys, collaborate with CBOs, and draw on publicly available datasets, such as the Area Deprivation Index.

Please read our post, 3 Ways Healthcare Organizations Can Advance Health Equity by Addressing Social Determinants of Health, for more information on the importance of addressing SDoH, how to use your available data, and how you partner with CBOs.

Racial biases drive poor health outcomes and must be explicitly addressed and dismantled internally. Despite years of improvement, racial biases persist in healthcare through structures, policies, and discrimination. COVID-19 evinced this clearly; racial minorities had over three times more premature excess deaths per 100,000 than white people in 2020, reflecting increased risk of exposure to COVID-19 due to socioeconomic disparities and barriers to care.

Racial biases are not always overt, but it’s critical to identify and minimize them. People may be willing to admit that biases exist at a system level, but may be hesitant to address them or acknowledge their own biases. For example, in a recent study from the Society of Maternal- Fetal Medicine, 84% of physician respondents agreed that race-based disparities in care delivery were negatively impacting their practice, but only 29% believed that their personal biases were affecting how they cared for their patients. To help expose and unseat even subconscious racial biases, HCOs can educate staff by holding recurring unconscious bias training sessions and seminars about racial biases.

If HCOs are serious about advancing health equity, they must deeply understand the issues facing their population of interest and consider collaborating with CBOs that systematically link healthcare and social services for the people at risk of poor outcomes. Rather than designing equity programs for their communities, HCOs should strive to design them with their communities. This may help to overcome historic distrust of medical institutions, more accurately address specific issues people are facing, and provide partnerships to more efficiently carry out intervention efforts.

The Health Improvement Partnership of Santa Cruz County is an excellent example of the impact HCOs can have when partnering with other health systems, CBOs, and community representatives. Since its founding in 2004, the coalition of HCOs and CBOs has successfully reduced the rate of uninsured residents in the county by 75%.

The Center for Health Care Strategies has also compiled a repository of resources to help organizations streamline CBO collaborations, including a partnership assessment tool. This tool is designed to help partners “understand progress toward benchmarks characteristic of effective partnerships, identify areas for further development, and guide strategic conversations.”

Launching Your Strategy

Developing a strategy that adheres to these five actions will foster more equitable health outcomes and will help to ensure that the pursuit of equity remains a central focus. However, creating an effective strategy to prioritize health equity and putting it into practice are two different things. Once your organization has established advancing equity as a guiding principle, it’s important to consider how you plan to drive long-term program success.

Strategy In Practice: Rush University Medical Center

Rush University Medical Center is an excellent example of an organization that successfully prioritized health equity and launched an effective, long-term initiative to improve outcomes. In 2016, the health system rolled out their strategy to address race-based life expectancy gaps in Chicago.

The city has the largest racial mortality gaps among the 30 largest U.S. cities, and Rush identified that a significant portion of premature mortality is caused by chronic conditions in racially segregated neighborhoods. In fact, there is a 14-year life expectancy disparity between neighborhoods just two metro stops away, and Rush identified that economic factors are partially responsible for this disparity. Downtown Chicago has a median household income of $107,000, whereas Garfield park, a nearby, predominantly Black neighborhood, has a median household income of $22,000.

To address the SDoH factors that contribute to this health disparity, Rush adopted an organizational health equity strategy centered on eliminating structural racism and economic deprivation in their community. Their strategy incorporated the IHI’s five key actions despite being developed independently. Rush launched the strategy with support from a dedicated senior executive team, collaborated with the community and local organizations, looked internally to confront biases, developed measurable goals, and listened to the people most negatively affected by SDoH to drive the initiative.

In 2017, Rush began to execute on their strategy by hiring, purchasing, investing, and volunteering locally to help ameliorate the financial pressures they had identified as driving mortality. As one of the largest employers in west Chicago, their first mission centered on using their business units to promote economic activity, wealth-building, and community health. Moreover, they aggressively targeted financial inequities that existed within their organization, as many of their employees resided in economically disadvantaged communities and were experiencing financial hardships.

To ensure the success of their overall health equity strategy and their initial, economic-focused initiative, Rush created an employee resource group and held regular discussions with employees and community members. This had the effect of embedding a focus on health equity in the heart of the organization, ensuring it represented more than a one-time effort, and helping to guide internal practices. Since the strategy was adopted, Rush has:

• Opened 16 employment application hubs in targeted communities to support local hiring
• Increased hiring from these communities by 2% as of FY 2021
• Increased the percentage of employees contributing at least 6% of their income to retirement by 12%
• Raised hourly wages, launched a career development program, and began offering financial wellness and credit training
• Allocated $6 million (1% of its unrestricted reserves) to invest in local community projects and organizations

Rush also launched several initiatives to address SDoH and ensure their impact on achieving health equity was transparent, measurable, and consistently increasing. To this end, Rush established a multidisciplinary health equity oversight committee that provides input on organizational performance improvement projects involving race, ethnicity, gender, and age-related inequities. They also introduced new screening tools to capture SDoH data, such as food and housing security and access to transportation.

With measurement and an internal oversight committee guiding their efforts, Rush began an ambitious project to tackle SDoH, eliminate the life expectancy gap, and advance health equity: the foundation of a coalition that enlisted every health system and community-based organization on Chicago’s west side. West Side United (WSU) was created to standardize goals and processes related to health equity and share best practices across these organizations.

The organization was founded with input from local residents, continues to be driven with their guidance, and follows a key principle: Chicago’s disadvantaged neighborhoods know best about the challenges they face. Every WSU project is designed and executed collaboratively with input from the community it is intended to benefit, and the organization carefully avoids dictating change from a position of power. As one resident put it, “Don’t make top-down decisions and then invite everyone to something that’s already been decided.”

WSU has made remarkable progress towards eliminating health inequities and uniting healthcare organizations and community members behind a shared goal. In the first three years since its foundation, WSU has:

• Invested $7.6 million in west side projects that address SDoH
• Raised $3 million to establish career pathways in healthcare for disadvantaged Chicago residents
• Launched health outcomes initiatives targeting chronic diseases and maternal infant outcomes that disproportionately affect people of color
• Hired more than 2,000 employees from Chicago’s west side
• Raised $835,000 to support local businesses and offered business development support to over 640 local businesses
• Supported 60 community-based not-for-profit organizations

Five Recommendations To Get Started

Rush developed a set of five overarching recommendations for successfully putting a health equity strategy into practice that they base on their experience combating structural racism, improving health outcomes for underserved communities, and collaboratively founding the WSU. For other HCOs aspiring to tackle inequities, they recommend:

These guidelines share the IHI’s focus on responsible leadership and supporting initiatives throughout the organization, but Rush also emphasizes the importance of measurement, clear goals, and accountability. Without explicit equity goals that leaders value as highly as key financial and performance metrics, initiatives may fall by the wayside. As Ruch states, “tackling systemic racism, economic inequities, and other social and structural afflictions…is simultaneously a necessary and daunting long-term task.” Success is dependent on resolute commitment.

Finally, everyone has a vested interest in creating a more just system, and Rush credits their progress thus far to following the guidance of their local communities. They share that “listening sessions in the community and a commitment to share decision-making with community leaders are foundational to [their] strategy and necessary to overcome historical mistrust. Rush’s community efforts are guided by the voice of the community: ‘Nothing about us without us.’”‍

Additional Resources

For more information about developing a strategy to prioritize health equity and partnering with local communities, please check out the American Public Health Association’s health equity resource page. They provide an excellent series of fact sheets that detail everything from COVID-19’s impact on housing instability to combating environmental health disparities for children.

This post is part of our health equity series. If you’re interested in learning more about health equity and what can be done to achieve it, please check out our comprehensive overview post, Why Health Equity Matters in 2022, and our other posts on health equity:

• Diving into CMMI’s New Health Equity Objective For 2022 and Beyond
• 3 Ways Healthcare Organizations Can Advance Health Equity by Addressing Social Determinants of Health
• How COVID-19 Exacerbated Health Disparities‍
• Four Steps to Measure and Mitigate Algorithmic Bias in Healthcare

The post How You Can Develop and Launch a Strategy to Prioritize Health Equity appeared first on WellSect.

COVID-19 simultaneously exacerbated existing health disparities and introduced entirely new ones. The pandemic disproportionately impacted people of color, and due to a combination of persistent health disparities and social determinants of health (SDoH), they are at higher risk for infection, severe illness, and death.

The Kaiser Family Foundation (KFF) found that early in the pandemic “Black people accounted for more cases and deaths relative to their share of the population in 30 of 49 states reporting cases and 34 of 44 states reporting deaths.” Moreover, they found “the COVID-19 related death rate among Black people was over twice as high as the rate for White people.” While this specific gap has narrowed since 2020, the KFF’s recent research reflects persistent health disparities. Age-adjusted data shows that Black, Hispanic, and Native American people are still more than twice as likely to die from COVID-19 compared to White people. Further, Hispanic and Native American people are at almost two times greater risk for infection than White people.

These disparities reflect historic barriers to care, structural racism, and socioeconomic forces that persistently disadvantage minorities. We’ll explore the most important of these barriers below.

Low-income populations and people of color are more likely to be uninsured. The rate of uninsurance for Native American and Hispanic people is nearly three times higher than the rate for white people. This gap is expected to widen; nearly 50% of coverage in 2019 was distributed through employers, and the pandemic sparked waves of business closures and mass-layoffs. Many workers in the service, manufacturing, and retail industries held positions that could not be performed remotely, and survey data reports that approximately 60% of Hispanic households and roughly 50% of Black households lost a job due to the pandemic, compared to just 40% of White households.‍

Isolation and quarantine are far more difficult for individuals living in shared housing conditions, and a higher percentage of racial and ethnic minorities live in crowded or multi-generational housing. Further, racial and ethnic minorities are more likely to be exposed to COVID-19 through their occupation or method of transportation. The CDC notes that these groups are “disproportionately represented in essential work settings such as healthcare facilities, farms, factories, warehouses, food processing, accommodation and food services, retail services, grocery stores, and public transportation.” These roles frequently require extended periods of close contact with many others, often in heavily trafficked spaces, resulting in increased risk of exposure.

Extensive travel times and lack of testing sites in one’s community delays testing and subsequent care for minorities at a greater rate than for White people. A study of geographic access to testing early in the pandemic found that 20 minutes was the median travel time to a testing site, but that counties with longer travel times had a higher percentage of the population that were from racial and ethnic minority groups and people living in rural areas. Testing availability has greatly improved since the first months of the pandemic, but inability to take off work, less available transportation, and fear of exposure when using public transportation still disproportionately delays testing for minority populations and low-income communities.

The pandemic is also imposing economic pressure on low-income communities and minorities. These populations are more likely to hold jobs that can’t be performed remotely, potentially leading to either unemployment due to pandemic-related shutdowns or increased risk of contracting COVID-19 due to greater exposure. The KFF reports that “larger shares of Black and Hispanic adults report experiencing negative economic impacts [during the pandemic] with about half saying they have had difficulty paying for basic expenses…compared to 31% of White adults.”

Economic pressure contributes to worse health outcomes and greater disparities, elevating stress and potentially leading to increased exposure and vulnerability. Affected individuals may be forced to work while sick, be unable to seek medical care, struggle to purchase medications, have difficulty maintaining insurance coverage, and may struggle to purchase food and meet other essential needs.

Racial minorities also have a disproportionate burden of chronic diseases, including obesity, diabetes, and kidney disease, and these conditions are a risk factor that may lead to more severe cases of COVID-19. These comorbidities contribute to increased hospitalization and death rates from COVID-19, but the pandemic has also impacted the care and management of chronic conditions independent of COVID-19.

With healthcare resources reserved for the pandemic and fear of increasing the spread, a substantial proportion of people deferred care or completely avoided engaging with health services. One study found that “an estimated 41% of U.S. adults had delayed or avoided medical care including urgent or emergency care (12%) and routine care (32%).” The full impact of COVID-19 on chronic conditions will not be clear for years to come, but this significant gap in diagnoses, routine care, and preventive interventions is likely to dramatically widen associated health disparities.

COVID-19 is further disadvantaging children from low-income families and widening academic achievement gaps that will eventually translate to increased risk for worse health outcomes. Education is a major SDoH, and adults with less education report worse general health, more chronic conditions, and more functional limitations than those with higher levels of education. School closures have the greatest negative impact on children that rely on sponsored programs, such as food and nutritional support initiatives, after-school care, and mental health or social services. Moreover, children without internet access and other technology may be completely left behind by the shift to online learning. Much like the pandemic’s yet-uncertain impact on chronic conditions, its ramifications on education-based health disparities are unclear but likely to be substantial.

Addressing COVID-19 Inequities Starts With Data

To combat disparities caused or exacerbated by COVID-19, healthcare organizations (HCOs) must collect and use data that reflects race, ethnicity, and SDoH. Unfortunately, much of the available COVID-19 data at the federal and state levels lacks this information. As the KFF reports, “nearly two years into the pandemic, we still lack comprehensive data to understand disparities in COVID-19 impacts and uptake of the vaccines.” These limitations make accurate estimates more difficult, and even now race and ethnicity data regarding vaccination is missing for nearly 30% of people who have received at least one dose. Similar gaps in demographic data also exist for breakthrough infection potential.

Effective data collection is an essential first step to addressing health disparities associated with COVID-19 and advancing health equity at scale. Without the necessary data, HCOs attempting to address the impact of COVID-19 will be unable to pinpoint health disparities, develop interventions, and precisely target their efforts to the people at the highest risk for preventable, negative outcomes.

Partnering With Community-Based Organizations to Drive Impact

Fortunately, community-based organizations are also working to document, track, and address disparities tied to COVID-19. Partnering with these organizations may enable HCOs to more effectively identify and help those with the greatest need. For example, NPR recently highlighted the efforts of the Black Equity Coalition, a Pittsburgh-based team, that identified the need for more rigorous data collection to address health disparities caused by the pandemic.

Members of the coalition understood that minorities in Pittsburgh were being disproportionately affected, but realized that testing efforts were not recording race. At the state and local levels there was no requirement in place to record it. Without data that reflected race, the coalition and Pittsburgh’s HCOs, “couldn’t target [their] attention and know who needed the help most.”

Together, the team mapped and analyzed the locations of COVID-19 testing centers. They found that the people most likely to have easy access to testing lived in neighborhoods predominantly populated by White people, were generally employed in white-collar jobs, and had the financial means to weather the pandemic with delivery services and remote work. Moreover, labor statistics revealed that essential workers keeping Pittsburgh afloat were overwhelmingly people of color. ‍

The Coalition started analyzing the existing data and pushing for improved data collection efforts from local HCOs. They used racial poverty data and geographic data for federally qualified health centers to help advise health authorities and prioritize testing in previously overlooked communities. Using county data, they uncovered that Black people were experiencing hospitalization and more severe cases of COVID-19 at much greater rates.

Their work made a difference; Allegheny county records now only omit race in 12% of positive cases. However, 37% of Pennsylvania testing records are still lacking race. Having helped to shine a light on racial disparities, they provided two key recommendations for HCOs trying to advance COVID-19 equity:

• Adopt new technology protocols to ensure race and demographic data is always collected.
• Share data with other local organizations and form partnerships to enact positive change.

Data-Driven Initiatives Are Effective on a National Scale

On a national scale, efforts to address vaccination disparities have also highlighted the importance of data collection and the potential for organizations to impact disparities through data-driven, coordinated initiatives. For the first few months the vaccines were available to the U.S. adult population, the Black and Hispanic vaccination rates consistently trailed White and Asian vaccination rates by upwards of 10%. These groups faced access barriers to vaccination, such as a lack of local vaccination sites, and historic racism drove vaccine distrust. In particular, the horrific Tuskegee syphilis study contributed to vaccine hesitancy for Black people. ‍

Vaccination rate data collection and subsequent efforts to close the gap have had a notable impact. As of February 2022, the Hispanic vaccination rate has matched the White rate, and the vaccination rate for Black people has made up nearly 8% of the initial deficit—although significant room for improvement remains. Progress thus far was achieved by identifying the gap and targeting specific populations with interventions designed to address their vaccine education and access needs. The New York Times (NYT) reported on a number of effective initiatives led by hospital systems and community groups, including:

• Door-to-door canvassing about the benefits of vaccination
• Hosting pop-up clinics in communities with insufficient access to vaccines
• Public awareness campaigns to improve vaccine education
• Providing free transportation to vaccination sites

One community organizer encapsulated the importance of these targeted interventions in their comment to the NYT. When discussing their rural, largely Black community, they said, “[the] group of people in this given area or this community don’t have the information or access they need to overcome their hesitancy.”

Without accurately identifying low inoculation rates across race and ethnicity and addressing the specific needs of these communities, vaccination disparity would simply persist. Efforts to eliminate it are ongoing, but HCOs and community leaders have demonstrated impact and made substantial progress. Ultimately, tackling this disparity and addressing the full spectrum of COVID-19-related health disparities hinges on effective data collection and data-driven, targeted interventions. ‍

How AI Can Help

As of February 2022, there have been 75 million reported cases of COVID-19 in the U.S. and nearly 900,000 deaths due to COVID-19. This suffering has disproportionately impacted racial and ethnic minorities and surfaced a plethora of health disparities. However, HCOs can prioritize race and ethnicity data collection, partner with community-based organizations, and develop effective interventions to mitigate and eventually eliminate disparities in their populations.

To this end, AI solutions can help HCOs improve their data collection processes, normalize disparate data, and proactively address disparities with predictive models. With AI, organizations can optimize use of limited care management resources and target interventions with greater efficiency.

For example, In March 2020, Healthfirst, one of New York’s largest not-for-profit health insurers, integrated their member data with WellSect’s COVID-19 Vulnerability Index (C-19 Index), a series of AI/ML models that predict vulnerability to severe complications from COVID-19. The C-19 Index showed that vulnerability to serious illness from COVID-19 was concentrated—5% of Healthfirst’s population experienced 55% of poor outcomes. As a result, Healthfirst was able to more precisely leverage their resources and better safeguard their most vulnerable members.

To date, the C-19 Index has helped organizations evaluate more than 10 million lives to identify individuals at higher risk of developing severe complications during the pandemic. Please click here to learn more about the C-19 Index and how WellSect is helping to fight COVID-19.

This post is part of our health equity series. If you’re interested in learning more about health equity and what can be done to achieve it, please check out our comprehensive overview post, Why Health Equity Matters in 2022, and our other posts on health equity:

• Diving into CMMI’s New Health Equity Objective For 2022 and Beyond
• 3 Ways Healthcare Organizations Can Advance Health Equity by Addressing Social Determinants of Health‍
• How You Can Develop and Launch a Strategy to Prioritize Health Equity‍
• Four Steps to Measure and Mitigate Algorithmic Bias in Healthcare

The post How COVID-19 Exacerbated Health Disparities appeared first on WellSect.

Social determinants of health (SDoH) profoundly affect a person’s overall health, and according to the Centers for Disease Control and Prevention (CDC), addressing them is one of the “primary approaches to achieving health equity.” SDoH, defined as “the non-medical factors that influence health,” encompass the conditions in which people are born, grow, work, live, and age, including:

• Individual and household income
• Early childhood development
• Level and quality of education
• Access to nutritious food
• Access to health services of sufficient quality
• Living conditions and housing status
• Level of social inclusion and access to social services

These factors are rooted in socioeconomic status, greatly impact health outcomes, and directly contribute to persistent health inequities. The World Health Organization (WHO) estimates that SDoH account for up to 30–55% of health outcomes globally and contribute more to overall population health than medical care or genetics. While there is no consensus on the exact impact of SDoH outcomes within the U.S., there is an extensive body of evidence indicating their overwhelming importance. One study concluded that “potentially avoidable factors associated with lower educational status account for almost half of all deaths among working-age adults in the U.S.”

Socioeconomic factors are also closely linked to the development of health-related behaviors that affect outcomes. For example, children exposed to smoking and substance abuse in their communities are more likely to adopt these practices later in life. Similarly, fewer opportunities to exercise and eat well as a child can lead to developing behaviors and habits that may increase risk for chronic diseases.

Addressing SDoH is an economic imperative. Health disparities cost hundreds of billions of dollars annually in direct medical costs and indirect costs tied to lost productivity. This is especially concerning, as national health spending is projected to reach $6.2 trillion by 2028, but 25% of total healthcare spending is estimated to be wasted. Given the estimated impact of SDoH relative to medical care, proactively addressing the socioeconomic factors propagating health disparities may significantly reduce this excess spending while improving outcomes.

An Example of SDoH: Where You Live Matters

The definition of SDoH is broad and examining a single subset of social factors can help contextualize how significantly they contribute to health outcomes and propagate health disparities. Consider just the environment in which people live. A person’s ZIP code is a stronger predictor of their overall health than other factors, and in some U.S. cities, a child’s life expectancy can vary more than 25 years between neighborhoods that are only a few miles apart. Where someone lives is associated with a variety of social factors that can negatively affect health, and may include:

Living space determines proximity to grocery stores and other sources of healthy foods, as well as the availability of reliable transportation required to reach these locations. This phenomenon, known as a food desert, may deprive people of access to affordable food options and leave them with insufficient nutrition, ultimately leading to higher rates of chronic conditions and adverse health outcomes. Research estimates that food deserts affect 19 million people in the U.S. (6.2% of the population) and disproportionately affect Black, multiracial, and low-income communities.

Living in communities with higher crime rates and violent incidents is associated with worse health outcomes. Even if an individual does not personally experience such events, they can still suffer indirect consequences. According to the Department of Health and Human Services, “children exposed to violence may experience behavioral problems, depression, anxiety, and post-traumatic stress disorder.”

According to the WHO, one third of all deaths from stroke, lung cancer and heart disease are due to air pollution. The WHO also notes the impact on children, stating that “worldwide, up to 14% of children aged 5-18 years have asthma relating to factors including air pollution.” Water pollution is just as impactful. Recently, the Flint water crisis led to 12 deaths and left tens of thousands of people to subsist on lead-contaminated drinking water. Unfortunately, the impact of environmental factors will only continue to increase as climate change accelerates.

Living spaces that don’t support regular exercise can also significantly contribute to worse health (e.g., greater rates of obesity and diabetes). Neighborhoods may suffer from a lack of safely walkable areas, a dearth of public transportation options, and limited or no recreational spaces, such as public parks, pools, and fields. Further, high levels of crime and violence may prevent residents from exercising in public spaces.

The significance of living conditions extends far beyond the examples listed above. Due to their place of residence, people may also lack access to broadband internet, health services, educational opportunities, and may be exposed to frequent tobacco use and substance abuse.

How Healthcare Organizations Can Address SDoH

Addressing SDoH is critical to achieving health equity and succeeding in value-based care. To this end, healthcare organizations (HCOs) can participate in CMS’s alternative payment models (APMs), which incentivize tackling the root causes of poor health and provide reimbursement for quality of care rather than quantity of services provided. Further, HCOs can collaborate with community-based organizations that systematically link healthcare and social services for people at risk of poor outcomes. Increasingly, these collaborations are forming to specifically address SDoH, and partnerships between clinicians, social service agencies, and health systems are on the rise.

Collaborate With Community-Based Organizations

One such collaboration is the Pathways Community HUB care management model (HUB). Originating in Toledo, Ohio, the HUB is a value-based approach to care coordination that specifically addresses SDoH by connecting at-risk individuals with local healthcare organizations and community health workers. Payment is predicated on risk mitigation and achieving specific outcomes; for example, meeting a six-month sobriety goal for a person that previously struggled with alcohol or ensuring a previously unemployed person has been hired and remains employed for several months.

The HUB care model targets individual, modifiable risk factors (e.g., lack of housing), and uses a technical tool called a “Pathway” that serves as a comprehensive checklist to address each factor. Pathways help to outline step-by-step improvements and ultimately lead to completing a specific outcome goal. While Pathways help to ensure meaningful progress, each individual Pathway is also considered in tandem with all of a person’s other risk factors. For example, “An expectant teenage mother at risk for a low-birthweight, preterm delivery who is simultaneously homeless, depressed, and without access to medical care must have all three factors addressed, since fixing one by itself is unlikely to make much difference in the outcome.”

Rather than creating Pathways for each issue and attempting to resolve them all at once, health workers identify the person’s most pressing need and address it first. Using Pathways, they work with people to make steady progress on their most urgent issues and eventually work towards less critical factors over time. Even if some Pathways aren’t fully completed through this approach, the data they produce is still valuable to HUB. Information on unfinished Pathways may demonstrate a need for greater investment in community infrastructure and help local HCOs identify specific social factors creating health disparities in their population.

Pathways Community HUB has received funding and support from the Agency for Healthcare Research and Quality (AHRQ), has established a national certification program, and continues to grow. Today the HUB model is helping to close health disparities for nearly 8,000 people monthly with approximately 400 certified health workers.

The Idaho Health Data Exchange (IHDE) is another example of HCO collaboration designed to address SDoH. A statewide data exchange that enables HCOs to coordinate care, the organization recently partnered with Aunt Bertha, a search and referral platform for social services, to educate its users on SDoH and bridge the gap between clinical environments and their local communities. Now, they’re enabling providers and payers to connect with community-based organizations across the state, improving quality of care and directly impacting social factors.

Three Steps to Get Started

1. Collect Data and Identify Specific Needs

Identifying specific social needs by collecting non-medical, demographic data is the first step towards launching a SDoH initiative or beginning a partnership with community-based organizations. Collecting SDoH data is key to pinpointing the most actionable areas for improvement and the population subsets at the highest risk for negative outcomes due to specific SDoH factors. To conduct this analysis, HCOs can distribute social needs assessment surveys, organize meetings with local leaders and social services organizations, and draw on publicly available datasets, such as the Area Deprivation Index.

To aid in this process, AHRQ has created a repository of tools designed to help HCOs assess social risks and needs. Their resources include:

• Screening tools and surveys that evaluate SDoH needs
• SDoH data collection, summary, and referral tools
• Tools to help organizations form clinical-community linkages
• Care coordination resources and tools for delivering culturally competent care
• Literature about federal data sources that can be used to measure SDoH

Collecting SDoH data and survey responses regularly is critical; social circumstances change over time and the issues people face may not be reported in clinical encounters. Surveys and other outreach methods that inquire about SDoH can help to foster discussion and encourage people to share their needs.

2. Research Local Community-Based Organizations

Once enough data has been collected to determine which modifiable social factors are having the greatest negative impact, HCOs will be able to take the next step—identifying and researching community-based resources that map to these factors. Several resources exist to help HCOs familiarize themselves with local community resources and discover organizations they may be unaware of:

• The American Academy of Family Physicians Neighborhood Navigator is a regularly updated database of community-resources organized by ZIP code. Neighborhood Navigator also includes tools to manage referrals and shareable literature that explains how individuals can use and benefit from specific CBO services.
• The National Association of County and City Health Officials provides HCOs with a directory of U.S. health departments that are categorized at the county and city level. Local health departments generally have long standing relationships with CBOs in the same area and can help connect HCOs to the most appropriate organizations.
• Aunt Bertha’s search and referral platform for social services enables HCOs to identify and contact a myriad of local organizations dedicated to addressing SDoH.

3. Form Partnerships and Set Measurable Goals

When engaging with CBOs that share a commitment to addressing a specific SDoH, HCOs should come to the table with an understanding of their available resources and how they can most effectively drive productive collaboration. Generally, HCO involvement will center on referring people to services offered by the CBO and supporting the CBO’s existing efforts. However, these actions require internal support and communication. HCOs should:

• Disseminate key information about the collaboration and ensure employees are well informed about the services offered. They must be able to clearly communicate the benefits to patients or members and easily navigate the referral process.
• Establish an easy-to-use communication channel that enables employees to inform CBO workers when they refer someone.
• Set appropriate expectations regarding the partnership’s scope and the expected results. Addressing SDoH is a complex process that hinges on factors beyond a traditional clinical setting, and HCOs should determine realistic goals that account for both organizations’ resource availability and staffing capacity.
• Develop key outcomes or metrics of progress to track performance, assess impact, and use this information to improve programs whenever possible. This data can be further augmented with stakeholder feedback and qualitative assessments that evaluate how specific processes, such as outreach messaging, could be improved and what has been most effective.

The Center for Health Care Strategies has also compiled a repository of resources to help organizations streamline collaborations that address SDoH, including a partnership assessment tool. This tool is designed to help partners “understand progress toward benchmarks characteristic of effective partnerships, identify areas for further development, and guide strategic conversations.”

Addressing SDoH With Technology

Driving new and existing initiatives with SDoH data and using this data to design and collaborate on interventions will help close health equity gaps. HCOs can also go a step further by implementing technology solutions to coordinate care, predict risk tied to SDoH on an individual level, and surface the specific SDoH factors that contribute most significantly to increased risk. In particular, AI solutions can help organizations use limited care management resources and target SDoH interventions with greater efficiency.

For example, Genesis Physicians Group, the largest network of independent primary and specialty care physicians in North Texas, is using WellSect to build predictive AI models that help the network mitigate SDoH contributors to health risk.

To learn more about how Genesis Physicians Group and WellSect are using AI to tackle SDoH and advance health equity, please check out this video interview with Dr. Jim Walton, President and CEO of Genesis Physicians Group, and Carol McCall, WellSect’s Chief Health Analytics Officer.

This post is part of our health equity series. If you’re interested in learning more about health equity and what can be done to achieve it, please check out our comprehensive overview post: Why Health Equity Matters in 2022, and our other posts on health equity:

• Diving Into CMMI’s New Health Equity Objective For 2022 and Beyond
• How COVID-19 Exacerbated Health Disparities‍
• How You Can Develop and Launch a Strategy to Prioritize Health Equity‍
• Four Steps to Measure and Mitigate Algorithmic Bias in Healthcare

The post 3 Ways Healthcare Organizations Can Advance Health Equity by Addressing Social Determinants of Health appeared first on WellSect.

Advancing health equity is chief among the Centers for Medicare and Medicaid Services’s  (CMS) new strategic objectives, and from 2022 on, achieving this goal will guide every aspect of CMS’s work. CMS established this new objective in October of 2021 through a complete strategic refresh for its Innovation Center (CMMI) that hinges on five strategic pillars:

1. Driving Accountable Care
2. Advancing Health Equity
3. Supporting Innovation
4. Addressing Affordability
5. Partnering to Achieve System Transformation

These objectives will drive major policy changes and serve as the foundation for the Biden administration’s overarching goal of “creating a health system that achieves equitable outcomes through high-quality, affordable, person-centered care.”

The new strategy was announced in a live webinar, led by CMS Administrator Chiquita Brooks-LaSure and CMMI Director Elizabeth Fowler, and an accompanying white paper: Driving Health System Transformation: A Strategy for the CMS Innovation Center’s Second Decade. Each core objective will be measured and considered in every stage of payment model design, testing, and evaluation, and CMS Administrator Chiquita Brooks-LaSure said, “everything we do at CMS should be aligned with one or more of our strategic pillars.”

Now more than ever, CMS is prioritizing its commitment to promoting health equity, as “ensuring health equity is embedded in every [payment] model” is the number one lesson learned over the past decade. Their new objective defines achieving health equity as “the attainment of the highest level of health for all people” and comes at a pivotal moment. Health disparities are persistent, and COVID-19 recently overturned years of incremental progress towards closing the gaps, exacerbating existing inequities while creating new ones.

Challenges to Achieving Health Equity

Over the past decade, CMMI experienced difficulties thoroughly integrating equity in value-based payment model design and reaching their desired level of impact. Now, the Center plans to incorporate learnings from their portfolio of over 50 payment models, which have reached millions of patients and over 500,000 distinct providers and health plans.

CMMI extensively reviewed past model design, participation, and results, stating that “each of the 50+ models launched in the past decade has yielded important policy and operational learnings” to help accelerate value-based care and produce more equitable outcomes.

Through these reviews and external feedback, CMMI identified the following challenges that have historically limited efforts to advance equity:

CMMI plans to address these challenges with insights gleaned from value-based payment models that directly address health inequalities in their design, such as the Community Health Access and Rural Transformation (CHART) and the Maternal Opioid Misuse (MOM) model. Further, the Center will draw on the CMS Equity Plan For Improving Quality in Medicare and maintain a focus on increasing awareness and understanding of disparities, developing and disseminating solutions to advance equity, and implementing sustainable strategies.

CMMI’s Four Key Efforts to Advance Health Equity

‍CMMI is focusing their efforts around four key areas to address the challenges detailed above. These actions will broaden the reach of models, provide previously unavailable measures of impact, and help to refine existing models with new learnings. We’ll review each of these four areas below.

According to CMMI, many of their prior models were designed with advancing health equity in mind, but their model portfolio as a whole did not systematically address it or include sufficient demographic measures and impact evaluation. Now, CMMI will thoroughly evaluate and identify opportunities to embed equity across all models—throughout design, testing, and evaluation.

Many past models assessed beneficiaries’ social needs, and CMMI will expand its efforts to utilize screening tools and coordinate with community partners to address these specific needs. This will potentially include testing for certain clinical conditions, evaluating care settings, and introducing models designed to remedy community-level issues associated with social determinants of health (SDoH).‍

Increasing beneficiary diversity and reaching underserved populations that prior models may not have included is paramount. CMMI notes that beneficiaries enrolled in the Next Generation Accountable Care Organization (ACO) model and in advanced primary care models are more likely to be White, less likely to be dual-eligible, and less likely to live in rural areas compared to other fee-for-service (FFS) beneficiaries in the same markets. Analysis of direct contracting models revealed similar findings.

To extend impact to these previously overlooked, underserved beneficiaries, CMMI will focus on engaging with local communities and public health leaders. Upcoming outreach efforts will target providers that may not have previously participated in value-based care, focusing on providers that disproportionately care for underserved populations. As such, CMMI will review their application and selection process to ensure providers don’t face barriers to engagement and are not disincentivized from participating.

Evaluating the challenges rural providers faced in past models also indicates that safety net providers often require financial and technical assistance to provide equitable care. These providers may lack the necessary infrastructure, and CMMI is considering a variety of incentives to drive and sustain their participation. Potential incentives include, upfront payment, social risk adjustment and payment incentives based on screening, and incentives for collaboration with community-based organizations.

Accurately and systematically assessing model impact on health equity is critical to ensuring long-term program success. In order to standardize evaluation, CMMI is developing new impact assessment measurements and evaluation requirements. Beyond individual models, the Center will also track impact across its entire portfolio and determine how to share model-specific findings with partners and participants. To this end, consistent evaluation will help CMMI iterate on models when applicable and integrate new learnings at scale.

The availability of demographic data that includes factors such as race, ethnicity, disability, and geography is essential to designing impactful models and evaluating their effectiveness. Going forward, CMMI will require new model participants to collect and report demographic data and measure disparity prevalence in their populations. CMMI is also considering new requirements, incentives, and data collection methods for participants in existing models. For example, they are evaluating the use of other federal data sources, such as the Transformed Medicaid Statistical Information System (T-MSIS).

CMMI also aims to share more data with participants and plans to use dashboards to directly deliver information. They are considering augmenting patient and provider data with area-level indices, such as the Area Deprivation Index (ADI).

CMMI’s Next Steps

CMMI shared details about their immediate next steps, providing a closer look at how they plan to execute on their four key efforts. In the near future, CMMI will:‍

• Address barriers to participation. Some aspects of model design and the application process have historically limited engagement from rural and safety net providers, and CMMI will ensure they are not disincentivized from participating.
• Create new data collection requirements. These new requirements will necessitate beneficiary-level demographic data and track model impact for underserved beneficiaries. This may also include financially incentivizing and supporting data collection needs when appropriate.
• Collect and leverage social needs data. CMMI will screen for social needs, coordinate with community-based organizations, and collect social needs data in standardized formats.
• Analyze and learn from program data. The characteristics of participating providers and beneficiaries will be evaluated and used to help ensure equitable reach of models.
• Create new quality measures. These measures will incentivize the reduction of health disparities and measure model and provider performance.
• Provide support for equity education. Model participants caring for underserved populations will receive training and technical support as appropriate, and CMMI will share best practices for partnering with community-based organizations.

Additionally, CMMI is conducting a series of roundtable discussions to incorporate external perspectives from providers, community-based organizations, and health equity experts. The first of these conversations was held in early December of 2021, and participating stakeholders recommended that CMMI focus on investing in geographically-focused multi-payer models and refining measurement.

While it remains to be seen exactly how CMMI will integrate recommendations and feedback from this discussion series, notable organizations have voiced their approval for both the new strategic objectives and CMMI’s focus on collaboration with partners. The National Association of ACOs (NAACOs) wrote a letter in response to the health equity roundtable, expressing their excitement about the Center’s aggressive stance on advancing equity. They also shared a series of recommendations and proposed several quality measurement changes that they claim will help drive increased program participation.

Ultimately, CMMI’s health equity objective presents an opportunity for HCOs to reduce health disparities across their populations and enroll in models designed to produce and financially incentivize more equitable outcomes. CMMI has stated that they are considering upfront payments, social risk adjustment, and payment incentives for reducing disparities, coordinating with community-based organizations to address social needs, and SDoH data collection. All of these considerations are intended to enable greater participation and may attract organizations that haven’t previously engaged in value-based care.

This post is part of our health equity series. If you’re interested in learning more about health equity and what can be done to achieve it, please check out our comprehensive overview post: Why Health Equity Matters in 2022, and our other posts on health equity:

• 3 Ways Healthcare Organizations Can Advance Health Equity by Addressing Social Determinants of Health
• How COVID-19 Exacerbated Health Disparities‍
• How You Can Develop and Launch a Strategy to Prioritize Health Equity
• Four Steps to Measure and Mitigate Algorithmic Bias in Healthcare

The post Diving Into CMMI’s New Health Equity Objective For 2022 and Beyond appeared first on WellSect.

In October 2021, the Centers for Medicare and Medicaid Services (CMS) launched a complete strategic refresh through their Innovation Center (CMMI) with the overarching goal of achieving equitable outcomes by providing high-quality, affordable, person-centered care. Their new strategy is centered around five core objectives that will guide alternative payment model design for years to come.

Advancing health equity is one of these initiatives.

WellSect and its partners consider advancing health equity an industry-wide imperative, and we’re committed to proactively embedding equity in every healthcare decision. Persistent disparities drive preventable suffering, produce negative outcomes, and erect barriers to achieving optimal health. We’re working to identify and eliminate these disparities with data, and we encourage you to join us in the pursuit of health equity.

As defined by the CDC, health equity is achieved “when every person has the opportunity to attain their full health potential and no one is disadvantaged from achieving this potential due to their social position or other socially determined circumstances.” This requires coordinated efforts to address pervasive inequalities, discrimination, and disparities.

According to the Healthy People 2020 report, health disparities are “closely linked with social, economic, and/or environmental disadvantage” and “adversely affect groups of people who have systematically experienced greater obstacles to health based on their racial or ethnic group; religion; socioeconomic status; gender; age; mental health; cognitive, sensory, or physical disability; sexual orientation or gender identity; geographic location.”

Health disparities are pervasive and have an immense impact on outcomes. For example, Black women are three to four times more likely to die a pregnancy-related death compared to White women, and they are up to 12 times more likely in some cities. Health disparities also result in avoidable, excess spending. Recent studies of racial equity estimate that $135 billion could be saved annually if racial disparities in health were eliminated, including $93 billion in excess costs of care.

Eliminating disparities and advancing health equity has always been a priority for CMS, but it’s now a major objective that will be measured and considered in every stage of payment model design, testing, participant and program evaluations, and ultimately reimbursement. All new value-based payment models will:

• Require participants to collect and report on demographics, and when applicable, social determinants of health (SDoH) data
• Include patients from underserved communities and identify opportunities to reduce inequities at the population level
• Ensure providers are not disincentivized from participating and potentially introduce new measures to assess impact on equity

CMMI’s new strategic objectives come at a pivotal moment, as the COVID-19 pandemic significantly exacerbated long-standing disparities. Life expectancy for Black people has consistently been lower than for White people, and while the gap shrunk over the past decade, COVID-19 increased it to a difference of six years—the largest gap since 1998. Minorities also had over three times more premature excess deaths per 100,000 than White people in 2020, reflecting increased risk of exposure to COVID-19 due to socioeconomic disparities and barriers to care.

With greater support from CMS, and COVID-19 widening the health disparity gap, there has never been a greater opportunity for healthcare organizations (HCOs) to develop and implement strategies that directly promote health equity. This comprehensive overview will enable you to prepare for upcoming policy changes, explore the root causes of health disparities, and begin developing an actionable health equity strategy your organization can employ. Specifically, we’ll cover:

• CMMI’s New Health Equity Objective
• How Social Determinants of Health Fuel Health Inequities and What You Can Do About it
• How COVID-19 Exacerbated Health Disparities
• Developing a Strategy to Prioritize Health Equity
• How to Measure and Mitigate Algorithmic Bias in Healthcare‍

Let’s get started.

Advancing health equity is a core objective of CMMI’s refreshed strategy to dramatically increase the expansion of value-based payment models that reduce costs, ensure quality care, and improve health outcomes for Medicare and Medicaid beneficiaries. Their new strategic objectives will heavily lean on the center’s policy and operational learnings, and advancing health equity draws directly from the number one lesson learned over the past decade: ensuring health equity is embedded in every payment model.

In outlining health equity as a core objective, CMMI detailed the challenges they’ve experienced, their main areas of focus moving forward, and the next steps they’ll take to create a more equitable system for everyone.

Historic Challenges to Ensuring Health Equity

Over the past decade, the center has experienced difficulties thoroughly integrating equity in payment model design and reaching their desired level of impact. To incorporate equity in every facet of model creation and implementation, CMMI conducted extensive reviews to identify the following challenges and issues that have historically limited progress:

CMMI’s Four Key Efforts to Advance Health Equity

CMMI is focusing their efforts around four key areas to address and overcome the challenges they’ve experienced. These actions will broaden the reach of models, provide previously unavailable measures of impact, and help to refine models with new learnings. CMMI will:

CMMI’s Next Steps

CMMI also detailed their next steps, providing a closer look at their key efforts and how exactly they plan to achieve this objective. In the near future, CMMI will:

• Address barriers to participation. Some aspects of model design and the application process have historically limited engagement from rural and safety net providers, and CMMI will ensure they are not disincentivized from participating.
• Create new data collection requirements. These new requirements will necessitate beneficiary-level demographic data and track model impact for underserved beneficiaries. This may also include financially incentivizing and supporting data collection needs when appropriate.
• Collect and leverage social needs data. CMMI will screen for social needs, coordinate with community-based organizations, and collect social needs data in standardized formats.
• Analyze and learn from program data. The characteristics of participating providers and beneficiaries will be evaluated and used to help ensure equitable reach of models.
• Create new quality measures. These measures will incentivize the reduction of health disparities and measure model and provider performance. ‍
• Provide support for equity education. Model participants caring for underserved populations will receive training and technical support as appropriate, and CMMI will share best practices for partnering with community-based organizations.

CMMI’s new models will directly incentivize advancing health equity. CMMI has announced that they’re considering upfront payments, social risk adjustment, and payment incentives for reducing disparities. This will enable greater engagement from HCOs that care for underserved populations and help to include organizations that have not participated in any of CMMI’s prior payment models.

For a more detailed breakdown of CMMI’s new health equity objective, please check out the full blog post: Diving Into CMMI’s New Health Equity Objective For 2022 and Beyond.‍

‍Social determinants of health (SDoH) profoundly affect a person’s overall health, and addressing them is key to advancing health equity. The World Health Organization (WHO) defines SDoH as “the non-medical factors that influence health outcomes. They are the conditions in which people are born, grow, work, live, and age, and the wider set of forces and systems shaping the conditions of daily life.”

SDoH have an immense impact on outcomes and fuel persistent health inequities. Studies suggest that SDoH account for up to 30–55% of health outcomes, and estimate that they make up a far greater contribution to overall population health than medical care or genetics.

There is also an economic imperative to address SDoH. Health disparities cost hundreds of billions of dollars annually in direct medical costs and indirect costs tied to lost productivity. This is especially concerning, as U.S. health spending is projected to reach $6.2 trillion by 2028, but 25% of total healthcare spending is estimated to be wasted. Given the estimated impact of SDoH relative to medical care, proactively addressing the socioeconomic factors propagating health disparities may significantly reduce this excess spending while improving outcomes.

An Example of SDoH: Where You Live Matters

The definition of SDoH is broad and it can be helpful to zoom in on a single subset of social factors to contextualize how significantly they contribute to health. Consider just the environment in which people live. A person’s ZIP code is a stronger predictor of their overall health than other factors, and in some U.S. cities, a child’s life expectancy can vary more than 25 years between neighborhoods that are only a few miles apart. Where someone lives is associated with a variety of social factors that can negatively affect health, including:

• Living in a food desert. Significant distance to grocery stores and other sources of food can deprive people of access to affordable options and leave them with insufficient nutrition.
• Exposure to violent crime. Communities with higher crime rates and violent incidents are associated with worse health outcomes.
• Local environmental factors. Local water and air pollution are detrimental to health, potentially leading to strokes and lung cancer.
• Barriers to physical activity. Neighborhoods lacking the infrastructure for regular exercise can significantly contribute to worse health (e.g., greater rates of obesity).

The significance of living space extends far beyond the examples listed above. People may also lack access to broadband internet, health services, educational opportunities, and may live in residences constructed with lead and asbestos.

How Healthcare Organizations Can Address SDoH

Addressing SDoH is critical to achieving health equity and succeeding in value-based care. To this end, HCOs can participate in CMS’s alternative payment models (APMs), which incentivize tackling the root causes of poor health and provide reimbursement for quality of care rather than quantity of services provided. In addition to partnering with CMS, HCOs can collaborate with community-based organizations that systematically link healthcare and social services for people at risk of poor outcomes. Increasingly, these collaborations are forming to specifically address SDoH, and partnerships between clinicians, social service agencies, and health systems are on the rise.

One such collaboration is the Pathways Community HUB (HUB). The HUB acts as a registry for at-risk individuals, pairs them with a coordinator with access to services, enables and monitors service delivery, and ties payments to milestones and improved outcomes.

The Idaho Health Data Exchange (IHDE) is another example, implementing a search and referral platform for its provider users to better understand SDoH and enhance the exchange of data. It also partnered with a social care network to help providers connect their patients with relevant social services and community resources.

To take the first step, HCOs can work to identify specific social needs by collecting non-medical, demographic data. This is key to identifying the most actionable areas for improvement and the population subsets at the highest risk for negative outcomes due to specific SDoH factors. To conduct this analysis, HCOs can distribute social needs assessment surveys, collaborate with community-based organizations, and draw on publicly available datasets, such as the Area Deprivation Index, to evaluate where barriers and disparities exist.

Driving new and existing initiatives with SDoH data and using this data to design interventions will help close equity gaps. HCOs can also go a step further by implementing technology solutions to coordinate care, predict risk tied to SDoH on an individual level, and surface the specific SDoH factors that contribute most significantly to increased risk.

To learn more about SDoH and developing intervention programs, please read our post: 3 Ways Healthcare Organizations Can Advance Health Equity by Addressing Social Determinants of Health.

COVID-19 simultaneously exacerbated existing health disparities and introduced entirely new ones. The pandemic disproportionately impacted people of color. Due to a combination of persistent health disparities and SDoH factors, they are at higher risk for infection, severe illness, and death.

The Kaiser Family Foundation found that “Black people accounted for more cases and deaths relative to their share of the population in 30 of 49 states reporting cases and 34 of 44 states reporting deaths.” Moreover, they found “the COVID-19 related death rate among Black people was over twice as high as the rate for White people.”

The pandemic is also imposing economic pressure on low-income communities. These populations are more likely to hold jobs that can’t be performed remotely, potentially leading to either unemployment due to pandemic-related shutdowns or increased risk of contracting COVID-19 due to greater exposure.

Survey data reports that approximately 60% of Hispanic households and roughly 50% of Black households lost a job due to the pandemic, compared to just 40% of White households.

The risks and severity associated with COVID-19 are further compounded by structural barriers to care and health resources, existing insurance coverage disparities, and an inability to take time off work without severe financial repercussions and potential loss of employment if infected.

Combating disparities caused or exacerbated by COVID-19 is paramount, but HCOs will only be able to maximize their efforts by collecting data that reflects race, ethnicity, and SDoH. Tackling COVID-19 health disparities begins with identifying them, developing a strategy that directly supports people at the highest risk for negative outcomes, and partnering with community organizations to close the gaps.

We dive deeper into the ramifications of COVID-19 on health equity and how HCOs are using data to address inequities in our blog post, How COVID-19 Exacerbated Health Disparities.

Achieving health equity necessitates establishing it as an organizational priority and developing a robust strategy. Critically, efforts will be more likely to succeed if they extend beyond a clinical setting and address SDoH in your local communities. That said, cultivating a culture around health equity and executing on an equity-centered initiative is challenging. Fortunately, The Institute for Healthcare Improvement has outlined five ways to make health equity a core strategy:

Developing a strategy that adheres to these five actions will foster more equitable health outcomes and will help to ensure that the pursuit of equity remains a central focus.

For more information about developing a strategy to prioritize health equity and partnering with local communities, please check out the American Public Health Association’s health equity resource page. They provide an excellent series of fact sheets that detail everything from COVID-19’s impact on housing instability to combating environmental health disparities for children. Our blog post, How You Can Develop a Strategy to Prioritize Health Equity, will also help you develop and carry out your equity strategy.

‍Amidst an ever-growing torrent of available data in healthcare and the shift to value-based care, artificial intelligence (AI) and machine learning (ML) increasingly play a central role. Instead of treating patients as adverse events and complications occur, HCOs are beginning to anticipate the future and working proactively to prevent these events and improve outcomes. AI/ML is essential to preemptively surface high-risk patients, predict exactly what they’re at risk of and why (e.g., an unplanned admission due to COPD exacerbation), and help care teams target interventions efficiently.

AI/ML helps HCOs identify and combat disparities, advancing health equity with more efficient resource allocation, reduced spending, and improved outcomes. However, if algorithms are biased, the AI solutions designed to improve care and equity can end up making things worse.

Unfortunately, algorithmic bias in healthcare is already pervasive. The Chicago Booth Center for Applied Artificial Intelligence states, “Algorithmic bias is everywhere…Biased algorithms are deployed throughout the healthcare system, influencing clinical care, operational workflows, and policy.”

High-profile cases of algorithmic bias have directly propagated racial health disparities. Optum’s algorithm managed 70 million lives and did not account for race—ultimately disadvantaging and producing worse outcomes for Black members while prioritizing White members for care and special programs, despite being less sick than Black members on average.

Ensuring Unbiased, Fair AI

Fortunately, algorithmic bias in healthcare is by no means inevitable, and organizations are taking major steps to ensure AI is unbiased, fair, and explainable. The University of Chicago Booth School of Business has developed a playbook to guide HCOs and policy leaders on defining, measuring, and mitigating bias. HCOs should strive to:

Please visit our resources page to discover more content that addresses algorithmic bias, and read our post, Four Steps To Measure and Mitigate Algorithmic Bias in Healthcare.‍

The Next Step: Advancing Health Equity With AI

If you’re interested in eliminating health disparities, excelling in CMS’s alternative payment models that prioritize health equity, and tackling SDoH gaps with unbiased AI, we’re here to help.

WellSect provides an end-to-end, healthcare-specific machine learning platform and pre-built content library that enables HCOs to quickly produce predictive models customized to their specific populations and available data sources. By accurately predicting which people are at highest risk of potentially preventable negative outcomes, WellSect can help you target care management or other human-led interventions more efficiently and effectively, potentially reducing disparities and leading to better, more equitable outcomes at lower costs.

To learn more about how data, alternative payment models, and predictive analytics are reshaping healthcare, please download our white paper: Precision Health Intelligence – The Key to Population Health Success, or request a demo.

The post Why Health Equity Matters in 2022 appeared first on WellSect.

According to a survey by the Society of Actuaries, 93% of healthcare organizations (HCOs) believe leveraging predictive analytics is key to the future of their business. This perspective is undoubtedly warranted. As an industry, healthcare is evolving, and to succeed in a system centered on value-based care, HCOs must be able to harness the ever-increasing stream of data to produce better outcomes with greater precision. Instead of treating patients as adverse events and complications occur, HCOs will need to anticipate the future and work proactively. To this end, AI is essential to preemptively surface high-risk patients, predict exactly what they’re at risk of and why (e.g., an unplanned admission due to COPD exacerbation), and help care teams to determine effective interventions with a long enough runway.

The benefits of AI extend far beyond singular use cases and permeate every aspect of healthcare. Leveraged effectively, AI has the potential to impact HCO business operations, clinical programs, marketing efforts, patient engagement, sales, contracting, rate setting, digital health implementation, and more. There is not a single facet of business or care delivery that doesn’t stand to benefit from AI-based insights, and the HCOs scaling it the fastest are beginning to solidify competitive advantages.

And yet, healthcare is lagging behind other industries when it comes to realizing the value of AI. A staggering 87% of all data science projects never make it into production, and HCOs implementing AI must do so while also accounting for healthcare-specific challenges. They must have the tools to handle everything from defining episodes of care with dynamic healthcare terminologies to addressing algorithmic biases which may exacerbate health disparities – all while ensuring adoption and explainability. Even if they’re among the 13% that manage to successfully deploy machine learning (ML) models, they still require comprehensive capabilities to maintain them in production. Put simply, successfully operationalizing ML models for healthcare is difficult.

The need for AI-driven insights is clear, but for many HCOs, the steps they need to take to avoid having their AI initiatives end up as failed science fair projects are unclear. How can they ensure their investments result in improved decision-making and better outcomes? Answering this question necessitates a holistic evaluation of data science processes, key considerations for healthcare-specific use cases, and an understanding of the required ML capabilities. Ultimately, HCOs must ask themselves, “What can we specifically focus on to operationalize models, streamline adoption across the organization, and drive impact at scale?”

Start Small and Get SMART

HCOs that try to boil the ocean from the outset are bound to fail. Unrealistic expectations regarding the initial scope of AI projects set the initiative up for failure and leave stakeholders and data scientists disappointed. In turn, this failure can leave organizations hesitant to attempt subsequent AI projects. To demonstrate success and begin realizing improvement, it’s imperative that data scientists work closely with both clinical and business stakeholders to determine specific use cases and ensure models are developed with that context in mind. This means establishing a complete understanding from all parties of how models will be used in production, how results will impact outcomes, and how improvement will be measured. Moreover, it means establishing consensus about the specifics from the outset. Data scientists and stakeholders need to thoroughly answer the question: “What problem are we trying to solve and how?”

SMART (Specific, Measurable, Achievable, Realistic, and Timely) is an acronym used in the goal-setting process that does an excellent job of framing real-world problems in a machine learning context. This framework enables stakeholders and data scientists to consider the problem they’re attempting to address practically and set clear goals and expectations rather than attempting to solve broad issues with vague criteria for success. For example, consider the following two statements that both attempt to frame the same AI initiative:

“We need to reduce our inpatient admissions for asthma.” VS “Compared to the national average of 5.5 per 10,000, our inpatient admission rate associated with poorly managed asthma is 8.4. By implementing the asthma management program recently published in Annals of Allergy, Asthma & Immunology we expect a 15% reduction in asthma-related admissions within six months of the initial diagnosis of asthma.”

The first statement is vague and would need to be substantially fleshed out to begin addressing the problem at hand. Conversely, the SMART statement provides a much clearer foundation, with specific reference points, measurable goals, a set time-period, and expectations for effectiveness. It also drives constructive discussion and pinpoints specifics to provide essential guidance for designing and training models. As a result, data scientists can purposefully conceptualize and build models to target specific pain points, help facilitate intervention programs, and accurately reflect the needs of care teams.

AI projects will never reach deployment without clear communication, specificity, and defined goals that are realistically achievable. Rather than tackling a broad, widely interpretable issue, such as inpatient admissions, organizations need to actively engage in identifying specific use cases and collaboratively making them actionable.

Ensure Explainability

Explainability is crucial to promote adoption and enable the shift to value-based care. Clinical stakeholders simply won’t use model results if they aren’t able to verify the integrity of predictions and intimately understand why they were made. To this end, “black box” models are insufficient, delivering no more than minute improvements when integrated in clinical workflows.

To trust models, clinical stakeholders must be able to unpack exactly which variables were used and which were the most impactful in making predictions. This level of specificity is necessary for validation at the population health level, (e.g., understanding how frailty contributes to risk of fall-related injuries) but stakeholders need even greater explainability at the individual level. They need the ability to identify which patients are at high risk for a given outcome, the modifiable risk factors that had the greatest impact on this prediction, and which of these patients are most likely to benefit from intervention efforts.

Diagnosis-code level insight enables clinical stakeholders to deliver personalized interventions that maximize value. For example, a given patient may be flagged as high risk for heart failure in the coming months, but it may be difficult to understand what distinguishes them from the larger population without granular explainability. However, if AI-based models are able to surface the raw data from this patient’s health history, such as a decline in their left ventricle ejection fraction, care teams will better understand impactability. Ultimately, this specific evidence is essential to personalize interventions and improve health and financial outcomes.

Managing, Monitoring, Auditing

Deployment is a new beginning, not an ending. Once models are in production, data science teams have a tendency to adjust them less regularly than they did in development and monitor performance infrequently. This is a critical mistake; as time progresses, accuracy degradation, feature drift, and unforeseen bias will inevitably begin to jeopardize model performance. Moreover, data scientists must also keep deployed models constantly updated and running reliably in production amidst frequent changes across the healthcare industry, such as shifting code systems, program eligibility requirements, and chronic condition prevalence. Case in point, a model to predict respiratory illness that was trained on 2019 data and never retrained would be wildly inaccurate due to the COVID-19 Pandemic.

To evolve along with healthcare demands that HCOs have the capability to seamlessly manage deployments, monitor performance, and audit models in production. This requires ML Ops. ML Ops is a set of practices that enhances the DevOps procedures used to deploy traditional software with the unique requirements of ML systems. It provides a suite of capabilities to not only manage deployments, but automate model performance monitoring and effortlessly retrain models, with support for audit and governance. This enables HCOs to better manage the explosion of healthcare data, anticipate and address algorithmic bias, maintain model accuracy, and explore new strategic opportunities.

The Future of Healthcare

The ability to operationalize ML models and improve outcomes from their predictions is quickly becoming a given. HCOs must adapt to value-based care, the inundation of available data, and the rise of alternate payment models. Today’s healthcare leaders will simply be left behind if they are too slow to invest in the required technologies and adopt end-to-end model development processes. Going forward, success will be predicated on the ability to contextualize models for specific healthcare use cases, establish trust in predictions, and rapidly adapt to emerging challenges.

For a more thorough analysis of healthcare-specific data science challenges, please download our white paper: Six Mistakes You Can Avoid In Healthcare Data Science.

The post Avoid Failed Science Fair Projects – Operationalizing ML Models for Healthcare appeared first on WellSect.

I’d like to talk about a crucial component of any healthcare analytics system — terminologies. These are the underlying codes for diagnoses, drugs, procedures, devices, and other healthcare services that are used ubiquitously in healthcare data. At WellSect, we maintain and utilize many terminologies within our data science platform. We’ve found Fast Healthcare Interoperability Resources (FHIR) — a standard describing data formats and elements and an application programming interface for exchanging health data — to be both incredibly helpful and incredibly frustrating for this task. FHIR provides a consistent structure for these terminologies, which were often previously managed as file and database tables with different formats and conventions. FHIR is great at managing data, but is not designed for analytics. In this post, I’ll talk about some of the issues we faced when trying to build a purely FHIR-based terminology service, and how we eventually migrated to an approach that uses FHIR to maintain and manage the source data and then built our own terminology graph on top of that to power analytics queries.‍

What is FHIR?

FHIR has revolutionized healthcare data interoperability. Applications that exist solely to facilitate the exchange of healthcare data need a common language — and FHIR is that common language. It has enabled the success of applications like Apple HealthKit, CMS BlueButton, and many more. An entire subsection of the FHIR specification deals with terminology services and describes how the myriad diagnosis, procedure, and other codes used throughout healthcare records are used within FHIR. You might think that for those doing analytics on healthcare data, FHIR terminologies would be incredibly useful. Having all of the codes standardized and accessible from a single FHIR server could make building and maintaining analytical queries much easier. Unfortunately, FHIR was built for interoperability and has two serious shortcomings that make it very difficult to use for analytics.

Figure 1: FHIR Terminology Resources overview. Reproduced from http://hl7.org/fhir/2021Mar/terminology-module.html‍

To understand these shortcomings, it’s useful to have a bit of background on how terminologies work within FHIR. As shown in Figure 1, FHIR defines 3 key resources for managing terminologies: CodeSystems, ValueSets and ConceptMaps. A CodeSystem, unsurprisingly, is a set of unique codes that are part of some controlled terminology. ICD-10 diagnosis codes are a CodeSystem, as is SNOMED, and even the official list of FHIR languages. CodeSystems can be hierarchical, so that individual codes within the code system can have children underneath them. The ICD-10 code J45 for “asthma” has a child of J45.2 for “mild intermittent asthma”. The CodeSystem itself has an identifier so if you see a code in FHIR you know exactly where it came from.‍

Along with CodeSystems, FHIR defines 2 other resources, ValueSets and ConceptMaps. ValueSets define groups of codes that can be used in a particular place. For example, you might have a particular field that can be either a CPT code or an ICD-10 procedure code. The ValueSet for that field would be the combination of those two codes. ConceptMaps define relationships between codes. For example, a ConceptMap could relate an RxNorm concept code to the list of NDCs associated with that concept.

Challenges in Using FHIR for Analytics

The first major issue with using FHIR for analytics is that CodeSystems, ValueSets, and ConceptMaps each define relationships between codes, but do it in a different way with different rules and different query functions. This makes it very difficult to perform even basic analytics queries that involve aggregating codes from multiple CodeSystems into higher level groupings.

In a simple example, assume you want to find medical claims with diagnoses for a certain set of infectious diseases. You have a high level list of codes for these diagnoses. One of those is CCS category 1.1.1 for “Tuberculosis”. This CCS grouping includes ICD-10 code A15. A15 has several subcodes for different kinds of tuberculosis, A15.0, A15.4, etc. You’d like to be able to take your initial list of codes and determine all the ICD-10 codes that match your query. That is, you’d like to get all the child codes of CCS 1.1.1 and this needs to include ICD-10 A15.0 and ICD-10 A15.4. To capture this in FHIR, you’d have a ValueSet that defines your initial list of codes, which would include CCS 1.1.1. The mapping of the CCS to diagnosis codes would be done through a ConceptMap, and the relationship between A15 and its child codes would be stored in the hierarchy of the ICD-10 code system.

‍The limitations of FHIR become clear when you try to resolve your initial list to a set of diagnosis codes; you’d need to run 3 separate FHIR queries. You’d first have to expand the ValueSet, and then traverse the ConceptMap, and finally get all the child codes for the ICD-10 code. What’s worse is that this combination of FHIR queries is specific to this particular structure of the data. If the structure of the ConceptMap or CodeSystem hierarchy were different, you might have to perform a different set of queries.

If you found the previous example confusing, don’t worry. The point of that example was to show how difficult it is to try to use FHIR queries for analytics. For analytics, a much more natural view of terminologies is as a directed graph, where each code is a node and the relationships between them are edges. Any sort of aggregation is then a straightforward graph traversal.

The second major issue is that the FHIR approach to versioning makes it difficult to work with historical data that may span multiple versions of prior terminologies. In analytics, we are generally looking at a data set that contains historical data, sometimes data that goes back years. A particular code may be outdated today, but it was valid when it was used several years ago and you don’t want to throw all those codes out when doing analytics. For analytics, you want a historical view of terminologies that includes both current and expired codes and works with them seamlessly. In FHIR, you either need to specify a specific version, or just choose the latest. You can’t combine a search across versions.‍

Terminology Graphs to the Rescue

To address these two issues we have developed a hybrid approach to terminologies, the WellSect Terminology Graph. The WellSect Terminology Graph is a high performance terminology graph that can quickly resolve code relationships. It sits in front of our FHIR server and provides a graph-based query API that is much faster and simpler to use for analytics than standard FHIR endpoints. The performance is achieved by storing the graph structure in memory, using a highly-compressed format that just stores the node and edge structure, and the underlying information on each code is retrieved from the FHIR server as needed. This graph has the ability to merge multiple versions of FHIR resources so that it can contain a complete historical code set. By default the graph returns the most recent version of any code, but in cases where a code has been deleted, it will retrieve that code from the most recent version in which it was active.

‍The WellSect Terminology Graph gives us the best of both worlds. FHIR is an excellent way to store and maintain CodeSystems, ConceptMaps, and ValueSets, and we use it for that purpose. On top of FHIR we provide a graph-based view of terminologies that can span multiple historical versions to enable fast and convenient queries for analytics. While it was a challenge to build our own graph and find a way to keep it synchronized with our FHIR server, the benefits in the end outweighed the cost. It allows us to remain standards compliant, but still provide an excellent analytics experience. FHIR may at some point incorporate more analytics workflows and address these issues. Until it does we have a solution.

Interested in learning more about data interoperability or the WellSect Platform? Read Data at The Point of Care – CMS Opens up New Opportunities for Data Sharing.

The post Why FHIR Terminologies Don’t Work for Analytics (And what to do about it) appeared first on WellSect.

Remote patient monitoring (RPM) is quickly establishing itself as one of the most effective tools for chronic disease management. RPM is the collection and transmission of patient health data to providers via connected devices outside of a conventional care setting. These monitoring devices record vital information, such as blood pressure, heart rate, or oxygen levels, and are especially helpful for easily tracking patient health without frequent visits and examinations. But before the pandemic, widespread RPM adoption was hindered by the relatively few reimbursement incentives available and saw limited use despite its immense potential. That all changed in the blink of an eye.

COVID-19 necessitated a complete paradigm shift in the delivery of care and dramatically accelerated the implementation of remote technology solutions. Throughout the pandemic, healthcare resources were under far greater strain than ever before, and non-essential appointments were deferred or conducted via telehealth to reduce the burden on providers, lower the risk for more vulnerable patients, and slow the spread. As a result, more healthcare organizations (HCOs) adopted RPM to enable chronic disease management in the absence of in-person encounters. CMS responded accordingly and announced sweeping regulatory changes, which included expanded reimbursement for telehealth and RPM. While these interim reimbursements have since been revised, far greater support for RPM will remain post-pandemic and continue to grow.

RPM Will Become the Norm

The potential for RPM to improve chronic care is clear and its adoption will undoubtedly continue to increase. Providers expect RPM use to match or surpass in-patient monitoring within the next five years, and stakeholders expect the market to double over the same period. Moreover, telehealth continues to grow, and the data RPM provides on vital signs and other clinically relevant information is a critical component of virtual appointments. While the viability of remote care was somewhat dubious before the pandemic, it’s now a proven commodity and well on its way to becoming ubiquitous.

Forward-thinking patients now expect to have access to convenient virtual care options and HCOs that haven’t yet adapted long-term to this shift in preferences are likely to see this change reflected in their bottom line. This will be especially true for patients with chronic conditions that require frequent checkups. While these patients might have previously been content with the ability to simply connect with any physician or discipline-specific specialist, the proliferation of virtual care programs and RPM is raising the bar. As patients become increasingly comfortable with virtual care, they expect greater convenience and flexibility, increased control over virtual encounters, and the ability to selectively connect with their primary care provider and specific specialists of their choice. Rather than scheduling routine office visits, patients with chronic conditions will increasingly endeavor to connect more immediately through virtual care and seek experts that they feel provide the best disease management options digitally.

Simply implementing these virtual care programs no longer provides a competitive advantage. Now, they must also effectively leverage digital technologies to streamline engagement, provide personalized care, and ensure virtual care meets the same expectations patients have for in-person interactions. Patients are likely to explore other options if HCOs can’t meet these requirements, and doing so requires RPM. Virtual appointments don’t allow for many of the examinations that are essential for accurate diagnoses and treatments. Instead, the practitioners rely on remote monitoring to collect critical patient health data.

The need for RPM to facilitate effective virtual care programs is evident, but how HCOs should incorporate the myriad of novel data sources RPM includes to best improve care is much less clear. Answering this question is critical for chronic condition management, especially because RPM has the potential to significantly augment existing disease management practices and help address gaps in care. Ideally, RPM data can enable HCOs to proactively identify key patients for engagement, and the increase in available data can help to anticipate changes in behaviors and conditions. To this end, HCOs will need to ask, “How can RPM be harnessed to predict individual health outcomes?”

Data Collection Is Just the First Step

RPM is so much more than an essential complement to effective telehealth visits. It represents a tremendous, ongoing stream of novel data that has the potential to dramatically increase insight into patient health. In tandem with the right technology stack, it presents an opportunity for HCOs to revolutionize their approach to patient engagement, anticipating declines in health without face-to-face examinations and providing personalized care remotely.

Collecting the data is merely the first step, not the end goal. If RPM is to deliver value for HCOs and improve health outcomes, it must be paired with analytics capabilities that can integrate these novel data streams into clinical workflows, generate accurate predictions, and identify individual patients for outreach. HCOs will need to quickly ingest data from connected devices and leverage it to abstract high-level information about individual patients. To achieve this efficiently, they will need to exploit machine learning and artificial intelligence to channel a deluge of raw data into personalized predictions.

RPM and AI in Practice

The potential for RPM and AI extends far beyond just surfacing raw patient health data into physician-facing reports and interfaces; it enables proactive care and the ability to anticipate adverse events without consistent visits. For example, an older patient with COPD may have had their regular, in-person appointments disrupted by the pandemic and be reluctant to frequently connect with their primary care physician virtually. During this period, their self-management practices may worsen. Without the combination of RPM and AI, it would be exceedingly challenging for a care team to identify this patient’s increase in risk and accurately anticipate potential complications.

However, the synthesis of RPM and AI enables proactive identification and intervention. If the patient’s remote monitoring device detects a decline in blood oxygen saturation and this data source is integrated with predictive analytics, the care team can initiate preventive measures. AI can analyze this decline in combination with existing clinical data to predict that the patient is highly likely to experience an ER visit within the next three months due to exacerbation. This insight and the specific evidence that supports it are then surfaced to help prioritize this patient for engagement. As a result, the adverse outcome is prevented, significantly reducing costs and improving the patient’s long-term health.

The combination of remote monitoring and AI can also promote healthy behaviors without clinical intervention. Digital health solutions that ingest remote monitoring data can leverage AI to establish intelligent feedback loops and provide precise recommendations. These solutions can help individuals with chronic conditions gain greater insight into their health and improve their own self-management practices with personalized insights. For example, an AI-driven digital health app may identify the need for more regular exercise from an increase in blood pressure, but personalize its recommendation to improve engagement. It may encourage long walks with a dog rather than jogging if it has learned that the individual is more receptive to such a suggestion. Similarly, HCOs can synthesize RPM and AI to improve long-term disease management programs and determine which interventions and methods of engagement most effectively foster healthy behaviors.‍

Engaged and Personalized Care

The pandemic catalyzed greater implementation of RPM and definitively proved the viability of virtual care programs. Moving forward, HCOs will be able to amplify the success of these care management programs through proactive engagement. With RPM and AI-based models, HCOs can predict high-risk patients to promote earlier diagnosis of chronic conditions, mitigate disease progression, and anticipate adverse events. In conjunction with AI, RPM enables care teams to surface specific, modifiable risk factors from raw data and measure engagement effectiveness to support the best possible health and financial outcomes.

Effective RPM implementation will not only help to better allocate limited healthcare resources to the patients that will benefit most, it will also continue to support fundamental changes to care delivery. At its core, it presents an efficient approach for patients and their care teams to continuously share health data remotely and enable engagement outside of in-person interactions. At the other extreme, RPM offers the opportunity to rethink care delivery entirely and to replace existing paradigms with a better vision of patient engagement and the pursuit of health. Ultimately, they can enjoy an improved quality of life, knowing that their health data is being seamlessly transmitted and assessed to support proactive, personalized care.

Interested in learning more about remote patient monitoring and other rapidly growing healthcare technologies? Check out these related blog posts and other content:

• Use Case Library – Remote Patient Monitoring
• Data at The Point of Care – CMS Opens up New Opportunities for Data Sharing
• Precision Health Intelligence — The Key to Population Health Success

The post Remote Patient Monitoring: Coming of Age in an AI-driven World appeared first on WellSect.

March 23, 2020, marked the 10th anniversary of the Affordable Care Act (ACA), whose passage ignited the most comprehensive set of changes in US healthcare.  Ever.  The industry had been changing for decades, but the ACA accelerated both the speed and the magnitude of change consuming the industry.  ‍

When measured in ‘legislative years,’ the ACA is still young.  Even so, it has already left its mark.  It created game-changing progress on coverage, rewrote the rules of the insurance market, showed evidence of real impacts on health, and transformed the policy and legal landscape.  Despite being sometimes hailed as a ‘big deal’ while at other times cursed that it simply didn’t work, no recent legislation has provoked more controversy or been as resilient. The ACA has survived more than 70 ‘repeal and replace’ efforts and has been to the Supreme Court five times, including watershed disputes involving Congress’s power to regulate economic activity, Congress’s obligations when using states and private implementers, the appropriate way to interpret Congress’s wordings, and the extent to which ACA could press against religious freedoms.¹

The ACA has also shifted the baseline of public attitudes and with it, future policy making.  One such attitude is that Americans should not lose health coverage because of pre-existing health conditions.  The litmus test of policy alternatives is now a comparison of coverage numbers, with proposals judged on whether they cover roughly the same population. Given how recently ACA was passed, the significance of this cannot be overstated.  Protections for pre-existing conditions are now table stakes. The debate has moved on.

Alternate Payment Models Change Everything

The more visible effects can be seen in the delivery system where ACA has accelerated the arrival of value-based healthcare that is aligned with – and paid based on – health outcomes.  The ACA contained several such initiatives, including the Hospital Readmissions Reduction Program, Bundled Payments for Care Improvement, Medicare Shared Savings Program (MSSP), and Accountable Care Organizations (ACOs), which have seen remarkable growth.  It has grown tremendously.  As of June 2019²:

• Nearly 1,000 ACOs now cover almost 44 million lives
• ACOs are now physician-led (43%) more than hospital- or jointly- led, a reversal from early on
• More ACOs are starting to take downside risk, with physician-led ACOs taking downside risk more often than hospital-led ACOs

This growth is not a surprise, since part of reform included the funding of the Center for Medicare & Medicaid Services Innovation Center (CMMI), a $10 billion investment in developing and testing multiple payment and delivery model innovations and alternate payment models (APMs).  Since then, CMMI has devised 90 alternate models, ranging from P4P, bundles, and global payments.³ ‍

To help drive APM adoption, CMS established the Health Care Payment Learning & Action Network (LAN) in 2015.  The LAN’s mission has been to accelerate the transition to APMs by combining the innovation, power, and reach of the public and private sectors.  The LAN’s landmark achievement has been the APM Framework (see graphic, below). The APM Framework establishes core APM design principles, classifies APMs into distinct categories, and establishes a common vocabulary and pathway for measuring successful models.⁴

The framework also forms the basis of the annual APM Measurement Effort which measures progress in the adoption of APMs. In 2018, APM Measurement reported 36% of healthcare payments flowing through Categories 3 & 4.⁵  Despite significant uptake, nearly two-thirds of payments remain rooted in fee-for-service (FFS).  Experts agree this will fail to drive fundamental change and that the next stage must focus on APMs that support population health management, community engagement, and other value-adding activities.

CMS intends to accelerate the percentage of payments made under two-sided APMs even further, with targets of 100% of Medicare and 50% of Commercial and Medicaid by 2025.⁶  To drive adoption, CMS made one of the largest changes to the Medicare ACO program since its inception.  Dubbed Pathways to Success, it includes a major overhaul of MSSP.

They also launched several new models. One is the Primary Cares Initiative, a set of APMs that create options for primary care physicians to be paid for keeping patients healthy and out of the hospital. The agency considers these to be potential game-changers and hopes they will transform primary care.⁷ In fact, when introducing them, CMS said the goal was to dismantle fee-for-service payments. They have aggressive goals for this program and are targeting to cover 11 million Medicare beneficiaries.⁷ ⁸

Another new model is the Community Health Access and Rural Transformation (CHART) Model with a focus to accelerate opportunities in rural communities.⁹  This effort makes investments that enable rural communities to pursue innovative financial arrangements and gain the operational and regulatory flexibility they need to pursue value-based arrangements.

Irrespective of details, the transformation to value-based care is disrupting nearly every aspect of how care is organized, delivered, measured, and reimbursed.  It is reshaping the industry and its key segments, including where profit pools lie and who gets them. Today’s healthcare organizations, including its leaders, need to aggressively adapt or risk long-term viability.¹⁰

End Notes

1. Emanuel E, Abbe G. The ACA at 10: Health Care Revolution | Health Affairs. Healthaffairs.org. February 2020.
2. Muhlestein D, Bleser W, Saunders R, Richards R, Singletary E, McClellan M. Spread of ACOs and Value-Based Payment Models in 2019: Gauging the Impact of Pathways to Success | Health Affairs. Healthaffairs.org. October 2019.
3. EHRIntelligence. Decade-Defining Moments in Healthcare Innovation, Reform. EHRIntelligence. https://ehrintelligence.com/news/decade-defining-moments-in-healthcare-innovation-reform. Published December 20, 2019
4. HCP-LAN. Alternate Payment Model – APM Framework. The MITRE Corporation; 2017:1-45. https://hcp-lan.org/workproducts/apm-refresh-whitepaper-final.pdf.
5. HCP-LAN. Roadmap for Driving High Performance in Alternative Payment Models – Health Care Payment Learning & Action Network. https://hcp-lan.org/apm-roadmap/. Published December 18, 2019.‍
6. Muhlestein D, Bleser W, Saunders R, Richards R, Singletary E, McClellan M. Spread of ACOs and Value-Based Payment Models in 2019: Gauging the Impact of Pathways to Success | Health Affairs. Healthaffairs.org. October 2019.
7. Centers for Medicare and Medicaid Services (CMS. HHS To Deliver Value-Based Transformation in Primary Care. HHS.gov. https://www.hhs.gov/about/news/2019/04/22/hhs-deliver-value-based-transformation-primary-care.html. Published April 22, 2019.
8. M. Brady, “Americans ‘fed up’ with high healthcare costs, surprise billing, Verma says,” Modern Healthcare, September 10, 2019
9. CMS Innovation Center. CHART Model | CMS Innovation Center. Cms.gov. https://innovation.cms.gov/innovation-models/chart-model. Published 2020.
10. PwC Health Research Institute. Top Health Industry Issues of 2020: Will Digital Start to Show an ROI? PwC Health Research Institute; 2019:1-53. https://www.pwc.com/us/en/industries/health-industries/assets/pwc-us-health-top-health-issues.pdf.

Interested in learning more about upcoming payment models and the importance of AI to succeed under value-based care? Check out these resources:

• Upcoming CMS Payment Models And the Importance of AI
• Avoid Failed Science Fair Projects – Operationalizing ML Models for Healthcare
• Podcast – What Happens if Healthcare Rejects New Tech like Machine Learning

The post Here to Stay: Alternate Payment Models appeared first on WellSect.