Achieving Equity in Healthcare Modeling

In Blog by BHI

As the pandemic has made painfully clear, healthcare utilization and cost do not always indicate true healthcare needs. People of color are contracting the novel coronavirus and dying at rates that are staggeringly out of line with their share of the U.S. populace.

Recent social injustice events in the U.S. have heightened industrywide discussion of healthcare equity. BHI’s data scientists already have been accounting for disadvantaged persons’ lack of access to care, excessive underlying health conditions, and other socioeconomic factors that are needed to more accurately assess the resources required to provide essential care.

BHI has put a process into action that assures our predictive models are adjusted for equity. Our algorithms account for a range of healthcare barriers, such as lack of transportation and access to primary care. We then use these algorithms to contrast a disadvantaged member, such as an ethnic minority who lives in a predominate low-income area, with a non-disadvantaged member.

In the following example, even though their medical profiles are the same (males of the same age who have coronary disease and are high ED utilizers), claims data revealed lower costs for disadvantaged members because of their barriers to care. To address this situation, BHI applies equity adjusters to our various cost prediction tools to derive a prospective health need score. This score gives a more accurate account of each member’s healthcare needs and the resources required to ensure those needs are met equitably.

To express other societal influences on access to healthcare, BHI’s also uses our proprietary Healthcare Barrier Index (HBI) to rate each member’s incidence of socioeconomic and behavioral factors. The factors used in our HBI are derived from characteristics found in BHI’s immense claims repository (e.g., number of sick members in a family.)

We then add publicly sourced information to impute socioeconomic factors based on the community where a member lives. Together, these factors allow us to produce a health access score that informs more appropriate interventions for members regardless of their healthcare utilization.

Solving the problems of healthcare inequities is a long-term task for the larger health system, policymakers at all levels, and other institutions. Leveraging data is a good place to start the process of fixing one of society’s most glaring examples of discrimination.