A recent report by the Urban Institute seeks to quantify and explain measures of disparity in healthcare, one of the most important yet challenging prospects for pursuing equitable healthcare.
Racial and ethnic disparities have been a long-standing problem in the U.S. healthcare system, yet efforts to eliminate these inequities have been limited by a lack of research and evidence-based practice. The report’s authors note that a key limitation of many studies on disparities in health care is the lack of a clear definition of the problem and an analytical framework, which can lead to estimates that are ill-defined and easily misinterpreted.
A definition of health disparity provided by the Institute of Medicine (IOM) was used for the analysis, which suggests that racial and ethnic disparities in access to and use of health care for reasons other than clinical need and patient preference are considered disproportionate and thus constitute difference.
“Based on this definition, differences due to age, sex, health status, and patient preference were considered acceptable or justified,” they explain. “While this is not the only valid definition, it provides a clear example of how to clearly define difference and use a conceptual framework to define fair and unjust drivers of difference to guide interpretation.”
The report offers 5 recommendations for generating and interpreting estimates of racial and ethnic disparities in healthcare:
- Include a clear and unambiguous definition of the difference in health care benefits, and corresponding methods for applying that definition to estimate the size and/or drivers of the difference
- Provide a conceptual framework that considers or incorporates the role of systemic racism as an injustice driver of group differences to guide the interpretation of findings
- Discuss data limitations when applying given definitions of difference, including measures of race and ethnicity
- Estimate the model using a comprehensive set of covariates consistent with the conceptual framework and use the appropriate components of the estimated model to calculate the size of the defined difference
- Investigate models that go beyond recording differences to analyze influencing factors
The authors also provide several empirical examples using National Health Interview Survey data to further illustrate the value of these recommendations. The main finding of their empirical analysis is that difference estimates can vary substantially when different definitions of difference are used, suggesting that difference is explicitly defined using a supporting conceptual frameworkârather than implicitly based on the covariates included in the regression model. Define the difference – essential for interpretation.
Furthermore, estimates of individual disparity definitions, such as those from the IOM and their components, vary slightly across different estimation methods. However, they say their example suggests that approaches using more comprehensive models with rich covariates are less likely to underestimate IOM differences.
Finally, if the data indicate differential effects of socioeconomic status or other covariates on racial group outcomes, estimates of the differences and drivers of the differences can be improved by estimating separate models by race, they said.
“Clear, high-quality measurements of racial and ethnic disparities in healthcare use, with clear definitions and explanations, are critical to understanding disparities and exploring their causes,” the authors said.
“Clearly defined and interpretable estimates are necessary to generate more actionable information to address disparities, guide and evaluate interventions to reduce and eliminate disparities in healthcare use, and to shape and prioritize an equitable policy agenda for policymakers and the public of.”
The report was funded by the Robert Wood Johnson Foundation.
refer to
Clemans-Cope L, Garrett B, McMorrow S. How should we measure and explain racial and ethnic disparities in health care? city.org. Published January 11, 2023. https://urbn.is/3wxCKiL
