Abstract

The study used cross-sectional analysis using nationally representative secondary data to estimate the economic burden of racial and ethnic health inequities in the US. The data included 2016-2019 data from the Medical Expenditure Panel Survey (MEPS), state-level Behavioral Risk Factor Surveillance System (BRFSS), 2016-2018 mortality data from the National Vital Statistics System (NVSS), and 2018 IPUMS American Community Survey (ACS).

87,855 survey respondents to MEPS, 1,792,023 survey respondents to the BRFSS, and 8,416,203 death records from the NVSS were exampled to determine the sum of excess medical care costs, lost labor market productivity, and excess premature death (before age 78) costs by race and ethnicity compared to health equity goals.

Background

Most racial and ethnic health inequities exist because there are differences among these groups in their exposure to economic, social, structural, and environmental risks and their access to healthcare. This study was commissioned by the NIMHD and updates our previous estimates of the US’s economic burden of health inequities. We focused on American Indian or Alaska Native (AIAN), Asian, Black or African American (Black), Hispanic or Latino (Latino), and Native Hawaiin or Other Pacific Islander (NHOPI) because their health is impacted by disparities in the health care delivery system, public health system, employment, education, transportation, public safety, and other systems that are social determinants of health. The study shows that health inequities are also an economic concern for society.

Methods

Detailed Description of the Methodology

To compute excess medical care expenditures and lost labor market productivity, we used an incremental approach based on regression models for medical care costs, hours worked, wages, and sick days. We simulated the outcomes using the prevalence rates from the MEPS or BRFSS data and health equity goals. We computed the costs of health inequities as the difference between the predicted outcomes using the actual health conditions and the healthy equity targets. To estimate the cost of premature death, we computed crude death rates by race and ethnicity for ten age groups: under 1, 1-9, 10-19, 20-29, 30-29, 40-49, 50-59, 60-69, 70-79, and 80 and over. We then estimated the number of deaths that would have occurred for each racial/ethnic group if every group’s death rate were equal to the health equity target death rate within each category. The difference between the actual number of deaths and the estimated deaths represents “excess deaths.” We computed the number of years of life lost in each racial/ethnic group by assuming that all persons would have lived to age 78 had they not died prematurely since that was the average life expectancy in the US in 2018. We then computed the premature death cost by multiplying the years of life lost by $100,000. (View Figure 1)

Figure 1. Health Inequity vs Health Disparity

Definition of Health Equity Perspective

We established a health equity target of reaching the 90th percentile of the predicted prevalence rates for health conditions and the 10th percentile of crude deaths. Using the MEPS adult sample, we estimated prevalence models for each health condition, including fair/poor health, diabetes, joint pain/arthritis, depression, any limitation of activities, hypertension, high cholesterol, stroke, heart attack, coronary heart disease/angina, asthma, emphysema/chronic bronchitis, and cancer. The sample was divided into 14 cohorts based on sex and seven more age groups: 18-24, 25-34, 35-44, 45-54, 55-64, 65-74, and 75 and over. Within each cohort, we identified the risk of each health condition at the 90th percentile. To compute years of life lost due to premature death, we ordered the states from lowest to highest crude death rates for each category, identified the 5th state with the lowest rate, and used the crude death rate for that state as the healthy equity goal for that age category.

Estimating the Direct Cost of Medical Care Due to Excess Morbidity

Modeling Medical Care Expenditures

Using the MEPS data, we built a regression model from the 2016 survey. We used a two-part estimation technique for the medical costs. The first part was a logistic regression model to estimate the impact of health conditions on the probability of having any time of medical care expenditures, and the second part was a generalized linear model to estimate the impact of the conditions on levels of expenditures for individuals with positive expenditures. Our dependent variable was total medical expenditures, and the predictor values were the health conditions.

Predicting Excess Medical Care Expenditures

Using the MEPS and BRFSS data, we simulated medical care costs using the reported health conditions and assigned each group the target prevalence rate for the condition within the age/sex cohort. The respondents in each cohort were randomly assigned the conditions using a uniform distribution, so the prevalence rate for the condition was at the 90th percentile. We randomly chose 1,000 samples to get one predicted probability and one predicted mean for the models. We repeated this exercise 1,000 times to get 1,000 predicted probabilities and 1,000 predicted means by race and ethnicity.

Estimating the Value of Lost Labor Market Productivity

To compute the value of lost productivity, we built three separate two-part labor market models. We estimated the impact of health conditions on three outcomes: disability and illness on sick days, annual hours of work, and wages for working adults ages 25-64.

We predicted disability days off from work, annual hours worked, and hourly wages using the reported health conditions from the MEPS and BRFSS data. We then simulated the labor market outcomes by assigning each group a health condition.

Estimating the Costs of Excess Premature Death

Premature deaths are deaths that may be preventable through lifestyle modifications, better access to health care, and basic resources, such as food and housing. Typical measures of premature death estimate potential life lost when death occurs before one would expect according to population average death rates.

Calculating the Costs of Premature Death based on Willingness to Pay

Willingness to pay is a comprehensive measure of the private valuation individuals place on small reductions in risk of death. Generally, an individual’s willingness to pay has been assessed using surveys. Conceptually, the measure captures everything that would contribute to a person’s well-being, including time preference, non-labor income, the value of leisure, and the value of pain and suffering.

Analytic Approach

The estimate of costs of health inequities has three components: excess medical care expenditures, lost labor market productivity, and premature death costs. To compute excess medical care expenditures and lost labor market productivity, we used an incremental approach based on regression models for medical care costs; hours worked, wages, and sick days simulating costs using the MEPS or BRFSS data. To estimate the cost of premature death, we computed crude death rates by race and ethnicity for ten age groups: under 1, 1-9, 10-19, 20-29, 30-29, 40-49, 50-59, 60-69, 70-79, and 80 and over. We then estimated excess deaths and subsequent years of life lost reality to the health equity goals. We valued a year of life lost at $100,000 to reflect recent estimates, which have valued willingness to pay at $95,000 to up to $264,000.

Results

In 2018, the overall economic burden of failing to achieve the health equity goals was $1.03 trillion. This included $421.1 billion for racial and ethnic minorities and $608.7 billion for the White population. For racial and ethnic minorities, approximately two-thirds of the economic burden was attributable to premature death, while excess medical care costs were 18% and lost labor market productivity were 14%. The study also found that most of the costs were attributable to the Black population; however, costs attributable to Native Hawaiian or Other Pacific Islander and American Indian or Alaska Native populations were disproportionately greater than their share of the population.

View All Tables and Figures

Discussion

The economic costs of racial and ethnic health inequities are unacceptably high. In some states, the economic burden of racial and ethnic health inequity is over twice the growth rate of the nation’s economy. Therefore, federal, state, and local policymakers must invest resources to develop research, policies, and practices to eliminate these inequities in the US. Additionally, state health policymakers and state officers of minority health should use this information to guide their policies and programs to address the inequities in their state. The magnitude of the costs is not the only thing that needs to be considered, but also how the state’s costs compare to the size of their economy.

The COVID-19 pandemic has increased the cost of health inequities, given the pandemic’s disproportionate impact on Black, Latino, and AIAN populations. With the higher excess death rates due to the pandemic, an increased number of hospitalizations, and disruptions to the labor market that employ large numbers of Black and Latino workers, we anticipate the cost of health inequity will have increased in 2020 and 2021 due to its disruption of follow-up visits, screening tests, and even treatments that would have likely resulted in reduced early detection of cancers and costly chronic illnesses. Furthermore, worsening mental health, substance misuse, and violence will exacerbate existing disparities in the area. This disruption in care may have reduced medical care expenditures in the short term, but the delay in treatment may increase care costs in future years.

With our study, there were several limitations. First, we could not always compute estimates for the AIAN and NHOPI populations, even when we pooled the data across three years, because the sample sizes were too small. Second, the MEPS excludes persons residing in institutions and persons in the military, so our medical care and labor market estimates exclude costs associated with these groups. We also excluded children since they face different health inequities than adults and are not typically in the labor force. However, the premature death calculation does include children, persons in institutions, and persons in the military. Third, our labor market outcomes model excluded adults over 64 and under 25. Fourth, our simulation assigned health conditions to individuals as the occurrence of one condition is unrelated to the others, even though it is well-known that some conditions are more likely to co-occur with others, such as diabetes, hypertension, and heart disease. It is unclear how this lack of accounting for the covariance biases our excess medical care costs and lost labor market productivity estimates.

The costs and racial and ethnic health inequities to the US economy are substantial and more than justify the societal investment in developing policies and programs to eliminate health inequities and larger data sets for smaller racial and ethnic subpopulations. Even a modest reduction in health inequalities can save the nation billions of dollars in medical spending and lost labor market productivity annually.