Rethink Race in Pooled Cohort Equations for ASCVD, REGARDS Data Suggest

Neither removing race from the pooled cohort equations (PCE) nor adding social determinants of health (SDOH), appears to have an impact on predicting a patient’s 10-year risk of atherosclerotic cardiovascular disease (ASCVD), a new analysis of the REGARDS study suggests.

The study authors, led by Arnab K. Ghosh, MD (Weill Cornell Medical College, New York, NY), say their findings support calls for removing race as a covariate from the PCE, replacing it with more nuanced ways of capturing biological and social risk factors.

“We appreciate and we definitely understand that race plays an important role in whether someone develops a heart attack or a stroke, [but] the way that the equations work suggests to us that the impact of race works through other factors in the equation,” Ghosh told TCTMD. Those other factors could include things like diabetes, hypertension, current smoking, or medication use for a comorbid condition.

“I think this just speaks to the fact of the nature of race and the disparities that we see playing a role throughout the life course of an individual and thus leading them to having an increased risk of developing heart attacks and strokes,” he added.

In an accompanying editorial, Sadiya S. Khan, MD, and Clyde W. Yancy, MD (both Northwestern University Feinberg School of Medicine, Chicago, IL), point out that race is an “aggregate proxy” of both social and cultural constructs that impact life and lived experiences. Khan and Yancy say there are growing opportunities to expunge race and embrace SDOH, which they see as being more precise variables in risk prediction, despite the latter not making much of an impact in this study.

“It is critical to highlight that the limited impact of SDOH observed on prognostic performance do not equate to a null effect of SDOH on CVD risk for several reasons,” they write. Not only are SDOH upstream of proximal causal risk factors, they were highly prevalent in the current data set: more than one in two Black adults were on antihypertensives and more than one in six had diabetes.

Another consideration, Khan and Yancy say, is that the available tools for assessing SDOH are “formative, omit any assessment of discrimination, and are awaiting further development.”

Poor Model Performance Seen in Men

To understand the role that race plays in clinical prediction tools, Ghosh and colleagues analyzed data from 11,638 participants (mean age 61.8 years; 58% female) from the Reasons for Geographic and Racial Differences in Stroke (REGARDS) study. The study enrolled community-dwelling Black and white adults over the age of 45 who had no ASCVD but had a LDL-cholesterol level of 70 to 189 mg/dL or non-HDL-cholesterol level of 100 to 219 mg/dL at baseline. All were followed to 10 years for incident ASCVD, including MI, coronary heart disease death, and fatal and nonfatal stroke.

Compared with white participants, Black men and women had low income (US $35 000 or less per year), more chronic diseases, and higher rates of smoking.

Variables included in the model for 10-year risk of ASCVD were low income, zip code in poverty (more than 25% of households below the federal poverty level), low education (less than high school), Area Deprivation Index (ADI), lack of health insurance, and residing in a state with poor public health infrastructure, including a health professional shortage.

Whether you control someone's blood pressure or cholesterol, [you] have to be mindful of the environment that they live in. Arnab K. Ghosh

Across models that included Black men and women and white men and women, C-statistics did not change for the original PCE; best-fit, race-stratified equations including the same variables as the PCE; best-fit equations without race stratification; or best-fit equations without race stratification but including SDOH.

While there were minimal differences between Black and white men in correct risk assessment, there were significant differences between male and female participants, regardless of race, across all the models. In race-stratified equations including the same variables as in the PCE, for example, the C-statistic was the same at 0.68 for Black and white males versus 0.71 for Black females and 0.77 for white females.  

Sensitivity analyses indicated that race was not significantly associated with ASCVD risk, nor did adjusting for race affect the performance of the race- and sex-stratified models.

Race-Free Predictors, With Mindful Assessment

Ghosh and colleagues say these findings “suggest that drawbacks to the continued use of race as a covariate in one of the most widely used risk predictors in the country, including a misconceived conception of race as a biological construct, likely outweigh any incremental benefits in terms of predictive value.”

They argue for novel risk prediction that accounts for the ability of both biological and social factors to uniquely confer added ASCVD risk. “These include modeling ASCVD outcomes using more sophisticated methods such as machine learning and expanding beyond ASCVD risk factors and minimal demographics in the variables incorporated,” they write.

According to Khan and Yancy, there is reason to believe that race-free or race-neutral prediction tools are the future, citing a recent American Heart Association advisory on Cardiovascular-Kidney-Metabolic Health. The advisory advocates a race-free calculator that incorporates SDOH into risk prediction and encourages addressing SDOH as part of the clinical care model.

“This is an active area of inquiry, and we have to be thoughtful about which social determinants of health matter,” Ghosh told TCTMD.

An important consideration, he added, is ensuring that changes to risk prediction models don’t affect the user experience of clinicians who rely on the tools to make decisions for their patients about blood pressure or cholesterol control, but do make them think a bit differently.

“These ideas of thinking about race as a biological construct are important to bring to mind within the field of medicine amongst our trainees, amongst our fellows, and even amongst seasoned professionals,” Ghosh said. “Whether you control someone's blood pressure or cholesterol, [you] have to be mindful of the environment that they live in. This research is one of the ways that we are trying to do that.”