As part of a focused effort to assess current cardiovascular treatment algorithms for racial bias, the American Heart Association, the single largest non-government supporter of heart and brain health research in the U.S., is funding three new scientific research projects at $50,000 each.
Clinical algorithms are formulas, flow charts and computerized "calculators" that work behind the scenes to analyze health data and help determine a person's risk for heart disease or guide their treatment decisions. Age, weight, information from blood tests, personal health history, health habits — like physical activity and smoking — are among the types of data used by clinical algorithms. Some algorithms include race or ethnicity in their analysis, but recent evidence has suggested that race is often an inadequate substitute for genetics.
"By inappropriately including race as a proxy for biological characteristics, algorithms may unintentionally perpetuate disparities in care," said Nav Persaud, M.D., M.Sc., co-chair of the American Heart Association's De-biasing Clinical Care Algorithms (DECCA) Expert Advisory Panel and a physician in the Department of Family and Community Medicine at St. Michael's Hospital in Toronto, Ontario, and an assistant professor of medicine at the University of Toronto.
"Studying how race is incorporated into algorithms is an important endeavor in health equity by separating race from the social determinants of health that drive the relationship between race and disease," adds Judy Wawira Gichoya, M.D., M.S., an American Heart Association volunteer and assistant professor in the Department of Radiology and Imaging Sciences at Emory University School of Medicine and an interdisciplinary researcher studying informatics in medicine. Gichoya co-chairs the DECCA Expert Advisory Panel with Persaud.
The teams of scientists who received funding for "Assessing Race in Clinical Research Models" are from Stanford University School of Medicine in Stanford, California, Mayo Clinic in Jacksonville, Florida and The University of Texas Southwestern Medical Center in Dallas. Support for these studies is part of a two-year scientific research strategy funded in part by a grant from the Doris Duke Foundation to study the complex issue of how race and ethnicity factor into clinical care algorithms and risk prediction tools.
Specifically, the researchers are charged with (1) assessing potential bias in risk models and identifying factors for the bias (e.g., sampling bias, selection bias, missing data values and potential risk factors); (2) developing statistical methods and advanced models that correct or mitigate against algorithm bias to support equitable care and treatment.
The three research projects, which began July 1, 2023, and are funded for up to two years each, include:
- Evaluating Cardiovascular Risk Algorithms Among Single and Multiracial/Multiethnic Asian People – led by Adrian Bacong, Ph.D., M.P.H., a postdoctoral research scholar with the Stanford University School of Medicine and associate program director for the Stanford Center for Asian Health Research and Education Team Science Fellowship in Stanford, California. In a three-part project, this team will study the use of heart disease calculators for Asian-Americans who identify as "Asian only" and those who identify as multi-racial. They will first determine the prediction accuracy of Asian-specific heart disease calculators for people who identify as a single Asian race. Second, they will test which race adjustment (Asian or White) better predicts heart disease risk among Asian-Americans who identify as more than one race. Third, they will determine which clinical or lifestyle factors best predict heart disease among Asian-Americans. The team anticipates that their findings could be used to create a more accurate heart disease risk calculator for Asian-Americans.
- Evaluating the Impact of Race-Specific Pooled Risk Equations for Cardiovascular Risk Prediction on Clinical Outcomes – led by Ramla Kasozi, M.B.Ch.B., M.P.H., a family medicine physician and senior associate consultant in the Department of Family Medicine and assistant professor of family medicine at Mayo Clinic College of Medicine and Sciences in Jacksonville, Florida. This study seeks to assess the potential influence of race on the performance of the atherosclerotic cardiovascular disease (ASCVD) pooled cohort equation (PCE) in a study population that is different from the original population that informed the creation of the PCE. They will assess in the cardiovascular disease risk results in the study population by sex and other races when using the race-specific PCE. They will also determine if use of the race-specific PCE in other races impacted clinical care. They expect to provide answers on the usefulness of including race in the PCE.
- Machine Learning-based Models with Social Determinants of Health to Improve Incident Atrial Fibrillation Prediction – co-led by Ambarish Pandey, M.D., M.S., assistant professor in the Department of Internal Medicine at UT Southwestern Medical Center in Dallas and Matthew Segar, M.D., M.S., a cardiology fellow at the Texas Heart Institute in Dallas. The researchers are investigating new ways to predict the risk of cardiovascular disease. Using machine learning and data from five large community cohort studies, they will create and test new models that include social determinants of health (SDOH) instead of race to predict new onset atrial fibrillation (AF). They will compare the new model to current risk models to evaluate biases. The investigators hypothesize that including SDOH can improve the model accuracy and they hope to determine which factors contribute most to the development of AF.
The American Heart Association has funded more than $5 billion in cardiovascular, cerebrovascular and brain health research since 1949. New knowledge resulting from this funding benefits millions of lives in every corner of the U.S. and around the world.
