New evidence of bias found in predictive healthcare tools

A study from researchers including Dr. David Stone (Res ’81, ’85), of UVA’s departments of Anesthesiology and Neurology, adds to recent research that has found evidence of bias in important healthcare decision-making tools. In an analysis of three scoring systems used to predict risk of death in hospital intensive care units, the study found that the tools overestimate the risk of death for all patients, but particularly for Black and Hispanic patients. Noting the pressure put on limited ICU resources during the coronavirus pandemic, the authors write that “such inaccurate predictions are concerning, particularly if treatment is withheld or care withdrawn on the basis of a false high predicted mortality.”

In more positive news, a large-scale project analyzing the genetic makeup of a global, diverse body of more than 53,000 individuals is providing a much broader view of genetic variants, including rare and previously unidentified ones. UVA genetics researcher Stephen S. Rich, who helps lead the project, notes that it holds the promise of broadening scientific knowledge and expanding the effort “to extend personalized medicine to everyone.”