Study Finds Racial Bias Embedded in Electronic Health Records

January 21, 2022

Although ostensibly objective observations of patient information, a new study in Health Affairs shows that electronic health records (EHR) reveal racial bias on the part of clinicians. According to the study, negative language was 2.54 times as likely to be present in the EHR of Black patients than white patients.

According to Stat News, “With the help of a panel from the Health Equity Commission of the Society of General Internal Medicine, the researchers chose a set of 15 negative words that doctors sometimes use. These included terms like “angry,” “aggressive,” “challenging,” “exaggerate,” and “noncompliant.” Then the researchers trained a machine learning model to comb through the records and find instances where patients were described in these ways. The researchers didn’t have enough data to compare additional racial groups, but the difference between Black and white patients was clear.”