Natural language processing (NLP) algorithms are used to train computers to understand and process natural language. In health care, clinical notes are used to train these NLP models. Clinicians’ use of language in documentation may contribute to transmission of bias that could lead to differences in care.

Cobert and colleagues explored mentions of race and ethnicity in clinical notes from 6 intensive care units. They found that

  • Notes for Black patients had nearly twice as many mentions of race as notes for White patients, and Black patients were more likely than White patients to have race mentioned in clinical notes.

  • Latinx patients did not have more mentions of race than White patients had.

Study findings support the literature on linguistic biases that occur in clinical notes. The authors acknowledge that stigmatizing language may influence clinicians’ perceptions, yet race and ethnicity documentation could also raise awareness for interventional care of marginalized...

You do not currently have access to this content.