By Stacy M. Brown, NNPA Newswire Senior National Correspondent
A ground-breaking study found glaring disparities in the language that healthcare professionals use when recording patient interactions, which could have an impact on the caliber of care that minority groups receive. Titled “Examining Linguistic Differences in Electronic Health Records for Diverse Patients with Diabetes: Natural Language Processing Analysis,” the study analyzed electronic health records (EHRs) of Black, white, and Hispanic or Latino patients treated by 281 physicians in a major metropolitan area. The findings highlight how racial and ethnic biases may permeate even the most intimate and routine aspects of healthcare.
The research, which Eden King, the Lynette S. Autrey Professor of Psychological Sciences at Rice University, led, sought to ascertain whether doctors use biased language when describing patients in post-visit reports. “Language and communication are central to social interactions across cultures, including the critical exchanges that occur between clinicians and patients,” King stated. “Our study sought to uncover whether the words physicians use in health records reflect biases, and the results are concerning.”
The study employed a sophisticated natural language processing tool, the Sentiment Analysis and Social Cognition Engine (SEANCE), to examine various linguistic markers in the EHR text. The analysis revealed that physicians’ notes for Black and Hispanic or Latino patients contained significantly more negative adjectives—such as “unkind,” “negative,” and “stupid”—and words associated with fear and disgust, including “intimidate,” “attack,” and “cringe.” In contrast, notes for white, non-Hispanic patients featured more positive language, including adjectives like “supportive” and “kind,” as well as verbs indicating trust, such as “affirm” and “advise.”
“These findings align with a growing body of research demonstrating that racial and ethnic minorities often receive inferior care, marked by less empathy, reduced rapport, and diminished patient trust,” King explained in a news release. The study’s results are consistent with previous reports, such as the 2018 National Healthcare Disparities Report, which found that Black and Hispanic patients frequently experience worse care on numerous quality measures compared to their white counterparts. This includes receiving less respect and attention from physicians, further contributing to health disparities.
Researchers assert that the implications of these linguistic biases are far-reaching. Language in EHRs not merely reflects a physician’s observations but also influences future medical decisions, as these records are reviewed and referenced in subsequent patient encounters. Bias in these records can perpetuate negative stereotypes and lead to ongoing disparities in care. The study’s authors emphasize that understanding and addressing these biases is crucial for improving health outcomes for minority populations.
King and her team hope their research will be a wake-up call to the medical community. “If we can develop and refine algorithms to detect such biases, we can raise awareness among clinicians during patient interactions,” King said in the release. “This heightened awareness could be a critical step toward more equitable healthcare.”
The study also points to the need for systemic changes in how medical professionals are trained and how EHRs are utilized. While EHRs are essential for documenting patient care, officials noted that the language used in these records can reflect unconscious biases that may influence patient outcomes. They said the research underscores the importance of ongoing education and training for healthcare providers to recognize and mitigate these biases.
The study’s findings open the door for further research into the relationship between biased language in medical records and patient outcomes. The research team, which includes experts from institutions such as the University of North Carolina at Charlotte, the University of Houston, and Duke University, plans to explore whether biased language correlates with poorer health outcomes for minority patients and whether interventions can reduce these disparities.
The study, supported in part by a grant from the Rice Race and Anti-Racism Research Fund, was published in JMIR Medical Informatics and is available online.
