News
Article
In 13M+ EHR notes from a Mid-Atlantic health system, clinicians were more likely to record doubt about Black patients’ sincerity and competency than White patients.
© M Einero/peopleimages.com - stock.adobe.com
Clinicians were more likely to question the sincerity or competence of Black patients in their medical documentation compared with White patients, according to a study published August 13 in PLOS One.
Researchers from Johns Hopkins University analyzed 13,065,081 clinical notes written between 2016 and 2023 for 1,537,587 adult patients by 12,027 clinicians spanning emergency medicine, internal medicine, OB/GYN and surgery across five hospitals and a network of ambulatory practices in the Mid-Atlantic U.S.
Using natural language processing (NLP) models, the team searched for words and phrases that can undermine patient credibility, like “claims,” “insists,” “adamant,” or “poor historian.” They also looked for language supporting credibility, like “good” or “reliable historian.”
Overall, 0.82% of notes contained language undermining credibility (106,523 total), split between sincerity-undermining terms (0.48%) and competence-undermining terms (0.40%). After adjusting for patient demographics, insurance, substance use disorder, severe mental illness, clinician role, department and interpreter use, notes about non-Hispanic Black patients had 29% higher odds of including credibility-undermining language than notes about White patients.
Notes about Black patients were also less likely to include language supporting credibility.
“Our findings accordingly signal larger underlying disparities in credibility assessments,” the authors wrote, noting that even infrequent biased language can influence how future clinicians view a patient and affect treatment decisions.
The disparities extended across both sincerity and competence:
When examining specific terms, each credibility-undermining word category appeared more frequently in notes about Black versus non-Black patients — except phrases related to narcotic/drug-seeking, which were less common in Black patients’ notes in exploratory analyses.
By comparison, notes about Asian patients had lower odds of language undermining credibility overall and higher odds of language supporting credibility.
Documentation for Hispanic/Latino patients showed mixed patterns. After full adjustment, sincerity-undermining language was less common than for White patients, while competence-undermining was more common.
The authors frame these findings as a form of “testimonial injustice,” in which prejudice shapes whether a person is believed. They note that dismissing patient-reported symptoms can lead to delayed diagnoses, inappropriate or missed treatments and medical errors, while eroding trust and engagement with care.
Because notes are read by multiple clinicians, biased or stigmatizing phrasing can shape subsequent decision-making. Prior work has shown that labels like “poor historian” can influence how later clinicians interpret symptoms and allocate attention.
The authors suggest replacing blanket judgment with specific, descriptive statements, like “patient unable to provide a complete history” or “patient is uncertain of some details,” to convey limitations without assigning blame that can follow a patient across encounters.
The authors recommend that medical education and documentation training address credibility bias alongside broader efforts to confront structural inequities and interpersonal attitudes that contribute to biased language in the electronic health record (EHR).
As artificial intelligence (AI)-assisted documentation tools become more common, the researchers argue that these systems should be designed to avoid propagating biased phrasing and to prompt more neutral, patient-centered language.
“For years, many patients — particularly Black patients — have felt their concerns were dismissed by health professionals,” the authors wrote. “By isolating words and phrases suggesting that a patient may not be believed or taken seriously, we hope to raise awareness of this type of credibility bias with the ultimate goal of eliminating it.”
Stay informed and empowered with Medical Economics enewsletter, delivering expert insights, financial strategies, practice management tips and technology trends — tailored for today’s physicians.