Is there implicit bias in patient clinical notes?

Inequities in clinical note language

The researchers looked at more than 18,000 clinical notes for patients admitted for labor and birth between 2017 and 2019. From there, the team used natural language processing to scan the clinical notes to detect stigmatizing language, which was subdivided into four categories, such as marginalized language or identities, difficult patient, unilateral or authoritarian decisions and questioning patient credibility.

The team likewise scanned the clinical notes to identify positive language categories, such as respecting patient autonomy or noting a patient's power or privilege.

All said, Black patients were more likely to have providers use both positive and negative language about them in their clinical notes. Specifically, Black patients were 25% more likely to have clinicians use stigmatizing language and 18% more likely to have positive language in their clinical notes compared to their White peers.

The researchers specified that language falling into the preferred language or autonomy category was frequently used in clinical notes about Black patients. In terms of stigmatizing language used, the team said Black patients were among the most likely to see more marginalized language and language describing them as difficult patients.

Meanwhile, Hispanic patients were 10% less likely to have their providers use positive language about them in their clinical notes. For Asian and Pacific Islander patients, that figure was 29%.

Risks to health equity

Racial differences in how healthcare providers write about patients in clinical notes is indicative of a larger industry problem about implicit bias, the researchers said. Take, for example, the higher odds Black patients have of reading negative clinical notes.

"Black patients had the highest frequency of stigmatizing language documented in their clinical notes," the researchers said. "This may be explained by implicit biases that influence clinician perceptions and documentation practices. These biases can affect a clinician’s decision-making for treatments and their interaction with patients, contributing to racial and ethnic disparities in documentation practices."

In addition to clinical decision-making, implicit bias can harm the patient-provider relationship. If stigmatizing language makes its way into patient-provider communication, patients might have poor healthcare experiences and even stop trusting their providers. That risks overall patient access to care.

To add insult to injury, patients have a legal right to read and review the clinical notes in question, per the information blocking rules under the 21st Century Cures Act. When patients read what could be perceived as unflattering language about themselves, the patient-provider relationship could be damaged.

Using education to improve clinician notes

Since the information blocking rule came into effect in 2021, healthcare providers have sought strategies to write clinician notes that preserve a good patient-provider relationship. This data provides more insights into where providers might be falling short and where provider education can be bolstered.

Although clinician education about using biased language is important, the researchers noted that it doesn't get to the root of the problem.

"However, simply removing language from documentation will not eliminate potential underlying biases, and the integration of bias-free templates and prompts in EHR to standardize documentation should be implemented with caution," they explained. "Efforts to implement software that flags stigmatizing language in real time, for example, should be paired with training to understand why it is being used."

Instead, cultural competence training and clinical documentation training that specifically considers implicit bias will be key, the team concluded.

Sara Heath has reported new related to patient engagement and health equity since 2015.