乌鸦传媒

NEW YORK CITY -- Clinical notes for patients with substance use disorder (SUD) contained stigmatizing language -- such as "junkie," "dirty user," and "this drunk" -- on a regular basis, according to a study of electronic health records (EHR).

A preliminary analysis of clinical notes from a random sample of 2,700 patients with SUD or opioid use disorder (OUD) diagnoses showed that 84.4% (n=2,279) of patients had notes that contained stigmatizing language, according to Jyoti Pathak, PhD, of Weill Cornell Medical College in New York City.

The most common terms found among patients' notes were abuse (77.1% of patients), opioid abuse (63.8%), addict (41.9%), and substance dependence (36.4%), Pathak reported at the American Psychiatric Association annual meeting. Stigmatizing language was based on identified by the National Institute on Drug Abuse (NIDA).

Pathak and colleagues also found that such language was applied to some patients more than others, and used by some care providers more than others.

"We would all agree that stigma persists in healthcare," Pathak said during a poster presentation. "Research has shown that the words that we choose in clinical documentation, in patient communication, and clinician-to-clinician documentation have consequences."

He explained that a 2022 study in the illustrated that stigmatizing language can frame the narrative around a patient's overall experience with treatment. A 2019 study in found that patients have negative attitudes toward terms like "addict" or "substance abuser," which may negatively affect their treatment outcomes.

Pathak reported that the two age groups who experienced the most common use of stigmatizing language in their clinical notes were those ages 33-48 (32.7% of all patients) and those ages 49-64 (34.1%). Similarly, Black or African-American patients made up almost a fifth of all patients who experienced stigmatizing language in their clinical notes (18.2%), he reported.

Social workers were far more likely to write clinical notes that contained stigmatizing language. Of the 390 clinical notes written by social workers in the analysis, all but one contained stigmatizing language. Comparatively, psychiatrists used stigmatizing language in 750 out of 877 clinical notes, Pathak reported.

"This is very consistent with what has been shown in other literature," he said. "There is absolutely a need for provider education in this space."

Pathak noted that people-first language is the recommended alternative. For example, instead of using "addict" a provider can use "person with substance use disorder." Similarly, the term "substance abuser" can be replaced by "patient." He pointed out that the NIDA has with learning and implementing the appropriate language in their clinical notes.

Pathak and colleagues collected 980,194 clinical notes from the random sample of patients with either an OUD or an SUD diagnosis from 2010 through 2023 at a major academic medical center in New York City.

Using the NIDA list of stigmatizing terms and phrases, the researchers developed a rule-based natural language processing algorithm to extract clinical notes with stigmatizing language, and analyzed the incidence rates of this language in those clinical notes. Demographic characteristics of patients who had stigmatizing language used on them were compared to those without.

The researchers reported that notes authored by female providers use SL [stigmatizing language] terms more frequently (50.0% in SL cohort vs 42.4% in no SL cohort) than those authored by male providers.

Pathak highlighted that the use of natural language processing to identify notes with stigmatizing language may open the door to the use of other artificial intelligence tools to potentially address this issue.

"[When] you write your [email in] Gmail, it autocorrects your language," he said. "Can we develop similar tools that can be embedded within your [EHR], so when you are writing your note it can provide recommendations [such as] instead of using the word 'addict,' perhaps [use] the right terminologies?"

Comment