About 85% of patients with opioid use disorder (OUD) are described in clinical notes as being abusers, addicts, junkies, or with other stigmatizing terms, preliminary results of a new study suggest.
Female healthcare providers and social workers used stigmatizing language at a relatively high rate, investigators found. The researcher also showed that demeaning language was used more often in medical records of Hispanic and Black patients, reaffirming previous research.
Investigators noted that words chosen by clinicians can contribute to patients developing a negative attitude toward their healthcare provider, affecting follow-ups and overall treatment outcomes. This is particularly important after the passage of the 21st Century Cures Act, which mandates that patients have free access to their personal medical records, they added.
“No matter what type of specialty you’re in, and no matter if you’re a male or female clinician, you should choose your words carefully,” study investigator Jyotishman Pathak, PhD, professor of psychiatry and of population health sciences, Weill Cornell Medicine, Cornell University, New York, told Medscape Medical News.
In physician-patient encounters, especially with patients of lower socioeconomic status or less education, “there’s already a power dynamic going on,” Pathak added. “And if patients come across phrases that are perhaps demeaning and stigmatizing, it makes it even worse.”
The findings were presented on May 5 at the American Psychiatric Association (APA) 2024 Annual Meeting.
Artificial Intelligence–Based Algorithm
Pathak and his colleagues used natural language processing based on artificial intelligence (AI) and machine learning to update a list of stigmatizing words put together by the National Institute of Drug Abuse‘s Words Matter campaign, which includes words like addict, abuse, dirty, junkie, and alcoholic.
They developed a preliminary version of an algorithm and applied it to close to 1 million clinical notes from electronic health records of a random sample of 2700 patients diagnosed with OUD or substance use disorder (SUD) between 2010 and 2023 at Cornell. The patient encounters included visits to primary care providers, emergency room physicians, social workers, psychiatrists, psychologists, or other clinicians.
Researchers divided patients into those with evidence of stigmatizing language in their clinical notes (SL group) and those with no evidence of stigmatizing language (No SL group).
About 85% of patients with OUD/SUD (n = 2279) had stigmatized language terms in 111,422 notes. The most common negative terms included abuser, addict, substance dependence, and alcoholic.
Individuals in the SL group tended to be older, and there were more people in the SL group than in the No SL group in each age category from ages 33 to 80 years.
The SL group also had more women (34% vs 29%), Black patients (18% vs 15%) and Hispanic patients (23% vs 17%) than did the No SL group. Such results are consistent with what has been reported in other studies and patient surveys, noted Pathak.
The study showed a greater percent of female healthcare providers in the SL group than in the No SL group (50% vs 42.4%), which is a new finding. It’s not clear whether these female clinicians are older and were trained at an earlier era, said Pathak, who also noted the increased focus on diversity, equity, and inclusion today in medical schools.
Investigators also found more social workers in the SL group than in the No SL group (6% vs 0.2%).
“We saw a very high use of stigma language in social worker notes,” said Pathak, adding this was “surprising” and “disappointing.”
“Patients are already very vulnerable, and you would expect that the language being used would have more empathy,” he said.
Researchers found the opposite situation for the psychiatrists who made up 12% of the SL group and 26% of the No SL group.
“We actually found that both psychiatrists and psychologists used less stigmatizing language in their clinical documentation,” said Pathak. “It looks like they’re more aware of what’s the right terminology to use.”
‘Active’ Labels Recommended
Pathak suggests replacing passive labels, such as drug abuser and junkie, with first-person active phrasing, such as “the patient has an opioid use disorder”; the terms alcoholic or drunk should be changed to “a person with an alcohol use disorder” or “a person who misuses alcohol or engages in unhealthy/hazardous alcohol use.”
The research team has already developed a second, more nuanced version of the algorithm.
The aim is to develop AI tools that help clinicians document clinical notes appropriately. This might involve an autocorrect feature that flags or corrects inappropriate language or prompts the user to change the wording, said Pathak.
He hopes that this will lead to more positive interactions between physicians and patients and better treatment outcomes and expects that this tool could facilitate improved clinician education.
Better Education Needed
Commenting on the research for Medscape Medical News, Howard Y. Liu, MD, chair, Council on Communications, APA, and professor of adult, child, and adolescent psychiatry, University of Nebraska Medical Center, Omaha, Nebraska, said that the study is a reminder that terms like addict and drunk are still being used by professionals.
“This can be hurtful to individuals living with a substance use disorder and can be a barrier to them seeking help” or being upfront about their drinking or drug use with their clinical team, said Liu.
The study showed less use of stigmatizing language among psychiatrists and psychologists but more frequent use by social workers, which suggests to Lui that “across professions, we need to educate ourselves about person-first language. He also shared and that those in the education system should act as role models.
“We need to ensure that in academic health centers where future psychiatrists and clinicians are trained, professors and peers are using person-first language and challenging stigmatizing language when it arises,” Liu said.
The study received no outside funding and there were no reported relevant conflicts of interest.
