Stigma and language: how a label changes clinical conduct

The handover lasts less than a minute per bed. "Bed seven, the dependent one, asking for analgesia again, he already knows how it goes." The night team takes over the patient with that sentence in mind. When he reports severe abdominal pain at four in the morning, the response is quick and economical. The next morning, a CT scan shows a perforation. No one acted out of malice. One word, said in a hurry, arrived before the physical exam and decided how much credit the report deserved.

NEXO is a daily medical diagnosis game: one clinical case a day, up to six progressive clues, and ICD-10 proximity feedback across five levels.

This text is about how language about substance use enters clinical reasoning. The setting is addiction, but the subject is cognition: how a label works as an anchor, how the medical record amplifies a judgment and how the choice of words changes the conduct of whoever reads it. You will see studies that measured this effect, a table of terms with descriptive alternatives and practices for the team and for teaching.

Key points

• In a randomized study, professionals who read the word "abuser" assigned more blame and recommended more punitive measures than those who read "person with a substance use disorder", for the same case.
• Labels work as anchors: they narrow the space of hypotheses and change the weight given to the person's report.
• The medical record propagates stigmatizing language from one professional to the next, and this language has already been associated with more negative attitudes in those who read it.
• Diagnostic overshadowing, when physical symptoms are attributed to substance use, is a reasoning error with serious consequences.
• Person-centered language and objective descriptions preserve clinical information and remove judgment.
• Changing a team's language goes through example, teaching and review of medical record templates.

The experiment of the two words

In 2010, Kelly and Westerhoff ran a simple experiment with mental health professionals. All of them read the same clinical vignette about a person with problems related to substance use. The only difference was one term: half read that the person was a "substance abuser", and half read that they "has a substance use disorder". Those who read the first term considered the person more to blame for their own situation and more deserving of punitive measures than of treatment.

What makes the study important for clinical reasoning is the audience. The participants were trained professionals with experience in the subject, who would probably say they act without bias. The word acted below the level of intention. Ashford and colleagues replicated the idea with health professionals and people in recovery and found stronger negative associations with terms such as "addict" and "substance abuser" than with person-centered alternatives. In another study, using an implicit association task, the same group showed that these terms activate automatic biases.

These findings fit what is already known about implicit bias in health care. The systematic review by FitzGerald and Hurst found evidence of implicit bias among health professionals across several social groups, with an influence on interaction and decisions. The post about implicit bias in diagnosis describes the same mechanism for gender, race and age.

What these studies measure, and what remains open

It is worth reading these results with the same critical eye you would apply to any study. Vignettes measure judgments in a controlled setting, and the real decision at the bedside involves time, fatigue, pressure and information that the vignette does not reproduce. Implicit association tasks measure automatic tendencies, and the relationship between these tendencies and concrete conduct is debated in the literature. The studies were also mostly conducted in English and in few countries.

These limitations do not erase the conclusion, but they adjust its size. What the data support is that words change professionals' judgments under controlled conditions, and that negative attitudes are associated with worse quality of care. What is still being studied is how much each change of language, on its own, improves outcomes. For practice, the asymmetry favors change: swapping the term takes little effort and preserves the clinical information.

The practical conclusion is uncomfortable and useful at the same time: good intentions do not protect. What protects is changing the stimulus, that is, the words that circulate in the team and in the medical record.

From label to bias: how language enters reasoning

A label such as "dependent" or "addict" carries a ready-made explanation. It suggests that the main problem is already known, that the person's behavior is predictable and that their complaints should be read with suspicion. Each of these suggestions corresponds to a cognitive bias described in the literature on diagnostic reasoning.

Anchoring

The first information received sets the starting point, and later adjustments are usually insufficient. Croskerry describes anchoring as one of the most frequent biases in emergency care. When the first information is a substance use label, the following hypotheses tend to revolve around it: intoxication, withdrawal, medication seeking.

Framing effect

The same situation described in different ways leads to different decisions. "Patient demands analgesia" and "patient reports severe pain and asks for relief" describe the same request, but the first frames the person as an adversary and the second as someone with a symptom.

Affective bias

Feelings toward the person, such as irritation, tiredness or fear, influence judgment. The review by van Boekel and colleagues showed that professionals' negative attitudes toward people with substance use disorders are common and are associated with less engagement, shorter consultations and less empathy. The post on affective bias shows how to recognize this effect in yourself.

Premature closure

When the label offers a sufficient explanation, the search for other explanations stops. It is the same trap as in any case, only accelerated by the feeling of "already knowing" the person.

Confirmation bias

Once the label takes hold, attention starts looking for what confirms it. A request for analgesia becomes proof of "seeking behavior"; a slightly altered pupil becomes proof of "recent use". Data that contradict the label, such as pain that follows a coherent anatomical pattern or an altered physical examination, receive less weight. The antidote is the same as in any case: actively ask "what finding would I expect if my hypothesis were wrong?" and look for it.

Taken together, these biases produce a recognizable pattern: less physical examination, fewer additional tests, less credit to the report of pain and more attribution of symptoms to substance use. Stigma, in this sense, is a problem of diagnostic quality, not only of courtesy.

There is an additional effect, less discussed: the label also changes what the team asks. Someone who has already decided that the pain is "medication seeking" stops asking about onset, radiation, relieving and aggravating factors. The history becomes poor, and a poor history confirms any hypothesis. Thus, bias does not only distort the interpretation of data; it reduces the amount of data available to interpret.

The medical record as an amplifier

Language spoken in a shift handover disappears with the shift. Language written in the medical record stays, is copied and is read by professionals who never saw the person. This makes the medical record an amplifier of judgments.

Park and colleagues analyzed clinical notes and identified recurring categories of stigmatizing language: questioning the credibility of the report ("alleges", "states that supposedly"), describing the person as difficult or uncooperative, using literal quotes to ridicule and attributing behaviors to character traits. Sun and colleagues showed that negative descriptors appear at a different frequency depending on the patient's race, which indicates that the medical record carries and reproduces social biases.

Two mechanisms make the effect lasting:

• Copy and paste: a phrase like "known user, polysymptomatic complainer" can survive years of updates without anyone reassessing it.
• Problem list: a diagnosis recorded without a date, severity or current status behaves like a permanent fact, even when the person has been in remission for years.

A good clinical note describes what was observed and what was reported, separates the two and avoids adjectives about character. The post on organizing clinical reasoning in the medical record offers a model that applies directly: explicit hypotheses, the data that support them and the data that weaken them. This format leaves little room for the label to do the work of reasoning.

Rewriting a note

A fictional example helps show the difference. The first version: "Chronic alcohol user, known to the service, polysymptomatic complainer, claims abdominal pain and demands medication. Reportedly says he has not drunk for a week." The second version: "Man, forty-eight years old, with alcohol use disorder, reports last drink seven days ago. Epigastric pain for twelve hours, intensity eight out of ten, radiating to the back, with vomiting. Asks for analgesia. On examination, abdomen tender to deep palpation in the epigastrium."

The second note is longer, but every extra word is a piece of data. It lets the next professional think of pancreatitis, perforated ulcer or cholecystitis, because it describes the pain instead of judging the request. The first note, by contrast, suggests a single course of action: distrust.

Who else reads the medical record

In many countries, patients have increasingly direct access to their own clinical notes. Reading "claims" or "difficult" about oneself affects trust in the team and the willingness to come back. Writing with the awareness that the person will read it is a simple test: if the sentence would embarrass you when read aloud in front of them, it probably carries judgment instead of information.

Table: term, alternative and why it matters

The table gathers frequent terms and alternatives suggested by reference documents, such as the text by Botticelli and Koh in JAMA, the article by Volkow and colleagues and the NIDA page on stigma. The alternatives serve as general guidance; each language and each service adjusts the vocabulary.

Stigmatizing term	Descriptive alternative	Effect on reasoning
"Addict", "junkie", "druggie"	Person with a substance use disorder	Separates the person from the condition and keeps other hypotheses open
"Alcoholic", "drunkard"	Person with alcohol use disorder, with severity and current status	Requires recording severity and whether the condition is active or in remission
"Drug abuse"	Substance use, risky use or use disorder, as applicable	Replaces moral judgment with a clinical category
"Dirty urine", "dirty test"	Test positive for substance X	Keeps the information and removes the connotation of dirt
"Clean", "got clean"	In remission, no use since date Y	Gives time and status, useful for reasoning
"Claims pain", "reportedly says"	Reports pain of intensity Z	Records the symptom without prejudging credibility
"Polysymptomatic complainer", "difficult"	Description of the complaints and the observed behavior	Preserves data that may point to another disease
"Relapsed", "failed treatment"	Resumed use; the current approach needs adjustment	Shifts the focus from blame to the approach

Look at the third column. Swapping words almost always adds clinical information: severity, date, substance, intensity. Person-centered language, besides being more respectful, is usually more precise.

Diagnostic overshadowing: when the label hides the disease

Diagnostic overshadowing happens when physical symptoms are attributed to an already known psychiatric condition or to substance use. The concept analysis by Hallyburton describes the phenomenon as the mistaken attribution of physical symptoms to preexisting psychological illnesses, with diagnostic delay and worse outcome.

In the context of substance use, some patterns appear often:

• Confusion attributed to intoxication when the cause is hypoglycemia, head trauma or central nervous system infection.
• Chest pain attributed to anxiety or to stimulants without the assessment anyone else would receive.
• Fever in a person who injects drugs attributed to a "drug reaction" when the diagnosis is endocarditis.
• Abdominal pain treated as drug-seeking when there is pancreatitis, perforation or ischemia.
• Weight loss and cough attributed to lifestyle when there is tuberculosis or cancer.

It is also worth remembering that people with substance use disorder have, for several reasons, an increased risk of infectious, cardiovascular, hepatic and traumatic diseases. This means the pre-test probability of several serious diseases is higher in this group, not lower. Diagnostic overshadowing reverses this logic: precisely those at higher risk receive less investigation. Correcting the reasoning here means applying the same rigor to this group as to any other and adding to it the attention that increased risk demands.

A simple question helps: "if this person did not have the label, what tests would I order?". If the answer differs from the current approach, the label is deciding in place of reasoning. Structuring the reasoning in patients with many complaints offers a complementary technique: list each complaint as an independent problem before trying a single explanation.

A fictional case in three clues

First clue: a man in his early thirties, with injection drug use recorded in the chart, arrives with fever and malaise. Many students stop here and think of "skin abscess" or "reaction to the substance". Second clue: there is a heart murmur that was not in previous records. Third clue: small hemorrhagic lesions under the nails. The hypothesis of infective endocarditis, which was available from the first clue, only appears to those who treated injection drug use as a risk factor, not as an explanation.

The difference is subtle and decisive. A risk factor increases the probability of certain diagnoses and asks you to look for them. A label ends the search. The same piece of data can serve both functions, and the language in which it reaches your reasoning influences which one prevails.

Changing language in practice: team, teaching and hidden curriculum

Knowing that language matters rarely changes how a team speaks. Change happens when the environment makes the new language easier than the old one. The WHO and UNODC international standards for the treatment of drug use disorders set ethical, non-discriminatory, person-centered care as a principle of treatment, and that principle begins with words.

Some practices that teams in different countries have adopted:

1. Review chart templates. Pre-filled fields with stigmatizing terms multiply with every note; changing the template changes thousands of records.
2. Correct with questions, not sermons. "What exactly did he report?" brings the conversation back to the clinical data without embarrassing the colleague.
3. Set the example in shift handovers. The language of those who coordinate is copied by those who are learning.
4. Include the topic in clinical reasoning teaching. Discussing the Kelly and Westerhoff study in a class on biases often has more impact than a class on etiquette.
5. Write teaching cases with person-centered language. Cases are models, and students copy the vocabulary they read.

For those who coordinate teams or teach, some transition phrases help swap the label for description without breaking the flow of discussion:

• "Let's describe what he reported, instead of how he asked."
• "What is the substance, the severity and when was the last use?"
• "If I take this information out of the history, what remains in the differential?"
• "What does the next team need to know so as not to repeat this assessment?"

Measuring also helps. Some teams periodically review a sample of progress notes for stigmatizing terms and share the result with the group, without identifying authors. The goal is to make visible a habit that, being automatic, goes unnoticed. The review works best when it accompanies the change of chart templates and the example set by those who coordinate.

The last point connects the theme to the hidden curriculum in medicine: what the school teaches without saying it is teaching. A professor who says "the junkie in bed three" teaches more about stigma than any formal class. The guide on how to write inclusive clinical cases offers practical criteria for those preparing teaching material.

Language in several languages

Much of the discussion about language began in English, but each language has its own loaded terms. In Portuguese, "drogadito" and "viciado" circulate in hospital corridors; in Spanish, "drogadicto" and "borracho"; in French, "toxico" and "alcoolo". Translating the English list word by word leaves out precisely the most used local terms. The useful exercise for each team is to identify its own expressions and think of the descriptive alternative for each one.

There is also the side of the person receiving care. People who feel judged tend to omit information, avoid services and come back later, already with complications. Stigmatizing language therefore worsens the quality of the data that reach the reasoning. A respectful conversation improves the history, and a better history improves the diagnosis.

How NEXO handles language in the cases

NEXO presents fictional cases with a real clinical presentation, in up to six progressive clues. When a case involves substance use, it appears as a piece of the history, described with person-centered language, alongside the other findings. The exercise is the same as in any case: integrate the data with the other clues and propose a diagnosis with an ICD-10 code, receiving proximity feedback at five levels.

This format offers an opportunity for particular practice. In cases where substance use appears in the first clue and the diagnosis is in another ICD-10 chapter, the distance between the initial hypothesis and the answer shows, concretely, the anchoring effect described in this text. Physicians who review and suggest adjustments to cases, including in language, can learn about the NEXO Curator program. The Collection page shows the scope of the diagnoses and the limits of use are at /economia.

Researchers interested in biases in diagnostic reasoning find the path in NEXO Researcher. Creators of medical content have a program in NEXO Creators, and teaching institutions find information at /closer. To learn about the design of the game, read what NEXO is.

Frequently asked questions

Is changing words just a matter of courtesy?

Studies show an effect on clinical judgments. In the experiment by Kelly and Westerhoff, a single term changed the attribution of blame and the preference for punitive measures among trained professionals. This makes language a matter of quality of care.

Does person-centered language erase clinical information?

In general, it adds. "Person with alcohol use disorder, moderate severity, no use for six months" conveys more than "alcoholic". The alternative forces the recording of severity, substance and current status.

What is diagnostic overshadowing?

It is attributing physical symptoms to an already known psychiatric condition or to substance use, failing to investigate other causes. The analysis by Hallyburton describes the concept and its consequences.

How can I correct a colleague without creating conflict?

Questions that return to the data work better than direct corrections. "What exactly did he report?" or "when was the last use?" bring the conversation back to the clinic and show, in practice, the alternative language.

Can stigmatizing language in the medical record affect other professionals?

Yes. The analysis by Park and colleagues describes how these expressions appear in the notes and are read by those who provide care later, and Sun and colleagues show that they are distributed unevenly across groups.

Is this text about treatment?

The focus is clinical reasoning and communication. Treatment follows guidelines and individual assessment. If you or someone close to you is facing substance use, seek the help you need from a health service.

Final checklist

• I described the condition with substance, severity and current status, instead of a label.
• I recorded the person's account without verbs that question credibility.
• I separated what I observed from what was reported to me.
• I asked which tests I would order if the person did not have the label.
• I reviewed the problem list for dates and remission status.
• During the shift handover I used the same language I would want to read in the medical record.

Play the case of the day: open NEXO and, with each clue, ask whether your hypothesis would change if that piece of the history were removed.

References

1. Kelly JF, Westerhoff CM. Does it matter how we refer to individuals with substance-related conditions? A randomized study of two commonly used terms. Int J Drug Policy. 2010. PMID 20005692.
2. Ashford RD, Brown AM, McDaniel J, Curtis B. Biased labels: an experimental study of language and stigma among individuals in recovery and health professionals. Subst Use Misuse. 2019. PMID 30945955.
3. Ashford RD, Brown AM, Curtis B. The language of substance use and recovery: novel use of the Go/No-Go Association Task to measure implicit bias. Health Commun. 2019. PMID 29863411.
4. FitzGerald C, Hurst S. Implicit bias in healthcare professionals: a systematic review. BMC Med Ethics. 2017. PMID 28249596.
5. van Boekel LC, Brouwers EP, van Weeghel J, Garretsen HF. Stigma among health professionals towards patients with substance use disorders and its consequences for healthcare delivery: systematic review. Drug Alcohol Depend. 2013. PMID 23490450.
6. Croskerry P. Achieving quality in clinical decision making: cognitive strategies and detection of bias. Acad Emerg Med. 2002. PMID 12414468.
7. Park J, Saha S, Chee B, Taylor J, Beach MC. Physician use of stigmatizing language in patient medical records. JAMA Netw Open. 2021. PMID 34259849.
8. Sun M, Oliwa T, Peek ME, Tung EL. Negative patient descriptors: documenting racial bias in the electronic health record. Health Aff (Millwood). 2022. PMID 35044842.
9. Botticelli MP, Koh HK. Changing the language of addiction. JAMA. 2016. PMID 27701667.
10. Volkow ND, Gordon JA, Koob GF. Choosing appropriate language to reduce the stigma around mental illness and substance use disorders. Neuropsychopharmacology. 2021. PMID 34276051.
11. Hallyburton A. Diagnostic overshadowing: an evolutionary concept analysis on the misattribution of physical symptoms to pre-existing psychological illnesses. Int J Ment Health Nurs. 2022. PMID 35718951.
12. National Institute on Drug Abuse. Stigma and discrimination. nida.nih.gov.
13. World Health Organization, United Nations Office on Drugs and Crime. International standards for the treatment of drug use disorders. who.int.