How to write inclusive clinical cases

You open the case document the student group will use tomorrow. The first line says: "older Black woman, diabetic, non-adherent, arrives at the appointment irritated". In two lines, the text has already stacked race, age, illness as identity, moral judgment and mood as a trait. The teaching goal was to teach ketoacidosis. What the student takes away, beyond the differential, is a stereotype. Rewriting the case is not "political correctness": it is pedagogy. Fictional cases with a real clinical presentation teach reasoning. They also teach, without announcing it, who deserves to be heard and who arrives already labeled. This text is a practical guide for those who write vignettes: preceptors, student groups, reviewers and creators. The focus is inclusion and accessibility in writing. The general method for building a case (goal, clues, difficulty) is already in how to write a clinical case for teaching.

Inclusion, here, is not decoration. It is choosing words that describe findings without reducing the person to a label; it is deciding when age, gender or race come in because they change the reasoning, and when they only reinforce bias; it is keeping the text readable for those who use a screen reader, those who read in a second language and those who study under high cognitive load. The ground overlaps with implicit bias in diagnosis, with hidden curriculum and with narrative medicine. Do not confuse this angle with the clinical software checklist in digital accessibility in medical software, nor with the guide to readers and magnifiers in assistive technology for physicians with visual impairment. Here the object is the text of the case.

Key points

• Writing an inclusive case is pedagogy: the vignette trains reasoning and, at the same time, models clinical language.
• Person-centered language is the starting point; communities such as the deaf and autistic communities often prefer identity-centered language. Ask or follow the preference of the community cited.
• Age, gender and race enter the case only with an explicit clinical or representational purpose; otherwise, they become a stereotyped shortcut.
• Clarity, structure and real text (not just an image) make the case accessible to screen reader users and to those who study under load.
• At NEXO, the Curator and the Creators flow work with fictional cases with a real clinical presentation and up to six clues; the statement at /accessibility describes the platform's accessibility resources. No claim of efficacy.

What makes a clinical case inclusive

A clinical case for teaching is a text (or a sequence of clues) that puts the student before a presentation and asks for hypotheses, tests or management. The case-based learning line explains why the vignette works: it anchors knowledge in a situation. Inclusion, in this guide, is the set of choices of language, structure and representation that keep the case from teaching exclusion along with the diagnosis.

Three layers are separated:

• Clinical layer: signs, symptoms, time, tests. It must be precise and sufficient for the teaching goal.
• Social and demographic layer: age, gender, race or ethnicity, occupation, disability, context. It enters only if it serves the reasoning or the deliberate representation of the class/population.
• Judgment layer: adjectives of morality, mood, "adherence", patient "difficulty". It is almost always left over; it almost never teaches pathophysiology.

The post on the general method (how to write a clinical case for teaching) covers the goal, the dosing of clues and review. This text covers the layer the general method usually leaves implicit: whom the vignette describes and how. In an analysis of clinical narratives in medical biochemistry books, the language often departs from good inclusive practices, with labeling by condition and negative descriptors (PMID 38887417). The same risk applies to the case you write in the student group or in the platform's bank.

Inclusive does not mean "case without identity" either. It means identity with purpose. A sickle cell anemia case may need to name ancestry or family context in a clinical way; a community-acquired pneumonia case rarely needs race as a shortcut. The useful question: if I remove this descriptor, does the differential change? If it does not change, why is it there?

Another simple test: ask a colleague to read only the opening and list what they "already know" about the person before any exam. If the list comes back full of moral judgment and empty of physiology, the opening failed. If it comes back full of plausible clinical hypotheses, the opening did its job. The same colleague can mark each adjective with C (clinical), R (deliberate representation) or J (judgment). Everything that falls under J is out in the next version.

Those who review large banks (league, course, platform) gain by keeping a short team glossary: how to name diabetes, how to name wheelchair use, how to cite an interpreter, how to avoid "non-adherent" without erasing the access barrier. A glossary does not make style rigid; it reduces ableist surprise from one case to the next. Combine this with the habit of writing the goal in one line at the top of the file ("train urinary-source sepsis in an adult", "train blood gas interpretation"), so that demographics do not steal the spotlight from the goal.

Disability language: person and identity

There are two broad families of language. Person-first language ("person with a disability", "person with diabetes") places the condition after the person. Identity-first language ("deaf person", "autistic person", in many communities "Deaf" with a capital D in English spelling) treats the characteristic as part of collective identity. Institutional guides (for example the NIH style) generally ask you to start with the person and respect the preference when the community declares it.

The preference is not uniform. In a sample of autism stakeholders in the United States, autistic adults predominantly preferred identity-first terms; professionals leaned more toward person-first language (Taboas et al., 2023). In research and practice, guidance texts ask you to avoid ableist language (for example "suffers from", "confined to", "special" as an empty euphemism) and to align vocabulary with what the community endorses (Bottema-Beutel et al., 2021). Disability-conscious medical education distinguishes teaching "about" disability as a list of diseases from teaching with rights and justice frameworks (Bowen et al., 2026).

For those writing the case, practical rules:

• Default: "person with [condition]" when you do not know the preference of the community cited.
• Deafness and autism: consider the community's identity preference; if the case is generic, a pedagogical footnote can explain the choice.
• Avoid: "the diabetic", "the epileptic", "wheelchair user" as the only label, "invalid", "suffers from".
• Describe the functional finding when it is clinical ("uses a wheelchair", "communicates in sign language", "reads with a screen reader"), not the moral value.
• Disability in the case is not an overcoming plot or a melodramatic obstacle; it is context, like hypertension.

Physicians, in national research, reported discomfort and biased perceptions in the care of people with disabilities (Iezzoni et al., 2021). The vignette you write trains the newer generation. If the only patient with a disability in the bank appears as "difficult" or "uncooperative", the hidden curriculum speaks louder than the ethics class.

Relate this to the diagnostic error axis: the Improving Diagnosis report reminds us that communication and representation matter in the diagnostic process (diagnostic error and the Improving Diagnosis report). Stigmatizing language in the case text is the first link in a chain that later shows up in the medical record.

In writing practice, separate three mental fields before typing the disability into the case. Field 1: the condition or mode of functioning relevant to the differential or the exam. Field 2: the care or environment adjustment (interpreter, extra time, adjustable table, written communication). Field 3: the life story that does not fit the teaching objective. Fields 1 and 2 can go in. Field 3 can almost always wait for another format (interview, workshop, long narrative), so the short vignette does not become an incomplete and biased biography.

Also avoid the opposite misstep: erasing every mention of disability when it changes the physical exam or communication. Erasing is not neutral; it is pretending the clinical world is made only of "standard" bodies and senses. Inclusion describes what matters for care and for reasoning, with respect and precision.

Stereotypes the vignette teaches without meaning to

Undergraduate vignettes can carry a hidden curriculum of gender: professions and care roles distributed in a stereotyped way, male doctors and female nurses as the norm, traditional families as the only model (Arsever et al., 2023). The student learns the diagnosis and, along the way, who "looks like" a doctor and who "looks like" a caregiver.

Stigmatizing language also travels from the chart to judgment. In a classic study, notes with stigmatizing language passed biases on to readers and affected attitudes (Goddu / Beach et al., 2018). In an analysis of records, physicians used stigmatizing language with patterns associated with already marginalized groups (Park et al., 2021). The teaching case is not a chart, but it trains the reflex of writing. If the vignette says "non-adherent and aggressive patient", the student practices the shortcut before practicing the hypothesis.

Signs of stereotype on first reading:

• Humor or morality in place of the finding ("irritable", "dramatic", "manipulative").
• Disease as the person's noun ("the 50-year-old hypertensive").
• Occupation and family care always in the same gender mold.
• The only mention of disability tied to exam "difficulty", with no clinical data.
• Race cited only in diseases already saturated with stereotype, with no mechanism.

Fixes: replace judgment with observable behavior ("raises his voice when the pain worsens"); replace the label with the condition ("person with hypertension"); vary care roles and professions; if the disability matters for the exam, state the adjustment ("needs an adjustable table", "prefers an interpreter"), not the adjective. The post on affective bias helps show how emotion and label distort the differential; the vignette does not need to hand over the bias ready-made.

There are "positive" stereotypes that get in the way too. The patient "always smiling despite everything", the "warrior" mother, the "inspiring" young person. Forced praise still reduces the person to a role. Prefer clinical facts and, if you want to train communication, train listening and open questions in another exercise, not in the adjective of the opening. Narrative medicine teaches how to hear a story with density; the short vignette teaches how not to invent character.

When the objective is precisely to discuss bias, state that in the preceptor's material (not necessarily in the text the student sees on first reading). For example: after the closing, the explanation can ask whether the race or mood cited in the opening changed the differential without evidence. That way the stereotype becomes an object of analysis, not silent fuel. Without that explicit goal, the descriptor is left over and teaches the wrong thing.

Age, gender and race with teaching care

In vignettes of preclinical multiple-choice questions, age and sex/gender appear in almost all stems, while race/ethnicity is much rarer; when race does appear, it often shows up as a genetic proxy, not as social context (Carey-Ewend et al., 2023). Sex/gender, despite being ubiquitous, was relevant to the correct answer in a much smaller fraction of items. In other words: the demographics are there, but they do not always do any work.

Medical schools have also been criticized for reproducing race as simplified biology in the curriculum, with the risk of spreading bias (Amutah et al., 2021). In first-year students, implicit race and class biases were frequent in association tests; in the vignettes of that study, clinical assessments did not always follow the implicit bias in a statistically linked way (Haider et al., 2011). The teaching point is not "never mention race". It is: mention it with a mechanism, honest local prevalence or an access barrier, never as the single shortcut to the diagnosis.

Quick guide:

• Age: useful when it changes probability (for example the pediatric range versus the elderly range). Avoid "elderly" as a synonym for frail without data.
• Gender and sex: separate what is biological sex relevant to the mechanism from what is a social role. Vary who provides care and who works.
• Race and ethnicity: only with a purpose (contextualized epidemiology, inequity, a well-founded specific risk). Avoid the racial list as an exam clue.
• Class and occupation: useful for exposure and access; dangerous as moral judgment ("uneducated").
• Sexual orientation and gender identity: included when care or the differential requires them; never as a joke or a final surprise.

Deliberate diversity in creating vignettes is an emerging topic, including with the support of assisted writing tools, as long as there is human review of stereotype and clinical accuracy (Bakkum et al., 2024). No tool replaces the question: does this descriptor teach Medicine or teach prejudice?

For the student who plays or reads the case, demographics should illuminate the differential, as in differential diagnosis with clues, not steal the scene. Up to six clues are enough when each one carries information; empty descriptors only take up working memory.

If the bank is international, avoid anchoring the case in exams, laws or jargon from a single country as if they were universal. Age and clinical findings travel better than local acronyms. When a health system context is needed, explain in one sentence what the student needs to know for the reasoning ("no regular access to insulin this month"), instead of assuming the whole class knows the same institutional arrangement.

Representation matters across the whole bank, not only in each isolated vignette. A single "diverse" case among ninety homogeneous openings does not correct the hidden curriculum. Plan the portfolio: who appears, who decides, who provides care, who is the specialist. This is curation work as much as writing a case.

Clarity and structure: text accessibility

Inclusion is also legibility. Principles of learning science applied to medical education call for reducing extraneous load: what does not teach gets in the way (Mayer, 2010). In the written case, extraneous load is an 80-word sentence, unnecessary jargon, a confusing time order and critical information only in an image without alternative text.

Good practices for accessible text in the case:

• Short sentences; one idea per sentence when the finding is decisive.
• Explicit time order ("3 days ago", "worse at night"), not a flashback without a marker.
• Findings in real text, not only in a figure; if there is an image, describe the essentials in the body.
• Avoid relying only on typographic color (red = urgent) without a word.
• Stable names and terms: if you called it "burning pain", do not switch to "smarting" without reason.
• Lists or blocks for vital signs and exams, instead of a single dense paragraph.

Someone using a screen reader hears the case in DOM order. Headings, lists and alternative text matter as much as the pathophysiology. Someone who enlarges the screen needs contrast and words, not a colored badge. These precautions align with the spirit of digital accessibility in medical software and of visual assistive technology, applied to the "vignette" artifact. The full flow of a playable case with a screen reader goes clue by clue; cognitive accessibility covers the load and language of the vignette. Here the minimum is: the case must exist as structured text.

Feedback after the case also inherits the language. If the explanation repeats the stereotype of the opening, the harm is confirmed. Prefer the pattern of useful feedback in clinical cases: criterion, evidence, next step. No character adjectives.

Quick textual accessibility checklist before publishing the case: (1) the HTML or markdown has real headings, not just bold text; (2) lists of vitals and exams are in list form, not a wall of text; (3) every image has useful alt text; (4) no meaning depends on color alone; (5) uncommon abbreviations are spelled out the first time; (6) the language of the case is the language of the student's interface. If the case is translated, review the demographics and adjectives in the target language: stereotypes get translated too.

Clarity is not oversimplification. You can keep precise technical terms ("hyperkalemia", "anion gap") and still write short sentences. What gets cut is ornament and judgment. The medical student needs the clinical lexicon; they do not need the moral vernacular of the shift.

Table: before and after

Use the table when reviewing a case of your own, from a student league or from a bank. The examples are fictional and serve only as a linguistic contrast.

Problem	Before (avoid)	After (prefer)	Why
Illness as identity	"54-year-old diabetic, non-adherent"	"54-year-old person with type 2 diabetes; reports difficulty getting insulin this month"	Condition + barrier, without moral judgment
Judgment of mood	"Hysterical patient in the emergency department"	"28-year-old person, agitated, rapid speech, reports fear of dying"	Observable finding
Disability as an obstacle	"Wheelchair user, hard to examine"	"Person who uses a wheelchair; needs an adjustable exam table"	Adjustment, not adjective
Race as an empty hint	"Black man with chest pain" (and nothing more)	Include race only if the goal is inequity, contextualized prevalence or a bias to discuss; otherwise, describe the pain and the clinical risk factors	Explicit purpose
Stereotyped gender	"Worried mother brings her son; father absent"	Vary who accompanies the patient; if the caregiver matters, state the role without moral judgment	Hidden curriculum
Inaccessible text	Critical finding only in a red image	Finding in text + image with description	Screen reader and CVD

After editing, read the case aloud or with a screen reader. If you hesitate at an adjective, it is probably judgment. If the demographics disappear and the differential stays identical, the demographics were noise.

How NEXO fits in

NEXO is the platform where you play the case of the day and where creators publish fictional cases with a real clinical presentation. Curator supports the curation and organization of teaching clinical material. The Creators flow is the path for those who write and submit cases to the community. Tools such as Closer and Researcher are part of the ecosystem for study and evidence lookup as the product describes; this post makes no claim about learning outcomes.

For those who write in the NEXO ecosystem, the practical rule mirrors this guide: up to six clues; language without judgment; demographics with purpose; structured text. The /accessibility page states the interface's accessibility features (including independence from color in proximity signals, among others). Accessibility of the case text and accessibility of the interface add up: a stigmatizing vignette in an accessible app still teaches badly.

If you are getting to know the platform, the post what NEXO is places the daily diagnosis game. Usage limits, when relevant, are at /economy. Nothing here is a claim of effectiveness.

For the creator, the suggested cycle is: clinical draft, a pass through this post's checklist, review by a second reader, adjustment of clues (up to six), publication in the Creators flow, and attention to community feedback without turning a hostile comment into "proof" that inclusive language gets in the way. Inclusive language only gets in the way of the shortcut. Clinical reasoning stays at the center.

Frequently asked questions

What is the difference between this guide and "how to write a clinical case for teaching"?

That post covers the general method: pedagogical goal, source material, dosing of clues, format and review. This one covers inclusion and accessibility in writing: disability language, stereotypes, purposeful demographics and readable text. The two complement each other; start with the method if you do not yet have the skeleton of the case.

Should I always use person-first language?

As a starting point, yes, when you do not know the preference. In deafness and autism, many people and communities prefer identity-first language. In generic material, state the choice or alternate with a pedagogical note. The criterion is respect for the community cited, not fashion.

Can I mention race in the case?

Yes, when there is a clinical, contextualized epidemiological or explicit pedagogical purpose (for example discussing inequity or bias). Do not use race as the only clue disguised as a differential. Review whether the sentence would survive without the racial descriptor.

How do I avoid "difficult patient" without losing realism?

Describe behavior and context ("refuses the exam until the pain is treated", "needs an interpreter"). Clinical realism does not require a moral adjective. Fictional cases with a real clinical presentation gain strength with precision, not with caricature.

What changes for those who use a screen reader?

Text order, headings, lists and text alternatives to images. If the critical finding is only in a colored figure, part of the class does not receive the clue. Structure the case as a document, not as a decorative slide.

Do Creators on NEXO need to follow this checklist?

It is the recommended editorial reference for those who write fictional cases with a real clinical presentation in the Creators flow, aligned with the Curator curation. The platform declares resources at /accessibility. This text does not evaluate the performance of published cases.

Final checklist

• Clear pedagogical goal (what the student should practice).
• No adjective of morality, mood or "difficulty" in place of a finding.
• Condition in person-first language, unless the community prefers identity-first language.
• Age, gender, race, class and disability only with purpose; otherwise, remove them.
• Varied care roles and professions, without a single gender template.
• Short sentences, explicit temporal order, findings in text.
• Images with a text description of the essential.
• Up to six clues; each one informs the differential.
• Feedback without repeating the stereotype from the opening.
• Review by a second person (preferably someone from the community represented, when the case centers on it).

Play the case of the day on NEXO. If you write cases, use this checklist before submitting through the Creators flow.

References

1. Carey-Ewend K, et al. Use of Sociodemographic Information in Clinical Vignettes of Multiple-Choice Questions for Preclinical Medical Students. Med Sci Educ. 2023. PMID 37501800.
2. Arsever S, et al. A gender biased hidden curriculum of clinical vignettes in undergraduate medical training. Patient Educ Couns. 2023. PMID 37595505.
3. Taboas A, Doepke K, Zimmerman C. Preferences for identity-first versus person-first language in a US sample of autism stakeholders. Autism. 2023. PMID 36237135.
4. Bottema-Beutel K, et al. Avoiding Ableist Language: Suggestions for Autism Researchers. Autism Adulthood. 2021. PMID 36601265.
5. Bowen L, et al. Dismantling Ableism in Undergraduate Medical Education: Promising Practices in Disability-Conscious Training. Teach Learn Med. 2026. PMID 39964131.
6. Bakkum MJ, et al. Using artificial intelligence to create diverse and inclusive medical case vignettes for education. Br J Clin Pharmacol. 2024. PMID 38016816.
7. Haider AH, et al. Association of unconscious race and social class bias with vignette-based clinical assessments by medical students. JAMA. 2011. PMID 21900134.
8. Iezzoni LI, et al. Physicians' Perceptions Of People With Disability And Their Health Care. Health Aff. 2021. PMID 33523739.
9. Amutah C, et al. Misrepresenting Race - The Role of Medical Schools in Propagating Physician Bias. N Engl J Med. 2021. PMID 33406326.
10. Goddu AP, et al. Do Words Matter? Stigmatizing Language and the Transmission of Bias in the Medical Record. J Gen Intern Med. 2018. PMID 29374357.
11. Park J, et al. Physician Use of Stigmatizing Language in Patient Medical Records. JAMA Netw Open. 2021. PMID 34259849.
12. George S, et al. Examining Inclusive Language in Clinical Narratives in Medical Biochemistry Textbooks to Model Equitable Patient-Centered Care in Preclinical Undergraduate Medical Education. Med Sci Educ. 2024. PMID 38887417.
13. Mayer RE. Applying the science of learning to medical education. Med Educ. 2010. PMID 20604850.
14. NIH Style Guide. Person-first and Destigmatizing Language. nih.gov.
15. National Academies of Sciences, Engineering, and Medicine. Improving Diagnosis in Health Care. Washington, DC: National Academies Press; 2015. nap.nationalacademies.org.