How to write a teaching clinical case: a practical guide

The professor has ten minutes before class to write the case that will open the discussion about chest pain. He remembers a patient from his shift, writes a paragraph with age, complaint, physical exam, electrocardiogram and troponin, and puts the question at the end: what is the diagnosis? In class, the first student answers in five seconds. The discussion dies there. The case had all the information, but no decision to make.

Writing a good clinical case for teaching is a skill of its own, different from reporting a case for publication and different from seeing the patient. A teaching case needs a clear objective, must release information in the right order, require decisions and leave room for error. This guide shows how to start from the learning objective, choose the raw material of the case, dose the clues, adjust difficulty and format, avoid biases that go unnoticed, use artificial intelligence with care and review the case before taking it to the classroom.

Key points

• Every teaching case starts with a learning objective, not with the diagnosis.
• Well-built cases measure and train reasoning in a way close to real practice.
• The order in which the clues appear decides whether the case requires reasoning or just recognition.
• Formats such as key features and script concordance help focus on the decisions that matter.
• Teaching cases can convey stereotypes without the author noticing.
• Artificial intelligence can speed up writing, but the case still needs review by someone who knows the clinical side.

Summary

• What a clinical case for teaching is
• Start with the objective
• The raw material of the case
• How to dose the clues
• Format and difficulty
• Biases the case can teach
• Artificial intelligence and review
• How NEXO builds its cases
• Frequently asked questions
• Checklist for writing a clinical case

What a clinical case for teaching is

A clinical case for teaching is the narrative of a patient built to provoke learning. It can be short, like a vignette of a few lines in an exam, or long, like a problem-based learning case that unfolds over several sessions. What defines it is not its size but its intention: every piece of information is there to lead the student to think, decide and, preferably, make mistakes in a safe environment.

Why cases work

Case-based teaching is old and widely used in the health professions. A systematic review by the BEME group analyzed more than a hundred studies on case-based learning and found that students and teachers like the method, although the evidence on its effect on performance is less solid than many assume. The text on case-based learning in clinical practice explores this discussion further.

There is also a cognitive reason. The way experienced doctors organize clinical knowledge into disease scripts, the so-called illness scripts, is formed from repeated contact with patients and cases. Each well-written case is an opportunity to build or refine one of these scripts.

Cases measure practice

Clinical vignettes are not only for teaching. In a prospective study in outpatient clinics, doctors saw simulated patients without warning and also answered vignettes with exactly the same picture. The quality measured by the simulated patients was 73%, and by the vignettes 68%, a close result, better than the one obtained by chart review. An earlier study by the same group had already compared vignettes, simulated patients and chart review as methods to measure quality. This shows that a well-built case comes close to what the doctor does in front of the patient.

Start with the objective

The most common mistake is to start with the diagnosis: "I'm going to write a case of pneumonia". The diagnosis is the destination, but the teaching case is about the path. Before writing, define what the student should be able to do at the end. Recognize an atypical presentation? Tell two similar hypotheses apart? Decide when to order a test? Interpret a finding that is usually ignored?

Objectives that call for a decision

Good objectives use action verbs tied to clinical decisions: prioritize, differentiate, justify, interpret, choose. Vague objectives, such as "know the disease", produce cases that become a lecture in disguise. A commentary on strategies to promote diagnostic reasoning suggests that teaching should let the student represent the problem, compare hypotheses and make their own reasoning explicit, and the case must be written to allow that.

An example of a well-written objective

Compare two objectives for the same topic. The first: "the student should know pulmonary thromboembolism". The second: "given sudden dyspnea in the postoperative period, the student should estimate the clinical probability before ordering tests and justify the choice of initial test". The second already suggests the patient, the clues and the questions of the case. It also shows the teacher what to watch for in the discussion: if the student skips the probability estimate and goes straight to the test, that is the point to work on. The more concrete the objective, the easier it is to write the rest.

One case, few objectives

A case that tries to teach everything teaches nothing. Choose one or two main objectives. If the chest pain case is about differentiating coronary syndrome from aortic dissection, the information about treatment can wait for another case. This discipline makes the case shorter and the discussion more focused.

• Write the objective in one sentence before anything else.
• List the decisions the student will need to make to reach it.
• Identify the most common mistake you want them to learn to avoid.
• Define the audience level: basic cycle, internship or residency.
• Only then choose the patient and the diagnosis.

The raw material of the case

Teaching cases can come from the service's appointments, from the literature or from the author's imagination supported by clinical knowledge. Each source has advantages and cautions.

Cases from practice

Cases from practice bring the richness of practice: false clues, findings that do not fit, misleading tests. An essay on teaching clinical reasoning argues that true cases are preferable to cases built from memory, precisely because they reflect these deviations, and that the presence of a tutor who points out errors on the spot increases the value of the experience. The essential caution is to protect the patient's identity: remove names, dates, places and any detail that could make them recognizable.

Fictional cases with a real basis

Often the best path is a fictional case with a real clinical presentation: the combination of data is invented, but each finding is plausible and consistent with the disease. This protects privacy and allows the case to be adjusted to the objective, adding or removing information. The risk is the case becoming too clean, with the textbook presentation that rarely appears on shift.

The patient as a person

A guide on writing cases for problem-based learning recommends creating immediacy and a sense of the patient as a human being, because this awakens the curiosity that drives learning. Profession, routine, fears and family context are not decoration: they make the case memorable and often carry diagnostic clues. The text on narrative medicine shows how the patient's story changes the reasoning.

A recent guide on case studies in research and teaching recalls that a good case explores a phenomenon in its real context, not isolated from it. For the teaching case, this means the patient's context is part of the problem, not the background.

How to dose the clues

The order in which information appears is what makes a text an exercise in reasoning. If the diagnosis is obvious in the first line, the student recognizes instead of reasoning. If the decisive information only appears at the end, they are forced to form hypotheses, sustain them and revise them along the way. That is what happens on shift, when data arrives little by little.

From the general to the specific

A useful structure is to start with the complaint and the context, continue with the history and background, then physical examination, initial tests and, finally, the most specific finding. Each step should allow plausible hypotheses and leave at least one alternative alive. The table below shows how each type of clue usually works.

Type of clue	Role in the case	Careful writing
Complaint and context	Open the differential diagnosis	Avoid terms that already give away the disease
History and background	Shift probabilities between hypotheses	Include relevant data and some neutral data, as in real life
Physical examination	Confirm or weaken hypotheses	Describe the finding, not the name of the syndrome
Initial tests	Narrow the differential	Give values, not ready-made interpretations
Distractor clue	Test whether the student revises hypotheses	Be plausible and not deliberately misleading
Specific finding	Close the case	Leave it for the end, when the decision has already been trained

An example in stages

Imagine a case whose goal is to differentiate causes of abdominal pain in a young woman. The first clue brings the complaint: lower abdominal pain for one day, with progressive worsening. Many hypotheses fit here. The second brings the context: menstrual delay of a few weeks and irregular use of contraceptives. The probability changes, but other causes remain possible. The third brings the examination: pain on palpation in the iliac fossa, without signs of diffuse peritoneal irritation. The fourth, an initial test with a value described without interpretation. At each stage the student records the main hypothesis and the most dangerous alternative. At the end, the discussion returns to the stages and asks at what point each hypothesis should have risen or fallen. It is this movement, and not the final diagnosis, that the case teaches.

Describe, do not interpret

Writing "aortic stenosis murmur" already gives away the interpretation. Writing "systolic murmur at the aortic focus, harsh, radiating to the carotids" forces the student to interpret. The same applies to tests: give the value and leave the reading to whoever solves the case. This simple care greatly increases the reasoning required.

Distractor clues with honesty

A finding that points elsewhere teaches you to revise hypotheses, but it must be plausible. Invented clues meant to mislead, which would never appear in an actual patient, teach distrust of the case, not reasoning. The guide on differential diagnosis with clues shows how to train this revision.

Format and difficulty

The format of the case must serve the goal. A case for group discussion can be open and long. A case for assessment needs precise questions and defensible answers. Two formats studied for decades help to think through this choice.

Key features

The key features approach starts from the idea that each clinical problem has a few critical steps, the ones where errors happen most. The case is written to test exactly those steps. A review of the validity evidence for this approach showed that it validly assesses clinical decisions and is used in licensing exams. For the case author, the lesson is to focus on the difficult decisions, not the obvious ones.

Script concordance

Script concordance tests present a situation, a hypothesis and a new piece of information, and ask whether the information makes the hypothesis more or less likely. They measure reasoning in situations of uncertainty, comparing the student's answer with that of a panel of experts. There are published guidelines for building these tests, which include choosing truly ambiguous situations and assembling a reference panel.

Adjusting the difficulty

For students in the basic cycle, the ideal is shorter cases, typical presentations and questions that guide the step by step. In the internship and residency, the case can have more noise, atypical presentations and more than one disease at the same time.

1. Define the audience level and the time available.
2. Choose between open discussion, key features or script concordance.
3. Write the clues in order from general to specific.
4. Formulate the questions at each decision point.
5. Write the expected answer and the acceptable alternatives.
6. Prepare the final explanation, with the step-by-step reasoning.

Biases the case can teach

Teaching cases carry messages that go beyond clinical content. Who the patient is, who the doctor is, what profession the patient has, who cares for them at home: all of this teaches, even without intention. It is a form of hidden curriculum, the topic of the text on hidden curriculum in Medicine.

What a study found

An analysis of the clinical vignettes used in teaching and assessment in the undergraduate Medicine program in Geneva identified more than two thousand vignettes, of which nearly a thousand entered the analysis. Patients' professions and family caregiver roles had a distribution strongly marked by gender, with male doctors and female nurses as the norm. The authors concluded that these vignettes conveyed a hidden curriculum with gender bias, which can affect patient care.

How to avoid it

• Vary patients' gender, age, skin color, profession and region across the set of cases.
• Avoid associating diseases with groups in a stereotyped way when the epidemiology does not justify it.
• Describe doctors, nurses and other professionals without fixed roles by gender.
• Ask a colleague to read the set of cases looking for patterns.
• Review old cases with the same eye.

Attention to representativeness also trains reasoning. Diseases that present differently in women, older adults or people with dark skin are precisely the ones that most often lead to error when the student has only seen the classic presentation. The text on implicit bias in diagnosis explores the topic in depth.

Artificial intelligence and review

Language models can already generate clinical vignettes in seconds, and many teachers use these tools to save time. The studies are recent and show both the potential and the limits.

What the studies show

In a mixed-methods study, general practitioners with experience in education evaluated clinical vignettes generated in Japanese by a language model for quality and accuracy of information, educational usefulness and diagnostic difficulty. In gynecology and obstetrics, a report described the use of structured prompting techniques to generate vignettes and questions, and tested whether residents and preceptors could tell questions written by people from those generated by artificial intelligence. Both papers point to usefulness, but also to the need for review by specialists.

Precautions when using AI

AI-generated cases tend to be correct on the surface and fragile in the details: doses, test values, temporal sequence and local epidemiology. They can also repeat the stereotypes present in the training data. Use the tool as a draft, never as the final version, and check every clinical datum. The text on artificial intelligence in clinical reasoning discusses these limits.

Review and test before class

Every case, written by a person or by a machine, improves when it is tested. Ask a colleague to work through the case without knowing the diagnosis and observe where they get stuck, where they get it right too early and where the information is ambiguous. Then apply the case to a small group of students and adjust. A case tested two or three times is usually much better than the first version. And prepare the final explanation carefully: the text on feedback in clinical cases shows what makes an explanation useful.

The final explanation

The explanation that closes the case deserves as much care as the clues. It should revisit each decision point, show which datum changed the probability of each hypothesis and explain why the alternatives were ruled out. It is also worth naming the most common error and the reason it happens. An explanation that only repeats the diagnosis and the treatment wastes the most valuable part of the case, which is the path to the answer. When possible, include a short reference for those who want to go deeper.

After class

Keep notes on how the case worked: which clue the class got stuck on, which hypotheses came up, which questions went unanswered. The next time you use the case, these notes indicate what to adjust. Over time, the teacher builds a bank of tested cases, each with its usage history, which can be shared with colleagues in the course. Reviewing this bank once a year, with attention to guideline updates and to the diversity of patients, keeps the cases correct and representative.

How NEXO builds its cases

In NEXO, the case of the day is presented in up to six clues, released one at a time, from the general to the specific. With each guess, the proximity in ICD-10 across five levels shows how far your hypothesis was from the diagnosis, which makes every miss a piece of information about your reasoning. The cases are fictional with a realistic clinical presentation, written with the same principles as this guide: a clear objective, clues that describe without interpreting, and an explanation at the end.

For those who write cases, playing the case of the day is also an exercise for the author: notice at which clue you formed the right hypothesis and consider whether the order could be different. The guide what NEXO is explains the basics, and the limits of use are on the NEXO economy page.

For those who coordinate internships, student leagues or residency programs and want to use cases in groups, there is NEXO Closer. Researchers in medical education can propose studies through NEXO Researcher, physicians who review cases take part through NEXO Curator, and those who create health content on social media can take part through NEXO Creators.

Frequently asked questions

What is the ideal length of a clinical case for teaching?

It depends on the objective and the format. Exam vignettes usually run a few lines; cases for group discussion can fill one or two pages, released in stages. The criterion is to include only what serves the objective.

Can I use an actual case from my service?

You can, as long as the patient's identity is protected. Remove names, dates, places and details that could make them recognizable, and follow your institution's rules on the use of patient data.

Is a teaching case the same as a case report?

No. A case report is written for publication and describes something rare or instructive. A teaching case is built to prompt reasoning, with information released gradually and questions at the decision points.

How do I know if the case is too easy?

If most students get the diagnosis right on the first piece of information, the case is giving it away too early. Move the most specific clue to the end and describe findings instead of naming them.

Can I write cases with artificial intelligence?

You can use it as a draft, but review every clinical detail, check the timeline for consistency and look for stereotypes. The final version needs to go through someone with clinical experience.

How many objectives should a case have?

One or two main objectives. Cases that try to teach everything become long and the discussion loses focus.

Checklist for writing a clinical case

• Write the learning objective in one sentence.
• List the decisions the student must make.
• Choose the patient and protect their identity, if the case is real.
• Order the clues from the general to the specific.
• Describe findings without naming syndromes.
• Include a plausible distractor clue, if it makes sense.
• Review gender, age and roles to avoid stereotypes.
• Test the case with a colleague and adjust it.
• Prepare the final explanation with the step-by-step reasoning.

Want to see in practice how clues build a diagnosis? Play the case of the day on NEXO.

References

1. Thistlethwaite JE, Davies D, Ekeocha S, Kidd JM, MacDougall C, Matthews P, et al. The effectiveness of case-based learning in health professional education. A BEME systematic review: BEME Guide No. 23. Med Teach. 2012. PubMed 22578051
2. Schmidt HG, Rikers RM. How expertise develops in medicine: knowledge encapsulation and illness script formation. Med Educ. 2007. PubMed 18004989
3. Peabody JW, Luck J, Glassman P, Jain S, Hansen J, Spell M, et al. Measuring the quality of physician practice by using clinical vignettes: a prospective validation study. Ann Intern Med. 2004. PubMed 15545677
4. Peabody JW, Luck J, Glassman P, Dresselhaus TR, Lee M. Comparison of vignettes, standardized patients, and chart abstraction: a prospective validation study of 3 methods for measuring quality. JAMA. 2000. PubMed 10755498
5. Bowen JL. Educational strategies to promote clinical diagnostic reasoning. N Engl J Med. 2006. PubMed 17124019
6. Kassirer JP. Teaching clinical reasoning: case-based and coached. Acad Med. 2010. PubMed 20603909
7. Bergemann AD, McMillin M, Gillispie-Taylor MC, Gingell G, Surles RT, Wallace R. Words of advice: how to write the story of a patient for a problem-based learning session in a healthcare education program. FEBS J. 2025. PubMed 39673066
8. Greenhalgh T. Case studies: a guide for researchers, educators, and implementers. BMJ Med. 2025. PubMed 41000468
9. Bordage G, Page G. The key-features approach to assess clinical decisions: validity evidence to date. Adv Health Sci Educ Theory Pract. 2018. PubMed 29777464
10. Fournier JP, Demeester A, Charlin B. Script concordance tests: guidelines for construction. BMC Med Inform Decis Mak. 2008. PubMed 18460199
11. Arsever S, Broers B, Cerutti B, Wiesner J, Dao MD. A gender biased hidden curriculum of clinical vignettes in undergraduate medical training. Patient Educ Couns. 2023. PubMed 37595505
12. Takahashi H, Shikino K, Kondo T, Komori A, Yamada Y, Saita M, et al. Educational Utility of Clinical Vignettes Generated in Japanese by ChatGPT-4: Mixed Methods Study. JMIR Med Educ. 2024. PubMed 39137031
13. Jackson FI, Keller NA, Kouba I, Kouba W, Bracero LA, Blitz MJ. Large Language Model Clinical Vignettes and Multiple-Choice Questions for Postgraduate Medical Education. Acad Med. 2025. PubMed 40550116