Clinical reasoning
Building a differential diagnosis
A differential diagnosis is a ranked list of the conditions that could explain the patient's problem. You build it from the presenting complaint, adjust it with the history, examination and results, and keep the dangerous causes in view until you have excluded them. Writing down why each item ranks where it does is what makes the list useful to the next doctor.
The Workup Desk team. Last reviewed November 3, 2026. General guidance on clinical documentation, not medical advice about any individual patient.
Start from the presenting problem
Define the problem as precisely as the history allows. Exertional central chest pain for six weeks gives you a shorter and better list than chest pain.
Use a systematic sweep to avoid anchoring on the first idea. Many doctors work through organ systems, or through categories such as vascular, infective, inflammatory, neoplastic, metabolic and drug related.
Weigh how likely each cause is before the tests
Pre-test probability is how likely a condition is before a given test, based on the patient's age, sex, risk factors and symptoms. It decides how much a test result should move you.
A normal result in a patient with a high pre-test probability may not exclude the condition. An abnormal result in a patient with a very low pre-test probability may be a false positive. Write down the probability you started from so the reader can follow your reasoning.
Keep the dangerous causes in view
Rank by likelihood, then check the list for conditions that would be serious to miss. Keep them on the list until you have a reason to remove them, and write down that reason.
Red flags are the history and examination features that raise the probability of a serious cause. Record the ones you asked about and the answers, including the negatives.
Use the findings to move items up and down
Each finding should change the list. Note which items a finding supports and which it makes less likely, rather than listing findings and conditions separately.
Be specific about what a test can and cannot tell you. A single normal troponin in a patient with weeks of exertional symptoms answers a different question from a troponin taken during an episode of pain at rest.
| Part | What to write | Example |
|---|---|---|
| Condition | The diagnosis, as specific as you can make it | Chronic coronary syndrome |
| For | The history and findings that support it | Typical exertional pain, risk factors |
| Against | The findings that make it less likely | None so far |
| Next step | What would confirm or exclude it | Specialist assessment and further testing |
Name a working diagnosis
The working diagnosis is the explanation you act on until new information says otherwise. Name it, say how confident you are, and say what would make you change it.
When you cannot name one, say so and say what you need. An honest undifferentiated presentation with a clear plan is more useful than a confident label.
Watch for the common reasoning traps
Most diagnostic errors come from a small set of patterns. Naming them in your own practice is the first defense.
- Anchoring on the first diagnosis and reading later findings to fit it.
- Premature closure, where you stop looking once one explanation fits.
- Accepting a diagnosis from a referral or an earlier letter without checking the evidence for it.
- Ignoring a result that does not fit, instead of explaining it.
- Attributing new symptoms to a known condition without considering a second cause.
Ask an AI tool for alternatives instead of answers
An AI tool can help you check a differential for gaps. Give it the sources and your list, and ask which conditions you have not considered and which findings in the record bear on each.
Treat the reply as a prompt for your own reasoning. Check every finding it cites against the source, and make the ranking and the decision yourself.
Questions to check in your own work
- 1.Have I defined the presenting problem as precisely as the history allows?
- 2.Did I write down the pre-test probability I started from?
- 3.Which serious causes are still on the list, and what would exclude them?
- 4.Does every item say what supports it and what argues against it?
- 5.Have I checked the diagnoses I inherited from earlier letters?
- 6.Is there a finding I have not explained?
Common questions
How many items should a differential have?
As many as are plausible and worth acting on. Three to five ranked items with reasoning are usually more useful than a long unranked list.
What is the difference between a differential and a working diagnosis?
The differential is the ranked list of possibilities. The working diagnosis is the one you act on now, usually the top of the list, until new information changes it.
Should the differential go in a referral letter?
Yes, briefly. The receiving doctor uses it to triage and to focus the first appointment.
Try it on one patient's documents
Keep each patient's sources in one workspace, ask questions of them, and draft documents you edit and export. Early access is by invitation.