Workflow: Drafting a Post-Myocardial Infarction Discharge Letter

This workflow takes you from a set of clinical bullet points to a finished post-myocardial infarction (MI) discharge letter. The steps are the same whether you use Claude or ChatGPT. The whole process takes around ten minutes the first time.

1. Write your clinical bullet points before opening the tool

   Spend two minutes writing the key clinical facts on a blank document: the admission presentation, the diagnosis, investigations, intervention performed (angiography, percutaneous coronary intervention), medications started, medications changed, discharge plan, and follow-up. Having these ready before you open the tool means the session is faster and more focused.

2. Open the tool and set the task

   Open claude.ai or chat.openai.com and start a fresh conversation. Tell the tool what you need before you give it the clinical content: "You are helping a cardiologist draft a post-MI discharge letter. I will provide anonymised clinical bullet points. Turn them into a structured discharge summary with professional medical language. Do not include specific drug doses. Use these headings: Admission Presentation, Investigations, Management, Discharge Medications, Discharge Plan, Follow-Up."

3. Provide your clinical bullet points

   Paste or type your anonymised clinical notes. The tool will produce a structured draft in under thirty seconds.

4. Read the draft for clinical accuracy first

   Read the draft as a clinician, not as a reader. Ask yourself: Is every clinical fact accurate? Is the timeline of events correct? Is anything that the tool has inferred from your bullet points actually wrong? Look specifically for: the description of the angiographic findings (the tool may have generalised or slightly misrepresented these), the medication section (check no drug names have been introduced that you did not include), and the follow-up plan (check it matches your actual plan).

5. Add drug doses and specific values yourself

   Copy the draft into your word processor or clinical system. Now add: specific drug names and doses from the discharge medication reconciliation, specific echocardiogram values if relevant, specific angiographic findings from the catheter lab report, and any other quantitative clinical data. These must come from the source documents, not from the AI tool.

6. Review the follow-up and safety net section

   The follow-up section is where patients and GPs most need clarity. Check that it includes: the cardiac rehabilitation referral status, the outpatient cardiology follow-up date and purpose, what the GP is being asked to do and when (renal function check, medication review), and what symptoms should trigger early contact with the cardiology team. Adjust anything that does not reflect your actual plan.

7. Check the tone for a patient-facing version if needed

   If this letter is going directly to the patient as well as the GP, read it again from the patient's perspective. Medical language that is appropriate for a GP letter may be confusing for a patient who has just had a cardiac event. Either ask the tool to produce a separate plain-language version, or adjust the language in the sections most likely to be read by the patient.

8. Add identifying details and finalise

   Back in your clinical system, add the patient's name, date of birth, hospital number, the date, and the names of the receiving clinicians. Review the complete letter one final time. Sign, dictate, or authorise as per your clinical system's requirements.

What this means for you

Post-MI discharge letters that are complete, clear, and sent promptly reduce re-admission. They ensure the GP knows exactly what happened, what changed, and what they need to do. They give patients something to read at home when they — or their family — are trying to understand what happened and what comes next. The time investment in this workflow is small. The clinical impact is not.

When not to use this workflow

For complex admissions involving multiple complications, transfers from another centre, or significant clinical uncertainty about the diagnosis or management, the discharge letter warrants more careful individual writing. Use this workflow for straightforward post-MI discharges where the clinical narrative is clear and the main task is efficient, accurate documentation.

Remember: AI is a helpful assistant, not a clinician. You make the call.