Workflow: Drafting a Post-Treatment Summary Letter

This workflow takes you from a list of clinical bullet points to a finished post-treatment summary letter. It assumes you are using a general-purpose AI tool such as Claude or ChatGPT. The steps are identical for either.

The whole process takes around ten to fifteen minutes the first time. Subsequent letters for similar patients take five.

1. Prepare your clinical bullet points before opening the tool

   Spend two minutes writing down the key facts you want in the letter. Include: the treatment that has been completed, the follow-up plan, what the patient should watch for, who to contact and when, and any late effects worth mentioning. Having this prepared before you open the tool means you spend less time in the tool and more time refining the result.

2. Open the tool and start a fresh conversation

   Go to claude.ai or chat.openai.com. Sign in. Start a new conversation each time you begin a new task — this prevents the tool from mixing up context from previous sessions. You will see a text box at the bottom of the screen. This is where you type.

3. Give the tool its instructions before the clinical content

   Begin by telling the tool what it is doing and the format you want. For example: "You are helping an oncologist draft a patient letter. I will give you clinical bullet points. Turn them into a letter — warm and clear, written for someone with no medical background. Use plain English. No specific drug names or doses. Maximum 400 words." Then press Enter and wait for the tool to acknowledge.

4. Provide your clinical bullet points

   Type or paste your bullet points into the conversation. Keep them factual and anonymised. The tool will produce a drafted letter in under thirty seconds.

5. Read the draft as a clinician, not as a writer

   Read the draft with clinical eyes. Check: Is every fact accurate? Is anything missing? Is anything ambiguous? Is there language about prognosis, likelihood of recurrence, or expected outcomes that should not be there? The tool sometimes adds reassuring language that implies a clinical judgement you have not made. Find it and remove it.

6. Adjust the tone for the specific patient

   Read the draft again, this time imagining the patient receiving it. Is the tone right for someone who has just finished an exhausting course of treatment? Is it warm enough? Does it acknowledge what she has been through? If not, ask the tool: "The tone is slightly clinical. Rewrite to be warmer and more personally acknowledging of the patient's experience, while keeping all the clinical content."

7. Refine with targeted follow-up instructions

   If specific sections need adjustment, tell the tool exactly what to change rather than rewriting from scratch. For example: "The section on follow-up appointments is unclear. Rewrite it in two sentences, making the timing and purpose of each appointment explicit." The tool will revise only what you ask.

8. Copy the draft into your clinical system

   Once you are satisfied with the draft, copy it into your word processor or clinical letter template. This is the point at which you add: the patient's name and address, your name and contact details, the date, and any local resource information such as nurse specialist contacts or helpline numbers.

9. Read the completed letter one final time before sending

   Read the finished letter from beginning to end, as if you are the patient receiving it. Check that nothing in the AI-generated sections inadvertently implies a clinical prediction or contains information that does not apply to this individual. Make any final adjustments. Sign and send.

What this means for you

The gain is not just speed — it is consistency. Every patient you discharge from active treatment receives a letter of the same quality, regardless of where that appointment fell in your day. A patient seen at the end of a long Friday clinic deserves the same care in communication as one seen on a fresh Tuesday morning.

When not to use this workflow

Do not use this workflow for letters that follow a consultation involving very difficult news — a recurrence, a change to palliative intent, a new diagnosis. Those letters carry greater emotional weight and require more personal care than a drafting tool can reliably provide without close review. Use it for standard post-treatment and follow-up correspondence.

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