Workflow: Producing a Patient-Facing Side-Effect Management Guide

A good side-effect guide does two things. It prepares patients for what is coming — reducing fear of the unknown. And it teaches them which symptoms to manage at home and which warrant a call to the clinical team. This workflow takes you through producing one using an AI tool.

1. Identify the treatment type and patient profile

   Before opening the tool, decide exactly what you are writing for. Which treatment category? Which patient — age range, any relevant comorbidities that affect side-effect experience? The more specific you are in your prompt, the more relevant the output. "Immunotherapy side-effect guide for a patient in his sixties with no prior significant medical history" will produce a more useful draft than "chemotherapy side effects."

2. List the key side effects you want covered

   Write down the side effects you want the guide to address — not an exhaustive list, but the ones that are most clinically important or most commonly cause patients to contact your team unnecessarily. Grouping them into "usually manageable at home" and "needs clinical contact" before you start makes the prompt much more effective.

3. Open the tool and write a specific prompt

   Open your AI tool and type your prompt. Be explicit about: the treatment type, the patient profile, the side effects to cover, the structure you want, and the reading level. Include the instruction: "Do not include specific drug names, doses, or clinical decision thresholds. I will add those separately."

4. Read the first draft as a clinician

   Read the draft with your clinical knowledge active. Ask yourself: Is this accurate? Are the side effects described correctly? Is the distinction between home management and clinical escalation appropriate for this treatment type? Does anything overstate or understate the likely experience?

5. Replace any generic thresholds with your unit's specific guidance

   AI tools often include general guidance such as "contact the team if your temperature is above 38°C" or "seek advice if you cannot eat for more than 24 hours." These may or may not match your unit's specific protocols. Replace every threshold or escalation criterion with your own unit's guidance. This step is non-negotiable.

6. Check the emotional tone

   Read the guide again, imagining a patient who has just started treatment and is frightened. Is the tone calm? Is it human? Does it acknowledge that treatment is hard, without being alarming? Ask the tool to adjust if needed: "Rewrite the introduction to be warmer and more acknowledging of how difficult starting treatment can feel."

7. Add local and patient-specific information

   Add your unit's contact details, nurse specialist name, and out-of-hours pathway. If the patient has a specific comorbidity that affects one of the side effects covered — for example, existing bowel disease and a treatment that causes diarrhoea — add a personalised note for that patient. The tool produces the template. You personalise it.

8. Consider a reading-level check

   If you are unsure whether the guide is accessible enough, ask the tool: "Is the language in this guide appropriate for someone who has not been to university? If not, simplify it." Or paste a paragraph and ask: "Rewrite this paragraph at a simpler reading level." Do not assume that because the tool wrote in plain English, it wrote at the right level for every patient.

What this means for you

When patients understand their side effects, they manage them better. They call the helpline for the right reasons rather than for every minor symptom or because they were not warned. A clear, accurate side-effect guide is not just a patient service — it also protects your team's capacity by reducing avoidable contacts.

When not to use this workflow

Do not use this workflow for treatments with complex immune-related or targeted side-effect profiles where the distinction between manageable symptoms and medical emergencies is particularly fine. In those cases, a drafting tool is useful for general framework but the clinical content requires especially careful review.

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