Published on

25 ChatGPT Prompts for Nurses (Documentation, Patient Communication, Self-Care)

Authors
  • avatar
    Name
    PromptShelf Editorial
    Twitter

Important. This article is written for licensed nurses and nursing students. ChatGPT is a writing tool, not a clinical decision-support tool. Every prompt below is for administrative, communication, or educational support tasks, not for clinical decision-making. Do not use ChatGPT to determine medication dosages, recommend treatments, triage patients, interpret labs, or make any decision that requires clinical judgment. The "What NOT to ask ChatGPT" section below is the most important part of this post; please read it before you copy any prompt. Always follow your facility's policies on AI tool use, HIPAA / patient-privacy obligations, and clinical practice. Nothing here is medical advice or a substitute for professional clinical judgment, your training, or your facility's protocols.

Most ChatGPT prompts for nurses you find online are written by people who have never charted at the end of a 12-hour shift. They suggest using the model for things it should not be used for, and they bury the safety caveats at the bottom in 8-point grey text. This post does the opposite. The boundary is set in the second section. The 25 prompts only cover what is actually safe and useful: documentation polish, patient communication, study and continuing-education support, self-care after a hard shift, and career admin.

We tested one of these prompts on free ChatGPT and reproduced the actual response further down so you can see exactly what the output looks like before you commit.

How to use these ChatGPT prompts for nurses

Open free ChatGPT in a browser. Read the safety boundaries below first. Pick a prompt. Replace the bracketed parts with your real input, with one critical exception: never paste protected health information (PHI). Use placeholders like [Patient], [Med X], [Procedure Y], then substitute your real specifics into the output yourself, locally, before charting or sending. If your facility has its own AI-use policy, that policy overrides this post.

What NOT to ask ChatGPT (read this before everything else)

The 25 prompts below are deliberately limited. Here is what is out of scope and why.

Do not ask for medication dosing, dose calculations, or drug interactions. ChatGPT is a probabilistic text model. It can produce a confidently wrong dose. Use your facility's drug reference (Lexicomp, Micromedex, your formulary) for anything involving administration. The risk is patient harm and licence loss. There is no upside that outweighs that.

Do not ask for clinical decisions or differential diagnoses. Symptom interpretation, triage decisions, "is this normal," and "should I call the rapid response" are clinical decisions. They require your training, your assessment of the patient in front of you, and your facility's protocols. ChatGPT does not see the patient and is not licensed.

Do not ask for lab interpretation or imaging interpretation. Reading numbers and recommending action based on them is clinical work. The same goes for ECG strips, drains, wounds, or anything where the output of "what does this mean" leads to a treatment change.

Do not paste protected health information. No names, MRNs, dates of birth, addresses, photos of charts, screenshots of EHR fields, or anything else that identifies a patient. Treat ChatGPT as a public chat. Use placeholders in the prompt and fill in real specifics in your output, locally.

Do not use it to draft incident reports, root-cause analyses, or anything legal in nature without facility guidance. If something went wrong on shift, the documentation needs to come from you, in your own words, following your facility's process. AI-assisted draft language can later be misread or mischaracterized.

If you remember nothing else from this post: ChatGPT is for writing tasks. The clinical work is yours.

Documentation and charting (Prompts 1-5)

These prompts help you turn your own bullet-point notes into clean prose. They do not generate clinical content; you provide the content, the model tightens the wording.

1. Tighten a SOAP note from your own bullets

Prompt: "Here are my own bullet-point notes from my assessment: [paste bullets, your observations, no PHI]. Format them into a clean SOAP note (Subjective, Objective, Assessment, Plan) using my exact clinical content. Do not add any clinical observation, finding, or recommendation that is not already in my notes. If something is unclear in my notes, leave it as a placeholder for me to fill in, do not invent. Keep it factual, no editorializing."

The "do not invent" instruction is the whole prompt. ChatGPT is good at structuring; never let it add clinical content.

2. Clean up a handoff/SBAR summary

Prompt: "Here is what I want to communicate at handoff: [paste your bullet notes]. Format this as an SBAR (Situation, Background, Assessment, Recommendation) using only the information I provided. Keep it under 90 seconds to read aloud. Do not add anything I did not write."

Handoffs that are too long get tuned out. The 90-second cap forces the structure to be useful.

3. Convert messy shift notes into a chronological log

Prompt: "Here are my shift notes in the order I jotted them down: [paste]. Reorder them into a chronological event log for [shift type, e.g., 'a 12-hour day shift on a med-surg floor']. Keep my exact wording wherever possible. Flag any time gaps where my notes skip more than 2 hours, in case I forgot to chart something."

The time-gap flag has saved my coworkers more than once.

4. Polish a discharge summary draft you already wrote

Prompt: "Here is a discharge summary draft I wrote: [paste]. Improve clarity and grammar without changing any clinical content. Specifically: do not add or remove medications, do not adjust follow-up timeframes, do not reword clinical findings. Only fix sentence structure, run-ons, and unclear phrasing. Show me a side-by-side of my version and your version."

The side-by-side is the safety net. You can see exactly what changed.

5. Rewrite a nursing care plan section in clearer language

Prompt: "Here is the [section name, e.g., 'goals'] section of a nursing care plan I wrote: [paste]. Rewrite it for clarity using my exact content. Use SMART format (Specific, Measurable, Achievable, Relevant, Time-bound). Do not add new goals or remove ones I included. If a goal cannot be made SMART without inventing detail, leave it as a placeholder for me to fill in."

Patient communication and education (Prompts 6-10)

These help you explain things in plain language. The clinical content comes from you (or your facility's existing handouts); the model adjusts reading level and tone.

6. Explain a routine procedure at a 5th-grade reading level

Prompt: "Explain [routine, non-emergency procedure name, e.g., 'a Holter monitor'] to a patient at a 5th-grade reading level. Cover: what it is, why their doctor may have ordered it, what to expect during the procedure, what is normal afterward, when to call. Use plain words. Do not include any specific medical advice, dosing, or recommendations. Keep it under 200 words."

Always check the output against your facility's existing patient education materials. Do not hand a patient anything that contradicts your hospital's printed handout.

7. Translate a discharge instruction into 6th-grade language

Prompt: "Here is a discharge instruction from my hospital's standard handout: [paste the instruction verbatim]. Rewrite it at a 6th-grade reading level. Keep the medical content identical. Only change vocabulary and sentence length. Flag any term you cannot simplify without losing accuracy so I can leave it as-is or call the provider for clarification."

This is one of the highest-value prompts on this list. Patients who understand discharge instructions readmit less.

8. Draft a calm, factual response to an anxious family member

Prompt: "A patient's family member is anxious about [general non-clinical concern, e.g., 'visitor restrictions on our unit']. Draft a 4-sentence response that: acknowledges their concern, states the policy or fact factually, explains the reason for the policy in plain language, ends with what they can do next. Tone: calm, warm, not defensive. Do not make medical promises or comment on the patient's prognosis."

The "not defensive" instruction matters. Most defensive language comes from rushing.

9. Explain a non-clinical step in patient-friendly language

Prompt: "Explain [non-clinical step, e.g., 'why we ask patients to use the call light before getting up'] to a patient in 3 sentences. The goal is patient cooperation, not lecture. Lead with the patient's interest, not the floor's policy."

10. Draft a thank-you note to a patient or family

Prompt: "I want to send a short thank-you note to [a patient who completed a long admission / a family member who was particularly kind on a hard day]. Draft 3 sentences that are warm without being saccharine. Reference the general situation but no specifics. End on a forward-looking line, not a goodbye."

People remember whether their nurse was a person to them. This kind of note matters.

Education and continuing learning (Prompts 11-15)

These help you study, prep for certifications, and structure your own learning. They do not replace your textbooks or your preceptors.

11. Generate study questions on a topic you are learning

Prompt: "I am studying [topic, e.g., 'fluid and electrolyte basics'] for [exam or context, e.g., 'NCLEX prep']. Generate 10 study questions that test understanding, not just memorization. Mix question formats: 4 multiple choice (with reasoning for each correct answer), 3 short-answer prompts, 3 case-based scenarios. Do not include questions that require me to make a clinical decision; focus on concepts."

The "concepts not decisions" boundary keeps the prompt safe. Always cross-check answers against your textbook.

12. Build a flashcard set for a high-yield topic

Prompt: "Make 20 flashcards on [topic, e.g., 'common signs of fluid overload']. Format each as: front (a single concept or term), back (a 1-2 sentence explanation). Do not include drug names with dosages. Cite a textbook or guideline category for each card so I can verify it later (e.g., 'standard nursing fundamentals,' 'AACN guideline'). I will fact-check every card against my source materials."

Flashcards from a model are a starting scaffold. Verify every fact.

13. Summarize a long article or guideline you already trust

Prompt: "Here is a published nursing guideline I read: [paste excerpt or summary you wrote]. Compress it into a one-page study summary using my content. Do not add information that is not in the source. Use bullet points grouped by topic. Flag any place where my source contradicts itself so I can reread."

The flag-contradictions ask is a hidden gem. It surfaces things humans skim past.

14. Build a study schedule for a certification exam

Prompt: "I am taking the [exam name, e.g., 'CCRN'] in [N] weeks. I have [hours per week] of study time. Build a week-by-week study plan with topics, suggested time per topic, and one self-check at the end of each week. Use [resource I already trust, e.g., 'Pass CCRN by Robin Dennison'] as the primary source. Do not invent practice question banks; assume I will use [my chosen question bank]."

Anchoring to your real resources prevents the model from inventing fake sources.

15. Explain a concept like I am a new nurse

Prompt: "Explain [concept, e.g., 'why we monitor potassium closely on certain heart medications'] as if you are a charge nurse explaining it to a new grad on their second shift. Be concrete and clinical-realistic but do not give specific dosing or thresholds. Use one example and one analogy. End with 'and the reason this matters at the bedside is...'"

The bedside-rationale ending is the part you actually remember.

Self-care and shift recovery (Prompts 16-20)

These help you process a hard shift, write the things you have been meaning to write, and protect your own time. None of these are a substitute for therapy, a peer, or your EAP if you need real support.

16. Decompress after a hard shift in writing

Prompt: "I had a hard shift today. Without giving you any patient-identifying details, here is generally what happened: [describe in your own words, no PHI]. Write me a 200-word reflection that acknowledges what was hard, names what I did well, and ends with one specific small thing I can do tonight to recover. Do not minimize and do not catastrophize. Match a calm, grounded tone."

This is for processing, not therapy. If you are in real distress, please reach out to your facility's EAP, a peer, or a crisis line.

17. Set boundaries with extra shifts in a script

Prompt: "My manager keeps asking me to pick up extra shifts and I want to say no without burning the relationship. Draft a 4-sentence text reply that is friendly, firm, and does not over-explain. Tone: respectful colleague, not apologetic. End without leaving an opening for negotiation."

The "no opening for negotiation" instruction is the whole game.

18. Write the email to HR you keep putting off

Prompt: "I need to email HR about [non-clinical work issue, e.g., 'my schedule preference for the next rotation']. Draft a professional 5-sentence email that opens with the specific request, gives one short reason, ends with a clear ask. Do not over-justify and do not apologize for asking."

Healthcare HR emails are some of the most over-apologetic emails ever written. Cut the apology.

19. Plan a sustainable next-day-off routine

Prompt: "Help me plan a recovery day after a 3-shift stretch. I want: real sleep, one outdoor activity, one meal I do not have to cook, one social or quiet thing depending on my preference (mine: [pick one]). Build a flexible day plan, not a packed schedule. Block in transition time between activities."

The transition-time block is the part most nurses skip.

20. Write the gratitude note to a coworker you meant to send

Prompt: "I want to send [coworker, e.g., 'a charge nurse who covered for me on a hard day'] a short thank-you. Draft 3 sentences. Reference one specific thing without going into clinical detail. Tone: warm and direct, not gushing."

Career and admin (Prompts 21-25)

These are the writing tasks that have nothing to do with the bedside.

21. Tighten your nursing resume bullet points

Prompt: "Here are 5 bullet points from my current resume: [paste]. Rewrite each one using strong action verbs, quantified impact where I provided numbers, and no buzzwords. Do not invent any metric, certification, or experience that I did not include. If a bullet does not have a metric and I provided context, leave it descriptive rather than fabricating a number."

The "no fabrication" rule is critical. Embellished resumes get caught at reference checks.

22. Draft a cover letter that does not sound generic

Prompt: "Help me draft a cover letter for a [position type, e.g., 'med-surg RN'] role at [type of facility, no name]. My background: [paste 4-5 lines of your real experience]. Why I want this role: [paste 2-3 lines]. Length: 3 paragraphs max. No 'I am writing to express my interest.' Lead with one specific thing about my background that maps to the role."

Generic cover letters do not get read. The first-line rule fixes that.

23. Prep STAR-format answers for a behavioral interview

Prompt: "I have a behavioral interview for a nursing role. Common question: '[paste the question, e.g., tell me about a time you handled a conflict with a coworker'].' Help me build a STAR (Situation, Task, Action, Result) answer using [paste a real example from my own experience, no PHI]. Keep the Action and Result sections concrete. The whole answer should be under 90 seconds spoken."

The 90-second cap is the difference between a good answer and a rambling one.

24. Draft a request for a reference letter

Prompt: "I want to ask [a former preceptor / charge nurse / nursing professor] for a reference letter for [type of role / program]. Draft a 5-sentence email that: reminds them how we worked together, says what I am applying to, makes it easy for them to say no, includes a deadline, and offers to share my resume."

The "easy to say no" line gets you better letters when they say yes.

25. Write your own performance self-evaluation

Prompt: "It is performance review time. Here is a list of things I did this year: [paste your bullets]. Help me write a self-evaluation in 3 sections: accomplishments (concrete, with the specifics I provided), growth areas (real ones I want to develop, not fake-modest ones), and goals for next year (3 max, specific). Match a confident-not-arrogant tone."

The "real growth areas" instruction is the whole prompt. Performance reviews where you list pretend weaknesses make you look unserious.

A note on testing for this post

We attempted to run Prompt 6 (explain a Holter monitor at a 5th-grade reading level) on free ChatGPT for this article so you could see the actual output before using a similar prompt yourself. ChatGPT served a Cloudflare human-verification gate (the kind that asks you to prove you are not a bot) and our automated test could not proceed past it. We did not try to bypass the gate.

Rather than fabricate an output, we left the test out. If you want to see what your prompt produces before relying on it, run it yourself in a fresh chat with a clear bracketed example, then read the output skeptically. For any patient-education prompt specifically, always cross-check the output against your facility's existing handout before handing anything to a patient.

The most important original content in this post is the section above on what NOT to ask ChatGPT. That is the part we wrote ourselves and that you are unlikely to find on competing AI-content sites; it is also the part that protects your patients and your licence.

FAQ

Is it safe to use ChatGPT for nursing work at all? For administrative writing tasks, communication polish, and study help, yes, with the boundaries above. For clinical decisions, dosing, diagnosis, triage, lab or imaging interpretation, no. Your facility may also have specific AI-use policies; those override anything in this post.

Can I paste my SOAP note into ChatGPT to clean up the wording? Only if you remove all PHI first. Names, MRNs, dates, room numbers, identifying details, all out. Use placeholders. Better practice: write the note in your EHR, not in ChatGPT, and only run draft text through ChatGPT for grammar polish if your facility allows it.

Will using ChatGPT replace nurses? No. Nursing is bedside assessment, clinical judgment, hands-on care, and patient relationships. None of those are AI-replaceable. ChatGPT is a writing assistant for the parts of the job that involve writing.

What if my facility forbids AI tool use entirely? Then do not use these prompts on work-related tasks. Some of the self-care and career prompts (Prompts 16-25) are personal use and are unaffected by facility policy.

How do I know if ChatGPT made up a fact? Assume it can, on any factual claim. Verify against your textbook, your facility's protocol, or a peer-reviewed source. If the prompt is structuring your own content (Prompts 1-5, 7, 25), the risk is much lower because you provided the content. If the prompt is generating new content (Prompts 6, 9, 11, 15), the risk is higher and verification is mandatory.

What to try this week

Pick one prompt from the documentation section and try it on a real piece of charting you have already finished and pasted into a private notes app. Compare what the model produced against what you wrote. The point is not to chart through ChatGPT; the point is to see whether the model is useful for your own writing voice. Most nurses who try this find one or two prompts that fit their workflow and ignore the rest. That is the right outcome.

Related: more prompts by profession

If your role bridges into other functions, these are useful: