Draft a prior authorization letter from documented facts
Insurance wants a letter of medical necessity and I have twelve other things to do.
Small simulated evaluation De-identified text only Individual access available
Source check: 2026-11-11 · Physician review: not yet recorded
Build an authorization letter from documented facts
Give the tool the payer’s criteria and your record; get a draft with a missing-information checklist.
Use the covered clinician workspace. Supply the actual payer policy, not a guessed requirement.
You provide
Fictional administrative example; not an actual payer policy. Requested service: physical therapy for a clinician-documented shoulder problem. Clinic note A: symptoms limit dressing and overhead reach. Clinician requests six PT visits for functional rehabilitation. Example payer criterion 1: document functional limitations. Example payer criterion 2: document the requested visit count. Example payer criterion 3: document a prior home-exercise trial. No home-exercise trial is documented. No diagnosis code supplied.
A useful result
Criterion-to-evidence check 1. Functional limitation: documented in clinic note A (dressing and overhead reach). 2. Visit count: six visits requested in clinic note A. 3. Prior home-exercise trial: not documented. Draft excerpt I am requesting six physical therapy visits for the shoulder problem documented in clinic note A, which limits dressing and overhead reach. The intended purpose is functional rehabilitation. Before submission Supply the verified diagnosis/code and any required home-exercise documentation, or explain why that criterion does not apply. This draft does not establish that all payer criteria are met.
Illustrative output written for this guide, not a live model run. Fictional example; clinician decisions are supplied in the input.
Draft a prior-authorization letter using only the supplied clinical facts and payer criteria. First create a criterion-to-evidence table with a source label for every fact. Mark unsupported criteria “not documented”; do not fill gaps. Then draft a concise letter. Do not invent diagnosis codes, prior therapies, intolerance, dates, severity, citations or medical necessity. Preserve the clinician’s stated rationale. Do not assert that a criterion is met if its documentation is missing. End with a list of information required before submission. DOCUMENTED FACTS AND PAYER CRITERIA: Fictional administrative example; not an actual payer policy. Requested service: physical therapy for a clinician-documented shoulder problem. Clinic note A: symptoms limit dressing and overhead reach. Clinician requests six PT visits for functional rehabilitation. Example payer criterion 1: document functional limitations. Example payer criterion 2: document the requested visit count. Example payer criterion 3: document a prior home-exercise trial. No home-exercise trial is documented. No diagnosis code supplied.
Replace the bracketed material in your AI tool. The site does not collect clinical text.
Draft a prior-authorization letter using only the supplied clinical facts and payer criteria. First create a criterion-to-evidence table with a source label for every fact. Mark unsupported criteria “not documented”; do not fill gaps. Then draft a concise letter. Do not invent diagnosis codes, prior therapies, intolerance, dates, severity, citations or medical necessity. Preserve the clinician’s stated rationale. Do not assert that a criterion is met if its documentation is missing. End with a list of information required before submission. DOCUMENTED FACTS AND PAYER CRITERIA: [paste your material here]
Before you use it
- Is every assertion traceable to the record or payer document?
- Resolve missing criteria and verify codes yourself.
- Confirm the correct patient, payer, service and submission requirements.
The prompts are editorial templates, not clinically validated tools. Keep responsibility for clinical decisions and final communication.
The workflow
- For a first test, use fictional or appropriately de-identified content. For identifiable clinical records, use a workspace covered by your organization’s patient-data agreement; removing names alone is insufficient.
- Gather the payer’s actual written criteria and the documented facts from the record; do not rely on a guessed requirement.
- Paste both into the reusable prompt in ChatGPT for Clinicians, which builds a criterion-to-evidence table before the letter.
- Do not ask the AI to find or cite guidelines or studies. Include a citation only if you supplied it and checked it against the original source; delete any reference the draft adds on its own.
- Leave any criterion without documentation marked “not documented,” and verify every diagnosis code yourself.
- Add the patient’s details in your own chart or letter template, confirm the payer and service, and sign.
Tools and cost
USD unless stated otherwise. Billing periods and eligibility matter. A tool listed here was not necessarily tested in the cited study.
Individual clinician verification
Eligible clinician workspace: sign in Settings → Agreements; authority to sign required.
What the evidence supports
In 29 synthetic nephrology scenarios, ChatGPT-generated authorization letters had the correct diagnosis code in 23 cases (79.3%). One cited an irrelevant trial. Real insurer approvals and time saved were not measured.
Primary study · Read the source ↗
Where it goes wrong
Check every code, factual assertion and reference against the chart and payer criteria before submission. The cited study tested ChatGPT letters in synthetic scenarios; it does not validate this editorial prompt or the current ChatGPT for Clinicians product.