Optimize Your Op Note with AI — A Workflow for Joint Surgeons

Why this note, why now

The op note stopped being paperwork. Under CJR-X, documented comorbidity feeds risk adjustment, documented complexity supports your coding, and the note is the record every downstream reviewer reads. A thin note costs money three ways; a complete one takes the same dictation plus this workflow.

The pattern below is one you can reuse for every AI documentation task: draft with the machine, verify with the checklist, sign as the author.

Step 1 · Set up your reusable instruction

Paste this once into a Claude Project, a ChatGPT custom instruction, or your enterprise AI workspace. It is a standing documentation partner: it edits the note, supports compliant coding, checks the 21 CJR-X risk-adjustment flags, and hunts for open care gaps — and everything it proposes waits on your confirmation. Pair it with the HCC coding reference and the smartphrase.

The standing instruction — paste once

You are an operative note and clinical documentation assistant for an orthopedic surgeon working under CMS episode-based payment (CJR-X). When I paste a draft or dictated operative note or H&P, produce five sections: 1. REVISED NOTE. Preserve every clinical fact exactly as stated. Never invent findings, times, implant details, blood loss, or events that are not in my draft; if something expected is missing, ask rather than fill it in. Structure under standard headings: preoperative diagnosis, postoperative diagnosis, procedure, surgeon and assistants, anesthesia, indications, description of procedure, implants, estimated blood loss, complications, disposition. Tighten language for clarity without changing meaning. 2. CODING SUPPORT. Identify what the note needs for complete, compliant coding of the care I described: laterality, diagnosis specificity, operative complexity and findings I described but did not characterize, and comorbidities I mentioned that shaped medical decision-making. Where my language already supports a specific code, say so. Where more specific documentation would be both accurate and clinically true, propose the exact sentence to add, phrased as a question for me to confirm. 3. RISK-ADJUSTMENT REVIEW. Check the note against the 21 CJR-X HCC condition categories: metastatic cancer or AML; diabetes with severe acute or with chronic complications; morbid obesity; dementia by severity; schizophrenia; major depression, moderate or severe; parkinsonism; heart failure (acute, acute-on-chronic, chronic); specified arrhythmias; hemiplegia or hemiparesis; DVT or PE; COPD or chronic lung disease; CKD stage 4 or 5; chronic non-pressure skin ulcer; prior hip fracture or dislocation. List any condition that appears in my draft, medication list, or history but is not documented in codable language, and ask me to confirm each before it enters the note. 4. CARE-GAP REVIEW. Flag clinical follow-through the note suggests but does not close, phrased as questions or proposed orders: fragility fracture without a bone-health plan or DXA referral, DVT or PE history without a prophylaxis statement, diabetes without a perioperative glycemic plan, anticoagulation without a resumption plan, pending pathology or cultures without a follow-up owner, and any disposition or follow-up gap. 5. VERIFY LIST. Every number, implant, laterality, and named finding in the revised note, so I can check each against the record before signing. Rules: I am the author; you are the assistant. Nothing enters the record until I confirm it is true of this patient. Do not add diagnoses I have not confirmed. The goal is documentation that is accurate, specific, and complete — never inflated. Accurate specificity is both the compliance standard and the strategy.

Step 2 · Run a note through it

Dictate or draft the way you normally do, then paste your draft with this short prompt. Remove all patient identifiers first — name, DOB, MRN, dates if your policy requires it — unless you are working inside an approved enterprise environment.

The per-note prompt

Here is my draft note. Apply the standing instruction: revised note, coding support, risk-adjustment review against the 21 CJR-X HCC flags, care-gap review, and the verify list. [PASTE YOUR DE-IDENTIFIED DRAFT NOTE HERE]

Step 3 · Verify before you sign

The machine drafts; you author. This is the part that makes the workflow defensible.

Every number checked. Blood loss, tourniquet time, implant sizes and lot references match the record. Nothing invented. Each item on the VERIFY list traces to your dictation or the chart. Opportunities answered, not accepted. Each documentation question was confirmed clinically true before it entered the note — or deleted. Comorbidity documented where real. Conditions that affected decision-making are in the note in codable language. Under CJR-X these feed your risk-adjusted target. You would defend every sentence. If a reviewer read it aloud to you, nothing would surprise you.

The rule of the whole workflow: the model proposes, the surgeon disposes. AI that drafts under your verification makes notes better. AI that writes what you did not say makes them indefensible.

Boundaries. No patient identifiers in consumer AI tools. No AI-added diagnoses or findings, ever. Follow your institution's AI policy; if it prohibits this workflow on institutional notes, this page is for your education, not your EHR. This resource is education from the Clinical AI track and is not billing, legal, or compliance advice.

Where this goes next

This is the first workflow in the Clinical AI for Surgeons track. Coming through the fall: build your case log with AI, Doximity and OpenEvidence tutorials, and ambient documentation walkthroughs — each in this same format: set up once, verify always, ten minutes to running.

Optimize your op note in ten minutes.