Rethinking Ortho Documentation at Scale: St. Cloud Orthopedics’ Playbook for AI Scribing
12 December 2025•10 min read
Heidi is the best notetaker you’ve ever had. The notes that were done within 24-48 hours—usually in just a few hours. That was unheard of before.”
Chad Ritter
Director of Clinical Operations & Physical Therapist
Key outcomes:
Transitioned from costly transcription and error-prone dictation to AI-driven notes
Achieved near-universal provider adoption through careful piloting and champion-led rollout
~90–95% of encounters use Heidi output directly; the remainder still leverage Heidi as a high‑fidelity transcript for quick dictation.
Reduced turnaround time for notes from 7–10 days to same‑day (often within 4–5 hours) and consistently -48 hours ~80% decrease in turnaround time
Increased clinical efficiency by speeding up prior auths, referrals, and billing workflows
Improved note quality, depth, and accuracy across specialties, from orthopedics to therapy
Read on if:
You’re relying on dictation or human transcription but still struggling with accuracy and cost. Your providers are facing backlogs in signing or editing notes, and documentation delays are slowing down patient care coordination and prior authorizations. You want to roll out AI scribing clinic-wide, but you need a clear roadmap for change management to make it successful.
Background
St. Cloud Orthopedics is a high-volume specialty practice with clinics in Sartell and South St. Cloud. They cover nearly every subspecialty—spine, sports medicine, shoulder/elbow, hand/wrist, hip/knee, foot/ankle, and joint replacement—supported by urgent care, imaging, and on-site PT/OT. Procedures range from injections, casting, and carpal tunnel release to ACL reconstructions, rotator cuff repairs, total joint replacements, and complex foot/ankle reconstructions. That breadth translates into equally diverse documentation: detailed MSK exams with laterality, imaging reports, injection/procedure notes, pre- and post-op surgical documentation, implant logs, and therapy evaluations, progress notes, and discharge summaries. With multiple providers rotating across seven rooms simultaneously, documentation was one of their biggest operational challenges.
For years, the clinic relied on transcription at 10 cents per line, costing thousands each month. The notes came back somewhat polished, but the delays were significant. Switching to dictation tools cut turnaround time but not errors. Providers often sent out letters that “made us look like we had a third-grade reading level,” as one leader put it.
The clinic needed a solution that combined accuracy, speed, and scalability—without adding to staff workload.
Challenges
1) Costly transcription, slow turnaround
Transcribed notes could take 7–10 days to hit the chart, slowing imaging authorizations and downstream care. Costs piled up with each line transcribed.
2) Dictation without a safety net
Without a human editor, the dictation tool routinely mangled specialty terms like “cortisone injection” or “sinoscopy” and produced letters the team “hated sending.” Without human transcriptionists to clean them up, notes sometimes went out error-ridden.
3) A workflow that doesn’t fit the mold
Traditional “one provider, one mic” models don’t fit a room‑anchored, team‑handoff clinic running multiple rooms at once.
4) End-of-day cognitive drag
EOD dictation meant reconstructing exams from memory, adding stress and variability even when the typing time itself seemed short.
Solution
St. Cloud Orthopedics approached Heidi adoption as a staged rollout designed to prove value, train staff, and build templates gradually.
1) Start with pilots, expand with confidence
The team pieced together a few weeks of self‑serve trials plus 3 more weeks of vendor trial support. That time let them iron out hiccups, build templates/snippets, and prove fit across roles before paying. Clinical Ops lead + lead PA ran the experiments, then converted skeptics with 4‑hour blocks of real use (most skeptics switched after one session).
“We didn’t rush. By go‑live, most of the hiccups were already solved.” – Kelli Penrose, Physician Assistant and Implementation Lead
2) Multi-provider session setup
Sessions were designed so that nurses, PAs, therapists, and surgeons could contribute to the same record as the patient moved through their visit. Pause/resume etiquette kept the notes clean and ensured each role’s input was captured. Providers can log in later to each hallway’s queue to finish notes without hunting through charts.
3) Cross‑team alignment
Nursing initiates sessions and drops procedure snippets; PAs/surgeons resume to add history/exam/plan; business offices can verify charges against the exact injection/consent wording in the finished note. Everyone touches the same session, so handoffs are cleaner.
4) Keep templates minimal; move precision into the Context
Rather than overwhelm clinicians with dozens of rigid templates, the team found success with versatile templates (new, follow-up, phone) with conditional sections for procedure/consent/exam supported by macros. Any wording that must be exact—e.g., injections, consent language, PT goal phrasing—lives as snippets pasted into Contextbefore generation. Heidi then weaves them into the note naturally. For example:
Procedures: Injection doses (e.g., 40 mg Depo-Medrol, 5 cc lidocaine) dropped in via snippets in context. The generated note always includes exact dose/agent, satisfying billing/compliance without the provider re‑typing details.
Consents: Consent statements must be word‑for‑word to match signed forms. The snippet is copied verbatim into Context and the template tells Heidi to quote it exactly in the Procedure/Consent section.
Therapy goals: Stock goals (e.g., return‑to‑run progression, ROM thresholds, time‑bound milestones) live as snippets; therapists still tailor content conversationally, but the generated plan lands with precise timelines and metrics.
This combination of broad templates and targeted snippets allowed Heidi to produce natural, accurate notes with minimal editing. Gender‑neutral snippets (they/them) are auto‑resolved by Heidi to “he/she/they” correctly, improving readability without duplicating variants.
“We first tried pasting stock language after generation. Better result was adding those snippets to Context before we generate—the note flows naturally and the exact words are there for billing and consent.”
Workflow snapshot
High‑volume injection block
Nurse rooms patient, opens Heidi on the room account, and adds the injection snippet via quick‑type (e.g., /inj_knee_depo) in context
PA evaluates, resumes Heidi, confirms indication and laterality, orders X‑ray if needed, pauses.
Tech applies splint or escorts to imaging; staff resume/pauses as needed.
Surgeon resumes, performs injection; template pulls the exact dose/agent from Context into the Procedure section.
Final note is ready for billing same day, with precise diagnosis codes and consent text included.
Therapy evaluation + follow‑up
PT runs all initial evals with Heidi live; for bustling gym follow‑ups (too much ambient noise), PT starts a post‑visit dictation using Heidi to produce a SOAP note that’s cleaner than Dragon and requires minimal edits.
Oddball visits
On rare complex/atypical encounters the provider may revert to a quickly edited dictation, but still uses the Heidi transcript as the authoritative record—dramatically reducing time spent reconstructing the visit.
“Even if you don’t use the draft verbatim, Heidi gives you a perfect memory of the room. Editing that is a whole different (easier) job than creating from scratch.”
Impact
“Our notes don’t read like third grade anymore” – Stronger, more credible documentation
Notes are richer and more consistent; Specialty terms and numeric details (e.g., dose/volume, laterality, device sizes, ROM values) are captured the first time. Business office stopped bouncing injection notes for missing details.
“Our letters no longer read like we have a third‑grade reading level. The prose is clean, and the right numbers are in the right places.”
“Editing on my couch beats blank pages at 7 p.m.” – Stress and burnout down
Clinicians report a dramatic drop in end‑of‑day cognitive load. Skeptics converted after a single 4‑hour clinic block; most now report finishing sooner and with less stress.
“Reading and tweaking a solid draft on my couch beats starting from a blank page at 7 p.m.”
“From a week to a workday” – Faster turnaround that speeds care
Notes are typically ready within 4–5 hours, and almost always <24 hours. Prior auths and imaging requests move days faster because documentation is already in the chart.
A playbook for scalable adoption
Minimal templates + context snippets = less clicking, fewer variants to maintain, and scalable onboarding for new staff. By transitioning away from outsourced transcription, the clinic redirected spend toward care operations while improving note quality and reliability.
“It removes the hard minutes: the remembering, the hunting for labs, the retyping of doses. That’s what burns people out.”
What other clinics can learn
Keep templates lean, rely on snippets for precision: Generic templates keep flexibility; snippets ensure compliance and exact language.
Design sessions around team workflows: Shared pause/resume etiquette ensures every role’s contribution is captured.
Trial for workflow fit, not just features: Allow time for champions to refine workflows and train others. Let skeptical clinicians trial Heidi during real clinics; seeing their own notes is the best driver of adoption.
What’s next
St. Cloud Orthopedics continues to refine templates and explore better ways to share snippets and memory across teams. Leadership sees opportunity to expand Heidi into surgical letters, appeals, and therapy workflows.
“Heidi is the best notetaker you’ve ever had. It makes the end of the day so much easier. That’s what makes it stick.”
Share this post
Rethinking Ortho Documentation at Scale: St. Cloud Orthopedics’ Playbook for AI Scribing | Heidi AI
