By Sagar Shankaran, Founder of CallSphere
Nine files, a carrier policy bulletin and one question: how oral surgery practices catch the missing criterion before the orthognathic packet is submitted.
Key takeaways
Ask the surgical coordinator to pull the combined orthodontic-surgical cases from the last twelve months and count two things: how many came back from the medical carrier with a denial or a request for more information on the first submission, and how many of those were about a document that was already somewhere in the chart. In most oral and maxillofacial practices the second number is uncomfortably close to the first.
It is almost never a case that failed on medical merit. It is a missing note about six months of conservative management. It is a cephalometric superimposition that was in the orthodontist's chart and not in yours. It is a sleep study that was done in 2023 at a different facility. It is a functional impairment statement that the carrier's policy bulletin requires in a particular form on page 22, and the letter you sent said the same thing in different words.
Nine files, give or take, and none of them small. Eighteen to twenty-four months of progress notes from the referring orthodontist, usually exported as a long PDF. The cone-beam study report from the CS 9600 or the i-CAT. Cephalometric analyses at two or three time points with the tracings. A full photographic series, facial and intraoral. A model or digital scan analysis showing the occlusal discrepancy in millimeters. The sleep study with the apnea-hypopnea index, if the case has an airway component. The surgeon's letter of medical necessity. The plan's certificate of coverage, including the exclusion language that decides whether this is a dental or a medical claim at all. And the carrier's own medical policy bulletin for orthognathic surgery, which runs thirty to sixty pages of criteria and is revised without warning.
Somebody has to hold all of that in their head at once and answer a single question: does this case, as documented, meet every criterion this carrier states, and if not, what is missing? That is a person's whole afternoon, and it is an afternoon your surgical coordinator does not have during third-molar season.
Practices tried. The problem was that you had to feed the material in pieces, a chapter at a time, and by the time it read the policy bulletin it had lost the progress notes. You spent your time telling it which page you meant. What changed with Claude Opus 4.6 is that you can hand over the entire file — every note, every image report, every page of the carrier's bulletin — in one question, instead of chopping it up and hoping it remembers page 40 when it reaches page 300. No selecting excerpts, no summarizing first, no losing the one line in the orthodontist's note from fourteen months ago that documents the failed non-surgical attempt.
That single change turns an afternoon of cross-referencing into a question you can ask three times before lunch, with the bulletin swapped for the current version each time.
Hear it before you finish reading
Talk to a live CallSphere AI voice agent for dental practice in your browser — 60 seconds, no signup.
flowchart TD
A["Nine files handed over as one question"] --> B["Cross-check chart against the carrier orthognathic policy bulletin"]
B --> C{"Any stated criterion unsupported by a document in the file?"}
C -->|Yes| D["Name the gap: conservative care note, ceph superimposition, sleep study"]
D --> E["Surgical coordinator requests it from the referring orthodontist"]
E --> B
C -->|No| F["Draft medical necessity letter citing the page each criterion is met on"]
F --> G["Surgeon edits and signs, coordinator submits prior authorization"]
A twenty-three-year-old with a skeletal Class III and a five-millimeter reverse overjet has been in pre-surgical orthodontics for sixteen months. The referring orthodontist says decompensation is done and wants a surgery date in the fall. Historically the coordinator would block Thursday afternoon to assemble the packet.
Instead, at 9:15 she drops all nine files in together with the carrier's current bulletin and asks one question: go criterion by criterion through this policy and tell me, for each one, whether this chart supports it and on which page. By 9:40 she has a list. Twelve criteria supported with page references. Two not supported: there is no documentation of a trial of non-surgical management, and the masticatory function statement never quantifies anything. One ambiguity: the bulletin requires skeletal maturity confirmation and the chart has it in a hand-wrist film reference the orthodontist mentioned but never sent.
She sends three specific requests to the referring practice instead of a vague "can you send everything." The surgeon spends twenty minutes on Friday editing a drafted letter that already quotes the right criteria in the carrier's own language, rather than ninety minutes writing one from scratch. The packet goes out the following Tuesday instead of five weeks later, and it goes out complete.
Illustration figures for a two-surgeon practice. Use your own denial log.
| Line | Today | After a whole-file check before submission |
|---|---|---|
| Combined orthodontic-surgical cases per year | 14 | 14 |
| Denied or returned for more information on first submission | 5 | 2 |
| Surgical coordinator hours per appeal | 5.5 (27.5 total) | 5.5 (11.0 total) |
| Surgeon hours per appeal | 1.5 (7.5 total) | 1.5 (3.0 total) |
| Average delay to authorization when returned | 38 days | 38 days on 2 cases |
| Cases abandoned by the patient after a second denial | 1 | 0 to 1 |
| Surgical professional fee at risk on an abandoned case | about $9,400 | |
Add it up: 16.5 coordinator hours and 4.5 surgeon hours returned, roughly 114 days of receivable delay removed across three cases, and a fair chance of keeping one case a year that would otherwise have walked. The delay number is the one owners underweight. An orthognathic case that authorizes in July gets a September surgery date, which the patient's employer and the orthodontist's finish schedule can both live with. The same case authorizing in September slides to a January date that half of patients never take.
Medical necessity is a clinical judgment and it belongs to the surgeon. A whole-file check tells you whether a criterion is documented. It does not tell you whether the case should be done, and no letter should go out over a surgeon's signature that the surgeon has not read line by line. Carriers know the difference between a letter written by the operating surgeon and a letter assembled around them.
Two more limits. Policy bulletins change and get republished, so a check run against last spring's bulletin is worse than useless — pull the current version each time and note the date on your copy. And a chart is protected health information. This belongs on a business account under a business associate agreement with administrative controls, never a personal login, and never in a consumer chat window on someone's phone in the parking lot.
Still reading? Stop comparing — try CallSphere live.
See the dental practice AI agent handle a real call — complete, industry-specific, and live in your browser. No signup.
Finally, the appeal itself. If the first submission is denied on merit rather than paperwork, that is a peer-to-peer conversation between your surgeon and the carrier's dental or medical director. Nothing about 2026 changes who makes that call.
Somewhat, but the payoff is smaller because the pile is smaller. A third-molar case billed to medical turns on a handful of documented findings — pericoronitis episodes, pathology, position on the panoramic — and a good coordinator holds those in her head. The orthognathic and pathology cases are where 340 pages sit and where a missed criterion costs five figures.
It can draft one that quotes the carrier's own criteria and points to the chart page supporting each. That draft is a starting document, not a submission. The surgeon rewrites the clinical reasoning in their own words and signs it. Practices that skip that step end up with letters that all sound identical, which carriers notice.
Which is exactly why the gap list matters. Instead of asking the orthodontic office for "the records," you send three named requests — the hand-wrist film, the twelve-month superimposition, the note documenting the non-surgical trial. Their treatment coordinator can fill a specific request in an afternoon. A vague one sits for two weeks.
Require page references on every criterion and spot-check four of them. If a claim about the chart cannot be traced to a page, treat the whole run as suspect. That habit takes ten minutes and it is the difference between a tool you can rely on and one you cannot.
Take one case you already lost — a denial from the last year where you still have the whole file and the bulletin that applied. Run the criterion-by-criterion check on it and see whether it names the gap the carrier named. That is a one-hour test with a known answer, and it tells you more than any demonstration will.
Everything upstream of that file is still a phone call: the referring dentist's office asking whether you take the patient's medical plan, the patient calling about the surgery date, the pre-op questions the week before. CallSphere builds AI voice and chat agents that answer those calls and the web chat around the clock, book consults, and route referral calls with the details captured, so your surgical coordinator's afternoon stays on the file in front of her.

Written by
Sagar Shankaran· Founder, CallSphere
LinkedInSagar Shankaran is the founder of CallSphere, where he builds production AI voice and chat agents deployed across healthcare, hospitality, real estate, and home services. He writes about agentic AI, LLM engineering, and shipping voice agents that handle real calls in production.
See how AI voice agents work for your industry. Live demo available -- no signup required.
Prior-authorization phone tag drains Madison psychiatry practices. An AI agent absorbs payer callbacks, pharmacy rejections, and patient status-check calls.
A credit union's core agreement plus nine amendments now fits in one question. Find the notice window, the escalators and the exit fees before renewal.
How a 340-page residential chart, the payer policy and the denial letter now go into one question - and what that does to your appeal rate and per diems.
Seven in ten PT referrals are routine. Route comp, PIP and delegated Medicare Advantage to the strong model, and let the denial report draw the line for you.
How an imaging center's prior-auth desk changes when Claude Cowork and ChatGPT Work return finished eviCore and RadMD packets instead of chat answers.
Split the prior-auth queue: fast model for clean continuations, strong model for biologic starts and appeals - with the 91-hours-a-month arithmetic behind it.
© 2026 CallSphere Inc. All rights reserved.
Made within San Francisco
Watch how CallSphere handles real customer calls, schedules appointments, and processes payments — live.
Try Live DemoBook a DemoCalculate Your ROI