The Paper Trail an Authorization Leaves in the Chart

The authorization paper trail at most imaging centers lives in three places at once. A fax confirmation in a shared folder, a number written into an appointment note, and somebody’s memory of the phone call. All three are fine until the day the claim needs the number and the person who took the call is on vacation.

This is not a filing preference. The approval is the thing that makes the scan payable, and the proof of it has to be attached to the patient rather than to a workflow that has since moved on.

Imaging makes it worse than most settings because the volume is high and the episode is short. A patient is authorized, scanned, and gone inside a couple of weeks, and the front office has already moved to the next hundred studies.

Then a claim comes back needing the reference number, or a patient returns for a follow-up study and nobody can tell whether the earlier approval covers it. The work was done. Finding the evidence of it is now its own project.

The number has a shelf life, and the shelf life is the reason to file it

Newer review programs attach an explicit identifier and an explicit expiry to the answer. Under the CMS WISeR model, once a prior authorization request is approved, the unique tracking number associated with it is valid for 120 calendar days starting from the decision date. CMS spells the arithmetic out: a request affirmed on January 5, 2026 covers dates of service through May 4, 2026, and after that the practice needs a new request and a new tracking number.

When multiple affirmed services will be billed on the same claim in a hospital outpatient department setting, CMS requires them to have been included on the same request and to carry the same tracking number.

So the artifact you are filing is not a confirmation that something was approved. It is an identifier, a date range, and a scope, and all three have to travel with the claim.

A number written in an appointment note does not travel. It stays attached to an appointment that may get cancelled, rebooked, or absorbed into a different order.

Where the evidence should actually live on athenaOne

There are two records here and they do different jobs. The authorization record holds the structured facts, the request, the plan, the service, and the current status. The administrative document holds the artifact itself, the payer’s letter or the fax confirmation that a human will want to look at when a claim is questioned.

Both of these records are stored against the patient rather than against the appointment. That single choice fixes most of the retrieval problem, because a patient chart survives every reschedule and every cancelled order.

The filing has to be predictable enough to search. A document class that means what it says, a title that carries the service and the date range rather than fax_2026_09_02, and the identifier in the record where the billing team looks rather than in a note somebody wrote for their own benefit.

What the automation does is unglamorous and it is exactly the part humans skip when the day is busy. It takes the inbound approval, files it as an administrative document against the right patient, writes the identifier and the validity window into the authorization record, and confirms the document actually landed rather than assuming it did.

The rescheduling problem, which is where most trails go cold

Rescheduling silently breaks the link between an appointment and its authorization. The new appointment is not attached to the existing approval, and an appointment whose authorization has not come back can only be pushed later rather than pulled earlier.

At an imaging center that happens constantly and for ordinary reasons. A patient reschedules around work. A scanner goes down and half a day moves. A referring office changes the order after the patient is already booked.

Each of those events leaves the approval intact and the association broken. The paperwork is still in the chart and nothing points at it from the new date, so at checkout the study looks unauthorized.

The fix is to treat the reschedule as a trigger rather than an update. When an appointment moves, the automation re-reads the authorization record for that patient, checks whether the new date still falls inside the validity window, and re-attaches the reference. When the new date falls outside the window, that is not something to resolve quietly. It goes to a person with the old date, the new date, the expiry, and the next in-window opening already attached.

Why this reads as a denial problem in your reports

Filing failures do not report as filing failures. A January 6, 2026, MGMA Stat poll found the biggest revenue cycle leaks for practices today are denials and appeals at 48%, followed by front end issues at 23%, billing and collections at 14%, coding at 13%, and charge posting at 2%.

A study that was properly authorized and cannot prove it lands in the first bucket. Somebody works it as an appeal, finds the approval eventually, and resubmits. The queue records a denial recovered rather than a document that was never filed.

That is worth naming because it changes what you fix. If the reports say denials, the instinct is to staff appeals. If the underlying cause is that the evidence was never attached to the patient, more appeals capacity just processes the same avoidable work faster.

A cheap test: pull twenty recently appealed imaging claims and count how many were authorized all along. If most of them were, the problem is upstream of the appeal.

The line the automation does not cross

Everything above is document handling and record keeping. It is worth being explicit about what it is not.

The automation does not read the clinical content of what it files, does not judge whether the study is warranted, and does not assemble or evaluate the supporting material a payer is reviewing. Those questions belong to the ordering clinician and to the staff who support them.

Its scope is the logistics: which patient, which service, which identifier, which dates, filed where the next person will look. Moving paper and keeping the index honest.

When something falls outside that, the handoff carries context rather than an alert. Not a document needs review, but this patient’s approval expires in nine days, the study is booked for the fourteenth, and here are the two openings inside the window.

What to ask before you buy any of this

The demo version of authorization tooling is a dashboard with a green checkmark. The version that matters is retrieval six months later.

Ask to see a patient chart in a live athenaOne environment where an approval was filed by the tool, and find the identifier and the date range without help. If it takes more than a few seconds, staff under pressure will not find it either.

Then ask what the tool does when an appointment moves. Re-checking the pair on every reschedule is the single behavior that separates a filing system from a filing habit.

That kind of joined-up reading needs real platform depth. PGA works across 440+ of athenahealth’s roughly 800 endpoints, which is what makes it possible to hold the authorization record, the administrative document, and the appointment in one view instead of three.

Key Takeaways

File the approval against the patient rather than the appointment, because the chart survives every reschedule and cancelled order.
Store the identifier and the validity window as structured facts, not as free text in an appointment note nobody else will read.
Title administrative documents so they can be found by service and date range, since retrieval months later is the whole point.
Treat every reschedule as a trigger to re-read the authorization and re-attach the reference, because moving an appointment breaks the link silently.
Audit twenty recently appealed imaging claims for how many were authorized all along, which tells you whether the problem is appeals capacity or filing.
Keep the automation on moving and indexing paper, and leave every question about whether the study is warranted with the ordering clinician.

An authorization you cannot produce is worth about as much as one you never got. File it against the patient, keep the identifier and the window as real fields, re-attach the link every time the appointment moves, and test the whole thing by trying to find one six months old.

Sources