- A records request is a deadline, not a conversation. The clock starts on the date of the letter, not the date you opened it.
- Reviewers decide on the documentation in front of them. Care you delivered but cannot evidence did not happen, as far as the review is concerned.
- Most recoupments in behavioral health turn on four things: missing signatures, missing authorizations, cloned notes, and incomplete responses.
- An extrapolated finding turns a sample of twenty claims into a demand covering a whole population. That is where the number stops being survivable.
A Unified Program Integrity Contractor does not call. A letter arrives, addressed to your billing entity, listing patients by name and date range, and asking for the complete medical record supporting the claims you already submitted and were already paid for.
Most programs read that letter as an accusation. It is better understood as a test of one specific capability: can you produce a complete, organised, contemporaneous record on demand? Programs that can, usually keep their money. Programs that cannot, usually do not — regardless of whether the care was excellent.
What the letter actually contains
Additional documentation requests vary by contractor, but the structure is consistent. You will be given a list of claims, identified by beneficiary and date of service. You will be told what documentation is expected. And you will be given a response deadline, commonly thirty days from the date on the letter, with the date on the letter often several days before it reaches you.
The expected documentation is broader than most people assume. For a behavioral health episode it typically includes:
- The assessment or evaluation that established the diagnosis and the level of care.
- The treatment plan, and every subsequent update or review.
- Every progress note for the dates billed, signed and dated by the rendering clinician.
- Documentation of the credential and supervision status of whoever delivered the service.
- Prior authorizations, continued-stay approvals, and any utilization review correspondence.
- For facility claims, the census or bed-day record supporting each day billed.
Note what is not on that list: an explanation. The reviewer is not asking you to argue. They are asking you to produce.
The care was delivered. The notes were written. If the response arrives late, incomplete, or as a pile of loose PDFs, the payer recoups anyway.
The four ways programs lose
1. Missing or late signatures
An unsigned note is not a note. Reviewers routinely disallow services where the documentation was not signed, or where the signature date sits weeks after the date of service in a way that suggests reconstruction. This is the single most preventable finding in the category, and the most common.
2. Authorization gaps
Services delivered outside an authorized window are rarely recoverable on appeal, because the payer's position is not that the care was poor — it is that the care was never approved. If your authorization tracking lives in a spreadsheet that is reconciled monthly, you will find these gaps at the same time the payer does.
3. Cloned documentation
When several notes for the same patient are substantially identical, a reviewer reads it as evidence that the sessions may not have been individualised — or, in the worst reading, that they did not occur. Template language is expected and fine. Identical clinical content is not.
4. An incomplete response
This is the quiet one. A program produces most of the record, misses four notes because nobody signed them, and submits anyway because the deadline is tomorrow. Every claim tied to a missing document is at risk, and the response itself becomes evidence that the record is not reliable.
Extrapolation is where the number gets serious
A review of twenty claims that finds a high error rate does not necessarily end at twenty claims. Where an error rate is high enough, a contractor may extrapolate the finding across a larger universe of claims from the same period, and issue a demand based on the projected overpayment rather than the sampled one.
This is the mechanism that turns a manageable finding into an existential one, and it is the strongest argument for treating documentation completeness as an operational metric rather than a clinical courtesy.
Timeframes, appeal rights, and extrapolation rules differ between Medicare contractors, Medicaid programs, and commercial payers, and they change. Read the letter you actually received and, where the exposure is significant, involve counsel early rather than after the demand arrives.
What a complete response looks like
A reviewer opening your response should be able to find any requested record in under a minute. In practice that means four things:
- A cover sheet naming the requesting payer, your billing entity, the patient and identifier, the date range, and the total record and page counts.
- An index listing every record with its date, type, and page range.
- The records themselves, inline and in chronological order, paginated continuously.
- An explicit note about anything that could not be produced, and why — silence about a gap is worse than disclosing it.
The part nobody budgets for
Ask a program what their last records request cost them and you will usually get a shrug. Ask how many staff days it consumed and the number is rarely under three: someone exporting documents one at a time, someone else chasing signatures, someone numbering pages by hand, and a director reading the whole thing on the last evening.
Those days are not recoverable and they are not billable. They are also entirely a function of how the record was stored, not how the care was delivered — which is the whole argument for building the response capability before the letter arrives rather than after.