Contract-Based Underpayment Recovery
The two directions of a payment error are not mirror images. A payer that identifies an overpayment recovers it by withholding the amount from a later payment — a mechanism built into the relationship, requiring nobody's agreement. A practice that identifies an underpayment has no equivalent. It has to prove the correct amount against a document the payer also holds, and do it inside a period the same payer largely set. The asymmetry is not a grievance; it is the design constraint that decides how recovery has to be organized.
Updated 9 min read
On this page
Key takeaways
- Recovery is an evidentiary problem. The practice must prove the correct amount; the payer, recovering the other way, simply reduces the next payment.
- The proof is the contract and whatever it incorporates — which is why a loaded, validated fee schedule is a precondition for recovery rather than an administrative nicety.
- The hardest part is usually invisible: the rule the payer actually applied is not on the remittance, so the practice is arguing against reasoning it cannot see.
- For an ERISA plan, and where a valid assignment exists, there is a route to that reasoning — a right to the internal rule relied upon, and to the documents relevant to the determination.
- "Relevant" is defined broadly enough to include material generated during the determination whether or not it was relied upon.
- Those limits are real: ERISA plans only, the right belongs to the claimant, and an anti-assignment clause can defeat a provider's standing to use it.
- Recover in populations rather than one claim at a time. A single claim invites a one-off adjustment; a proven pattern invites a configuration fix.
The asymmetry, stated plainly
It is worth setting out because it explains why underpayment recovery feels disproportionately hard, and because the explanation is structural rather than a matter of anyone's attitude.
| Payer recovers an overpayment | Practice recovers an underpayment | |
|---|---|---|
| The mechanism | Reduce a later payment and apply the amount to the debt. Self-executing. | Ask, prove, and wait. There is no equivalent power to withhold. |
| What has to be established first | That there was an overpayment, by the payer's own determination. | The correct amount, from the contract and everything it incorporates, claim by claim. |
| How it becomes visible | It arrives as a smaller deposit — a fact underpayments and overpayments covers as a posting problem. | It does not. The payment posts, the account closes, and nothing on the remittance says the amount was short. |
| The clock | Runs on the payer's rules and its own identification of the error. | Runs on a term in the agreement, frequently starting from the payment date rather than from discovery. |
The last row is the expensive one. A recovery window measured from the payment date means a systemic underpayment found through analysis — which is how they are found — may already be partly out of time on the day it is discovered. That single fact is the argument for a scheduled review rather than an occasional one.
What the proof actually consists of
An underpayment claim is arithmetic plus a document. The arithmetic is straightforward once the expected amount exists; producing the expected amount is where most practices are stuck, and it is not a recovery problem but an upstream one.
The expected rate, derived from the agreement
Not from an average of what this payer usually pays, which proves nothing about what was owed. Fee schedule load and validation is the precondition for everything here: a practice that never loaded the contracted rates has no expected amount to compare against, and therefore no recoverable underpayment — only a suspicion.Everything the agreement incorporates
A rate can be modified by a document the contract names rather than contains. Incorporation by reference is what makes the operative rate potentially different from the one in the exhibit, and it cuts both ways — it is also where a payer's justification for the lower amount may legitimately live.The version in force on the date of service
Rates and referenced documents change. The comparison has to use what applied when the service was furnished, which means the practice needs a dated record of its own fee schedules rather than only the current one.The remittance detail, kept
The claim, the line, the allowed amount and the adjustment reasons as received. A summarized net figure cannot support an argument about a specific line months later, which is one of the reasons a system conversion can quietly foreclose recoveries nobody had got to yet.The rule the payer applied — which is the missing piece
The remittance says an amount and a reason category. It does not say which internal policy produced that amount, and without it the practice is arguing against reasoning it cannot see. The next section is about getting it.
For an ERISA plan, there is a route to the reasoning
This is the part practices most often do not know they have, and it changes an underpayment discussion from an assertion into an examination. Under 29 CFR 2560.503-1 (opens in a new tab), where an internal rule, guideline, protocol or other similar criterion was relied upon in making an adverse determination, the notification must provide either the criterion itself, or a statement that one was relied upon and that a copy will be provided free of charge on request.
And the access right is broader than the rule
Three limits, and they are not small
Running recovery as a program rather than a reaction
Individual recoveries are worth pursuing and rarely worth organizing around. The return comes from treating this as a standing process with a cadence, because the failure mode is a window closing on money nobody had looked for yet.
- Recover in populations, not in claims. One claim invites a one-off adjustment and leaves the cause in place. A proven pattern — same code, same term, same payer, over a period — invites a configuration fix and a batch reprocessing, which is the outcome worth the effort.
- Find the recovery term before you need it. How long the agreement gives you, and — the part that matters more — what starts the clock. A window running from the payment date behaves very differently from one running from discovery, and only one of them survives contact with an analysis-driven finding.
- Check whether the term is reciprocal. Agreements sometimes give each side a different period to raise a payment error. That is a real term with real consequences and it is worth knowing which way yours runs, because it is also a negotiable one at renewal.
- Route it correctly. A payer applying its published policy exactly as the contract permits is not underpaying, however unwelcome the result — that is a policy argument. Where the payment contradicts the agreement, payer contract dispute escalation covers the ladder it belongs on and the clause that decides the route.
- Keep the claim ladder running. A recovery discussion tolls nothing. Appeal deadlines on the individual claims keep running while the larger argument proceeds, and the corpus records repeatedly that more is lost to pausing appeals during a promising conversation than to conversations that fail.
The cheapest recovery is the one that never has to happen
Common questions
Why can the payer just take money back when we cannot?
Because recoupment operates by reducing a payment the payer is about to make, and the practice has no comparable lever — there is no next payment flowing the other way to reduce. It is a structural asymmetry rather than a policy choice by anyone, and it has a practical consequence worth accepting rather than resenting: recovery has to be built as an evidentiary process, with the expected amount established in advance and the review running on a schedule, because none of it happens automatically the way recoupment does.
How do we prove the correct amount?
From the agreement and everything it incorporates, using the version in force on the date of service — not from an average of what the payer usually pays, which establishes only what has been happening. This is why loading and validating the fee schedule is a precondition rather than an administrative task: without a contracted expected amount, a practice has a suspicion rather than a recoverable underpayment. Keeping dated copies of superseded schedules matters for the same reason, because the comparison has to be against the rate that applied then.
Can we make a payer tell us what rule it applied?
For an ERISA plan, and where the practice holds a valid assignment, there is a real route. Where an internal rule, guideline, protocol or similar criterion was relied upon in an adverse determination, the notice must give either that criterion or a statement that one was relied upon with a copy available free of charge on request. Beyond that, the claimant may request reasonable access to and copies of all documents relevant to the claim, and relevance is defined broadly enough to include material generated during the determination whether or not it was relied upon. The limits matter: it applies to ERISA plans, the right is the claimant's and reaches a provider only by assignment — which an anti-assignment clause can defeat — and it produces documents rather than payment.
Is it worth pursuing a single underpaid claim?
Sometimes, and it is rarely where the value is. A single claim usually produces a one-off adjustment that leaves the cause untouched, so the same shortfall recurs on the next batch. A demonstrated pattern — the same code, the same contract term, the same payer, across a period — can produce a configuration change and a reprocessing of the affected population, which is worth an order of magnitude more for roughly the same analytical work. The practical rule is to work the pattern and use individual claims as its evidence.
How long do we have?
That is a term in your agreement rather than a general rule, and no honest number could be given here — periods vary by contract, sometimes differ for each party within the same contract, and interact with state law. What is worth establishing today, before anything is in dispute, is two things: how long the period is, and what starts it. A window running from the payment date rather than from discovery is the one that hurts, because a systemic underpayment surfaced by analysis may already be partly out of time on the day you find it — which is the whole argument for reviewing on a schedule rather than when something feels wrong.
Key terms in this article
Defined once, on their own pages.
Continue learning
Where the variance is found, and the route an argument takes.
Underpayments and Overpayments
The variance itself — why both hide, and how a recoupment appears on a deposit.
Zero Balance Review
How a shortfall absorbed into an adjustment is found after the account has closed.
Fee Schedule Load and Validation
The precondition — without a loaded expected rate there is nothing to recover against.
Payer Contract Dispute Escalation
Where an argument goes once the payment contradicts the agreement.
Contractual Variance Calculator
Compare expected against paid, from your own contracted rates.
Authoritative sources
- 29 CFR § 2560.503-1 — Claims procedure (ERISA employee benefit plans) (opens in a new tab)
Sets the requirements for benefit claims procedures under ERISA plans. Where an internal rule, guideline, protocol, or other similar criterion was relied upon in making an adverse benefit determination, the notification must provide either the specific criterion or a statement that one was relied upon and that a copy will be provided free of charge to the claimant on request. The claimant is separately entitled, upon request and free of charge, to reasonable access to and copies of all documents, records, and other information relevant to the claim. Paragraph (m)(8) defines relevance to include material that was relied upon in making the determination; that was submitted, considered, or generated in the course of making it without regard to whether it was relied upon; that demonstrates compliance with the administrative processes and safeguards required by the section; or, for a group health plan or a plan providing disability benefits, that constitutes a statement of policy or guidance concerning the denied treatment option or benefit for the claimant's diagnosis, without regard to whether it was relied upon.