Medical bill with no CPT code: where bill review stops
A medical bill with no procedure code on it cannot be reviewed. No code means no fee schedule allowance and no usual, customary and reasonable benchmark, and therefore no number an expert can defend. This is the most common reason a bill review case stalls, and it is almost never the reason anyone gives.
For eight years I prepared New Jersey PIP arbitration files as a senior paralegal, and for three of those years I also built the billing analyses that medical billing expert reports are written from. Both jobs are the same job underneath: turn a stack of documents into one sheet of numbers somebody is willing to sign. This is about the thing that stops that sheet more often than anything else.
The bill that stops everything
The most common blocker in a medical bill review is a bill that names a procedure but never states its code. A line reads as a description in words, with a charge next to it, and nothing that ties it to the coding system every benchmark is organised around. The file is complete in every other respect and the review still cannot start.
What follows is a stretch of back and forth. You ask the referring attorney for a coded bill. The attorney asks the provider. The provider sends the same document again, or sends a ledger instead, or sends nothing. Meanwhile the case sits. In the worst version, the coded bill does not exist in any form anybody can produce, and the honest answer is that the service cannot be reviewed at all.
Turning work away over a missing code is a real outcome and it happened. It is also a strange one to accept, because the provider certainly knew what it billed for. The code exists. It is just not on the paper in front of you.
Why a missing code is not a small gap
Every benchmark in a bill review is indexed by procedure code. A New Jersey fee schedule allowance is looked up by code. A usual, customary and reasonable figure is compiled by code. Without the code there is no allowance and no benchmark, so there is no defensible number, and an opinion with no defensible number underneath it is not an opinion anybody should sign.
It is worth being precise about how those two benchmarks relate, because they are often described as alternatives and they are not. Where the New Jersey automobile medical fee schedule carries the code, the schedule governs, and that is not a matter of preference. Anything the schedule does not reach falls to usual, customary and reasonable pricing instead. That means a missing code does not close one door and leave the other open. It closes both, because both are keyed to the same thing.
Once the code is in hand, the lookup itself is mechanical, which is exactly why the NJFS and UCR lookup returns the correct New Jersey allowance for a given code. The hard part was never the lookup.
Where the code actually is
Facilities publish their own charges, and those published files carry the procedure description and the procedure code side by side. Matching the description on an uncoded bill against that facility’s own published file recovers the code from the provider’s own document rather than from anybody’s judgment.
This is the same class of file that hospitals post to satisfy price transparency requirements, and most people who know about those files think of them as a benchmarking resource. They are also an authority on what a given facility calls a given service, which is a completely different use of the same document. The mechanics of finding one, including the standard address that locates it, are in the walkthrough of hospital price transparency files.
A facility that publishes a charge file has, in effect, published a dictionary translating its own wording into codes. When its bill says one of those things in words, the file says what code it uses for it. That is not an inference about clinical intent. It is reading the provider’s own mapping.
The rule: prove it, or do not use it
A recovered code is only usable when it can be proved from that provider’s own official material: its published charge file, its own definitions, its own documentation. Guessing is not permitted and neither is a close-enough match from a generic crosswalk. Where no provider document establishes the code, the line does not get a number and the report says so.
The reason for the rule is the other side of the case. A code that was inferred by judgment is exactly the thing a defence expert should attack, and they would be right to. A code read out of the billing provider’s own published document is a different proposition entirely, because the provider is the authority on what the provider billed. Nobody gets to argue that a facility does not know its own charge description.
That is also why this is stated as a refusal rule rather than a success rate. There is no percentage to quote here, and I am not going to invent one. Either the provider published something that establishes the code, or it did not, and the second case is a line that cannot be reviewed.
What arrives that does not help: one service, told three ways
Sending more documents for the same service makes the review slower, not faster. One surgery routinely arrives three times: as a HCFA-1500, as a provider ledger, and as an explanation of benefits. Each is legitimate, each says something the others do not, and none of them can simply be discarded, so all three have to be reconciled into one line.
| Document | What it establishes | What it does not |
|---|---|---|
| HCFA-1500 | What the provider billed: codes, units, charges | What was paid, and why |
| Provider ledger | Running balance, adjustments, patient responsibility | Often the codes themselves |
| Explanation of benefits | What the carrier allowed, paid and denied | The provider's own description of the service |
Scroll the table sideways on a phone.
The work this creates is not arithmetic. It is deciding that three differently worded records describe one event, which means reconciling provider names that are written differently on each document, and categories that do not line up, before any dollar figure is even in play. Counting the service three times overstates the claim. Dropping two of the three throws away the payment history or the codes.
So the ideal delivery is one clean coded set per provider, and everything else supplied as backup rather than as more bills. Sending every version of every document feels thorough. It is the expensive option.
What actually predicts how long a review takes
How the file arrives predicts the timeline far better than how large it is. A case delivered as 200 unsorted PDFs requires somebody to look at every page and decide, one page at a time, whether it is a bill or a record. The same page count arriving with the bills already grouped skips that step completely.
Separating the billing pages out of a mixed stack is the least skilled and most time-consuming part of the whole engagement, and it is entirely created by the way the file was packaged. Nobody enjoys it, nothing about it requires expertise, and it happens before any analysis can begin. Where the rest of that time goes once the sheet is under way is the subject of the four days before the opinion.
Which produces the single most useful thing a referring attorney can do, and it costs nothing: send the bills as their own set. That one act removes the largest block of unskilled time in the engagement. It also happens to remove work that we would otherwise be doing, which is worth saying out loud rather than leaving for you to notice.
What we built for the part that remains
Three things survive even a perfectly packaged file: finding the billing pages, transcribing every charge line without error, and tying each line back to the page it came from. Those are the three the engine covers, and none of them involves a model deciding what a charge should have been.
The bills are located and separated from the records automatically, the charge lines are extracted into the sheet, and every row carries a link back to the exact PDF and the exact page the figure was read from, so a reviewer checking a number lands on its source rather than searching for it. A newer addition links a billing line to the related medical record where one exists, which is the difference between verifying medical necessity in a click and reading a stack. That pipeline is what Expert Report Engine turns scanned medical bills into defensible damages describes, and the drafting layer on top of it is Expert Report Coder, which builds the report from a coder’s own notes. Cases that arrive as an unordered pile of records first go through RecordFlow, which puts the record set in date-of-service order.
Where this does not help
Recovering a code does not make a charge reasonable, and it does not decide medical necessity. It restores the ability to ask those questions. A line whose code has been established from the provider’s own file still has to be priced, compared and argued, and the expert who signs the report is the one who does that.
This also is not a coding service and it is not legal advice. I prepared these files as a paralegal, not as an attorney and not as a certified coder, and what is described here is a documentation practice for review work rather than a recommendation about assigning codes to claims. Where the provider has published nothing that establishes a code, there is no version of this that produces a defensible number.
Where objective evidence in the records is the question rather than the charges, the relevant compilation is ROM Reports, which gathers every range-of-motion measurement in the record. And where the same billing sheet is headed for a New Jersey PIP arbitration rather than an expert report, the filing side of that work is described in how to file a NJ PIP arbitration demand on Forthright.
Frequently asked questions
What do you do when a medical bill has no CPT code?
A bill with no CPT code has to have the code recovered from a source the provider itself published, most often the machine-readable charge file a facility posts on its own website, which lists the procedure description alongside the code the facility uses for it. Matching the description on the bill to that file recovers the code from the provider's own authority rather than from a guess.
Why can't a medical bill be reviewed without a CPT code?
Every benchmark in a bill review is indexed by procedure code. A fee schedule allowance is looked up by code, and a usual, customary and reasonable benchmark is compiled by code. Without a code there is no allowance and no benchmark, which means there is no number an expert can defend and nothing for an opinion to rest on.
Is it acceptable to infer a CPT code from a procedure description?
Inference alone is not acceptable, because a code that was guessed can be attacked by anyone on the other side and should be. What is defensible is reading the code out of the provider's own published document, where that provider maps its own description to its own code. The provider is the authority on what the provider billed. Where no such document exists, the correct answer is to say the line cannot be reviewed.
What is the difference between the fee schedule and UCR?
The New Jersey automobile medical fee schedule governs wherever it carries the code, and it is not optional in those cases. Usual, customary and reasonable pricing is the fallback for services the schedule does not cover. Both are indexed by procedure code, so a missing code blocks both routes equally.
What slows down a medical bill review the most?
How the file arrives, not how large it is. A case delivered as 200 unsorted PDFs requires every page to be examined before the billing pages can be separated from the records, while the same page count delivered with the bills already grouped skips that step entirely. Sending the bills as their own set is the single cheapest thing a referring attorney can do to shorten a review.
Burak Tamac spent eight years as a senior paralegal at a New Jersey personal-injury firm preparing PIP arbitration files, and three of those years building the billing analyses behind medical billing expert reports. He now builds document-processing systems for law firms and expert practices at S2Reason. He holds a Ph.D. in political science and is an adjunct professor at Montclair State University. He is not an attorney.