Medical records and billing review for injury and liability claims
Does this charge involve the injury complained of?
Etiology answers that question for every charge on the bill: from the records, charge by charge, with the record page cited under every call. It fits workers' compensation compensability, liability and medical-malpractice causation, and subrogation review.
Records are shared through a secure, access-controlled channel, set up before any file moves.
| Charge | Determination |
|---|---|
| Open reduction and internal fixation, right distal radius$18,442.00 | InvolvesOperative note, p. 04317 |
| Screening colonoscopy$2,915.00 | Does not involveOrder and consent, p. 11206 |
| Inpatient pharmacy, second admission$6,120.44 | MixedResolved by itemization: per-dose MAR detail, p. 07993 |
The problem with volume
A decision no one can check is a decision made on an estimate.
A single claim file can run to thousands of pages of medical records. Claims professionals, coverage counsel, and those with an oversight duty cannot read all of it, and the number of files makes that worse rather than better. So the question of whether a charge belongs to the injury complained of gets answered by inference, and the answer is only as good as the time available to form it.
Etiology reads the record and cites it: every charge classified against the injury complained of, every call anchored to the page it rests on, so a reader can pull that page and check it.
The decision stays with the payer. What changes is that it rests on the record instead of on an estimate of what the record probably says.
The firm
Built to be checked, not taken on faith.
Etiology is a Louisiana records-review firm, founded in 2026 and based in Metairie. It was started by a healthcare lawyer with over twenty-five years of experience defending providers in professional liability actions. That work is where the method comes from. What Etiology delivers does not depend on it. The method is written down, applied the same way to every file, and checkable against the record, which is the only thing an adjuster has to trust. Claim after claim, one pattern held: the charges that involve the claimed injury and the charges that do not are separable from the record itself, and adjusters were being asked to make that separation while buried in thousands of pages. Etiology was built to do that read. It is a records-review firm, not a law firm, and provides no legal advice.
Every charge gets two answers: does it involve the injury complained of, and would the claimant have needed the care anyway. Each call cites the record page it rests on, so any reader can pull the record and see what we saw. The read runs on Anthropic's Claude models under zero data retention terms, human-supervised under a written protocol authored by the firm's founding healthcare attorney, and independently verified before delivery. No client record trains any model. There is no valuation, no settlement recommendation, and no legal opinion. Etiology supplies the factual input, verified and checkable, and stops there.
Every record Etiology receives is confidential medical information, and the review environment is built for that. The obligation comes from the engagement contract and from professional duty. HIPAA reaches an engagement only where the client is itself a covered entity, it does not reach liability, malpractice or state-fund work, and where it does, a Business Associate Agreement governs as well.
The test
A chart question, answered from the chart.
Identify the injury complained of. Then ask, of each charge: does it involve treatment of that injury? The answer is in the records, and each determination cites the page it rests on.
We also answer the question every payer has to ask next: would the claimant have needed this care anyway? Each charge carries both calls, the chart determination and the causation screen, and a charge the record genuinely leaves in dispute is called GRAY and flagged with what would resolve it, never forced into a bucket.
The analysis runs on Anthropic's Claude models under zero data retention terms: nothing a client sends is kept by the model after the review, and no client record is ever used to train any model. The models operate under a written clinical and records protocol authored by the firm's founding healthcare attorney, who spent over twenty-five years defending providers in professional liability actions. The work is human-supervised, and every citation, subtotal, and determination is independently verified against published clinical references before delivery. None of it rests on trusting the reviewer.
Method
Built for productions that arrive in pieces.
Hospital records arrive as thousands of pages across many overlapping deliveries. The method exists so that nothing, and no page, falls between them.
- 01
Inventory
Every production logged as it arrives: source, delivery date, page count, format. Nothing is reviewed until everything is accounted for.
- 02
Working set
One sequentially labeled set built from all productions, page labels verified four independent ways. Duplicates are proven before they are excluded, never assumed.
- 03
Scanned-page sweep
Fifteen to twenty-five percent of a hospital production is scanned images with no searchable text. Keyword search passes over those pages silently. We render and read them.
- 04
Page-to-page read
The record is read against a coverage ledger: every operative note, discharge summary, H&P, radiology and pathology report, MAR, nursing flowsheet, and physician order inspected visually.
- 05
Classification
Each charge resolves to one of three calls, with the record page cited under every call. A MIXED call names the itemization that would resolve it, we do not guess percentages.
- 06
Verification
Independent verification of every citation, every subtotal, and scope compliance before anything is delivered.
From delivered reviews
Client matters, by the numbers.
Three findings, each computed from the delivered files: the time the read takes by hand, the demand set against the record, and the difference between the two.
The records, and the read
- 315
- medical record productions reviewed
- 99,229
- pages of medical records received: 95,492 unique after duplicates were proven
- about 1,985 Hours
- for one reviewer to read all 99,229 pages by hand, at 50 pages an hour for clean records
- about 248 Days
- of page-by-page reading, at eight hours a day
- 1 to 12 business days
- the published Etiology turnaround, set by the size of the file
The demand against the record
3,625 charges classified: $5,411,472.36 in billed line items, each call cited to the record page it rests on. That is $1,544,781.39 more than those demands itemized: the three call columns below divide every charge the review located in the productions, which is not the same pool as the schedule a claimant asserts. Demanded and classified are set out separately for that reason: the three call columns add up to Classified, never to Demanded. One matter's claimant filed two damages schedules that disagree; the lower is carried in every figure on this site until that is resolved, so no number here rests on the larger of their own two versions. The Demanded and Difference figures cover the 13 of 14 matters whose demand the claimant stated as a single number. One matter is excluded from those two columns and included in the call columns, because the claimant produced more than one damages schedule and the review declined to adopt either as the demand.
Past medicals
- Demanded‡
- $3,923,886.64
- Related
- $372,114.11
- Mixed
- $46,172.01
- Gray — open
- $3,695,720.47
- Not related
- $1,297,465.77
- Classified
- $5,411,472.36
Future care, as demanded†
- Demanded‡
- $4,584,853.06
- Related
- $160,000.00
- Mixed
- $19,896.00
- Gray — open
- $4,374,957.06
- Not related
- $30,000.00
- Classified
- $4,584,853.06
- Related: the record ties the charge to the injury complained of.
- Gray: the record genuinely leaves the charge open; every GRAY call names what would resolve it.
- Mixed: the provider billed two different kinds of care as a single line, so the charge has not reached the causation screen at all — there is nothing yet to screen. Not a fourth answer to the causation question but an answer to a different one. Every MIXED group names the itemization that would separate it; we do not guess percentages.
- Not related: care unconnected to the injury claimed, or care the record shows the claimant would have needed anyway; resolved outright.
‡ Claimant demanded will not equal the three calls beside it, and the gap is a finding rather than an error. Demanded is what a claimant's schedule asserts on a pleading page. The call columns divide what the review actually located in the billing productions, so they add to Classified, every time, and to Demanded, never. Across this ledger the two pools separate by $1,544,781.39 of billed charges the review located in the productions that those demands never itemized, which carries Classified above Demanded; $57,195.67 by which a stated demand exceeds the sum of its own line items, an addition error in the claimants' own schedules, which carries Classified below Demanded; and one matter whose charges are classified in full but whose demand is carried at nothing: on one, no demand has been made and no petition filed, so there is no figure anywhere to test. Future care separates for a further reason: a demand can assert care that has no billed line behind it yet, so there is nothing to classify against it.
† Future figures are shown as demanded, which is normally billed charges. A payer does not pay billed charges for future care: medical items are priced on the workers compensation fee schedule, in practice roughly sixty percent below billed, then capped at a ten-year look-out, with the life-expectancy figure carried alongside as the upper bracket. Future attendant care runs on a different basis again: a sitter rate, and no payment for hours the claimant is asleep. A life-expectancy bracket is assessed from the claimant's own documented morbidity, never from an actuarial table. Where a demand lumps its future care, the productions are read for an actual price on each item. What each file turned up under this method is set out below.
The difference
Claimants demanded a total. Read against the record, most of that demand is either not tied to the injury complained of or cannot be settled either way on the records as they stand.
- $532,114.11
- of the medicals demanded is tied to the injury complained of by the record, charge by charge, each call cited to the page it rests on
- $7,976,625.59
- is the rest: medicals demanded, past and future, that the record does not tie to the injury. This is a finding about the record, not a figure to pay or withhold.
- $8,070,677.53
- of that rest is GRAY, which is not money saved. It is money still in play: the record settles it neither way, so the adjuster has a position to argue.
The difference is the demand less what the analysis identified as related: a factual finding, cited to the record, not a recommendation on what to pay.
GRAY is the part of that difference the record leaves genuinely open. It is not a saving and not a result; it is the range the adjuster negotiates in, and each GRAY call names what would resolve it. Nothing is forced closed to make the difference larger.
A future item can screen related and still carry no dollar figure, because the demand never priced it separately. No figure is derived where the demand states none.
The past difference accounts to the penny
- $1,297,465.77the analysis identified as not related
- $3,695,720.47the analysis called GRAY: open on the record as it stands, and still in play
- $46,172.01the analysis called MIXED: billed as one line covering two kinds of care, and unscreenable until the bill is separated
- $57,195.67by which a stated demand exceeds the sum of its own line items: an addition error found in the claimants' own schedules
- less $1,544,781.39of billed charges the review located in the productions that the demands never itemized
What each file turned up
One entry per completed review, most recent first. Each names what was distinctive about that analysis and the figure it rests on. No client, claimant, provider, or date of service appears in these entries.
- Matter 152026-09
Almost all of it treated the injury. Almost none of it is related.
- Not related: $102,817.44
- Care that treated the very thing complained of and still screens not related, because the second question asks whether the claimant would have needed it anyway. Imaging found the joint already worn through to bone before the event.
- Gray, open: $2,783.00
- Charges in encounters that treat the injury and a separate ongoing complaint at once, where the record does not divide them.Resolves on an itemization that splits the combined visits
- Matter 142026-09
The largest line on the claim is contradicted by its own invoice.
- Not related: $12,951.93
- The elective cosmetic operation the claimant chose, plus travel. Nearly 35 percent of the past claim, and neither is treatment of an injury. The operation is billed at full price while the practice's own invoice closes it at a balance due of zero.
- Gray, open: $9,919.46
- Charges the record cannot resolve either way. The reason is structural: every records request behind this production was written for service dates from the index year forward, so no absence in the file can prove a competing cause did not exist.Resolves on records predating the index date from every treating provider, and in particular any prior eye care
- Mixed, unscreenable: $5,729.00
- Charges inside encounters that involve the injury and other care at once, so they cannot take a causation screen until the bill is separated. A classification outcome, not a fourth causation answer.Resolves on itemized bills from the payees that produced none, and the claimant's explanations of benefit
- Related: $8,553.40
- Screened as related in the main, being the specialist eye care that follows the event complained of.
- Matter 132026-09
The most certain money in this file is the part nobody claimed.
- Not related: $275,858.29
- Care delivered before the event complained of, across eight hospital encounters in the preceding eight and a half months, six of them for heart failure. It cannot have been caused by something that had not happened yet. Almost none of it is claimed.
- Gray, open: $152,460.30
- Twenty-nine rows, and 89 percent of what is actually claimed. The claim covers the days between the fall and the death, which is exactly where the record supports both readings: three fractures needing no operation, in an admission treating four other things.Resolves on the treating hospital's daily physician progress notes and the discharge summary's own statement of what extended the stay
- Related: $15,192.00
- Screened as related to the injury complained of on the face of the record.
- Matter 122026-09
Half the claim is billed as one line covering two kinds of care.
- Mixed, unscreenable: $40,443.01
- More than half the classified total. Lines the hospital billed as one entry covering two kinds of care, so they cannot take a causation screen until the bill is separated. Not a fourth causation answer; it answers a different question.Resolves on an itemization that splits the combined lines
- Related: $33,753.02
- Screened as related to the injury complained of. The counterfactual is not a shorter admission but no admission: booked as outpatient surgery, with the discharge order written four minutes after extubation.
- Not related: $3,638.97
- Care the record shows would have been needed regardless of the event complained of.
- Gray, open: $1,092.00
- The causation screen's own third value, where the record genuinely supports both readings. On two of the eight gray items no further document will resolve it.Resolves on the operating surgeon's full note for the catheter attempt, which the production does not hold; on two of eight items nothing will
10 earlier matters carry notes in the same form.
Page accounting
One review recovered a decisive report that keyword search had missed: it existed only as a scanned image, with no searchable text. That page is why the scanned-page sweep is mandatory.
Illustrative and de-identified. Fictitious names, dates, providers, and figures.
Why Etiology
The case for sending it out.
A payer can put an adjuster or a reviewer on the file and reach the same calls. The question is what that read costs in time, and what the calls rest on when someone challenges them. Three things separate the Etiology read from doing it by hand.
Time
A single claim can arrive as thousands of pages across overlapping productions. One matter on this page ran to 18,612 pages, 18,612 of them unique once duplicates were proven — and every one of the rest had to be read to prove it. Careful record review is cited at roughly 50 pages an hour for clean records and 20 for handwritten or scanned ones. Even at the faster pace, that read runs past 372 reviewer-hours, about 9 forty-hour weeks for one reviewer, and it runs longer because the 15 to 25 percent that arrives as scanned images returns nothing to a keyword search and has to be read on sight at the slower rate. Etiology does the full page-by-page read and returns the finished chart, so the file is not sitting in a queue while the pages are turned one at a time.
Defensibility
Every call names the record page it rests on. An adjuster, a supervisor, or opposing counsel can pull that page and see what the reviewer saw. A determination built that way does not depend on trusting the reviewer. It is reproducible from the record, so the payer can stand behind the decision it informs and show the basis for it on demand. Every determination, the page it cites, and the record of its independent verification are producible, and the written protocol they follow is documented, so the payer can show how a call was reached, not only that it was made.
Gray areas
Some charges are genuinely in dispute, and Etiology does not force them into a clean answer. A contestable charge is called GRAY, set against the published clinical reference that bears on it, with the itemization or record that would resolve it named. The payer argues from the clinical literature instead of guessing, and knows exactly what is unsettled and why.
Pricing
A flat fee, by the page, published.
One flat fee per claim file, set by the size of the record. You are quoted up front, from the production you send, and the fee is figured on unique pages, the count left after duplicates are proven and removed, not the raw page count. Proving duplicates can move a file to a lower tier, never a higher one. Every fee covers the four work products: the causation analysis that classifies each charge on both determinations, the medical chronology, and the past and future medical analyses, which classify care against the record without putting a value on it.
| Records volume, unique pages | Flat fee | Turnaround |
|---|---|---|
| Up to 500 | $2,750 | 1 business day |
| 501 to 1,500 | $5,500 | 3 business days |
| 1,501 to 3,000 | $8,000 | 5 business days |
| 3,001 to 5,000 | $10,000 | 8 business days |
| 5,001 to 10,000 | $15,000 | 12 business days |
| Over 10,000 | $1.75 per unique page | Quoted with the fee |
| Rush handling, any tier | Flat fee plus 50 percent | Half the committed turnaround |
Turnaround runs from the point the complete records are in hand, and can vary with the volume and condition of the production. Rush handling is agreed at intake, when Etiology confirms the compressed turnaround is achievable; on a first-tier file, rush means same-business-day delivery, measured in hours. High-volume and ongoing programs are quoted at a program rate agreed in advance. Unusually complex productions are noted and agreed before work begins, never billed after. A first engagement fits the first tier: one file, one business day, every call cited to a page you can pull. Schedule effective August 2026 and subject to change; your engagement letter controls.
Common questions
Asked and answered.
- What does Etiology do?
- Etiology reviews medical records and billing for insurance claims payers and classifies every charge on the bill. Each charge gets two determinations: a chart determination (does the charge involve the injury complained of) and a causation screen (would the claimant have needed the care anyway), with the record page cited under each call. The work fits workers' compensation compensability, liability and medical-malpractice causation, and subrogation review.
- What does a review cost?
- A flat fee per claim file, set by the number of unique pages in the record after duplicates are proven: $2,750 up to 500 pages, $5,500 for 501 to 1,500, $8,000 for 1,501 to 3,000, $10,000 for 3,001 to 5,000, and $15,000 for 5,001 to 10,000. Files over 10,000 pages are quoted from the production at $1.75 per unique page. Rush handling is half the committed turnaround for an additional 50 percent of the flat fee. You are quoted up front from the production you send.
- How fast is turnaround?
- 1 business day for files up to 500 unique pages, 3 business days up to 1,500, 5 business days up to 3,000, 8 business days up to 5,000, and 12 business days up to 10,000. Turnaround runs from the point the complete records are in hand.
- Is there a sample review?
- Yes. A de-identified sample review is available to download as a PDF. Its names, dates, providers, and figures are fictitious and illustrative.
What we don't do
We answer one question.
No settlement valuation. No recommended authority or reserves. No opinions on what a payer owes, and no recommendations on what to pay.
The boundary is deliberate. The adjuster applies the law; Etiology supplies the factual input: determinations an adjuster can rely on precisely because they decide nothing else.
Start with the file.
Send the claimed schedule and the record productions. What comes back is the determination chart: every charge classified, every call cited to the page. If the file is unusual or the deadline is tight, say so in the first email and we will tell you plainly whether we can meet it.
Security and confidentiality
Every record is confidential medical information.
Records are handled that way from transfer through delivery. What follows is how, in terms a risk officer can check, not a seal to take on faith.
Confidentiality, and where it comes from
Etiology is not itself a covered entity, so HIPAA reaches an engagement only where the client is one, and there a Business Associate Agreement is executed before any record changes hands. On liability, malpractice and state-fund work it does not reach, and the duty runs instead from the engagement contract, from professional obligation, and from Louisiana's breach-notification law. The handling standard is the same either way. The engagement serves one purpose, the review, and the data is used for nothing else.
Named AI subprocessor, zero data retention, no training
The analysis runs on Anthropic's Claude models under zero data retention terms: nothing a client sends is kept by the model after the review, and no client record is ever used to train any model. Anthropic is the only AI provider in the pipeline.
Encrypted transfer through a controlled channel
Records move through an established, access-controlled service, not an ad hoc channel and never ordinary email. Data is encrypted in transit and at rest, access is limited to the review, and the working copy is destroyed on a defined schedule after delivery.
Auditable by the client, cited to the page
The work is human-supervised under a written protocol authored by the firm's founding healthcare attorney, who spent over twenty-five years defending providers in professional liability actions, and every determination is independently verified before delivery. Each call cites the record page it stands on, so a client can pull the page and confirm the call without taking the reviewer's word for it.
No certification yet, said plainly
Etiology holds no SOC 2, HITRUST, or other third-party security attestation today, and none is in progress. We would rather say so than imply one. What protects a client's records instead is a set of controls a reviewer can check: a written confidentiality obligation in place before any record moves, analysis on Anthropic's Claude under zero data retention with no training, encryption in transit and at rest, access limited to the review under a minimum-necessary rule, independent second-reviewer verification of every determination, and destruction of the working set on a defined schedule after delivery. When a client engagement calls for a formal attestation, we will pursue it.