What is a claim pre-audit?
A claim pre-audit is the hospital's own review of a claim after it is closed and before it is sent to the health plan. It checks whether each billed item was authorized, documented in the medical record, coded and priced according to the contract. It is the retrospective audit done on the hospital side, as explained in the Medical claims auditing guide.
In Brazil's private health system (saúde suplementar), regulated by the national agency ANS (Agência Nacional de Saúde Suplementar), health plans audit claims to decide what to pay. The hospital audits so the claim arrives right. An error fixed before submission does not become a denial, does not require an appeal and does not lengthen the collection period. Pre-audit adds to prospective review (before the procedure) and concurrent review (during the stay), covered in the guide Types of claims audit.
Why audit 100% of claims instead of a sample?
Because a sample lets small, repeated errors through. Reviewing only high-value claims, or a random fraction, puts effort where each claim weighs most. But an expired authorization, a quantity above what was authorized or a duplicate item shows up in claims of any size. Added up over a month, these errors matter.
The problem is that manually reviewing every claim is usually not feasible and delays submission. The answer is to split the work into two layers:
- Layer 1, automated rules on 100% of claims. The system checks what is objective: eligibility, authorization number and validity, authorized quantities, duplicates, TUSS code (Brazil's standard billing terminology), contract price and submission deadline.
- Layer 2, risk-based human review. Billing staff, nurse auditors and physician auditors look only at claims the rules flagged, high-value claims and types of care with a history of clinical denials.
Every claim is checked, but only some wait for a person.
How the pre-audit flow works
The flow starts when the claim is closed and ends with the health plan's statement, which feeds back into the rules.
- Claim closing. After discharge (or at the end of the period, for interim claims), the claim brings together charges, authorizations, medical record entries and contract rules.
- Automated rules on 100% of claims. Every claim goes through the objective checks, by health plan and type of care.
- Risk triage. A claim with no flag and below the value cutoff goes straight to submission. Flagged, high-value or sensitive claims go to the review queue.
- Human review. The flag decides who reviews: price and code go to billing; nursing documentation to the nurse auditor; clinical justification to the physician auditor.
- Correction and submission. The claim is adjusted (item added, removed or corrected) and sent in an XML batch under the TISS standard (Brazil's mandatory electronic data exchange standard), which returns a receipt protocol [5].
- Statement and new rules. Each denial received carries a table 38 code. A recurring reason, by health plan, becomes a new rule in step 2.
The balance point is step 3. If the cutoff sends too many claims to human review, the queue grows and billing slows. If it sends too few, errors get through. A model trained on the denial history can rank that queue by risk and value; the guide AI for denial prevention shows how this works in practice.
What should you check on each claim?
The table links each checkpoint to the rule that can be automated, to who decides when there is a flag and to the denial code the failure usually triggers. Codes come from TUSS table 38, detailed in the guide TISS denial codes [6].
| Checkpoint | Automated rule (layer 1) | Who decides on a flag (layer 2) | Denial it prevents (table 38) |
|---|---|---|---|
| Authorization and validity | Every item that needs an authorization number has one; service date within validity | Billing, with the authorization team | 1414 — authorization validity expired |
| Authorized quantity | Billed quantity ≤ authorized quantity (days, sessions, implants) | Billing; nurse auditor if documentation is missing | 3292 — quantity above authorized |
| Duplicates | Same item, date and time do not appear twice; no period repeated across interim claims | Billing | 3209 — duplicate item; 3232 — period already billed in another interim claim |
| Price | Amount matches the contract price table in force on the service date | Billing | 1705 — amount billed above agreed |
| Medical record | Billed item has a matching administration record, order or progress note | Nurse auditor | 3217 — no or insufficient justification |
| Clinical justification | High-cost item or longer-than-expected stay triggers review | Physician auditor | 3040 — clinical denial (to be specified) |
| Code and documents | Active TUSS code; complete TISS forms and attachments in the current version | Billing | Batch rejection or administrative denial |
| Submission deadline | Claim within the contractual submission period | Billing coordination | Loss of the billing deadline |
The table is a starting point. Each health plan has its own contract rules, and ANS Normative Resolution (RN) 503/2022 requires the contract to describe the audit routine and the grounds for denial [4]. The hospital's rules should mirror that contract.
Who does the human review
Human review follows each profession's rules. In nursing, auditing is reserved to the registered nurse within the nursing team, and the opinion must show the nurse's Coren (regional nursing council) number, under Cofen Resolution 720/2023 [3]. Checking administration records, orders and nursing notes is therefore the nurse auditor's job, not the biller's.
For medical audit, CFM Resolution 2,448/2025 (Federal Council of Medicine) applies. The official text marks several provisions as suspended by a court decision in case no. 1017752-74.2026.4.01.3400, including art. 2, which defined medical audit as an act reserved to physicians [2]. Not marked as suspended are art. 8, which requires medical audit companies, their technical directors and auditors to register with the CRM, and art. 15, XV, which makes the technical director responsible for ensuring auditor pay is not tied to denials [2].
For the pre-audit, the practical effect is clear. Clinical justification is reviewed by the physician auditor; a company hired to audit claims needs CRM registration; and automated rules or AI tools prepare the analysis but do not issue the opinion. The guide AI in hospital billing covers that limit.
How do you avoid slowing billing?
Pre-audit slows billing when it becomes a single, manual queue with no deadline. At Anahp hospitals, the average collection period, from invoice to payment, was 77.35 days in 2025 [1]. That figure does not include the time between discharge and invoicing, which is where the pre-audit sits. Every day a claim waits in audit adds to it.
Four practices reduce the risk:
- Rules running at charge entry, not only at closing. Duplicates and quantities can be checked while the patient is still admitted, as in concurrent review.
- A queue with deadlines. Each flagged claim has an internal review deadline; overdue claims escalate to the coordinator.
- A value cutoff adjusted with data. The amount above which every claim goes to human review is tuned to results: if reviewed claims rarely need corrections, the cutoff can rise.
- Fixes at the source. Each finding goes back to whoever entered the charge or the record, with the reason, so the error is not repeated. The guide Medical record audit shows how.
Pre-audit outcome indicators
The indicators measure three things: how much of billing goes through review, the effect on denials and how much time the review costs. They make sense compared with the hospital's own history, by health plan and type of care, not against a generic target.
| Indicator | How to calculate | What it shows |
|---|---|---|
| Rule coverage | Claims run through automated rules ÷ claims submitted | Whether 100% of claims are actually checked |
| Flag rate | Flagged claims ÷ claims checked | Whether rules are calibrated (too many flags clog the queue) |
| Correction rate | Reviewed claims with adjustments ÷ reviewed claims | Whether human review targets the right claims |
| Adjusted value | Amount added, removed or corrected ÷ value of claims checked | Size of the error that would be denied or left unbilled |
| Initial denial by reason | Amount denied per table 38 code ÷ amount billed, by health plan | Whether failures covered by rules are falling |
| First-pass acceptance | Amount accepted without denial ÷ amount billed | Combined effect of the pre-audit |
| Pre-audit time | Days between claim closing and release for submission | Whether review is delaying billing |
| Recurring denials without a rule | Reasons that repeat in statements with no matching rule | Where to build the next rule |
When comparing with the sector, use the right denominators. Anahp's initial denial (15.93% in 2025) is calculated on amounts receivable from health plans; accepted denial (1.72% in 2025) is calculated on gross revenue from health plans [1]. Do not divide one by the other. The guide Denial rate explains each formula, and Hospital average collection period shows how to track timing.
What we see in operations
At Accrua, pre-audit reads every health plan claim, not a sample, and works in two layers. In the first, Accrua OS reviews the whole claim automatically: medical record, prescriptions, items, OPME (implants and devices) and bundles, checked against the contract, the current price table, the authorization rule and that health plan's denial history for that item.
In the second, only flagged claims go to a biller or auditor with nursing training or specialization, who receives each alert with its source. Anything that changes the value of the claim is that person's decision. Claims with no alert move on to XML validation and submission, which is how review avoids trading denials for delays, the risk this guide describes.
We measure pre-audit by what it holds and what it releases: alerts per claim, adjustments made and time from claim closing to submission. When the same alert keeps coming back, the fix goes to the source, and what we learn in each appeal feeds the rules for the next claim.
Where to start
- Measure current coverage: how many claims go through any review before submission, and how many days it takes.
- List denial reasons from recent months by health plan, from the statements, and turn the most frequent into the first rules.
- Define triage: what sends a claim to human review (flag, value, type of care) and who reviews each flag type.
- Set deadlines for the review queue and track pre-audit time weekly.
- Review the rules every statement cycle, adding a rule for each recurring denial without coverage.