Payer document processing is the work a health plan does to receive, identify and route the paper and electronic documents that arrive with or about claims: claim attachments, medical records, appeals and member correspondence. Each document is matched to a member and a claim or case, labelled by type, and sent to the team that acts on it.
Which documents a payer receives every day
A health plan, also called a payer, receives four broad groups of documents every day: claim attachments, medical records, appeals and correspondence. Each group arrives through several channels at once, such as mail, fax, portal uploads and electronic submissions, and each group belongs to a different team inside the plan.
Claim attachments are documents that support a claim the plan is deciding. They include operative reports, invoices for supplies, therapy plans of care and itemized bills. Some arrive before anyone asks for them. Medicare, for example, describes a process in its Medicare Program Integrity Manual, chapter 3 (section 3.2.3) in which providers can mark a claim with a PWK modifier and send documentation unasked. Medicare Administrative Contractors (MACs) then allow 7 calendar "waiting days" from receipt of the claim for that documentation to be submitted, or 10 if it is mailed, before they send a request for records.
Medical records arrive when the plan asks for them. A request can come from medical review, from a quality program such as HEDIS, from risk adjustment or from a special investigations unit. The same provider may receive several requests from the same plan in one month, each for a different purpose.
Appeals and reconsiderations are disputes about a decision the plan already made. They usually combine a letter, a copy of the denial and supporting records, so one appeal can be thirty pages of mixed content.
Correspondence is everything else: member letters, provider address changes, coordination of benefits questionnaires, accident questionnaires and returned mail. It is a small share of the pages but a large share of the routing questions, because a letter often has no claim number at all.
Health insurance document processing: intake, classification and routing
Health insurance document processing moves each incoming document through three steps. Intake records what arrived, when and through which channel. Classification decides what type of document it is and which member, claim or case it belongs to. Routing sends it to the queue that owns that type, with the identifiers attached.
Intake is where the received date is set. That date matters later, because many deadlines on both sides are counted from it. A fax that sits unopened for three days still arrived on the day it came in. Intake also splits multi-document packets, so that an appeal letter and its records are not filed as one item.
Classification answers two questions. The first is type: is this a claim attachment, an appeal, a requested medical record or a member letter? The second is identity: which member ID, claim number or case number does it belong to? A document with a type but no identity cannot be routed. A document with an identity but the wrong type goes to the wrong team.
Routing applies a rule to each classified document. The routing table below is a fictional example of how one plan might map document types to queues. Every value in it is made up.
| Document type (example) | Signals that identify it | Destination queue (example) |
|---|---|---|
| Claim attachment | Claim number on a cover sheet; a claim marked with a PWK modifier; operative report or invoice layout | Claims pending documentation |
| Requested medical record | Plan's request letter on top; request ID such as REQ-2026-04417 | The program that asked: medical review, quality or risk adjustment |
| Appeal or reconsideration | Words such as "appeal" or "reconsideration"; copy of a denial letter; member or provider signature | Appeals and grievances |
| Accident or other-insurance questionnaire | Questions about injury date, other coverage or an attorney | Coordination of benefits and recovery |
| Provider demographic change | Tax ID, new address or NPI on a form | Provider data management |
| Member letter with no identifiers | Handwritten, no claim or member number | Manual research queue |
The last row matters most. Every routing table needs a place for documents that match nothing, staffed by someone who can research them, or they disappear into a general inbox.
Risk adjustment chart review at a health plan
Risk adjustment chart review is a payer's coding review of medical records to confirm that the diagnoses it reported for its members are supported by the provider's documentation. Plans that are paid based on member health status, such as Medicare Advantage plans, need each reported diagnosis to be backed by a record of a visit.
The HIPAA Privacy Rule's definition of payment in 45 CFR 164.501 on the eCFR lists "risk adjusting amounts due based on enrollee health status and demographic characteristics" among payment activities. That is why a plan's request for charts for risk adjustment is treated as part of payment, not as a separate kind of request.
CMS checks this work. The CMS Medicare Advantage Risk Adjustment Data Validation page says that during a RADV audit, "CMS confirms that any diagnoses submitted by an MAO for risk adjustment are supported in the enrollees' medical records," and that CMS may collect overpayments when they are not. The regulation behind it, 42 CFR 422.310 on the eCFR, paragraph (e), requires plans "and their providers and practitioners" to submit a sample of medical records for validation.
For document processing, this means risk adjustment records need their own routing rule. A chart requested for risk adjustment carries a request ID, a member and a date range, and it should land with coders, not with the claims team. A chart that arrives without its request letter is the most common reason it cannot be matched.
Medical records in payer subrogation cases
Subrogation is the process by which a health plan recovers what it paid when another party, such as an auto insurer or a liable business, should have paid for the care. Medical records show the link between an injury and the services billed, so they sit at the center of a subrogation file.
The same HIPAA definition of payment in 45 CFR 164.501 names "adjudication or subrogation of health benefit claims" as a payment activity. In practice, a recovery team uses records to answer two questions: which services relate to the accident, and on what dates. An emergency department note that records how the injury happened, followed by therapy notes that refer back to it, is the kind of chain the team looks for.
Subrogation files come together from several sources. A claim with an injury diagnosis code may trigger an accident questionnaire to the member. The answer may name an attorney or another insurer. The plan then requests records for the dates after the injury. Each of those documents must be filed against the same case, which is hard when the questionnaire carries a member ID, the attorney letter carries only a name and the records carry the provider's account number.
For the document team, the lesson is to give recovery cases a case number early and to print it on every outgoing request. Records that come back with that number on the cover page route themselves.
Where does payer document processing break down?
Payer document processing breaks down at the points where a document loses its identity or its pages. Three failures account for most of the rework: attachments routed to the wrong queue, documents that cannot be matched to a member, and pages lost or separated between intake and the reviewer.
Misrouted attachments. A therapy plan of care sent for a pending claim lands in the appeals queue because its cover letter used the word "reconsider". The claim then denies for missing documentation while the attachment waits elsewhere. A rule that checks for a pending claim number before it looks at keywords prevents most of these.
Unmatched members. A record arrives for a fictional member, Dana Whitfield, born 03/14/1961, with the provider's account number 88213 and no plan member ID. The plan has two members with that name. Without a date of birth match and a request ID, the document waits for research. Asking providers to return the request letter as the first page solves this better than any matching logic.
Lost pages. A 60-page fax arrives in two transmissions, and the second half is filed as a new document. A reviewer sees pages 1 to 31, decides the record lacks a signature, and the signature page sits in another queue. Page counts on cover sheets, and a check that the received count matches, catch this early.
Duplicate requests. Quality, risk adjustment and medical review teams may each ask the same office for the same chart in the same season. HEDIS medical record review is one of those programs. Its measures and the hybrid method are explained in the guide to what HEDIS medical record review measures. Coordinating requests across programs reduces the volume of incoming paper and the chance of filing a chart twice.
The provider side of this exchange has its own discipline. A Medicare contractor's request has a fixed list of contents, and the Medicare ADR checklist for a complete response shows how a practice assembles one so it arrives complete and matched.
Which payer documents can software read reliably?
Software reads typed, structured payer documents most reliably: claim forms, typed letters with printed identifiers, and records printed from an electronic health record. It reads handwritten notes, faxed copies of copies and mixed packets less reliably, so those need a person to check the fields that drive routing.
Typed claim forms are the easiest case. A CMS-1500 or UB-04 puts each value in a known box, so the member ID, dates of service and codes can be read with a high degree of certainty. Printed EHR records come next, since their headings and dates are consistent within one system.
Handwritten correspondence is the hardest case. A member's letter may include the claim number in the body text, written partly from memory. Software can find a number that looks like a claim number, but a person should confirm it before the letter is routed on that basis.
Mixed packets fall in between. An appeal that combines a typed letter, a denial notice and forty pages of records needs to be split before its pages can be classified. The splitting step decides whether the rest of the process works.
The practical approach is to treat each extracted field as a value with a level of certainty, not as a fact. Fields read with high certainty route automatically. Fields with low certainty go to a person who sees the page. How confidence scores and review queues work is covered in the human review category. Schema Care follows that model for the document types it supports, including CMS-1500 and UB-04 claims and medical records: each field carries a score and a link to its spot on the page, as the list of supported documents shows. How payer decisions flow back into a provider's billing work is covered in the revenue cycle automation category.