Oncology authorization packets can include pathology reports, genomic reports, discrete results, and notes that describe the same test in different ways. A reviewer needs to know exactly which result supports a payer criterion, which specimen and date it belongs to, and whether the report is final or amended. The current HL7 mCODE v4.0.0 guide offers patterns for representing oncology data, but it does not itself decide what a payer requires or whether a treatment is clinically appropriate.

Start with the actual policy question

Before mapping a result, record the payer, plan, requested service or therapy, policy version, effective date, and the precise documentation criterion. One policy may ask for a test report, another for a particular result, and another for an explanation of prior testing. The workflow should ask only for evidence the applicable policy actually calls for.

Create a criterion-to-evidence row that identifies the candidate report, result, source system, and reviewer. Leave the row unresolved when the policy wording or evidence is ambiguous. A structured code is useful for retrieval, but it is not a substitute for reading the relevant report and policy.

  • What exact test or finding does the criterion name?
  • Does it require a complete report, a discrete result, or both?
  • Does the report concern the same patient, disease context, and requested episode of care?
  • Is the evidence current under the policy's own wording?

Distinguish genomic findings from other tumor markers

mCODE v4.0.0 separates genomics results from its TumorMarkerTest profile. Its Genomics group describes GenomicsReport, GenomicVariant, and GenomicRegionStudied; the guide places gene-expression tests and serum-based measurements in the tumor-marker area. This distinction matters when a search for a marker name could retrieve results from different methods or specimens.

Do not assume every biomarker is a sequenced variant or that a positive or negative string carries enough context. Capture the test name, method when documented, report identifier, result value and units or interpretation, specimen, and clinically relevant date. If the source only provides a PDF or narrative, retain that document and mark any extracted field for human verification.

Keep the report and its atomic results connected

FHIR R4 DiagnosticReport can carry report-level context and refer to Observation results, specimens, conclusions, and an attached presented form. In an authorization workflow, the report is often the best human-readable source while individual Observations support precise lookup. Preserve both when they exist, with a link between the criterion and the exact result used.

Check the result's status before reusing it. A preliminary value, final report, and later amendment should not be silently treated as interchangeable. Store the source version and receipt time, then route a changed interpretation for review if it could affect an active request. A missing result should remain missing, rather than being inferred from an order or a note that testing was planned.

Preserve specimen and region context

mCODE's genomics guidance notes that a tumor-normal test can involve two specimens, while a cancer gene panel may involve one. It also defines a GenomicRegionStudied profile and explains that a report may represent variants found in a tested region. For evidence matching, record which specimen and tested region a finding belongs to when those details are available.

This matters most when a patient has multiple tests over time or when reports from different specimens are present. The automation should offer candidate matches with context, not collapse all biomarker mentions into one patient-level fact. A qualified reviewer must resolve uncertainty about whether a result answers the policy criterion.

Use a small evidence checklist before submission

A useful checklist can be applied to each policy criterion: identify the source report; confirm patient and specimen; capture test and result; check status and date; retain the report or attachment; and record reviewer confirmation. Keep a separate state for not found, conflicting, or awaiting review. Those states generate different follow-up tasks.

During a pilot, sample cases with both structured and document-only results. Compare the selected evidence with expert review, count wrong-report and wrong-specimen matches, and measure how often reviewers must request another document. The aim is a more traceable packet, not an automated claim that a therapy meets medical necessity criteria.

Primary references