Adrastia Analytics

Government Health Program Intelligence

What an Adrastia validation pilot involves

A validation pilot is a limited, source-traceable review of public pharmacy reimbursement and benchmark data. It is designed to let an authorized agency, payer, counsel, or program-integrity team test Adrastia's method against concrete NDC-level records before deciding whether to expand the work.

Published 28 July 2026 · Last updated 30 July 2026

Pilot objective

The pilot answers a narrow question: can public reimbursement records be matched to public acquisition-cost and product-reference files in a way that produces reproducible review candidates worth validating against the customer's own records?

The output is decision support. It is not an allegation, audit finding, legal opinion, actuarial certification, clinical recommendation, or final agency determination.

Typical scope

ElementTypical pilot boundaryReason
DatasetOne state, one carrier, or one defined NDC portfolioKeeps validation concrete and reviewable
RecordsRanked candidate file plus controlsShows whether the method handles high, low, and neutral records
SourcesCMS NADAC, Medicaid SDUD, Medicare Part D public files, FDA NDC Directory, and related public references where applicableAllows independent reproduction from public data
Time periodA stated quarter, year, or source vintagePrevents mixing reimbursement periods and benchmark files
Customer dataNone for the standard pilot unless separately authorized in writingLimits privacy, procurement, and security scope

Deliverables

A pilot package commonly includes:

Validation steps

  1. Confirm authority and scope. Identify the agency, payer, program, state, carrier, portfolio, period, and permitted use.
  2. Freeze source vintages. Record the exact public files used so results can be reproduced later.
  3. Normalize product identity. Convert NDCs to a common 11-digit format and validate product identity against FDA reference data.
  4. Align unit basis. Confirm that reimbursement and benchmark prices are compared on the same denominator.
  5. Compute variance. Calculate percentage variance and benchmark multiple for matched records.
  6. Select controls. Include high, low, and neutral records so reviewers can test the method rather than only inspect outliers.
  7. Review limitations. Label data gaps, exclusions, specialty/340B concerns, dispensing-fee assumptions, and vintage sensitivity.
  8. Customer validation. The customer checks candidates against its own claims, contracts, fee schedules, audit authority, and legal process before any external action.

What a pilot does not require

The standard pilot does not require Protected Health Information, member-level claims, patient identifiers, clinical records, third-party system credentials, or production access to a customer's private systems. If a later engagement requires any restricted data, the parties should define that scope in a written agreement before transmission and use the necessary data protection terms.

What success looks like

A successful pilot does not mean every candidate becomes a recovery or enforcement action. It means the customer can confirm that the method is reproducible, the source rows are traceable, the limitations are explicit, and the candidate list is useful enough to justify a deeper internal review.

Pilot outputs identify analytical review candidates. Adrastia does not decide whether a price was improper, whether a party acted wrongly, whether money is recoverable, or whether any government, legal, contractual, clinical, or enforcement action should be taken.