A BuiltForAI Property
Reproducibility-grade documentation · reviewed quarterly

Every number we publish, reproducible.

The full methodology behind RxFinder — how we join eight federal datasets at the NDC level, how the leak opportunity is calculated, what we model, and what we deliberately don't. If you want to reproduce a number on your own infrastructure, this page has the recipe.

Last updated
Review cadence: quarterlyMaterial-change re-review: on federal methodology updateNext review:
How it works · in order

Every figure on RxFinder traces back to a federal data source and a public calculation. There are no proprietary models, no black-box scores, and no rebate assumptions. The goal is reproducibility — if you have access to the same federal files, you can recompute any number on the site to the cent.

The 3,186-drug universe

RxFinder benchmarks 3,186 distinct molecules that meet two criteria: (1) appear in the CMS Part D Spending Dashboard or CMS Part D Utilization file for any of the last three calendar years, AND (2) have a price record in the CMS NADAC weekly file. Drugs that are exclusively Part B billed (most physician-administered injectables), exclusively inpatient billed, or have no NADAC record are excluded. The universe is rebuilt monthly as CMS publishes new files.

Joining the federal datasets

All eight source datasets are joined at the NDC level (National Drug Code, 11-digit format normalized to remove leading-zero ambiguity). Joins:

· openFDA NDC Directory → canonical NDC, drug name, dosage form, route, manufacturer
· CMS NADAC → per-NDC unit acquisition cost, weekly effective date
· FDA Orange Book → ANDA filings per drug, patent and exclusivity expiry, reference-listed flag
· CMS Part D Formulary → tier placement, prior-auth, step-therapy, quantity-limit counts across plans
· CMS Part D Spending → annualized plan spend per drug
· CMS Part D Utilization → claim counts, beneficiary counts, supply days
· FDA Drug Shortage Database → shortage status, reason, dates, manufacturer
· HRSA OPAIS → 340B contract pharmacy and entity counts per drug

The leak opportunity score

For any drug in your input list, the gross annualized savings opportunity is computed as:

Opportunity = (lives ÷ 50,000,000) × Part D claims (CY24) × avg days supply × (avg NADAC − lowest NADAC)

Where:

· lives = your covered-life count
· 50,000,000 = total Medicare Part D enrollment (scaling denominator)
· Part D claims (CY24) = total CMS claim count for that drug in calendar year 2024
· avg days supply = average days per claim, defaulted to 30, configurable in the tool
· avg NADAC = volume-weighted average across all NDCs of the drug
· lowest NADAC = minimum NADAC across all NDCs of the drug

The opportunity is filtered to drugs where (1) generic manufacturer count ≥ your minimum threshold, (2) annual Part D spend ≥ your minimum threshold, and (3) opportunity is greater than zero. The default thresholds (5 mfrs, $500M spend) are tuned to produce actionable lists.

What we deliberately don't model

RxFinder is intentionally transparent about what is NOT in scope:

· PBM rebate retention. Brand drugs flow rebates to PBMs and (partially) to plans. The leak report shows gross pre-rebate opportunity. Net opportunity requires your plan's actual rebate-retention rates per molecule — not publicly available.
· Specialty pharmacy markup. Drugs dispensed through PBM-owned specialty pharmacies often carry markup well above NADAC + dispensing fee. We don't have visibility into contract-level markups.
· 340B-purchased product flow. Tracked at the contract pharmacy level in OPAIS but not joined back to commercial claims. Per-plan 340B leakage requires claims-level analysis we can't do from public data alone.
· Manufacturer copay assistance. Reduces member out-of-pocket but doesn't affect plan acquisition cost. Not modeled.
· Clinical equivalence between molecules. We report federal data on cost and competition; we do not assert that switching across molecules is clinically appropriate. That is a P&T committee decision.

Refresh cadence per data source

· CMS NADAC: weekly (every Wednesday), pulled within hours
· FDA Drug Shortage Database: daily at 06:00 UTC
· FDA Orange Book: weekly (Monday)
· openFDA NDC Directory: weekly
· CMS Part D Formulary file: quarterly
· CMS Part D Spending Dashboard: annual
· CMS Part D Utilization: annual
· HRSA OPAIS: quarterly

Reproducibility

The full SQL queries that produce every published number, the data join procedures, and the snapshot dates of each source file used are available on request. We can issue a reproducibility packet for any specific drug or class on request — useful for plan-sponsor due diligence, vendor RFPs, or academic citation.