Erin Rose

Founder & Methodology Architect, CareCost Estimate

I built CareCost’s billing and Medicare-coverage references myself, reconciling the primary CMS and FDA record into one cross-checked resource where every figure traces back to a document you can open. My name is on it for a plain reason: a wrong J-code or a missed covered diagnosis is a denied claim, and a bad estimate is a patient’s surprise bill.

LinkedIn → editorial@carecostestimate.com

Why you can trust these pages, without taking my word for it. Every covered-diagnosis list, J-code, unit rule, and reimbursement figure on this site traces to a named, dated primary source: the CMS Medicare Coverage Database, CMS pricing files, or FDA labeling. Each one is cited on the page, so you can check it in a click. The rigor is the credential. I’m not a certified coder, and CareCost doesn’t claim clinical credentials it doesn’t have. What I bring is the data architecture and a strict sourcing standard; the coding and clinical calls are reviewed and corrected by practicing specialty-pharmacy billers, with every change logged in the open.

Why these pages exist

They come out of real complaints. In beta, the billers using CareCost told us the painful part wasn’t the arithmetic; it was cross-checking a single claim across a scatter of CMS articles, manufacturer PDFs, and payer portals. So we built one place that holds the reconciled answer with its source attached, for the drugs they touch every day.

What went into this

These pages aren’t opinion. They reconcile the primary record. The coverage references cross-reference the CMS Medicare Coverage Database (every Billing & Coding Article and the LCD behind it) against FDA-approved labeling and the ICD-10-CM code set, drug by drug and MAC by MAC. The pricing sits on 11.3 million triangulated payer negotiated-rate records, drawn from more than 42 billion rows of Transparency-in-Coverage data published under 45 CFR Part 147 and joined to CMS authoritative files: the ASP Pricing File, the NDC-HCPCS Crosswalk, and the Physician Fee Schedule. Getting usable numbers out of roughly 180 terabytes of that data meant separating real negotiated rates from ghost rates, triangulating across payers and hospital disclosures, and resolving prices down to the individual NPI, the finest grain the data allows.

Pulling those sources into one consistent, verifiable picture, then re-checking it against CMS every quarter, is the real work, and it’s most of what I do. To my knowledge it is the most thoroughly cross-referenced free reference of its kind, and every figure in it is checkable against the source it came from.

Here’s one example of why that reconciliation is hard: billing units. A single drug ships in many presentations, with different vial sizes, concentrations, and pack quantities, and Medicare’s billable unit rarely equals the milligrams on the label. Normalizing dose to billable units, and the JW/JZ wastage that follows, across hundreds of products is hard, and it’s a spot the reviewers and I came back to again and again until it was right.

How it stays accurate

Nothing here is typed from memory, and nothing is invented. The coverage pages are compiled programmatically from the CMS sources above; the billing references pull live CMS ASP pricing at the active quarter. My part is the standard. I define the methodology, check each page against its source article, and re-verify the coverage data against CMS every quarter. When a working biller flags something that’s wrong, I fix it and log the change in the open. The full process, the source-conflict hierarchy, and the update schedule are in the Methodology and Editorial Policy, and every correction is public on the Corrections page.

Why I’m the one who built it

A dataset this size, accurate enough to quote a patient’s bill before treatment, is a data-infrastructure problem before it is a healthcare one. That’s the work I’ve done for twenty years, building data and AI systems for national media organizations. Earlier, at the healthcare agency RKD, I led campaign development for Susan G. Komen and MD Anderson Cancer Center, which put me inside healthcare economics and the mechanics of reimbursement. That mix is uncommon, and it’s what this project needed.

What I focus on

Specialty (buy-and-bill) drug billing — J-codes, units, JW/JZ wastage, NDC-to-HCPCS
Medicare Part B coverage — covered diagnoses, LCDs and Billing & Coding Articles, MAC jurisdiction, SAD (Part B vs. Part D)
Payer price transparency — turning raw Transparency-in-Coverage files into usable negotiated-rate benchmarks
Patient out-of-pocket estimation and manufacturer copay-assistance matching

References I maintain

A sample of the billing and coverage references I own on CareCost — the full library is at the reference hub.

Background

Role: Founder & methodology / data architect, CareCost Estimate (CareCost LLC)
Experience: Two decades building large-scale data and AI systems for national media organizations
Earlier: Senior Manager, RKD — healthcare campaign development for Susan G. Komen and MD Anderson Cancer Center
Speaking: AI and healthcare data at Google healthcare summits and industry conferences
Based in: Portland, Oregon
Page last reviewed: July 2026
Contact: LinkedIn · editorial@carecostestimate.com