Launching OpenPrescriber: 1.38 Million Medicare Prescribers, $275.6 Billion in Drug Costs, Every Pill Tracked

Today we're launching OpenPrescriber — a free, open-data platform that makes Medicare Part D prescribing data accessible, searchable, and analyzable. The site covers 1,380,665 prescribers, $275.6 billion in drug costs, and includes something no other public tool offers: specialty-adjusted fraud risk scoring and machine learning detection trained on confirmed fraud cases.

Why This Matters

Medicare Part D is the federal prescription drug benefit covering approximately 52 million Americans. In 2023 alone, the program spent $275.6 billion on prescription drugs through 1.38 million healthcare providers — that's over 1.6 billion individual prescriptions. Despite this enormous scale, the existing tools for exploring this data are either outdated (ProPublica's Prescriber Checkup is stuck on 2016 data) or locked behind paywalls.

OpenPrescriber changes that. We process the complete CMS public use files — the same data the government uses — and make it explorable from every angle: by provider, drug, state, specialty, or risk profile.

What's on the Platform

OpenPrescriber is built for depth. The platform includes:

• 19,400+ individual provider profiles — Detailed prescribing data for the highest-volume Medicare prescribers, with risk scores, peer comparisons, and top drugs prescribed
• 500 drug profiles — Every major drug ranked by cost, with prescriber data and cost-per-claim analysis
• 62 state & territory pages — State-level analysis with opioid rates, cost trends, and 5-year trend charts
• 205 specialty breakdowns — How prescribing patterns vary across medical specialties
• 29 in-depth analysis articles — Data journalism covering opioid crisis, fraud patterns, drug costs, geographic disparities, and more
• 8 interactive tools — Risk calculator, drug lookup, state report card, provider comparison, savings calculator, and more

The Risk Scoring Model

What makes OpenPrescriber unique is the risk analysis layer. We built a 10-component statistical model that evaluates every prescriber across multiple dimensions:

• Specialty-adjusted opioid analysis — We compare each provider against their own specialty peers using z-scores. A pain specialist prescribing opioids isn't the same as a dermatologist doing so.
• Cost and brand-name patterns — Providers who consistently prescribe expensive brand-name drugs when generics exist receive elevated scores.
• OIG exclusion cross-reference — We match all 1.38 million providers against the OIG's List of Excluded Individuals/Entities. We found 372 excluded providers who appear in active prescribing data.
• Machine learning detection — A model trained on 281 confirmed fraud cases identifies 4,100+ providers with similar prescribing patterns that rule-based systems miss.

The result: 233 providers flagged as high-risk (score ≥50), 6,473 at elevated risk, and thousands more with individual risk factors worth monitoring. Every score comes with an explanation of what triggered it.

Key Findings

Some of what the data reveals:

• 450,343 providers prescribed opioids — one in three Medicare prescribers. 113,169 have rates above 20%.
• Drug costs grew 50% in 5 years — from $183B (2019) to $275.6B (2023). GLP-1 drugs like Ozempic are a major driver.
• Brand drugs are 13.4% of prescriptions but 67% of costs — the brand-generic gap costs taxpayers billions annually.
• Nurse practitioners are 19% of prescribers but 49% of flagged providers — structural factors in scope-of-practice laws and supervision contribute to this disproportionate representation.
• 6,149 providers co-prescribe opioids with benzodiazepines — a combination the FDA explicitly warns against due to respiratory depression risk.

The Portfolio Grows

OpenPrescriber is the 10th platform in TheDataProject.ai portfolio, joining OpenMedicaid, OpenMedicare, OpenFeds, OpenSpending, OpenLobby, VaccineWatch, OpenImmigration, OpenSubsidies, and WarCosts. Together, these platforms represent over 100,000 pages of public data made accessible and searchable for free.

Every platform follows the same philosophy: take public data that exists but is hard to use, process it into something meaningful, and give it away. No paywalls, no logins, no ads.