About the method

Two public federal datasets, joined on the prescriber's NPI:

• CMS Open Payments — industry payments to clinicians (who paid whom, for which drug).
• Medicare Part D — Prescribers by Provider and Drug — how much each clinician prescribed.

We scope to 50 drugs — the 49 most-promoted branded drugs that are actually prescribed at retail (diabetes, immunology, psychiatry, cardiology, dermatology, migraine, respiratory, GI), selected data-driven from the payments themselves — plus metformin as a zero-payment control.

For each doctor and drug we show their Part D claims next to the average claims of unpaid prescribers in the same specialty. Comparing within a specialty removes the obvious confounder (cardiologists prescribe more Eliquis andget more Eliquis money) — so the gap isn't just specialty mix.

Controlling for specialty, paid physicians prescribe more of a drug than their unpaid peers in 39 of the 49 branded drugs (p < 0.001) — up to +69% for blood thinners, COPD inhalers, and diabetes drugs (median around +23%). For ~10 drugs — mostly specialty biologics — there's no significant effect. See the Explore page.

Metforminis a cheap generic with no promotional payments — it's our control, and it shows no effect (only a $0 bar). And ~10 branded drugs (mostly specialty biologics like Stelara, Tremfya, Skyrizi, prescribed by a handful of specialists for locked conditions) show a flat, non-significant gap. We keep them in on purpose — a method you can trust is one that also reports where the effect doesn't appear.