Why US drug prices look like this

This is Eliquis (apixaban) — a single product, one NDC. Every price below is real, current, and public. None of them is what it costs to make.

There is no "the price" of a drug in America — there is a price per payer, and every gap between two rungs is somebody's business model. The rest of this page walks the gaps.

Compare what Medicare Part D plans pay at the counter with what the pharmacy paid to acquire the same drug, in the same quarter. Across 1,134 matched drugs, the median negotiated price is 1.68× acquisition cost, 38% of drugs are at 2× or more, and 4% are at 5× or more. For one of the most-prescribed examples, Atorvastatin Calcium runs 4.0× — and percentage coinsurance is charged on the marked-up number, not the real one.

Explore the spread tracker, drug by drug →

The VA buys drugs on published federal contracts. Medicare Part D plans buy the same drugs through PBM negotiation. Across 908 identical NDCs carrying both prices this quarter, Part D's median negotiated price is 1.36× the federal contract price. Same molecule, same package, same taxpayer.

Exhibit 3 set the federal contract price against Medicare's. Here it is against the pharmacy's own invoice cost. 3,610 drugs are on both a federal supply contract (VA FSS / Big4) and the NADAC pharmacy survey. For the typical one, the federal per-unit price is 95% of what a pharmacy pays for the same drug — and only 51% of matched drugs are cheaper on the federal schedule at all. The deepest negotiated discounts:

Example: 2% means the government's contract price is 2% of what a pharmacy pays to buy the same NDC — a 98% discount.

Medicaid and Medicare Part D are the two largest public drug payers, and they buy for different people under different rebate law. Here are the drugs each spends the most on — gross, before rebates — with the programs side by side rather than merged, because their source files key differently (Medicaid by NDC, Part D by drug name) and a row-level join would imply a precision that does not exist.

When the first generic arrives, the price doesn't drift down — it falls off a cliff. Median price of the molecule, as a share of its pre-generic price:

This is the honest two-sided truth of US drug pricing: the expensive problem is brands with no competition, and the shortage problem is generics with too much — prices erode until manufacturers walk away, markets thin to one or two makers, and supply snaps. Our shortage-risk model uses sustained price deflation as a predictor for exactly this reason.

Medicare Part D spent $288.7B (gross) on 3,625 drugs in 2024. Just 10 drugs took 29% of it — Eliquis alone was $20.8B. Meanwhile ~90% of prescriptions are generics that cost pennies. Drug spending isn't high everywhere — it is hyper-concentrated in a handful of protected brands.

Part D spending in full →

Everything above is the visible system. Behind it sits a layer no public dataset can see: manufacturer rebates, negotiated in confidential contracts between drugmakers and PBMs. Applying MedPAC's last program-wide measurement (rebates and fees ≈ 29% of gross Part D spending) to 2024's $288.7B implies roughly $83.7B a year flowing through terms nobody outside the contracts can audit — an estimate, because that is all anyone outside can have. The list price rises, the confidential net price doesn't, and the patient's coinsurance is computed on the list. The bubble between them is why "drug prices went up" and "drugmakers' net prices went down" can both be true in the same year.

Estimated from MedPAC's program-wide rebate aggregates (June 2023: DIR = 29% of gross Part D spending; brand rebates ≈ 25–40% of gross, generics minimal). Actual rebates are confidential and vary widely by drug — this is a range, not a fact about this drug.

And where does the revenue go? The 15 largest US-listed drugmakers, from their own audited 10-K filings: