US drug shortages averaged 5.3 years in 2025 — and the cheapest injectables are the ones that disappear

What is happening?

An annual US drug shortages report, released on 2026-06-09, counted 75 drugs in shortage at the end of 2025, down from 98 a year earlier. The count fell and the shortages got longer: the average shortage ran 5.3 years in 2025, against 4.3 years in 2024 and roughly two years in 2019. 64 percent of the drugs in shortage had been short for more than three years and 39 percent for more than five.

The failure is concentrated in one dosage form. Sterile injectables were 71 percent of all shortages, the largest of any form, with oral solids at 16 percent.

Price is where the pattern shows. A generic injectable not in shortage averaged 169 dollars; one in shortage averaged 20 dollars. In oncology the gap is wider still — a review published on 2025-09-25 put injectable cancer drugs in shortage at an average of 2 dollars per vial against 1,423 dollars for those not in shortage. Meanwhile manufacturers left: 170 products were discontinued in 2025 against 106 in 2024, a 60 percent rise and the highest level since 2019. Among the discontinued oral solids, 65 percent had been priced below one dollar per unit; no source we opened publishes the same price split for the discontinued injectables.

A second register tells a different-looking story. A quarterly count kept by a hospital-pharmacy professional society and a university drug information service put 227 shortages active in the second quarter of 2026, below its record of 323 in the first quarter of 2024. The two totals are not two points on one line — they come from two registers with different inclusion rules, and we did not confirm the rule that separates 75 from 227.

Whose problem is this?

The party that sets the price never appears on the ward. A purchasing group negotiates a vial down to two dollars, a manufacturer exits at that price, and the consequence arrives months later as a meeting in which oncologists decide who gets treated.

Where does this problem end?

The boundary matters here. This is not a claim that medicines are unavailable in general, because most prescriptions are filled without incident and the products that fail sit in a narrow band at the bottom of the price range.

What is the state now, and what should it be?

Now

Needs a new measurementthe target state. No source we opened states an official target: no maximum acceptable number of active shortages, no maximum acceptable duration, no required number of suppliers per essential injectable. What the annual report offers instead is a direction, that procurement should carry incentives valuing resilience rather than the lowest bid, which is not a number anyone can be measured against.

How big is it?

Both registers count products, not patients. The annual register reports 75 drugs; the quarterly register reports 227 shortages. Neither publishes how many people were treated with, or failed to receive, any product on those lists. The clinical evidence we found is also stated in units other than patients: three quarters of oncology pharmacists reported delays, dose reductions or regimen changes, and 97 percent of hospitals surveyed during the 2023 carboplatin shortage said it was affecting patient care. Those are shares of clinicians and of institutions.

One patient-level estimate appeared in search results, covering the carboplatin and cisplatin shortages, but the page carrying it could not be opened and the figure is therefore not used. The accompanying population file records not-derivable rather than a manufactured chain.

Needs a new measurementthe number of affected people cannot be derived from what these sources publish.

Under what conditions does it arise?

1. The product sells at or near the floor. Among 51 injectable drugs in shortage, 73 percent were priced below 15 dollars per unit and nearly half below 5 dollars. There is no margin left to absorb a quality investment, a redundant line, or a bad year. 2. One maker holds the whole supply. 48 percent of the new shortages recorded in 2026 were sole-source products. When that one line stops, the market has no second path. 3. Quality failure is the trigger and exit is the answer. One large generic sterile injectable plant in the United States closed in 2022 after years of FDA compliance findings, including an October 2021 FDA letter on mold contamination that led to a recall of more than 2.5 million vials. A drug-supply alliance assessment at the time put 24 generic sterile injectables at risk, including five essential medicines for which that single manufacturer held more than 15 percent of the market. This is a 2022 case included as the mechanism rather than as current status, and we did not trace which of those 24 products are on either shortage register today. 4. Approval is not entry. 37 percent of generics approved between 2013 and the first quarter of 2024 had not launched. Among drugs in shortage, 62 percent have an approved generic at all, and 84 percent of that subset have at least one approved generic that never reached the market. Injectables account for 75 percent of the shortages involving an unlaunched generic.

A shortage begins with a quality failure at one plant, and it persists because no other maker finds it worth expanding into a product that sells for a few dollars a unit.

What has been tried?

Nothing in this table changes what a vial earns. The register measures, the enforcement penalizes, the payment reimburses a storage cost, and the rationing distributes a deficit that already exists.

What was found?

Why is it still unsolved?

Incentive inversion — a medicine earns a price by being new, and earns a shortage by being old, cheap and indispensable. The reward runs opposite to the social value of the product.

The published price series is the clearest statement of it. The drugs that stay on the shelf average 169 dollars and the drugs that vanish average 20; among injectable cancer drugs the ratio is roughly 700 to one in the same direction. Nothing about a two-dollar vial says it matters less. A maker that exits a product priced at a few dollars a unit gives up almost nothing, and the loss lands on a hospital ward months later.

The 2025 numbers show both halves of the inversion moving at once. Shortages fell from 98 to 75 while discontinuations rose from 106 to 170. A discontinued product stops being in shortage, because it stops being a product. The count improves as the supply base thins, which means the headline number rewards exactly the behavior that causes the problem.

Policy has begun to touch the edge of this without reaching the center. The first tangible federal response, the CMS add-on payment effective October 2024, pays some hospitals to hold a six-month buffer of 86 essential medicines. It reimburses only the Medicare inpatient share of the cost, estimated at about 11 percent, it reaches only independent hospitals of 100 beds or fewer, and it explicitly excludes buffer stock for products that are already in shortage. The payment that was built to hold a reserve does not reach the drugs that are already gone.

What observation would mean it is solved?

Candidates — (a) a sustained fall in the number of active shortages (b) a fall in average shortage duration back toward the two years seen in 2019 (c) a fall in the share of essential injectables with a single manufacturer, currently visible as the 48 percent sole-source share among new 2026 shortages.

(a) is actively misleading and 2025 proves it — the count fell 23 percent in the same year discontinuations rose 60 percent. A product removed from the market leaves the shortage list.

(b) can be satisfied the same way. The longest-running shortages are the ones most likely to end in discontinuation, and removing them pulls the average down without a single extra vial reaching a patient.

(c) is the closest to the mechanism but is not self-executing. Counting approved manufacturers overstates supply, since 37 percent of approved generics never launched, and among the drugs in shortage that have an approved generic at all, 84 percent have at least one sitting unlaunched. The measure has to count who is shipping, not who is approved.

None of the three is sufficient on its own, because each of them can improve while the underlying supply base gets thinner.

What is it connected to?

Fills with researchthis appears to connect to hospital and oncology finance, Medicare and Medicaid payment design, and the geographic concentration of active ingredient production, with 44 percent of drugs in shortage relying on at least one key starting material made in a single country, typically China or India. Relation type and evidence grade were not confirmed in this round.

What these sources do not say

• How many patients. No source we opened gives a patient denominator for any product on either shortage list. The clinical impact is reported as shares of pharmacists and of hospitals, which cannot be converted into people.
• Which drugs. The 71 percent sterile injectable share is published, but the product-level list behind the 75 is not in any page we opened, so the injectable subset cannot be named or checked.
• The margin, as distinct from the price. Price per unit is published in detail. Cost of goods, and the margin at which a manufacturer decides to leave, are not — so the inversion is documented at the price level and inferred at the profit level.
• Why the two registers differ. 75 and 227 come from different inclusion rules and neither publication we opened states what those rules are or how they map onto each other.
• Whether the CMS buffer-stock payment changed anything. The rule took effect in October 2024. We found no evaluation of how many hospitals took it up, how much buffer stock is actually held, or whether any shortage was shortened by it.
• What the 170 discontinuations were. The dosage-form split of the products that left the market in 2025 is not given, so it is not possible to say how much of the sterile injectable base was lost that year. The one price detail published about them, that 65 percent were priced below a dollar a unit, is reported for the oral solids, which leaves the exit price of the injectables unstated.

See the evidence

No primary report PDF was read directly. The annual report itself was not opened; what was opened is the publisher news release plus three independent write-ups of the same report. That redundancy is doing real work here: the 71 percent sterile injectable share appears in three separately opened pages, and the 5.3-year average duration and the 169-dollar against 20-dollar price contrast each appear in two, so none of those figures rests on a single transcription. The news release itself is looser than its own write-ups on duration, giving it as over 5 years rather than 5.3, which is why the precise figure is attributed here to the pages that state it.

One smaller divergence is worth naming, because resolving it changed a number in this dossier. The 65 percent figure for discontinued products priced under one dollar per unit is worded as covering discontinued oral solids in two of the three write-ups we opened, and as covering discontinued drugs generally in the third. All three are reporting the same annual analysis. This dossier takes the narrower reading, because two independent pages state the oral-solid qualifier explicitly while the third simply omits it. That choice matters more than it looks: read narrowly, the figure says nothing about injectables, so it no longer sits in the price argument in block 1 and is labeled as an oral-solid figure everywhere else it appears.

One figure needed care in transcription and is worth stating plainly. The generic-approval analysis reports that 62 percent of drugs in shortage have an approved generic, and that 84 percent of that subset have at least one generic that never launched. The 84 percent is therefore conditional on the 62 percent, not a share of every drug in shortage; against the whole it would come to roughly 52 percent. This dossier carries the conditional form in all four places the number appears.

The plant-closure material is the one row here that is not current. That closure and the assessment of 24 at-risk injectables date from May 2022, and the source page describes the shutdown as already completed. It is included because it shows the mechanism running end to end, from a quality finding to an exit rather than a repair, which no 2025 or 2026 source we opened documents at the level of a single plant. It is not evidence about the composition of either shortage register today, and we did not check which of those 24 products are still short.

The two shortage totals are the one place where the sources genuinely diverge. 75 at the end of 2025 comes from the annual register; 227 in the second quarter of 2026 comes from the quarterly register. Both are reported here with their own register distinguished, and neither is presented as an update of the other. The 227 figure could only be confirmed through a syndicated copy of a news report, because the statistics page of that quarterly register and a trade-journal report of the same release both refused automated retrieval; the record of 323 in the first quarter of 2024, which anchors that series, was confirmed separately from a 2024 report that opened cleanly. This is a Path A output, so observation_refs is empty and provenance_mode is press-derived.

People affected

Estimated range Not derivable

The reason and what is missing are listed under “What is missing” below