What the USPTO Drug Patent Study Shows—and What the Agency Withheld
Alex Moss | October 26, 2026
Everyone knows prescription drugs are expensive, and that patents are a big part of why. But the precise relationship between patents and drug prices is often very difficult to untangle. I-MAK took on that challenge with comprehensive, rigorous research, and it revealed large numbers of patents and patent applications on blockbuster drugs.
Perhaps unsurprisingly, the industry responded by attacking that research. One complaint was that I-MAK counted patent applications along with issued patents. But applications are a standard part of licensing discussions. Companies use them both to obtain additional patents and to add leverage in negotiations, and that makes them effective barriers to entry.
In 2022, Senator Thom Tillis wrote two letters asking the USPTO and FDA to assess the data relied on by I-MAK and the UC Law SF Evergreen Drug Patent Database. The USPTO did not do that. As its 2024 report says, “the USPTO did not specifically assess their data or analysis.” Instead, it produced a Drug Patent and Exclusivity Study that offers “a baseline approach” for measuring the time from FDA approval to first generic launch, and it applied that approach to 13 active ingredients.
That study has nevertheless been used to push back on I-MAK’s research and to diminish the role patents play in driving up drug prices and delaying generic competition. On its own terms, the study supports neither. PIPLI is releasing a white paper—Breaking Down the USPTO Drug Patent and Exclusivity Study: What It Does and Does Not Say—that explains why in detail.
What the study covers
The USPTO study is narrow in ways that matter:
A small sample. It examines 25 drug applications covering 13 active ingredients, about 1% of the small-molecule drugs FDA approved from 1990 to 2019. The study itself says “the sample size limits generalizations that can be drawn from the study.”
No prices. “Drug pricing is outside the scope of this report.”
No biologics. Drugs like Humira and Keytruda are excluded, though biologics account for nearly half of US drug spending.
A cutoff in 2018. The study counts only patents listed in the FDA’s Orange Book from 2005 through 2018. Companies kept listing patents after that. Imbruvica now has 42 listed patents, 13 more than the study counted, all listed since March 2019.
Examples that show the problem. Revlimid, Kaletra and Lipitor each waited nearly 15 to 16 years after FDA approval for a generic, past the 14 years that patent term extension allows. Imbruvica, approved in 2013, still has no generic in the US.
The study explains its choices in general terms. It says it matched the 2005–2018 period to the sources in Senator Tillis’s letters and drew on top-grossing and most-prescribed drugs from 2017. It calls its 25 applications “a representative sample” while also saying they cannot support general findings. The more one looks at the study, the more one wonders how the USPTO arrived at the drugs and time frame it did.
What the agency released
PIPLI isn’t alone in wondering. A requester filed a Freedom of Information Act request for the documents and data the USPTO and FDA used to create the study, all drafts, comments on drafts, and communications between and within the agencies about it.
The USPTO responded this summer, in releases dated June 28 and July 2, 2026. The first release identified 5,550 responsive pages. Two of the released documents we have reviewed are withheld in full. Every page of a 124-page draft of the report reads only “b5,Draft.” So does every page of a three-page “Lantus Patent Landscape Analysis.”
“b5” refers to FOIA Exemption 5, which the USPTO applied here as the deliberative process privilege. That privilege protects an agency’s internal opinions and recommendations made while it is deciding something. It generally does not protect factual material, and FOIA requires agencies to release any portion of a record that can reasonably be separated from what is exempt. The USPTO’s letter asserts that the facts in these records “are not reasonably segregable.”
The Lantus analysis shows why that matters. Lantus, an insulin product, does not appear in the published study, but the agency apparently prepared a patent landscape for it. A patent landscape is a compilation of facts: which patents cover a drug, what they claim, and when they expire. Withholding all of it leaves the public unable to see what the agency considered or how it chose what to study.
The letters also contain errors. Both describe the request as made on “December 3, 2026,” months after the dates of the letters themselves.
What remains
The USPTO conducted its study in response to a public request from a US Senator to evaluate research whose data and methodology are fully public. The agency’s own process remains largely hidden, and that makes its approach harder to accept.
We don’t know what went on behind the scenes. We do know that the study does nothing to undermine what I-MAK’s research and many other studies have shown. Drug companies acquire numerous patents on their drugs, and those patents create barriers to competition. Americans pay more than three times as much for brand-name drugs as people in comparable countries, even after accounting for rebates. And US patent law allows thicketing tactics that other countries’ laws do not.