Papers end with a promise: "The authors declare no conflicts of interest." Nobody checks it. We compared those statements with the US government's public record of payments from drug and device makers to doctors. Half the papers pass cleanly. A third don't.
Click any square to open that paper's check.
One real case, anonymised
Each paper is a traveller at customs. The declaration is what the authors wrote. The X-ray is what companies were legally required to report paying them in the three years before publication.
How authors declare
How to read: the bar is every checked author whom companies reported paying. Green named every company. The two reds left at least one out. For a publisher, the darkest red is the cheapest win: a paper that says "no conflicts" while the record shows payments needs no judgement, only a query.
Click a segment to see those authors.
How much
How to read: each dot is one flagged author; further right means more undeclared money. For a publisher, everything right of the dashed line is a five-figure relationship readers never saw.
Each dot is one flagged author, placed by the undeclared amount in the three years before publication (log scale; 1 US dollar ≈ ₹100). Red: undeclared. Amber: paper declared only "no relevant" conflicts. Click a dot to open the case.
Where
How to read: each bar is one specialty's checked authors, red share on the left, green on the right. For a publisher, this says where to start: the top three specialties would catch the most gaps per paper checked.
Share of checked US authors in each field. Green: declaration matches the record. Amber: declared only "no relevant" conflicts. Red: undeclared payments. Click a field for its cases.
Who pays
How to read: left is authors whose paper declared that company, right is authors whose paper did not. Longer red than green means the company's money is more often hidden than shown. For a publisher, the company most often left out is a device maker, so checks should not stop at drug companies.
Checked authors with significant payments from each corporate group, split by whether their paper declared that company. Click a company for its undeclared cases.
Trend
How to read: each column is one publication year's checked authors. For a publisher, waiting is not a strategy: the gap is wider in the newest papers.
2023 covers October to December only. Older papers look back at earlier payment years, so part of the rise may reflect the data, not behaviour. Click a year for its cases.
So what
Run the declaration against the public payment record automatically, before review. It takes seconds per paper.
Send the author the evidence with a polite query. Most gaps are omissions, fixed with one updated line.
Readers, reviewers and regulators trust disclosures that have been checked. That is a quality mark competitors don't have.
Explore
How we checked
A red flag is a question for an editor, not a finding of misconduct. Journal policies differ on what must be declared, and some disclosures reach the editor without being printed. Only US clinicians appear in the payment record. Authors are anonymised to specialty and US region; amounts are rounded to $1,000, and amounts over $100,000 appear only as bands. A hand check of 20 random red cases found about 9 in 10 correct.