Corruption in the pharmaceutical industry does not rest on suspicion: it is documented through a few precise mechanisms, established by court decisions and by the medical literature. The four main ones are off-label marketing (promotion of unapproved uses), the ghostwriting of scientific studies, publication bias (the withholding of unfavourable trials), and regulatory capture (the partial funding of agencies by the very industries they oversee). None of these facts means that medicines are ineffective: they concern the way the product is sold and the evidence is produced, not the therapeutic value of treatments.
Definition. "Pharmaceutical corruption" here designates a set of commercial and scientific practices that are judicially or methodologically established — marketing fraud, distortion of evidence, structural conflicts of interest — and not a blanket accusation against medicine. The word "mechanism" matters: these are repeated, nameable, verifiable patterns, not an intent imputed to an entire sector.
In the United States, the FDA approves a drug for specific indications. A physician may prescribe "off-label" in the exercise of their judgement; but it is illegal for a manufacturer to actively promote a drug for an unapproved use. It is this promotion by the manufacturer — not the prescription by the physician — that formed the core of the largest legal settlements in history. In July 2012, GlaxoSmithKline pleaded guilty and settled for $3 billion — then the largest health-care fraud settlement in U.S. history — notably for promoting Paxil and Wellbutrin for unapproved uses and failing to report safety data on Avandia. In 2009, Pfizer had settled for $2.3 billion (including a $1.3 billion criminal fine tied to Bextra alone); in 2013, Johnson & Johnson's Janssen subsidiary pleaded guilty in a $2.2 billion settlement over Risperdal.
The mechanism is recurrent: funding of medical "key opinion leaders," slanted conferences, prescription kickbacks, targeting of vulnerable populations (children, the elderly) for unvalidated uses — the Risperdal settlement concerned precisely promotion to the elderly and to children. This repertoire describes a business model, independent of any particular product. Many of these cases came to light thanks to internal whistleblowers, through the U.S. False Claims Act (the so-called qui tam procedure), which rewards the reporting of fraud against public programmes. It is this legal device that partly explains why so many practices ended up documented before the courts.
The same repertoire was established judicially for opioids. In October 2020, Purdue Pharma, the maker of OxyContin, pleaded guilty to three federal counts — including a conspiracy to defraud the United States and to violate the Anti-Kickback Statute — in a resolution valued at $8 billion, the heaviest ever imposed on a drug maker. On the public-health side, the Centers for Disease Control and Prevention (CDC) link the first wave of the overdose epidemic to the rise in opioid prescribing from the late 1990s; between 1999 and 2023, nearly 806,000 people died of an opioid-related overdose. Commercial fraud and the health crisis are two distinct facts here, one judicial, the other epidemiological.
The second mechanism touches the very production of scientific evidence. Ghostwriting consists in having an article written by industry pens, then having it signed by academics whose names lend academic credibility. The physician Ben Goldacre, in Bad Pharma (2012), assembled an indictment grounded in the medical literature: negative trials left unpublished, ghostwritten articles, slanted protocols. His central argument is not that medicines are false, but that missing and slanted data distort the assessment that doctors and patients can make of them.
This mechanism reveals a structural conflict: the entity that funds, designs, and promotes a product is also the one that produces part of the information about it. As long as that information is not independently verified and published, the risk of distortion exists — not through systematic malice, but through the natural slope of commercial interest.
Publication bias is the tendency for positive results to be published more than negative or null ones. It is a measured phenomenon, and its most telling illustration concerns antidepressants. In 2008, Erick Turner and colleagues published in the New England Journal of Medicine an analysis of trials registered with the FDA: of 74 trials examined (12 antidepressants, more than 12,000 patients), 31% had not been published, and almost all the trials judged positive had been published while most negative trials had not, or had been presented in a favourable light. The apparent effect size, based on the published literature alone, was overestimated by 32% for the entire class. The literature available to prescribers therefore gave a more favourable picture than the full set of data actually available.
This finding echoes the paper by the epidemiologist John Ioannidis, "Why Most Published Research Findings Are False" (PLoS Medicine, 2005), one of the most cited texts in the biomedical literature. Ioannidis analyses how small samples, conflicts of interest, and the pressure to publish can inflate the proportion of published results that do not replicate. It is an internal, methodological critique of science — not a rejection of science.
The fourth mechanism is structural: it concerns the relationship between the laboratories and the agencies meant to oversee them. In the United States, since the Prescription Drug User Fee Act (PDUFA, 1992), a substantial share of the budget the FDA devotes to drug evaluation comes from user fees paid by the manufacturers who submit their products for approval; the FDA itself describes this mechanism as an authorisation, granted by Congress, to collect fees from companies submitting applications. The device was designed to speed up review; it also creates a partial budgetary dependence on the regulated entities, which is regularly debated.
The concept of "regulatory capture" — a situation in which an authority comes to serve the interests of the sector it regulates — was theorised in political economy, notably by George Stigler (1971). It provides a framework for analysis, not an automatic accusation. If it operates, capture does not manifest itself through grossly fraudulent decisions, but through discreet inflections: a slightly lowered bar, a safety signal handled slowly, a dialogue too intimate between evaluator and evaluated. The risk factors are documented (revolving doors, cross-funding, asymmetry of expertise); they do not, on their own, establish that any particular decision was corrupted.
The limit of the judicial remedy was, moreover, seen in the Purdue case. In June 2024, the U.S. Supreme Court, in Harrington v. Purdue Pharma L.P., struck down (5 votes to 4) the bankruptcy plan that would have shielded the Sackler family from any prosecution in exchange for a payment, holding that the Bankruptcy Code does not authorise the non-consensual release of non-debtor third parties. The decision illustrates how far the restoration of accountability remains an institutional battle, waged by the checks and balances themselves.
Recognising these mechanisms means demanding more science, not less. It is exactly the opposite of an anti-vaccine or anti-science stance. To denounce a publication bias is to defend the scientific method against what corrupts it. To demand the publication of all trials is to call for more complete data to decide better. The AllTrials campaign, launched in 2013 at the initiative of the BMJ, Cochrane, PLOS, and Sense About Science among others, demands precisely the registration of all clinical trials and the publication of their results; this requirement is today partly translated into regulatory obligations, the direct fruit of these criticisms from within medicine.
The discipline of the file consists in holding two truths at once: laboratories have gravely defrauded, and evidence-based medicine remains one of the greatest achievements of our civilisation. To say "everything is rotten" would be a mistake symmetrical to naivety — and, paradoxically, this blanket distrust protects the fraudsters by drowning their precise acts in a generalised fog. Precision is more devastating to fraud than any theory.
Does a condemned off-label marketing scheme prove the drug is ineffective? No. Risperdal, Bextra, and Paxil had legitimate indications. What was found illegal was the promotion of unapproved uses by the manufacturer, not the existence or efficacy of the molecule. Distinguishing the molecule from the marketing is the golden rule.
Does publication bias mean that science lies? No. It means that incomplete access to data degrades the quality of decisions. It is a methodological critique, formulated by researchers such as Turner or Ioannidis from within medicine, and one that calls for more transparency — not a rejection of research.
Does regulatory capture prove the FDA is corrupt? No. It is a structural possibility to monitor, not a verdict. The agencies have also withdrawn dangerous products and protected millions of patients. The remedy is not distrust, but the strengthening of their financial independence and transparency.
To understand how these mechanisms played out in the major cases — OxyContin and the Sackler family, the withdrawal of Vioxx, the multibillion-dollar settlements of GSK, Pfizer, and Johnson & Johnson — see our file on Big Pharma's convictions and fines.
To go further, the full investigation Big Pharma — Documented Corruption gathers the real sources (DOJ press releases, the Supreme Court's Harrington v. Purdue Pharma decision, Goldacre's Bad Pharma, Turner 2008, Ioannidis 2005, CDC data, AllTrials, PDUFA, Stigler) and offers six markers for distinguishing documented commercial fraud from the real value of treatments.
No. Risperdal, Bextra, and Paxil had legitimate indications. What was found illegal was the promotion of unapproved uses by the manufacturer, not the existence or efficacy of the molecule. Distinguishing the molecule from the marketing is the golden rule.
No. It means that incomplete access to data degrades the quality of decisions. It is a methodological critique, formulated by researchers such as Turner or Ioannidis from within medicine, and one that calls for more transparency — not a rejection of research.
No. It is a structural possibility to monitor, not a verdict. The agencies have also withdrawn dangerous products and protected millions of patients. The remedy is not distrust, but the strengthening of their financial independence and transparency.
Dossier : Big Pharma & souveraineté alimentaire
MINI15: Big Pharma — La Corruption Documentée
Accueil · Collection · Journal · Dossiers · Sources