Imagine a coin that lands on heads half the time. Now imagine someone flips it a hundred times but only tells you about the heads. You would come away convinced the coin was rigged in your favor. That, in simple terms, is what happens when drug trials with disappointing results quietly disappear while the successful ones are published, promoted, and taught to doctors.
This problem has a name: publication bias, or the “missing trials” problem. It is not a conspiracy theory. It has been measured in leading medical journals, admitted in court settlements, and addressed (imperfectly) by federal law. Understanding it is one of the most useful things a patient can do, because it explains why the evidence behind a prescription can look stronger on paper than it really is, and it gives you clear questions to ask.
Key Takeaways
- A landmark 2008 study in the New England Journal of Medicine found that 94% of published antidepressant trials looked positive, while the FDA’s own review rated only 51% as positive.
- A 2015 reanalysis of GlaxoSmithKline’s “Study 329” concluded that paroxetine (Paxil) was no better than placebo for depressed teens and was linked to suicidal behavior, contrary to the original 2001 publication.
- In 2012, GSK pleaded guilty and agreed to pay $3 billion in a U.S. fraud settlement that included misbranding Paxil for patients under 18.
- When independent researchers finally obtained the full Tamiflu trial data, they found it shortened flu symptoms by less than a day in adults, with no clear reduction in hospital admissions.
- Even under a 2007 U.S. law requiring results to be reported, a 2020 study in The Lancet found that only 41% of trials did so on time.
- Patients can protect themselves by asking about absolute benefits, looking up trials, checking industry payments, and seeking independent reviews.
How Missing Trials Distort Medicine
Doctors, guideline committees, and medical students rely on published studies. If negative results never reach journals, or are written up to sound positive, everyone downstream inherits a distorted picture. No individual doctor has to be dishonest for this to happen; the bias is built into what gets seen.
The Antidepressant Study That Opened Eyes
In 2008, psychiatrist Erick Turner and colleagues compared what the U.S. Food and Drug Administration knew about 12 antidepressants with what had appeared in medical journals. They examined 74 trials registered with the FDA. Of the 38 trials the FDA judged positive, 37 were published. Of the trials the FDA judged negative or questionable, most were either never published or were published in a way that made them look positive; only three were published honestly as negative.
The result: a reader of the medical literature would believe 94% of trials were positive. The FDA’s own figure was 51%. Nearly a third of the trials, involving more than 3,400 patients, never appeared in print at all. Those patients had accepted the risks of research, and their results were effectively hidden.
Study 329: Rewriting the Story
In 2001, a journal article reported that the antidepressant paroxetine (Paxil) was effective and generally well tolerated in adolescents with major depression. The study, known as Study 329, was funded by the drug’s maker, SmithKline Beecham (later GlaxoSmithKline).
In 2012, the U.S. Department of Justice announced that GSK would plead guilty and pay $3 billion to resolve criminal and civil allegations. Among them, the government alleged that GSK helped prepare and distribute a misleading journal article reporting that Paxil worked for depression in patients under 18 when the study had failed to show that, and that it withheld data from two other pediatric studies in which Paxil also failed.
In 2015, an independent team led by Joanna Le Noury reanalyzed the original Study 329 data under the “Restoring Invisible and Abandoned Trials” initiative and published the results in The BMJ. They found that neither paroxetine nor the comparison drug imipramine was significantly better than placebo on any of the prespecified outcomes, and that paroxetine was associated with increased suicidal thinking and behavior. The same data told a very different story once independent eyes could see all of it.
Tamiflu: Billions Spent Before the Full Data Was Public
Governments around the world stockpiled the flu drug oseltamivir (Tamiflu) during pandemic preparations. Yet for years, independent Cochrane reviewers struggled to get the full clinical study reports from the manufacturer, Roche. After a long public campaign, they obtained thousands of pages of regulatory documents covering 83 trials.
Their 2014 analysis in The BMJ, by Tom Jefferson, Carl Heneghan, Peter Doshi, and colleagues, found that in adults Tamiflu shortened the time to first relief of flu symptoms by about 16.8 hours. Treatment trials showed no difference in hospital admissions, and the drug increased the risk of nausea and vomiting. Those findings are a far cry from the impression many people had when the stockpiles were bought.
The Law Changed. Compliance Didn’t Fully Follow.
In 2007, the U.S. Food and Drug Administration Amendments Act (FDAAA) required many clinical trials to post their results on the public registry ClinicalTrials.gov, generally within one year of completion. Stronger rules took effect in 2017, with fines of up to $10,000 a day for violations.
In 2020, Nicholas DeVito, Seb Bacon, and Ben Goldacre of the University of Oxford published an analysis in The Lancet of more than 4,200 trials that were legally required to report results. Only 41% had reported on time, and about 36% had not reported at all. Industry sponsors actually did better (50% on time) than universities and other non-industry sponsors (34%) or U.S. government sponsors (31%). The authors estimated that more than $4 billion in fines could have been collected if the law had been fully enforced.
The lesson is that the problem is systemic. It involves companies, academic institutions, and regulators, and it will only improve when patients and the public keep asking for full transparency.
The Evidence at a Glance
| Case | What was found | Source |
|---|---|---|
| Antidepressant trials | 94% looked positive in journals vs. 51% in FDA review | Turner et al., NEJM 2008 |
| Study 329 (Paxil in teens) | No benefit over placebo; increased suicidal behavior on reanalysis | Le Noury et al., BMJ 2015 |
| GSK settlement | $3 billion; guilty plea included misbranding Paxil for under-18s | U.S. Department of Justice, 2012 |
| Tamiflu | About 16.8 hours faster symptom relief in adults; no difference in admissions | Jefferson et al., BMJ 2014 |
| U.S. reporting law | Only 41% of required trials reported results on time | DeVito et al., Lancet 2020 |
Why This Matters for Natural Health, Too
Publication bias is not unique to drugs. Any field can favor exciting positive results, including research on herbs and supplements, which is why this blog tries to report the size and quality of evidence, not just the headline. But there is an imbalance worth noticing: a patented drug can attract hundreds of millions of dollars in research and marketing, while a food, herb, or lifestyle change that nobody can own rarely gets that kind of investment. We explored that funding gap in our article on profit and research disparities, and in our look at how little nutrition doctors are taught.
How to Protect Yourself: A Practical Toolkit
- Ask for absolute numbers. “Cuts risk by 30%” can mean going from 3 in 100 to 2 in 100. Ask your practitioner, “Out of 100 people like me who take this, how many benefit, and how many have side effects?”
- Look up the trials yourself. Search the drug or treatment name on ClinicalTrials.gov to see registered studies, including those that finished without published results.
- Seek independent reviews. Cochrane reviews include free plain-language summaries and are among the most cautious sources available.
- Check industry payments. The U.S. federal Open Payments database shows payments that drug and device companies make to doctors. A payment doesn’t mean bad advice, but it is useful context.
- Ask what happens if you wait or try lifestyle changes first. For many conditions, diet, movement, sleep, and stress care are legitimate first steps or valuable additions.
- Report side effects. You can submit reports directly to the FDA’s MedWatch program. Patient reports help uncover problems trials missed.
- Never stop a prescribed medicine abruptly on your own. Some medicines, including antidepressants, can cause withdrawal effects. Bring your questions to your prescriber and make changes together.
For more on recognizing warning signs in the healthcare system, see When Medicine Betrays Patients.
The Bottom Line
Hidden and spun clinical trials are a documented, measurable problem. The antidepressant record, Study 329, Tamiflu, and the gaps in legally required reporting all show how the evidence presented to doctors and patients can look better than the full data. The answer is not to distrust everything, but to become an informed, engaged patient: ask for real numbers, look for independent reviews, and make decisions in partnership with practitioners who welcome your questions.
Frequently Asked Questions
Does publication bias mean all medications are ineffective?
No. Many medicines have strong, well-documented benefits. Publication bias means the benefits of some treatments may be overstated and their harms understated, which is why the full picture matters.
Has anything improved since these cases came to light?
Yes, somewhat. Trial registration is now expected by major journals, U.S. law requires results reporting, and European regulators publish more clinical study reports. Compliance is still incomplete, so public pressure continues to matter.
How can I tell if a study I read about is trustworthy?
Look at who funded it, how many people took part, whether it was randomized and placebo-controlled, and whether results are reported in absolute terms. Systematic reviews that pool many trials are generally more reliable than a single study.
What is the RIAT initiative?
“Restoring Invisible and Abandoned Trials” encourages independent researchers to publish or correct trials that were never reported or were misreported, as happened with Study 329.
Sources
- Turner EH, et al. Selective publication of antidepressant trials and its influence on apparent efficacy. New England Journal of Medicine, 2008 (Psychiatric Times summary).
- Le Noury J, et al. Restoring Study 329: efficacy and harms of paroxetine and imipramine in treatment of major depression in adolescence. The BMJ, 2015.
- U.S. Department of Justice. GlaxoSmithKline to plead guilty and pay $3 billion to resolve fraud allegations and failure to report safety data. 2012.
- Jefferson T, Jones M, Doshi P, Spencer EA, Onakpoya I, Heneghan CJ. Oseltamivir for influenza in adults and children: systematic review of clinical study reports and summary of regulatory comments. The BMJ, 2014.
- DeVito NJ, Bacon S, Goldacre B. Compliance with legal requirement to report clinical trial results on ClinicalTrials.gov: a cohort study. The Lancet, 2020 (University of Oxford summary).
This article is for educational purposes and is not a substitute for personalized advice from a qualified health practitioner.
