A doctor swallowed bacteria to prove a point, and consensus still took a decade

Key takeaways

• Barry Marshall swallowed H. pylori bacteria in 1984 to show they inflame the stomach, and an independent NIH panel endorsed antibiotics for infected ulcer patients in 1994.
• National Academies reports come from balanced, unpaid committees whose drafts face independent review, and sponsors don’t get to suggest changes.
• A 2009 paper tying a virus called XMRV to chronic fatigue syndrome was retracted in 2011, and a blinded NIH-funded test of 293 people found no infection.
• In 2015, an Institute of Medicine committee concluded that ME/CFS is a real, serious disease affecting up to 2.5 million Americans, showing that consensus can back what patients report.
• Gut-brain research is a promising hypothesis, but NIH’s own summaries stress that key findings come from mice and still need testing in people.

In 1984, a young Australian doctor named Barry Marshall swallowed a culture of bacteria². Ten days later, a biopsy showed his stomach was severely inflamed².

Marshall and pathologist Robin Warren had a theory that sounded far-fetched at the time. Most experts believed ulcers came from stress and lifestyle¹. The pair thought a germ, later named Helicobacter pylori, was the real culprit¹.

Their path was bumpy. Their first letters to The Lancet waited months for publication, and an Australian gastroenterology society turned down their conference abstract in 1983². In 2005, they won the Nobel Prize¹.

It’s a great story, and it’s often told as proof that outsiders beat the experts. The more useful lesson is how they won. They produced evidence other scientists could check, and the field changed its mind.

That’s the difference between a hypothesis, a proposed explanation still being tested, and scientific consensus, the point where evidence has converged enough that independent experts agree. Below, we look at how ideas move from one to the other, why some never make it, and how to tell where a claim sits today.

Consensus means the evidence converged, not that people voted

Consensus isn’t a poll of scientists or a decree from an agency. It’s what happens when many independent groups, using different methods, keep getting the same answer.

Think of a court case built from witnesses who never met. One witness can be mistaken. When lab studies, patient studies and treatment trials all point the same way, coincidence becomes hard to believe.

Consensus also comes in degrees. Some findings are settled for practical purposes. Others are leading explanations with known gaps. Good consensus statements tell you which is which.

How a consensus body reaches a verdict

Some groups exist to put consensus in writing. In the U.S., the best known is the National Academies of Sciences, Engineering, and Medicine. Its process has several built-in checks:

• A balanced committee. Members are chosen for a mix of expertise and serve unpaid, as individuals rather than as representatives of their employers⁴.
• Public scrutiny. A provisional member list is posted for public comment, and conflicts of interest are reviewed throughout the study⁴⁵.
• Clear labels. Reports separate findings, the facts and evidence, from conclusions and recommendations, the committee’s judgments⁴.
• Outside review. Independent experts review the draft and ask whether the evidence supports the conclusions, and the committee must respond⁴⁵.
• Room for dissent. A member who disagrees can file a dissenting opinion⁵.

The sponsor, usually a federal agency, pays for the work but doesn’t get to suggest changes to the report, according to an HHS description of the process⁵. No system is perfect, and you can still check panel members’ ties yourself. But these steps are built to keep any one person or funder from steering the answer.

NIH has used a similar approach. In February 1994, it convened an independent panel of 14 members, including public representatives, to weigh the evidence on H. pylori³. After about a day and a half of expert talks and audience discussion, the panel concluded that ulcer patients with the infection needed antibiotics along with acid-lowering drugs³.

Notice what the panel didn’t claim. It called the value of treating people with indigestion but no ulcer undetermined, and it said the link to stomach cancer needed more study³. Real consensus statements mark their own limits.

How the ulcer germ climbed from hunch to consensus

The H. pylori story shows the steps an idea has to climb.

1. Observation. Warren noticed curved bacteria in stomach biopsies from about half of his patients, always near inflammation¹.
2. Isolation. After failed attempts, Marshall grew the bacteria in the lab. The first colonies appeared on plates left over the Easter holiday in 1982¹².
3. Association. A study of 100 patients, finished in May 1982, tied the bacteria strongly to gastritis and ulcers².
4. A causal test. Marshall’s self-experiment showed the germ could inflame a healthy stomach².
5. A prediction that came true. Ulcers treated with acid-blocking drugs alone tended to return. Clear the infection, and they stayed healed¹.
6. Independent agreement. The NIH panel endorsed antibiotic treatment in 1994³, and the Nobel came in 2005¹.

The Nobel committee pointed out that nothing in their toolkit was exotic: a flexible scope to look inside the stomach, silver stains on tissue slides and ordinary lab cultures¹. With them, the pair built what the committee called an “irrefutable case”¹. They didn’t win by going around science. They won by doing more of it.

The timeline also shows why change can feel slow. From the first culture to the NIH panel took about 12 years²³. That’s frustrating if you’re the one who’s right. It’s also what protects everyone from ideas that are wrong.

Bold ideas that didn’t survive testing

For every vindicated outsider, there are bold ideas that fell apart when others tried to repeat them. Two recent examples show how science sorts them out, and both involve researchers testing their own ideas.

A virus and chronic fatigue. A 2009 report in Science raised hopes by tying chronic fatigue syndrome to XMRV, a retrovirus⁸. Other labs went looking for the virus in patients and, one after another, came up empty. The leading explanation became a lab mix-up, with mouse genetic material a likely culprit⁸. Science fully retracted the paper in December 2011⁷.

For many researchers, that ended it. Ian Lipkin, who led the follow-up, said the question was still open for patients⁸. So NIH paid for a decisive test, and the original study’s lead researcher joined as a co-author⁷⁸.

Doctors at six U.S. sites enrolled 147 patients and 146 matched healthy people, and the blood was tested blind, so labs didn’t know whose sample was whose⁷⁸. The 2012 results found no evidence of infection⁷⁸. One patient advocate summed up the outcome for the Wall Street Journal: the virus was out of the picture, but his illness was still there⁸.

Veins and multiple sclerosis. In 2009, Italian vascular specialist Paolo Zamboni proposed that poor blood drainage from the brain and spinal cord, which he called chronic cerebrospinal venous insufficiency, played a role in MS⁹¹⁰. Tens of thousands of Canadians with MS sought a vein-widening procedure, often traveling abroad, and three deaths were linked to the treatment¹⁰.

Zamboni then put his own idea to a much tougher test. In the Brave Dreams trial, run at six MS centers in Italy, 115 patients were assigned by chance to one of two groups⁹¹⁰. Seventy-six had their veins widened. The other 39 got a sham: a catheter was threaded into the veins, but nothing was widened⁹.

The results appeared online in JAMA Neurology in November 2017⁹. Twelve months on, the groups looked alike on walking, balance, hand skills, bladder function and eyesight⁹. Zamboni and his co-authors wrote that the treatment should not be offered to people with MS⁹.

That last detail deserves respect. Testing your own favorite idea, and reporting that it failed, is science at its best.

Being doubted doesn’t make you right

Some promoters answer criticism by comparing themselves to Galileo, or to Marshall. The reasoning goes like this: brilliant people were mocked, I’m being mocked, so I must be brilliant. Skeptics call it the Galileo gambit.

The flaw is simple. Being doubted is something the right and the wrong have in common. The XMRV and vein researchers faced doubt too.

What separated Marshall from them wasn’t the doubt. It was what happened when other people ran the tests.

So when someone says “they laughed at Galileo,” ask a better question: who has repeated your results? This comparison, our own synthesis of the cases above, can help.

Signs a hypothesis is gaining ground	Signs a hypothesis is stuck
Independent teams, including skeptics, get the same result	Only the original group or its allies find the effect
The effect holds up in blinded or sham-controlled tests	The effect shrinks as studies get larger and more careful
A mechanism emerges that explains how it works	A new explanation appears after each failure
Predictions come true, like cures after a germ is cleared	Supporters point to persecution instead of new data
Its champions run hard tests of their own idea	It’s sold to the public before it’s tested

Consensus changes, and that’s the system working

Consensus isn’t permanent, and it shouldn’t be. The 2010 Dietary Guidelines capped cholesterol from food at 300 milligrams a day¹². In 2015, the government’s advisory committee said the evidence no longer supported treating dietary cholesterol as a nutrient people overeat¹².

Advice on menopausal hormone therapy has shifted more than once. In November 2025, the FDA asked makers to remove boxed-warning language about heart disease, breast cancer and probable dementia¹³. It noted that women in the key trials averaged 63, while bothersome hot flashes usually start between about 45 and 55¹³. Heart and breast cancer warnings stay elsewhere in the labels, and estrogen-only pills keep a boxed warning about uterine cancer¹³.

In both cases, the shift came from new evidence or a closer look at who had been studied, not from someone shouting louder. Our articles on expert advice and on observational studies tell both stories in full.

Change can also bring overdue recognition. In 2015, an Institute of Medicine committee called ME/CFS a “legitimate, serious, and complex systemic disease”⁶. It estimated that 836,000 to 2.5 million Americans have it, and that 84% to 91% hadn’t been diagnosed⁶.

The committee also said the old name, chronic fatigue syndrome, invited dismissive attitudes, and it reported that 67% to 77% of patients waited more than a year for a diagnosis⁶. Patients who had felt brushed off for years found a consensus body on their side.

Emerging, not established: the gut-brain axis

Some of today’s most exciting health ideas sit in the middle. The gut-brain axis, the two-way signaling between your digestive tract and your brain, is one of them.

Your gut holds tens of trillions of microbes, and research suggests signals travel from gut to brain as well as the other way¹¹. In a 2011 experiment, mice given a specific probiotic acted more relaxed¹¹. In a 2026 NIH-funded study, giving young mice the gut microbes of older mice worsened their memory by weakening a signal along the vagus nerve, a major nerve linking gut and brain¹⁴.

NIH’s own summary of that study adds a caution: the findings matter for people only if they hold true in humans¹⁴. A 2025 review of mood disorders called early results from diet changes, probiotics and stool transplants encouraging, but said major advances are still needed¹⁵.

That’s an emerging hypothesis in good health. It has a plausible mechanism, a growing body of animal work and serious researchers testing it. What it still lacks is consistent results from human trials. A product that promises to “fix your gut-brain axis” is selling a conclusion science hasn’t reached.

Where does this claim sit? A six-rung ladder

Use this ladder, our own synthesis, to place any health claim. Ask what has been done, not what has been promised.

1. An idea or a story. A theory, a hunch or a personal experience.
2. Early signals. Lab or animal studies, or one small study in people.
3. Repeated in people. Other teams find the same thing in human studies.
4. Fair tests agree. Randomized, blinded or sham-controlled trials confirm it, and a mechanism makes sense.
5. Expert bodies weigh in. Systematic reviews and consensus panels endorse it.
6. Lasting consensus. It holds up through years of new challenges.

Natural-health traditions often hold ideas on the lower rungs that deserve testing, not dismissal. Respecting an idea means helping it climb, the way Marshall did. Until it climbs, try it only if it’s low-risk, and never in place of proven care for a serious condition.

What the evidence shows for the examples here

Claim	Evidence level	Basis
H. pylori causes most peptic ulcers, and clearing it cures them	Strong evidence	Treatment studies, an NIH consensus panel and a Nobel Prize13
The XMRV virus causes chronic fatigue syndrome	Evidence says no, or unsafe	A blinded multisite test found no infection; the original paper was retracted78
Widening veins that drain the brain and spinal cord treats multiple sclerosis	Evidence says no, or unsafe	A sham-controlled trial found no benefit; deaths were linked to the procedure910
Gut microbes shape mood and memory in people	Promising but early	Strong animal work; human trials still needed111415

Questions readers ask

The bottom line

Science does change its mind, and outsiders are sometimes right. But they’re proven right the same way everyone else is: by results others can repeat, predictions that come true and evidence that convinces independent experts.

Before you accept a claim or dismiss one, find its rung. A hypothesis deserves curiosity, and a consensus deserves weight. Neither deserves blind faith.

Related reading

• Why observational studies got smoking right and hormone therapy wrong
• When expert advice deserves weight, now that patients have the right to decide
• Drugs that quieted extra heartbeats raised deaths, a lesson in reading a study’s limits
• Mi’kmaw elders offer a better way to bring tradition and science together
• How to tell easing a symptom from treating a disease

Sources

1. Nobel Assembly at Karolinska Institutet. “Press Release: The Nobel Prize in Physiology or Medicine 2005.” NobelPrize.org, October 3, 2005. nobelprize.org
2. Van Der Weyden MB, Armstrong RM, Gregory AT. “The 2005 Nobel Prize in Physiology or Medicine.” Medical Journal of Australia, 2005. mja.com.au
3. NIH Consensus Development Panel. “Helicobacter pylori in Peptic Ulcer Disease.” NIH Consensus Statement 1994;12(1):1–23; conference held February 7–9, 1994 (archived government copy, marked historical).
4. National Academies of Sciences, Engineering, and Medicine. Committee briefing slides describing the consensus study process (Committee on Improving the Health and Wellbeing of Children and Adolescents Through Health Care System Transformation), February 28, 2023. nationalacademies.org
5. U.S. Department of Health and Human Services. “IOM Study Talking Points.” Chronic Fatigue Syndrome Advisory Committee meeting presentation. hhs.gov
6. National Academies of Sciences, Engineering, and Medicine. “Chronic Fatigue Syndrome/Myalgic Encephalomyelitis Is a Legitimate Disease That Needs Proper Diagnosis and Treatment, Says IOM Report.” News release, February 10, 2015. nationalacademies.org
7. Retraction Watch. “New Study Puts Speculation to Rest About Link Between XMRV and Chronic Fatigue Syndrome.” September 18, 2012. retractionwatch.com
8. Marcus AD. “Virus Disproved as Cause of Chronic Fatigue Syndrome.” Wall Street Journal, September 18, 2012 (reprinted by the ME Association). meassociation.org.uk
9. TCTMD. “Dreams Dashed: Intravascular Therapy for MS Flops in Sham-Controlled Trial” (on Zamboni P, et al., JAMA Neurology, 2017). December 5, 2017. tctmd.com
10. Maclean’s. “A Conflicted Clinical Trial Finds Treatment for MS Is ‘Safe’ but ‘Largely Ineffective.'” November 2017. macleans.ca
11. National Center for Complementary and Integrative Health. “4 Fast Facts About the Gut-Brain Connection.” NCCIH, NIH. nccih.nih.gov
12. ACP Internist. “Guidelines Diminish Cholesterol Worries.” American College of Physicians, June 2015. acpinternist.org
13. U.S. Food and Drug Administration. “FDA Requests Labeling Changes Related to Safety Information to Clarify the Benefit/Risk Considerations for Menopausal Hormone Therapies.” FDA, November 10, 2025 (content current February 2026). fda.gov
14. National Institutes of Health. “Weakened Gut-Brain Connection May Contribute to Memory Loss.” NIH Research Matters, April 28, 2026. nih.gov
15. Marano G, Rossi S, Sfratta G, et al. “Gut Microbiota: A New Challenge in Mood Disorder Research.” Life, 2025 (abstract via Università Cattolica del Sacro Cuore). pubmed.ncbi.nlm.nih.gov