Helicobacter pylori: treat it or live with it

The finding usually turns up in passing. A letter after a gastroscopy, a line on a lab report, a stool test someone ordered themselves. And then there is a name you had heard at most in passing before.

Helicobacter pylori. Two words that lead you within ten minutes on the internet into the neighbourhood of stomach cancer.

Many people know that evening. On one screen it says every stomach should be freed of this bacterium. On the next it says we are currently losing an ancient companion and are buying heartburn and allergies in exchange. Both sides quote studies, both sound certain.

So I am not starting with reassurance and not with a warning, but with a sorting. There are situations in which the answer is unambiguous: a gastric or duodenal ulcer, a MALT lymphoma, the state after removal of an early gastric cancer, an immune thrombocytopenia, an unexplained iron deficiency anaemia. There the guidelines do not say may, they say should. And next to that lies a broad grey zone in which two German professors have publicly disagreed for years. That grey zone is precisely the reason for this article.

These three numbers belong together. The factor alone frightens. The absolute risk alone makes you careless. And the number of people who have to be treated so that a single one benefits explains why the answer comes out differently in Japan than in Germany.

What this bacterium is, how many carry it and where it comes from

Picture the stomach as a basin of hydrochloric acid. On an empty stomach the pH sits at around 1 to 2. That is a place where almost nothing survives. Which is exactly why the stomach was considered sterile for decades.

Helicobacter pylori has three tricks for this basin. It is built in a spiral and carries flagella, small whips with which it drills through the thick mucus layer that lies on the stomach wall. Down there, right at the cell layer, the pH is almost neutral. Then it produces the enzyme urease, which splits urea into ammonia and carbon dioxide, and ammonia buffers the acid in its surroundings. So it lays a cloud of base around itself. And third, it attaches to the cells with adhesins and stays.

This urease, by the way, is not only its survival trick. It is also the basis of two of the most important tests, and we will come back to that later.

For Germany the S2k guideline of the German Society for Gastroenterology, Digestive and Metabolic Diseases names a prevalence of 35.3 percent, with a confidence interval of 31.2 to 39.4 percent. Within Europe the range runs from 18.9 percent in Switzerland to 86.4 percent in Portugal. The individual German cohorts differ: 44 to 48 percent seroprevalence in older adults, 28.9 percent in blood donors, and in German children between seven and nine years old without a migration background a stable figure of around 9 percent.

That last value is the most interesting one. It shows that the number is falling across the generations. Anyone who is seventy today lived as a child in a different world: more people per flat, different sanitary conditions, different numbers of children. In Western Europe the bacterium appears to be slowly receding on its own, without anyone having fought it.

Where the bacterium comes from, and why it was mostly childhood

Almost all infections arise in the first years of life. Not at a buffet, not on holiday, not through bad food. Rather within your own household, mostly before the sixth birthday.

For Germany there is a matching figure: in the ESTHER cohort in Saarland with 9,444 people between 50 and 74 years, the probability of infection rose with the number of siblings, from an odds ratio of 1.45 with four siblings up to 1.84 with seven or more (Gao L et al. Int J Epidemiol. 2010;39(1):129-134. PMID: 19596750 · DOI: 10.1093/ije/dyp250, [Cohort, n=9,444]). This is a question of your biography and your childhood, not of your hygiene. Nobody did anything wrong here.

The most common follow up question is: will I infect my partner, my children, my grandchildren. Transmission in adulthood is rare, and the reinfection rate after successful treatment in developed countries is below 1 percent per year according to the guideline. For children a separate rule applies: the German guideline explicitly advises against a test and treat strategy in childhood, considers serology unsuitable in children and requires resistance testing before a first therapy.

And now you know why the answer to the question of origin lies almost always in your childhood and not in your fridge.

Marshall, Warren and the self experiment that overturned a textbook belief

Into the nineteen eighties the matter was clear. A gastric ulcer arose from stress and too much acid. It was treated with bland food, with rest, with acid suppression, sometimes with surgery. And then you waited for the next ulcer, because it mostly came. That bacteria could live in a stomach was considered absurd.

That was not enough for the professional world. A bacterium found in sick stomachs can be a cause or a passenger. To separate the two there has been a rule since Robert Koch: you have to show that the pathogen triggers the disease in a healthy organism. Only nobody wanted to drink a bacterial culture voluntarily.

And now you know why I am doing two things at once in this text: naming the evidence clearly and leaving the open edges visible.

What it can do, and how often

The most honest answer has two parts. In the majority of people: nothing that ever becomes noticeable. In a minority: a chain of events that runs over decades.

The rule is chronic active gastritis. It is present in practically everyone infected and mostly causes no complaints. Several paths lead away from there, and which one a person takes depends partly on which part of the stomach the inflammation emphasises.

The ulcer: the best supported part of the topic

The least disputed part of the topic is the peptic ulcer, meaning the ulcer in the stomach or duodenum. Anyone who has had one and keeps the bacterium, according to the available data, often gets a new one.

One point belongs here: Helicobacter pylori is not the only cause of an ulcer. The second large group is painkillers from the NSAID group such as ibuprofen and diclofenac, and ASA, often taken over long periods and often without gastric protection. If your Helicobacter test is negative and the complaints remain, your painkiller use belongs on the table too. And where both come together, both risks belong addressed.

MALT lymphoma: the cancer where the guidelines think of the bacterium first

Gastric MALT lymphoma is rare. It is, however, the strongest single piece of evidence that this bacterium is more than a silent companion. The German guideline puts the relative risk at around a factor of 6.

The first description is one of the shortest and most consequential papers in gastroenterology: six people with confirmed low grade MALT lymphoma received antibiotic therapy, in all six the bacterium disappeared, and in five the control biopsy showed no lymphoma any more (Wotherspoon AC et al. Lancet. 1993;342(8871):575-577. PMID: 8102719 · DOI: 10.1016/0140-6736(93)91409-f, [Case Series, n=6]).

Gastric cancer: the classification and the order of magnitude

In 1994 the International Agency for Research on Cancer of the World Health Organization classified Helicobacter pylori as a class 1 carcinogen, meaning the highest category. The updated assessment calls it the causal and most important risk factor for gastric cancer, and around 90 percent of carcinomas outside the gastric cardia worldwide are attributed to this infection.

This classification says something about the strength of the association, not about your personal risk. That is exactly the place where it pays to look twice.

This Japanese figure is famous, and many texts stop here. I deliberately place the Western figure right next to it.

Why the numbers between Japan and Germany lie so far apart has at least three reasons: the frequency of stomach cancer in the underlying population, the dietary history with salt and preserved foods, and the bacterial strains themselves. For CagA positive strains the odds ratio in a review by authors from the cancer research agency was 2.87 compared with 2.31 for the infection as such, and the German guideline names an 18 to 20 fold increased risk for persistently detectable CagA antibodies (Park JY et al. Toxins (Basel). 2018;10(4):163. PMID: 29671784 · DOI: 10.3390/toxins10040163, [Mechanism Review]). So the bacterium is not one thing but a family of strains with differing aggressiveness.

Two reasons for testing that surprise many people

Besides stomach complaints, quite different specialties also think of this bacterium. The German guideline names immune thrombocytopenia, meaning a lack of platelets through an immune reaction, and the unexplained or treatment resistant iron deficiency anaemia. In both cases the wording is should.

For iron deficiency anaemia an honest addition belongs here: the guideline cites meta analyses with odds ratios between 1.33 and 2.22 and at the same time records that in five randomised studies eradication did not meaningfully improve haemoglobin and ferritin. Why iron can be lost in the gut or fail to arrive at all is covered in the article on iron deficiency and absorption problems.

A second point belongs exactly here, because it is often missed in this constellation. Behind an unexplained iron deficiency anaemia there can also be coeliac disease, and both work ups often run alongside each other. The order matters: anyone who eats gluten free on their own initiative before the coeliac work up has been done can make the diagnosis impossible, because antibodies in the blood and the tissue sample can come back falsely negative on a gluten free diet. So get tested first, then change the diet, and have the change medically accompanied. How this work up runs is covered in the article on recognising coeliac disease.

A small association is also found with bowel cancer: in the German DACHS study the odds ratio after full adjustment was 1.18 (Zhang Y et al. Am J Epidemiol. 2012;175(5):441-450. PMID: 22294430 · DOI: 10.1093/aje/kwr331, [Cohort, case control, n=3,381]). That is an observation and not proof of causation. And if the test is negative and the complaints remain, it is worth looking at the text on irritable stomach and functional dyspepsia.

And now you know why the question is not whether this bacterium can be dangerous, but how large the danger is for you personally.

What treatment can do for the cancer risk

Now comes the core. If the infection raises the risk, does treating it lower the risk again. And if so, by how much.

This question is better studied than most people assume. It just cannot be answered with a single number.

And now a part that rarely turns up in texts written for the public. The same seven studies, evaluated with a different statistical model.

So how much benefit sits in a treatment in concrete terms. For that there is a figure I like a great deal, because it is honest: the number of people you have to treat so that a single one of them benefits.

Two groups in which the evidence is very clear

Is there a point at which it is too late

In many forums this question counts as settled. It is not. A meta analysis of ten evaluations from eight randomised studies with 7,955 people found an overall relative risk of 0.64. In people without intestinal metaplasia it fell to 0.25, while with metaplasia or dysplasia already present it was 0.88 and therefore without a detectable advantage (Chen HN et al. Gastric Cancer. 2016;19(1):166-175. PMID: 25609452 · DOI: 10.1007/s10120-015-0462-7, [Meta-analysis, k=8]).

The German guideline explicitly counters this subgroup analysis by saying that the argument that beyond the age of fifty the cascade can no longer be reversed may be considered refuted. And in 2018 Choi showed an improvement in the degree of atrophy even after an already removed early cancer. Both sides belong side by side, and both stand here.

Some context comes from the East Asian data. There the incidences in the control groups were 272.7 per 100,000 person years with atrophic gastritis and 1,790.7 per 100,000 person years after cancer surgery (Sugimoto M, Murata M, Yamaoka Y. World J Gastroenterol. 2020;26(15):1820-1840. PMID: 32351296 · DOI: 10.3748/wjg.v26.i15.1820, [Meta-analysis, cohort data]). With starting numbers like these, a relative halving moves very many people. In Germany the starting number is smaller, and the same relative effect moves correspondingly fewer.

And now you know why the question of benefit cannot be answered with a single percentage, but depends on which group you are standing in.

The other side: are we losing an old companion?

There is a second story about this bacterium, and it does not come from the internet but from research. Helicobacter pylori has accompanied humans for tens of thousands of years, as genetic analyses suggest, and its variants appear to follow human migration routes. And within a few generations it is disappearing in Western countries.

In parallel, reflux disease, Barrett's oesophagus, oesophageal cancer, asthma and allergies have increased in the same countries. The question that follows is legitimate: are we losing something there. I am running this debate here because the German guideline runs it itself. And I am running it in a way that makes clear at the end where the data stop.

On asthma and allergies there is an often quoted cross sectional analysis with 7,412 participants from a large American health survey: in 3 to 13 year olds the odds ratio for current asthma was 0.41, and inverse associations were also found for wheezing, allergic rhinitis and eczema (Chen Y, Blaser MJ. J Infect Dis. 2008;198(4):553-560. PMID: 18598192 · DOI: 10.1086/590158, [Cohort, cross sectional, n=7,412]). A cross section is the weakest design when it comes to cause and effect, and it says nothing about what a treatment would do.

Why no recommendation follows from this

First. Association is not causation. The network analysis shows this on itself: the association disappears exactly at the point where it counts clinically, namely at the step from Barrett's to carcinoma.

Second. The guideline writes that for adenocarcinoma of the oesophagus a plausible causality has so far not been presented, and that there is no association between eradication and the occurrence of Barrett's or adenocarcinoma. That is the decisive difference: the question is not whether infected people have Barrett's less often, but whether a treatment triggers Barrett's. And for that there is no evidence.

Third. On the worry that heartburn gets worse after treatment, the guideline puts it carefully: in the majority of earlier studies a negative influence could not be shown, newer meta analyses suggest that an erosive inflammation of the oesophagus can occur, and overall the evidence is contradictory. If the topic interests you in its own right, it is covered in the article on heartburn and acid blockers.

One further point is notable: in the data of the same American health survey, the infection was not associated with increased overall mortality. And in the Cochrane analysis, overall mortality under treatment likewise stayed unchanged. Both observations belong in the honest balance sheet.

And now you know why the counterarguments have to appear in a good text and why they still do not carry the conclusion often pinned on them online.

The tests, and the mistake that happens most often

There are five common ways to detect this bacterium, and they do not all answer the same question. That is the first point at which a lot goes wrong.

There are more situations in which a test comes back falsely negative. Statement 2.6 of the guideline names, besides acid blockers and recent antibiotics, an acute upper gastrointestinal bleed, a partial gastric resection, extensive atrophy with intestinal metaplasia, gastric cancer and MALT lymphoma. With an acute bleed the sensitivity of all direct tests is reduced by around 5 to 30 percent.

When the test comes back falsely positive, and what that has to do with stomach acid

The opposite mistake is rarer, but it fits this cluster very well. Two of the tests, the breath test and the rapid urease test on the biopsy, do not measure the bacterium itself but its urease activity. So they measure an enzyme.

And Helicobacter pylori is not the only bacterium that forms urease. The German guideline explicitly names that a bacterial overgrowth of the stomach or small intestine with other urease formers can lead to falsely positive results. This is favoured by delayed gastrointestinal motility, by hypochlorhydria, meaning too little stomach acid, and by ongoing therapy with proton pump inhibitors.

This is one of the points where my view of digestion and the guideline meet surprisingly well. In my practice, with upper abdominal complaints I first ask whether enough acid is arriving up there, before I go any further. Why that question makes sense is covered in the article on low stomach acid and betaine HCl, and what happens when bacteria take over in the small intestine is covered in the article on SIBO. One more distinction, because it is often confused: the Helicobacter stool antigen test is something entirely different from a microbiome stool test, and there is more on that in the article on stool testing and dysbiosis diagnostics.

And now you know why a negative finding sometimes only means that the test was done at the wrong moment.

The treatment: regimens, resistance, adherence, follow up

Once the decision for a treatment has been made, it comes down to one practical question: how do you manage to make the first attempt the only one.

The German S2k guideline recommends in recommendation 5.4 that first line therapy should preferably use a bismuth containing quadruple therapy over at least 10 days. The triple therapy with clarithromycin that used to be common is no longer recommended first line. The American ACG guideline of 2024 arrives at the same regimen where susceptibility is unknown, though over 14 days, and names a rifabutin triple therapy or a dual therapy with a potassium competitive acid blocker as alternatives. The substance meant there, vonoprazan, is not approved in Germany, so that alternative is not available here. The European consensus report Maastricht VI goes further and treats the infection fundamentally as a disease, with its own entry in the international classification of diseases.

Which regimen comes into question for you is decided by previous treatments, allergies, concurrent medication and the regional resistance situation. This choice belongs in medical hands and is nothing that can be derived from an article.

Why the resistance situation stands above everything

For Germany the guideline names, from the data of 2015 to 2018, values of 11.3 percent for clarithromycin and 13.4 percent for levofloxacin, but records that at least 15 percent must now be assumed. And now comes the number that changes everything.

The German guideline arrives at the same orders of magnitude: after one failed therapy around 60 percent clarithromycin resistance, after two failures around 80 percent, and more than 60 percent of isolates are then resistant to clarithromycin and metronidazole at the same time. For an existing clarithromycin resistance, the guideline puts the success rate of the classic triple therapy 66 percent lower.

The guideline names in statement 5.1 exactly three modifiable success factors: adherence, smoking and the extent of acid suppression. Statement 5.2 records that for the eradication indication itself no absolute contraindications are known, and that simply repeating a regimen that has already failed is not an option. That expressly does not mean that the antibiotics used are harmless. Every single active substance has its own contraindications, warnings and interactions, and the prescribing practice checks those for you.

The third point surprises many people. In this therapy the acid blocker is not an accessory against heartburn, it is part of the treatment. The more stably the pH in the stomach is raised, the better the antibiotics can probably do their job. One pointer to that: in a meta analysis the first line eradication rate was higher under the more potent potassium competitive acid blocker vonoprazan than under a proton pump inhibitor, relative risk 1.13. Important for Germany: vonoprazan is not approved here, so the figure shows a principle and not a treatment option in this country (Simadibrata DM, Syam AF, Lee YY. J Gastroenterol Hepatol. 2022;37(12):2217-2228. PMID: 36181401 · DOI: 10.1111/jgh.16017, [Meta-analysis, k=19, n=7,023]).

Side effects, honestly quantified

Typical are a metallic or bitter taste, nausea, loose stools, abdominal pressure and fatigue. Under bismuth containing regimens the stool turns dark, which is to be expected and should not be confused with tarry stools. Anything new, anything severe or anything that worries you still belongs in a conversation with the prescribing practice.

The check afterwards, and what the course does to the gut

The German guideline says: the confirmation of treatment success should take place at the earliest 4 weeks after the end of the antibiotics and at least 2 weeks after the end of acid suppression. The American guideline requires a test of cure after every therapy. Why this counts you read above: in the veterans cohort only confirmed eradication lowered the later risk. Anyone who does the course and skips the check has taken the side effects and not secured the benefit.

On reinfection: in developed countries the rate according to the guideline is below 1 percent per year, provided treatment and confirmation were done correctly. Routine reinfection checks should not be carried out.

And now you know why with this topic the care taken in the first round counts more than any fine detail in the regimen.

What can make sense alongside, and what cannot

Few topics attract as many recommendations as this one. Broccoli sprouts, mastic gum, cranberry, probiotics. The studies behind them really do exist, their results are simply more modest than they look at first glance. I go through them in the order of their strength of evidence, with the real numbers, so that you can judge for yourself.

Probiotics: the most honest case in the whole field

Compared with standard therapy alone, Lactobacillus looks good: in 11 randomised studies with 724 people the eradication rate was higher, relative risk 1.16, and taste disturbances occurred less often (Yu M et al. PLoS One. 2019;14(10):e0223309. PMID: 31577828 · DOI: 10.1371/journal.pone.0223309, [Meta-analysis, k=11, n=724]). This comparison has a weakness, however, and the next paper shows it.

In children the data look more favourable. In 15 randomised studies with 2,156 children the eradication rate with Saccharomyces boulardii was 87.7 versus 75.9 percent, and side effects 9.2 versus 29.2 percent (Liu LH et al. BMC Infect Dis. 2023;23(1):878. PMID: 38102568 · DOI: 10.1186/s12879-023-08896-4, [Meta-analysis, k=15, n=2,156]). A placebo comparison was missing here, though. Which form and which strain may make sense is covered in the article on probiotics in spore form versus capsule.

Sulforaphane from broccoli sprouts

A practical note on that figure: raw sprouts can carry germs. The German Federal Institute for Risk Assessment advises pregnant women, small children, older people and people with a weakened immune system against eating raw sprouts. The 70 grams are a study figure and not a recommendation from me. If you want to try something like this, it belongs discussed beforehand.

Mastic gum and cranberry

On mastic gum there is exactly one randomised study: 52 infected people, four groups. The bacterium was subsequently no longer detectable in 4 of 13 in the first mastic group, in 5 of 13 in the second, in 0 of 13 in the combination with an acid blocker and in 10 of 13 under standard therapy (Dabos KJ et al. Phytomedicine. 2010;17(3-4):296-299. PMID: 19879118 · DOI: 10.1016/j.phymed.2009.09.010, [RCT, n=52]). A measurable effect in humans, and it is not enough. Notable and rarely mentioned: the combination with the acid blocker was the only group without a single success.

An older study from the same research region with 189 people arrived at a similar picture: after 35 days, 14.43 percent in the cranberry group were negative compared with 5.44 percent on placebo (Zhang L et al. Helicobacter. 2005;10(2):139-145. PMID: 15810945 · DOI: 10.1111/j.1523-5378.2005.00301.x, [RCT, n=189]).

And now you know why, on current evidence, broccoli sprouts are no replacement for an indicated treatment.

Who should be tested and treated

Now to the practical sorting. The German guideline distinguishes three levels, and this distinction takes a lot of the heaviness out of the topic.

On functional dyspepsia one more number, because it otherwise sounds too friendly: in the same evaluation, side effects occurred with a relative risk of 2.19, and treatment discontinuations because of side effects with 2.60 (Ford AC et al. Gut. 2022;71(10):1967-1975. PMID: 35022266 · DOI: 10.1136/gutjnl-2021-326583, [Meta-analysis, k=29, n=6,781]). Both numbers belong in the conversation, not only the friendly one.

What is explicitly not a reason for testing

Statement 3.7 of the guideline is remarkable in its brevity: gastro oesophageal reflux disease on its own is not an indication for testing for Helicobacter pylori. So anyone asking because of heartburn whether they should have the bacterium tested gets, according to the German guideline, a no for the time being, and the reasoning for that stands earlier in this text.

The sentence that moves everything forward

I find this sentence remarkably clever, and it is almost never quoted. It moves the actual decision to the right place: before the test, not after it.

Because what happens otherwise. Someone orders a self test out of curiosity, the result is positive, and from that moment on there is a finding in the room that has to be dealt with. Then a decision is made under pressure that could have been discussed calmly.

And now you know why the most important decision with this topic is not made after the finding, but before it.

Frequently asked questions

Where this topic connects to the rest of the digestive system

A bacterium in the stomach does not stand on its own. It hangs on the acid question, on iron absorption, on what happens in the gut after a course of antibiotics, and on the question of how a person deals with an unclear finding.