What a SIBO Breath Test Can and Cannot Tell You

A breath test is easy to order, easy to do at home, and produces a graph with a line on it. That combination makes the result feel like a verdict. The diagnostic accuracy literature describes something considerably less decisive.

What the test measures

Of the gases present in the gut, hydrogen and methane are exclusively produced by microbial fermentation, and that fact is the principle behind clinical breath testing. Gut microbes readily digest carbohydrate, producing these gases, which diffuse into the abdominal venous circulation and are transported to the lungs, where they can be detected in exhaled breath.³ A test therefore gives a fasted subject a measured carbohydrate substrate and samples breath at set intervals over the following hours.³ If bacteria in the small intestine ferment that sugar before it reaches the colon, the rise appears earlier than it otherwise would, and that early rise is what the test looks for.³

A 2017 North American consensus meeting of clinician-scientists standardized the specifics. Consensus doses were 10 grams for lactulose, 75 grams for glucose, 25 grams for fructose and 25 grams for lactose. A rise in hydrogen of at least 20 parts per million above baseline by 90 minutes during a glucose or lactulose test was considered positive for small intestinal bacterial overgrowth, and a methane level of at least 10 parts per million was considered methane positive.³

The same consensus group described glucose and lactulose breath testing as the least invasive alternatives available for identifying SIBO, and considered breath testing useful in evaluating carbohydrate maldigestion, methane-associated constipation, and bloating and gas, but not useful for assessing oro-cecal transit.³

Those thresholds exist because reasonable experts sat down and agreed on them. They are a convention, and knowing that changes how a number just past the line should be read.

The accuracy numbers

A systematic review with meta-analysis in the Journal of Neurogastroenterology and Motility selected 14 studies in which breath testing had been compared against jejunal aspirate culture as the reference standard.⁴

Pooled sensitivity was 42.0 percent for the lactulose test and 54.5 percent for the glucose test. Pooled specificity was 70.6 percent for lactulose and 83.2 percent for glucose. Using the delta-over-baseline cutoff above 20 parts per million of hydrogen, glucose test sensitivity was 47.3 percent and specificity 80.9 percent.⁴

Sensitivity in the forties and fifties means the test misses roughly half of what the reference standard identifies. The authors concluded that the glucose test appears to work better than the lactulose test, and that breath tests performed best in patients with surgical reconstruction of the gastrointestinal tract, where pooled glucose sensitivity and specificity were 81.7 and 78.8 percent, against 40.6 and 84.0 percent in subjects without any predisposing condition.⁴

That last contrast is the practically important one. The test performed best in exactly the population where the pre-test probability was already high, and worst in the population most likely to order one for unexplained bloating.

The reference standard is itself unsettled

Comparing a test against a gold standard assumes there is one. A systematic review of diagnostic tests for SIBO reviewed 71 papers in detail and reported that studies were very heterogeneous in patient populations, test definitions, sample size, and methods.⁵

Tradition had accepted small bowel aspirate above 10⁵ CFU per millilitre as the standard, but the review found that this threshold was usually indicative of stagnant loop conditions rather than the milder presentations now labelled SIBO. Colony counts appeared elevated in most gastrointestinal diseases compared with controls. Not even the culture papers met published quality standards for diagnostic test studies. The conclusion was direct: there is no validated diagnostic test or gold standard for SIBO.⁵

This matters twice over. It limits what any accuracy figure means, because sensitivity and specificity are calculated against a comparator whose own validity is contested. It also means the accuracy figures above may understate or overstate breath test performance in ways nobody can currently quantify.

What a positive result means in a population

The most useful way to read any test result is to ask how common a positive is among people who do not have the condition of interest.

A meta-analysis of 25 case-control studies covering 3,192 patients with irritable bowel syndrome and 3,320 controls answered that directly. With breath testing, SIBO prevalence was 35.5 percent in patients with IBS and 29.7 percent in controls.⁶

Roughly three in ten control subjects tested positive. The odds of SIBO were higher in IBS patients than controls, at an odds ratio of 3.7 overall and 4.9 in studies using only healthy controls, so the association is real. But a test that flags nearly a third of controls is not separating two clean groups.⁶

The method used changes the answer substantially. In the same analysis, culture-based studies found SIBO in 13.9 percent of IBS patients and 5.0 percent of controls at one colony count cutoff, and 33.5 percent versus 8.2 percent at another. Prevalence diagnosed by lactulose breath test was 3.6 times greater in IBS patients and 7.6 times greater in controls than prevalence diagnosed by glucose breath test.⁶

The same patients, the same underlying biology, and a positivity rate that moves several-fold depending on which sugar was used. The authors rated the overall quality of evidence as low, attributing this largely to clinical heterogeneity.⁶

What the guidelines say about the concept itself

The 2020 American College of Gastroenterology guideline defines SIBO as the presence of excessive numbers of bacteria in the small bowel causing gastrointestinal symptoms, and evaluates the criteria for diagnosis using the GRADE process, developing key concepts by expert consensus where the evidence was not suitable for a formal recommendation.¹

An AGA clinical practice update published the same year is franker about the state of the concept. It notes that thanks to ready access to hydrogen breath testing, SIBO is now commonly diagnosed among people presenting with a wide variety of gastrointestinal and even non-gastrointestinal symptoms, and is increasingly implicated in lay press and media in the causation of a diverse array of disorders. Its first best practice advice states that the definition of SIBO as a clinical entity lacks precision and consistency, and that true prevalence is accordingly undefined.²

The same document lists bloating, diarrhea, and abdominal pain or discomfort as the symptoms traditionally linked to SIBO, notes that a major impediment to defining SIBO is limited understanding of normal small intestinal microbial populations, and states that controversy remains concerning the role of SIBO in the pathogenesis of common functional symptoms such as those regarded as components of irritable bowel syndrome. It also advises that there is insufficient evidence to support using inflammatory markers such as fecal calprotectin to detect SIBO.²

How to read your own result

A few things follow from the above.

A positive result raises the probability that small intestinal bacterial fermentation is contributing to symptoms. It does not establish it, given that nearly 30 percent of controls also test positive.⁶

A negative result does not clear the question. With pooled sensitivity in the 42 to 55 percent range, a substantial share of people who meet the reference standard will test negative.⁴

Which sugar was used matters, and glucose and lactulose are not interchangeable. Glucose testing showed higher pooled sensitivity and specificity, and lactulose testing produced several-fold higher positivity rates in both patients and controls.⁴⁶

A result close to the threshold is a result close to a number that a consensus panel selected. Twenty parts per million by 90 minutes is a defensible convention, not a biological boundary.³

The practical read

Breath testing is inexpensive, non-invasive, and standardized enough to be reproducible, which is why it is used. It is also being applied far beyond the populations in which it performs best, against a reference standard that has never been validated, to define a clinical entity whose own definition specialists describe as lacking precision.²⁵

A breath test result is one input to a clinical conversation, and it belongs in front of a clinician who can weigh it against the history and the alternatives rather than being read as a diagnosis on its own.