Sugar alcohols are sold as the gentle option. They are also the only sweetener class that appears in a gastroenterology guideline as a category of carbohydrate you are asked to cut back, not to substitute. Both things are true at once, and the reason is worth understanding properly, because most of what circulates about them is one or the other without the other.
They are a FODMAP category, and FODMAPs are a treatment target
A FODMAP is a fermentable oligo-, di- or monosaccharide or polyol. Polyols are the last category in that acronym, and they got there on the evidence: sorbitol, maltitol, mannitol, xylitol and erythritol reach the colon largely unabsorbed and are fermented by resident bacteria.
The clinical position follows directly from that. The first-ever American College of Gastroenterology clinical guideline, built with GRADE methodology after twenty-five clinically important questions, recommends a limited trial of a low-FODMAP diet for patients with irritable bowel syndrome to improve global symptoms. It says limited, and a 2022 review is explicit about why: implementation should be supervised by a trained dietitian, and a “bottom-up” approach has been proposed precisely to avoid altering the gut microbiome and nutritional status.
The trial that made the diet respectable is the 2014 crossover study from Melbourne. Thirty IBS patients and eight matched healthy controls, 21 days on each diet, a washout of at least 21 days, all food provided, and a goal of under 0.5 grams of FODMAPs per meal. Global gastrointestinal symptom score in the IBS patients: 22.8 millimetres (95 per cent confidence interval 16.7-28.8) on the low-FODMAP diet against 44.9 millimetres (36.6-53.1) on the Australian comparator diet, P below 0.001. Bloating, pain and wind all came down. In the controls, symptoms were minimal and neither diet moved them.
Two details in that result matter more than the headline. First, the effect is in IBS patients, not in healthy people — the control arm had nothing to fix. Second, only the diarrhoea-predominant subtype had its stool frequency and consistency actually change. If you are reading a study summary that says FODMAP restriction helps everyone, it has compressed the trial past its own content.
Malabsorption is common; disease is not
There is a specific trap here, and gastroenterology has written about it explicitly. How much sorbitol is absorbed depends on how much of it there is, how concentrated it is, and how much absorptive surface there is. So a positive breath test is expected at a high load in a lot of people, and it means less than it looks like: the finding of sorbitol malabsorption is not the expression of any specific cause of intestinal damage.
Worse, the test itself is soft. The 2013 review states that interpretation of the breath hydrogen test is difficult, and that neither studies comparing it with a gold standard nor validated doses and concentrations are available. It also names the classic misdiagnosis: small intestinal bacterial overgrowth ferments carbohydrates nonspecifically, produces an abnormal hydrogen output, and gets recorded as sugar malabsorption when what is actually present is a bacterial overgrowth.
There is one more correction to the popular story, and it is in the 2022 review: different studies show no differences in the frequency of sugar malabsorption between IBS patients and healthy controls, even though the severity of symptoms after a sugar challenge is higher in patients. Malabsorption is not what separates the two groups. Symptom severity is.
The microbiome claim: some trials found dysbiosis, most did not
The 2023 review on non-nutritive sweeteners and the gut microbiota covers aspartame, acesulfame-K, sucralose and saccharin rather than the polyols specifically, but it is the clearest available accounting of how solid this literature is. Preclinical results conflict, for reasons that are mundane: how the compound was administered, and how differently species metabolise it. In human trials, a dysbiotic effect showed up in some; many other randomised controlled trials reported no significant impact on microbiome composition. The studies also differed in participant numbers, dietary habits and lifestyle, all of which feed into the baseline composition and how people respond.
The review’s conclusion is worth quoting as the finding rather than as a caveat: the scientific community has no unanimous consensus on which outcomes and biomarkers can accurately define these effects. That is a real gap, not a rhetorical hedge.
For scale, the same review puts the microbiome at more than 1,500 species, with 99 per cent of bacteria coming from about 30 to 40 of them, and up to 100 trillion bacteria in the colon alone. A single sweetener compound landing in that system, without a validated biomarker, does not produce a conclusion.
The cardiovascular finding, and why it is not the end of the story
The loudest dataset on this class is the 2023 Nature Medicine paper on erythritol, and it is worth reading carefully because it is easy to over-read in either direction.
The design was staged. An untargeted metabolomics study in 1,157 patients undergoing cardiac risk assessment found that circulating levels of several polyol sweeteners, erythritol especially, were associated with three-year risk of major adverse cardiovascular events. Then targeted metabolomics in two independent validation cohorts — 2,149 patients in the United States and 833 in Europe, all stable patients undergoing elective cardiac evaluation — confirmed it. Fourth versus first quartile, adjusted hazard ratio 1.80 (1.18-2.77) in the US cohort and 2.21 (1.20-4.07) in the European one.
Then the mechanistic arm: at physiological levels erythritol enhanced platelet reactivity in vitro and thrombosis formation in vivo. And a small prospective pilot intervention in 8 healthy volunteers found that ingestion produced marked and sustained increases in plasma erythritol, lasting beyond two days, well above the thresholds associated with the platelet findings.
Three things that study is not, and one thing it is. It is not a randomized trial of cardiovascular outcomes — nobody was assigned to consume erythritol and then followed for heart attacks. The cohorts are people already undergoing cardiac risk assessment or elective evaluation, which is not the general population. The intervention arm is 8 people. And the association could run in either direction: the paper is as consistent with high erythritol being a marker of altered metabolism as with erythritol being a driver.
What it is: a consistent association across a discovery cohort and two independent validations, a gradient across intake quartiles, and a plausible mechanism in the same paper. The authors’ own conclusion is that studies assessing the long-term safety of erythritol are warranted — which is a request for evidence, not a verdict.
Appetite: a small trial with a genuinely interesting split
The one randomized crossover trial here from 2025 is small — 20 healthy participants of normal weight, four visits — and it measured appetite, not gut symptoms. Participants took a preload of 33.5 grams of xylitol, 33.5 grams of sucrose, or 0.1675 grams of acesulfame potassium in 300 millilitres of water, or plain water as control, then ate an unrestricted test meal 15 minutes later.
Total energy intake was lower after xylitol and after acesulfame than after sucrose, with no difference between xylitol and acesulfame or between either and water. Plasma cholecystokinin was higher after xylitol than after sucrose, acesulfame or water. GLP-1 did not differ across conditions. Glucose and insulin were lower after xylitol than after sucrose — and, the detail worth sitting with, xylitol still raised insulin relative to acesulfame and water.
Two readings are possible and the trial does not separate them. Either the satiety signal is doing work, or the insulin bump is doing work, or both. With 20 people and one test meal, it cannot tell you which, and it certainly does not tell you what happens over months. It does tell you that “sugar alcohols do not spike you like sugar” is a defensible summary of a single acute meal, and that “sugar alcohols are metabolically inert” is not.
Where the class does have measured benefit: teeth
The dental literature is the one place sugar alcohols have a real, quantified result, and the honest version of it comes from a Cochrane review of 10 studies and 5,903 participants.
The best-supported finding: over two and a half to three years, a fluoride toothpaste containing 10 per cent xylitol may reduce caries by 13 per cent compared with a fluoride-only toothpaste — prevented fraction -0.13, 95 per cent confidence interval from -0.18 to -0.08, in 4,216 children, on low-quality evidence. One study reported that a xylitol syrup at 8 grams a day reduced caries by 58 per cent (95 per cent CI 33 to 83 per cent) in 94 infants against a syrup at the lower intake at 2.67 grams a day over a year, also low-quality.
Read the quality assessment before the percentages. Of the ten studies, one was rated at low risk of bias, two were unclear, and seven were at high risk. Several other comparisons — lozenges against no treatment, sucking tablets against no treatment, tablets against sorbitol tablets — had confidence intervals compatible with both a reduction and an increase in caries. So the class has a headline number that is small, and that the review states with a modal verb rather than as a claim.
The adverse effects listed in that same review are the ones that connect this article back to the top of the page: mouth sores, cramps, bloating, constipation, flatulence, and loose stool or diarrhoea. Not in every study — four reported no adverse effects and two found similar rates between arms — but they are the documented ones.
What this literature does not show
It does not show that sugar alcohols are healthier than sugar for the general population. The dental benefit is real and the cardiovascular finding is a warning sign. Those point in opposite directions and nobody has reconciled them.
It does not show that a polyol-rich diet causes a measurable problem in healthy people. The IBS trial found nothing in its control arm. The microbiome review found no significant impact in many randomised trials. The erythritol cohorts are patients, not the public.
It does not show that “no calories” means “no consequence.” The 2025 trial shows an insulin response to xylitol that is real and different from water.
And it does not show that a breath test equals a diagnosis. With no validated doses and no gold-standard comparison, a positive hydrogen test for sorbitol is a weak fact about the person who ate a lot of sorbitol, and it is one of the easiest things in this area to misread.
The honest summary
Two true sentences that almost never appear in the same paragraph. Sugar alcohols are among the better-evidenced dental actives available without a prescription — 13 per cent less caries, on thin evidence. They are also a FODMAP category that gastroenterology asks IBS patients to restrict on a time-limited basis, they carry the class with the strongest cardiovascular signal in modern sweetener research, and the trials measuring their effect on the microbiome are unresolved enough that nobody has agreed what to measure.
Which of those sentences matters more for you depends on which column of the evidence question you are actually asking. It is worth being clear that both are the current state, not one of them being the pessimistic take.
