Two cohorts, one top quartile
The comparison is the same in both cohorts and the follow-up is not. Among 17,710 people in the Canadian Longitudinal Study on Aging and in EPIC-Norfolk, those whose blood xylitol sat in the highest quartile were more likely to die or to have a myocardial infarction or a stroke than those in the lowest: by 57 per cent over 6 years in the Canadian cohort and by 18 per cent over 30 years in the British one. Both cohorts pointed the same way at every level in between, so the pattern is dose-dependent rather than a single threshold effect. The full analysis reaches ESC Congress 2026 for presentation, and that is where it first opens to questions.[1]
What the release withholds is the arithmetic a reader would need to weigh it. There are no event counts, so neither percentage can be turned into a number of people; there is no blood concentration marking the boundary between the quartiles, so nobody can tell whether the top group is unusual or ordinary; and the two increases sit on windows five times apart in length, which makes 57 and 18 quantities that cannot be laid side by side. A relative risk without its denominator is a direction of travel and a hint of magnitude, and that is all it can be until the presentation supplies the rest.[1]
The mechanism was tested before the population was
Behind Witkowski's remark that earlier work showed xylitol can enhance the ability of platelets to form clots sits a 2024 paper in the European Heart Journal, and it is worth reading before the new numbers. In a validation cohort of 2,149 people, plasma xylitol in the third tertile carried an adjusted hazard ratio of 1.57, with a confidence interval of 1.12 to 2.21, for major adverse cardiovascular events over 3 years. Then 10 healthy volunteers drank 30 grams of xylitol, and platelet responsiveness rose in every one of them while plasma levels went up roughly 1,000-fold. A defined intake, a measured blood level and a measured platelet endpoint, in that order, in people.[1]
The exposure the new cohorts measure deserves its own sentence, because it is a blood level and not a habit. The body makes small amounts of xylitol on its own, while the amounts added to foods, beverages and oral care products run often more than 1,000-fold higher. Somebody in the top quartile may therefore be a heavy consumer of the sweetener, or somebody whose own metabolism runs the level high, and those two people carry different risks for reasons that have nothing to do with each other. Until the analysis separates them, a circulating polyol that marks the metabolic state already driving cardiovascular events remains as good an explanation as a polyol that drives them.[1]
What the presentation still has to show
Atar, speaking for the ESC Communication Committee, put the standing of the result plainly: causation is difficult to infer from any observational study, and this work shows that further studies are warranted. Witkowski said the same about his own findings while naming who is affected, that artificial sweeteners are consumed by people who think they are healthier than sugar. That is the group carrying the practical weight of an unsettled question. They are choosing a product on a safety judgement made by regulators, and this analysis neither confirms nor overturns that judgement.[1]
Two things would move this from a signal to a finding, and the 2024 paper shows what the third one already looks like. The presentation can give the event counts and the concentrations that divide the quartiles, which lets anyone convert 57 per cent into people. It can also show whether the two cohorts still differ threefold once age and follow-up are matched, or whether the shorter, older Canadian cohort was catching events that 30 years of Norfolk averaging diluted. What no longer needs establishing is the mechanism: an intake of 30 grams already moves a platelet endpoint. The missing study is the outcome trial, and anyone waiting for it is being asked to act on an unproven risk, which calls for a different answer from a proven one.[1]