Regulators call xylitol safe but Pacific Islander rates disagree

Because the Charité team drew blood instead of handing out food diaries, the figures presented at ESC Congress 2026 in Munich describe exposure. Marco Witkowski brought data on 17,710 people enrolled in two prospective observational studies, the Canadian Longitudinal Study on Aging and EPIC-Norfolk. At Messe München the congress opened on 28 August and closed on 31 August. Six years of follow-up in the Canadian arm separated the highest quartile of blood xylitol from the lowest by 57% on a composite of death, myocardial infarction and stroke, after adjustment for six baseline risk factors.

Across thirty years the Norfolk arm gave a narrower gap of 18% between the extreme quartiles, with the middle quartiles falling in rank order between them. The Canadian follow-up lasted six years, the Norfolk thirty. Dan Atar set the ESC caution. There is no causal claim in observational data of this kind. Atar wants further study into a question of this weight. Witkowski put it plainly. Nobody yet knows much about the cardiovascular safety of xylitol, though the quantity going into food and dental products keeps climbing.

Regulators have called the sugar alcohols safe. Shoppers do buy them on that rating. In the human body xylitol turns up anyway, a byproduct of glucose metabolism and a trace constituent of a few natural foods. Manufacturers push sweetened foods and dental products a thousandfold past that trace. Munich unsettles the shopper's assumption without overturning it, because an association across two observational studies stops short of cause.

In April 2026 the Cardiovascular Disease Among Asians and Pacific Islanders study reported from the electronic records of two health systems, Palo Alto Medical Foundation in Northern California and Kaiser Permanente Hawaiʻi. Investigators drew 807,643 adult records covering 2007 to 2018, then followed 303,958 people from January 2011 to December 2018. Diabetes at baseline lifted myocardial infarction incidence to 3.7% from 0.9%. Stroke incidence rose to 4.8% from 1.8%.

Non-Hispanic White adults with diabetes recorded 5.5 myocardial infarctions and 7.6 strokes per 1,000 person-years. Below that mark fell most Asian adults with no second race on file. Multiracial patients did the opposite. Among them the Pacific Islander, Asian and White combination produced 16.6 infarctions and 12.1 strokes, while Pacific Islander with Asian returned 14.6 and 11.1. Pacific Islander and White adults showed 13.0 and 12.1, more than double the White infarction rate. Aggregation cancels those margins.

Comparing the findings means converting between units. Munich supplies relative risk, 57% and 18% above a bottom quartile. The Californian and Hawaiian tables give event counts instead, 16.6 per 1,000 person-years at the top. A relative figure gives direction without magnitude, the shape of most relative reduction claims. An absolute rate tells a patient the count of infarctions in a thousand people over a year.

Since 27 December 2020, section 208 of division CC of the Consolidated Appropriations Act, 2021 has obliged the states and the District of Columbia to give full Medicaid benefits to compact citizens of the Marshall Islands, the Federated States of Micronesia and Palau. A drafting decision in the 1996 federal welfare statute had limited those citizens to emergency treatment. CMS gave state programs State Health Official Letter 21-005 on 18 October 2021. Four months earlier its engineers had rebuilt the Federal Data Services Hub logic. State eligibility screens began returning the correct immigration status on 13 June 2021.

A state program mails enrolment notices and appeal rights after a Medicaid determination, in English only. In Honolulu the family opening that envelope may speak Kajin M̧ajeļ, the first language of most compact citizens from the Marshall Islands. Majuro supplies patients to the Hawaiian and Californian systems behind the study. Investigators found the highest infarction rates among multiracial Pacific Islander patients there.

Drawing on a common English source, the team locked the terminology first. Screening and risk material looks like ordinary prose until the drafter gets to the instructions themselves. Across several Pacific languages the high-stake fields went through a check, field by field. Repeated phrases kept identical wording. Beside the consent forms the team cut spoken scripts for a synthetic voice. First-language reviewers audited orthography and pronunciation, then took two plain-language rounds over the instruction sections of a health-promotion set.

A reviewer struck a hedged line about partial protection and replaced it with a flat declarative naming the measure protecting against the two outcomes together. A surviving hedge softens a categorical instruction. Elsewhere a colleague rejected a technical term outright and supplied the wording of an ordinary Marshallese speaker. Neither catch would surface in a fluency score. The graded side-by-side comparison on Marshallese places the strengths of a machine draft beside the corrections from a first-language check.

Utugrafihan CHamoru, codified at 5 Guam Code Annotated Chapter 88, writes the glota in medial and final position, sparing only the start of a word. A voice dropping it turns baba, meaning open, into ba'ba', meaning whip. On a ward a nurse says that difference out loud. Synthetic speech repeats the error in identical form. A fluency score rates the sentence as clean either way. Patients from Guam and the Northern Marianas are US citizens by birth, outside section 208 altogether.

A 2024 report from the group implicated the platelets. Xylitol raised their ability to form clots. High blood levels tracked cardiovascular events in high-risk individuals over three years. The Munich analysis widens that finding to thirty years and to the general population. For patients at 16.6 infarctions per 1,000 person-years a wrong safety rating costs more than it does at 5.5.

A registration screen lists speakers of Gagana Sāmoa, Lea Faka-Tonga, CHamoru and Kajin M̧ajeļ under the Pacific Islander heading. Investigators pulled that heading apart. In a health-plan file it stays whole.

Adjustment for body mass index, hypertension, diabetes and lipid levels left the association intact. Nor did age and sex at baseline change the gradient. Those four conditions send a shopper to the low-calorie shelf in the first place. Confounding by indication does not explain the result away. Witkowski reported no conflicts of interest. The funders were the Canada First Research Excellence Fund, the Corona Foundation and the Berlin Institute of Health.