A new analysis posted to the preprint server medRxiv on July 8, 2026 asks whether artificially sweetened beverages, diet sodas and the like, track with liver problems in people who already have MASLD. The team drew 50,562 participants with MASLD from the UK Biobank, followed them for a median of 12.8 years, and recorded liver-related adverse events. Over that window, 292 events occurred, and the pattern across drinkers was dose-dependent.
Beverage intake came from 24-hour dietary recalls, and people were grouped as non-drinkers, up to one serving a day, or more than one serving a day. Compared with no intake, more than one serving daily was tied to a higher rate of liver-related adverse events, a fully adjusted hazard ratio of 1.40 (95% CI 1.02 to 1.93). The one-serving-or-less group did not reach statistical significance, and the test for trend across categories was significant, meaning the risk rose with the dose.
The authors ran the analysis hard. A restricted cubic spline showed a positive, roughly linear relationship between intake and liver events, and a competing-risk model, which treats liver-related death as a competing event, returned the same hazard ratio of 1.40. Sensitivity analyses held the association steady, which is what a dose-response pattern is supposed to do: it survives the alternative ways of cutting the data.
The honest framing has to come first, because this is a preprint. It has been posted for review but not yet peer-reviewed, and the data are observational, drawn from self-reported diet recalls. That combination means the finding is a strong hypothesis and a caution, not a verdict. A reader should treat it as one more reason to reach for water, not as proof that one can of diet soda scars the liver.
The result also sits inside a wider conversation. A separate Nature Food analysis of 173,840 people linked both sugary and artificially sweetened drinks to worse liver outcomes, and the UK Biobank paper extends that thread specifically to people with established MASLD. The two lines of evidence agree on the direction: beverages are not metabolically neutral, and swapping them for water or unsweetened tea is the safest read of both.
The mechanism under discussion is plausible but unfinished. Some studies link artificial sweeteners to shifts in the gut microbiome, glucose handling, and insulin responses, while others find neutral effects, and the field is far from settled. The preprint does not add mechanism; it adds a large population signal, and it is the population signal that deserves the cautious attention.
For a person managing fatty liver, the practical math is simple. Replacing a daily diet soda with water or unsweetened tea removes a variable with a murky risk profile and adds a habit with a clean one. No part of the evidence says diet drinks protect the liver, so there is nothing being sacrificed, and the substitution is the kind of change that costs a week of adjustment rather than a lifetime of effort.
The 24-hour recall method deserves a skeptical footnote. One or two recalls capture a day, not a decade, and people misremember and under-report what they drink. If anything, that noise usually blurs associations toward zero, so a signal that survives it may be understated rather than overstated, but the point stands that the exposure measurement is not precise.
The outcome definition matters too. Liver-related adverse events include serious endpoints like decompensation and liver cancer, so the study is not tracking a blood test moving a few points, it is tracking clinical harm. That is why the preprint, despite its limitations, has drawn attention: the exposure is common, the population is huge, and the endpoint is the one patients fear.
The dose-response detail is the part worth remembering. The risk did not appear at one serving a day; it appeared above that level, and the spline was roughly linear. That shape is exactly what a cautious reader wants to see, because it means the message is moderation and substitution rather than total prohibition, which is a message people can actually keep.
None of this replaces the clinician. A person with MASLD should review beverage habits, blood work, and any concerning symptoms with the care team, and the preprint is not a reason to make sudden changes on one report. It is a reason to prefer water and unsweetened tea, which is advice no clinician will dispute while the reviewed evidence catches up.
The preprint also reinforces the site’s stance that zero-calorie does not mean zero-signal. The liver responds to the whole metabolic environment, and a drink with no calories still lands in the middle of that environment, alongside the microbiome, the insulin response, and the food eaten with it. The cautious read is to treat diet drinks as occasional, not as a daily default, and to let water do the daily work.
Readers will see this headline again when the reviewed version appears, and the numbers may shift slightly under formal peer review. The direction is unlikely to flip, because the dose-response and sensitivity analyses are already consistent. Until then, the takeaway is stable and cheap: when the drink does not need to be sweet, keep it plain, and when it does, treat the sweet version as the exception rather than the routine.
The substitution list is easy to build from the pantry. Water, sparkling water with lemon, unsweetened tea, and coffee cover nearly every occasion where a diet soda used to appear, and none of them carries the murky signal the preprint describes. For readers who miss the fizz, sparkling water plus a slice of citrus is the bridge, and for readers who miss the sweetness, a piece of fruit after the drink is the honest swap.
The preprint also argues for reading labels with fresh eyes. Many products labeled zero-sugar still sit in a sweetened-drink category, and the cohort grouped them all together, which means the caution applies to the category, not to one brand. A reader who treats the whole sweetened aisle, diet and regular alike, as the occasional choice rather than the daily default is following the most defensible reading of the data.
What did the study find?
Over a median follow-up of 12.8 years, people with MASLD who drank more than one artificially sweetened beverage per day had a higher rate of liver-related adverse events than non-drinkers, a hazard ratio of 1.40 (95% CI 1.02 to 1.93). One serving or fewer per day did not reach statistical significance.
Is this proof that diet drinks harm the liver?
No. This is a preprint, not yet peer-reviewed, and the data are observational, so it shows association rather than cause. The dose-response pattern is consistent, but confirmation in a reviewed, prospective analysis is still needed before anyone should change habits on this alone.
What should a reader do?
Use the finding as a reason to lean on water, unsweetened tea, or coffee instead of any sweetened drink, diet or regular. The site treats both added sugar and heavy reliance on diet drinks as signals to check, and this cohort gives the caution a large nudge.