A two-year trial found similar weight loss with diet drinks and water. See how program support, missing follow-up and industry funding shape the findings.
On September 29, 2026, the University of Liverpool reported two-year findings from the SWITCH trial comparing water with nonnutritive-sweetened beverages. The researchers found no statistically significant difference in weight change between the groups within a behavioral weight-management program.
The study was published in the British Journal of Nutrition in 2026. It offers evidence for a narrower conclusion than either side of the diet soda debate might prefer: participants assigned to either beverage maintained some weight loss during the study.
Researchers enrolled 493 adults with overweight or obesity. They assigned 246 to water and 247 to nonnutritive-sweetened beverages, often shortened to NNS beverages. The comparison was randomized and open-label, meaning participants knew which beverage they had been assigned.
The useful question is not whether diet soda deserves a health halo. It is whether this beverage choice meaningfully changed weight outcomes within the particular program researchers tested.
The mechanism examined here is best understood through the relationship between beverage choice and behavior. This was not an announcement about a newly identified fat-burning pathway. Researchers tested whether assigning different drinks changed weight outcomes while participants received the same type of behavioral support.
According to the trial report, participants were asked to drink at least two 330-milliliter servings of their assigned beverage each day. The NNS options included products sweetened with aspartame, acesulfame potassium, or sucralose. Those details describe the study design rather than an intake recommendation.
The behavioral program changed over time. Participants attended weekly sessions during the initial weight-loss phase and monthly sessions during assisted maintenance. There were no sessions during the final unassisted year.
That sequence matters because the findings span both supported and unassisted periods. The beverage assignment remained part of a broader weight-management effort rather than an isolated intervention. The study therefore does not establish that either beverage caused the observed weight loss on its own.
For a nonscientist, the distinction is straightforward: a drink can be part of a pattern associated with weight change without being the biological driver of that change. These results do not establish a particular effect on appetite or a specific metabolic pathway. They answer a practical comparison about assigned beverages within a program.
A headline saying diet soda drinkers lost as much weight as water drinkers captures the broad finding. But it compresses the study's statistical result and its research setting into a much simpler message. Several details determine how much confidence to place in that message.
Of the 493 participants who began assigned beverage consumption, 220 completed the week-104 assessment. That included 117 assigned to water and 103 assigned to NNS beverages, according to the published trial.
Fewer than half of the original participants therefore had measurements at the final assessment. The reported end-point averages describe those measured participants rather than a complete set of final measurements from everyone who started. This attrition is central to interpreting the two-year result.
The researchers designed SWITCH as an equivalence study. Their reported weight result was that the between-group difference was not statistically significant. That wording should not be translated into proof that the beverages were identical in every respect.
The result also applies to the tested beverages within this particular program. It is not a finding about every sweetener or product. Nor does it establish that every individual would respond in the same way.
The authors identified additional limits to generalizability. These included a single site in England and missing race and ethnicity data. They also noted possible effects of the COVID-19 period and the voluntary final phase.
Some measurements were collected remotely during COVID-19 restrictions. Some participants lacked blood-pressure or blood-sample data, and certain biomarker comparisons involved small samples. Those gaps matter when interpreting outcomes beyond body weight.
The American Beverage Association funded the trial. The paper states that the association did not design or conduct the study and did not analyze or interpret the data. It also reports that the association funded medical-writing support and had an opportunity to review the manuscript.
The Hill's coverage reported the industry funding alongside the researchers' statement that the work was conducted independently. That disclosure belongs beside the findings rather than being treated as an afterthought.
Industry funding is a reason to examine the methods and reporting carefully. It is not proof that the result is false. Equally, the stated separation from trial operations does not establish that funding had no possible influence.
The University of Liverpool described the findings as relevant to World Health Organization guidance advising against using non-sugar sweeteners for weight control. The university characterized that guidance as reflecting uncertainty in observational evidence.
Random assignment makes SWITCH different from research that observes people's existing beverage habits. Even so, one randomized comparison does not settle the broader guidance question. The university's policy framing should not become a claim that the trial overturned WHO advice.
For everyday decisions, the defensible interpretation is modest: this study offers no significant weight-loss advantage for either assigned beverage within the tested program. It does not make diet drinks necessary for weight management.
Preference and sustainability can therefore remain part of the conversation. Someone who prefers water has no reason to view these findings as an instruction to start drinking diet soda. Someone who prefers an NNS beverage need not read this trial as evidence that the choice prevented weight-loss maintenance.
Keep the beverage question alongside the larger pattern of daily habits, rather than letting it dominate every nutrition decision. WeightRestart's coverage of dietary energy density addresses another nutrition question beyond this comparison.
When beverage choices intersect with a medical condition or a treatment plan, a qualified clinician can help interpret the findings in that personal context. The trial is evidence to discuss rather than a personal prescription.
At week 104, participants with measurements had lost an average of 3.7 kg in the water group and 4.8 kg in the NNS-beverage group. The British Journal of Nutrition paper reported no statistically significant difference between those groups.
The larger numerical loss in the NNS group should not be presented as evidence that diet drinks performed better. The authors' reported comparison did not support that conclusion. The relevant finding is the statistical comparison rather than which average appears larger on the page.
Waist and hip circumference fell in both groups. Researchers found no statistically significant between-group differences in those measurements either.
Body-composition measurements came from a small subset: 23 participants assigned to water and 25 assigned to NNS beverages. Both groups showed reductions in measured fat mass and other composition measures. Between-group differences were not significant, but this subset does not provide a full body-composition picture of everyone enrolled.
Other metabolic measures were less uniform. Total cholesterol increased more in the NNS group. HbA1c and diastolic blood pressure decreased more in the water group.
The authors said the HbA1c and blood-pressure differences did not meet their stated thresholds for clinical meaningfulness. That is distinct from asking whether a statistical difference was present. It concerns whether the size of a difference met the authors' criteria for clinical importance.
Taken together, the outcomes support a weight-focused finding rather than a sweeping declaration that the beverages had matching effects on every aspect of metabolic health. Body weight and the other measured markers should remain separate parts of the interpretation.
The next useful studies would test whether this pattern holds across more varied populations and settings. Stronger participant retention would make long-term results easier to interpret. More complete body-composition and biomarker measurements would also help researchers assess whether similar weight outcomes accompany similar metabolic changes.
What to watch next: independent replication and clearer separation of beverage effects from the surrounding behavioral program. Researchers would also need evidence about individual responses before turning a group average into a beverage-specific strategy for standard practice. For now, SWITCH adds a measured finding to the debate: neither assigned beverage showed a significant two-year weight-loss advantage within the program studied.
WeightRestart shares research-led guidance on weight loss, metabolism, nutrition, strength, appetite, sleep and recovery. Our goal is to make complex health information clear, practical and useful for people building progress they can maintain.
Learn how to build a weight-management approach around better information, realistic expectations and habits you can keep using.
read the blog
