October 8, 2026 · Science · Nutrition
What the SWITCH Trial Measured
A two-year randomized trial comparing diet drinks with water could not show a measurable difference between the two choices. Over 104 weeks, 493 randomized adults lost 4.8 kg on non-nutritive sweetened beverages and 3.7 kg on water. The 1.1 kg gap between the two groups was not statistically significant, meaning the trial demonstrated an absence of difference rather than a victory for either beverage.
Converting those weights into imperial units gives a loss of about 10.6 lb on diet drinks and 8.2 lb on water, leaving a 2.4 lb difference. These conversions are arithmetic based on 2.20462 lb per kilogram rather than figures reported directly by the study investigators. The trial was funded by the American Beverage Association, and fewer than half of the enrolled participants completed the two-year period.
The investigation was published in the British Journal of Nutrition on August 4, 2026, under DOI 10.1017/S0007114526108150. Lead author Professor Jo Harrold, Dean of Psychology at the University of Liverpool, conducted the work alongside ten co-authors: Hill, Radu, Thomas, Thorp, Hardman, Christiansen, Masic, McGale, Cuthbertson, and Halford. Coverage of the publication also appeared via ScienceDaily and the Cambridge University Press platform at cambridge.org.
Formally titled SWITCH, an acronym for The effectS of non-nutritive sWeetened beverages on appetITe during aCtive weigHt loss, the protocol was registered as NCT02591134. The study followed an open-label, parallel-group design divided into three sequential stages. Participants completed a 12-week active weight loss phase, followed by a 40-week assisted maintenance phase, and concluded with a 52-week voluntary unassisted extension.
Adults with overweight or obesity were randomized to receive either water (246 participants) or commercially available beverages containing non-nutritive sweeteners (247 participants). The sweetened drinks contained aspartame, acesulfame-K, or sucralose. Because the design was open-label, every participant knew which drink group they had been assigned to throughout each phase.
The short version
Why No Difference Is the Finding
A statistically non-significant difference between trial groups indicates that the observed numerical variation could easily occur by chance. In the SWITCH trial, the 1.1 kg difference between the water arm and the sweetened beverage arm failed to reach statistical significance. That finding cannot be cited as evidence of an advantage for diet beverages.
Both trial arms achieved statistically significant reductions in total body weight from baseline over the two-year period. Both arms also showed statistically significant reductions in waist circumference and hip circumference. The investigators reported no statistically significant between-group differences for either waist or hip measurements, and no absolute centimetre figures were published in the trial reports.
The practical value of an equivalence result is in what it removes. Many individuals attempting to navigate a weight loss plateau struggle to replace sweetened beverages with plain water exclusively. If maintaining non-caloric sweetened drinks does not impair weight reduction compared with water, forcing an unwanted swap may not be required to maintain caloric deficit.
At the same time, this equivalence does not establish that non-nutritive sweeteners are better than water. The trial simply shows that within this specific experimental cohort, consuming diet beverages did not impede weight loss relative to water.
The Completion Problem
High participant dropout over long-term clinical trials introduces survivorship bias that complicates the interpretation of final results. Of the 493 adults initially randomized in the SWITCH trial, only 220 individuals completed the entire 104-week protocol. That completion rate represents roughly 45% of the starting cohort, meaning more than half of the participants withdrew before the final weigh-in.
Retention differed slightly between the two arms by week 104. In the water arm, 117 of 246 participants finished the trial, compared with 103 of 247 participants in the non-nutritive sweetened beverage arm. Those who complete a two-year trial are not a random sample of those who started.
The open-label structure of the trial further magnifies the potential impact of attrition. When participants know their assigned drink, their expectations regarding satiety, hunger, and satisfaction can influence their daily food choices. Furthermore, the trial investigators noted that the COVID-19 pandemic disrupted standard data-collection protocols during the study, adding another procedural constraint to the final numbers.
Who Paid for the Trial
The American Beverage Association funded the SWITCH trial. The paper discloses the funding directly, and states that the association was not involved in the design, conduct, data analysis or interpretation of the trial.
Author conflict disclosures extend beyond the institutional trial grant. Multiple authors reported direct research funding from the American Beverage Association. Furthermore, several authors disclosed personal professional relationships, advisory roles, or project funding from commercial beverage and confectionery entities outside this submitted work, including Coca-Cola, the International Sweeteners Association, and Mars.
Disclosing industry funding does not invalidate clinical data, but it provides necessary context for weighing scientific literature. As detailed in our research methodology, corporate sponsorship requires careful scrutiny of endpoints, statistical framing, and publication transparency.
The Biomarkers That Moved
The SWITCH trial identified statistically significant between-group differences in total cholesterol, HbA1c, and diastolic blood pressure, yet the study authors judged these changes clinically irrelevant. Specifically, the authors stated that these biomarker variations fell below established thresholds for clinical meaningfulness and are not clinically relevant. The published sources do not report the absolute numerical effect sizes for these three parameters.
This distinction highlights a common challenge when reviewing modern metabolic health data. A laboratory marker can show a mathematical divergence between groups that passes a p-value threshold without reflecting any meaningful improvement in long-term health. When the researchers who ran the trial call a difference not clinically relevant, that judgement belongs with the number wherever it is repeated.
At Longevity Benchmark, we track the distinction between a laboratory marker that moves and an outcome that improves health. The pages in our nutrition hub hold to that line: a clinical endpoint counts for more than an isolated laboratory shift. In the SWITCH trial, neither beverage demonstrated a clinically relevant advantage in cardiovascular or metabolic blood markers over 104 weeks.
What Would Change This Read
Three updates would alter this read. First, an independently funded trial conducted without beverage industry sponsorship would eliminate concerns regarding commercial bias. Second, a study achieving better participant retention across two full years would provide greater statistical confidence in long-term adherence.
Third, a trial with a broader demographic footprint would improve clinical applicability. The investigators noted a potential lack of generalisability as a stated limitation because the SWITCH trial was conducted at a single site in England and collected no race or ethnicity data. Different populations and age brackets often face varying metabolic challenges, as explored in our guide on the hardest age to lose weight.
Sources
- Harrold JA, Hill S, Radu C, Thomas P, Thorp P, Hardman CA, Christiansen P, Masic U, McGale L, Cuthbertson DJ, Halford JCG, "Effect of non-nutritive sweetened beverages v. water on body weight: long-term results of a randomised controlled trial," British Journal of Nutrition, August 4, 2026, DOI 10.1017/S0007114526108150, journal record at cambridge.org (accessed October 8, 2026). Source of the SWITCH acronym, the open-label parallel-group design and its three phases of 12, 40 and 52 weeks, the 493 randomized split 246 to 247, the 220 completers split 117 to 103, the sweeteners used, the 4.8 kg and 3.7 kg weight changes and the non-significant 1.1 kg difference, the waist and hip findings, the cholesterol, HbA1c and diastolic blood pressure differences with the authors' "not clinically relevant" judgement, the American Beverage Association funding statement, the author conflict disclosures, and the stated limitations.
- University of Liverpool release carried by ScienceDaily, October 2026 (accessed October 8, 2026). Source of Professor Jo Harrold's title as Dean of Psychology at the University of Liverpool.
- Trial registration NCT02591134 on ClinicalTrials.gov (accessed October 8, 2026). Source of the registration identifier.
Frequently Asked Questions
Did diet drinks beat water for weight loss in the SWITCH trial?
No, diet drinks did not beat water. Although participants assigned to diet beverages lost an average of 4.8 kg compared with 3.7 kg in the water arm, the 1.1 kg difference between the two groups was not statistically significant. Both groups achieved statistically significant weight loss compared with baseline measurements, and the trial demonstrated equivalence rather than a superiority win for either beverage.
How many people finished the trial?
Only 220 of the 493 randomized participants completed the full 104-week study, representing a completion rate of roughly 45%. In the water arm, 117 of 246 participants finished the trial. In the non-nutritive sweetened beverage arm, 103 of 247 participants finished. More than half of the enrolled participants withdrew before reaching the 104-week endpoint.
Who funded the SWITCH trial?
The American Beverage Association funded the trial. The published paper notes that the association was not involved in the design, conduct, data analysis, or interpretation of the study. Multiple trial authors also reported receiving funding or maintaining professional relationships with commercial entities outside the submitted work, including Coca-Cola, the International Sweeteners Association, and Mars.
Did the trial find any health differences between the two drinks?
The trial recorded statistically significant between-group differences in total cholesterol, HbA1c, and diastolic blood pressure, but the study authors concluded that these differences are not clinically relevant. The changes fell below established clinical thresholds for meaningful health impacts. The authors also found no statistically significant difference between the two groups for reductions in waist or hip circumference.