Non-nutritive sweetened beverages versus water after a 52-week weight management programme: a randomised controlled trial

The study addresses whether non-nutritive sweetened (NNS) beverages are equivalent to water in supporting weight loss and maintenance within a behavioral weight management program. Given concerns that sugar-sweetened beverages contribute to obesity and metabolic disease, guidelines recommend lower-calorie alternatives such as water or NNS beverages. However, long-term RCT data, especially directly comparing NNS beverages to water during weight maintenance, are sparse and findings from observational studies are mixed. Building on prior trials (including a 52-week RCT reporting superiority of NNS over water and the SWITCH trial’s earlier 12-week results showing equivalence), this study evaluates 52-week outcomes for weight and related health measures.

Prior observational studies have reported associations between NNS beverage intake and increased body weight/BMI, but may be affected by reverse causation and residual confounding. In contrast, meta-analyses and systematic reviews of RCTs generally show that NNS beverages reduce energy intake and modestly reduce weight versus sugar-sweetened beverages, with potential cardiometabolic benefits. Few long-term RCTs compare NNS beverages directly with water for weight maintenance. A Colorado/Temple 52-week RCT found NNS beverages superior to water for weight loss and maintenance. The SWITCH trial’s 12-week phase showed equivalent weight loss between NNS and water. Reviews highlight that many RCTs are short (≤3 months), underscoring the need for longer-term trials. Evidence syntheses suggest that RCTs tend to show neutral/beneficial effects on weight, whereas negative findings often come from observational designs.

Design: Parallel-group, open-label, randomized controlled equivalence trial conducted at the University of Liverpool in three phases: 12-week active weight loss, 40-week assisted weight maintenance (reported here), and a voluntary 52-week unassisted maintenance extension (to be reported separately). Ethical approval obtained; written informed consent provided. Compensation up to £300 for 52-week completion; additional compensation for optional assessments. Population: Adults aged 18–65 years, BMI 27–35 kg/m², consuming >3 cold beverages/week; excluded if <3 cold beverages/week, recent/current smoker, relevant diseases (e.g., diabetes, gastrointestinal or cardiovascular disease), food allergies, high alcohol intake, medications affecting weight, intense regular exercise, current dieting/major recent weight loss, or prior bariatric surgery. Recruitment within 50-mile radius of Merseyside, England. Randomization and stratification: 1:1 to NNS beverages or water via computer-generated sequence (blocks of 4 and 6), stratified by sex, age (18–35, 36–50, 51–65), BMI (<30, ≥30 kg/m²), and NNS naïveté (0–25% vs 26–100% of beverage choices over prior 5 years). Interventions: Assigned beverage consumption of at least two 330 mL servings/day. NNS group selected from ~20 branded options (≤5 kcal/8 oz; ≤2.1 kcal/100 mL); water group to consume bottled water for at least two daily servings (plus tap water as needed) and abstain from NNS beverages and sweeteners in hot drinks. All participants could consume water; sugar-sweetened beverages were allowed but dietary counseling aimed to reduce caloric intake. Beverages supplied by investigators. Behavioral program: Weekly group sessions during 12-week weight loss; monthly sessions during 40-week maintenance, led by a qualified nutritionist, focusing on maintenance topics (energy gap, plateaus, emotional/situational eating, physical activity). Self-monitoring via exercise and diet diaries; supportive resources and monthly weigh-ins. Up to three missed monthly sessions permitted before exclusion. Adherence and assessments: Adherence monitored via daily online beverage logs, collection/photographs of empty packaging, Sugar and Sweetener Food Frequency Questionnaire (SSFFQ) monthly during weeks 0–12 and quarterly during weeks 12–52, and 3-day food diaries at baseline, weeks 12 and 52. Outcomes: Primary—change in body weight (kg) from baseline to week 52. Secondary—changes in waist/hip circumference, glycemic control (HbA1c, fasting glucose, insulin), fasting lipids, liver enzymes (AST, ALT, GGT), hunger (0–100 mm VAS), SSFFQ sugar and sweetener scores, physical activity (Fitbit steps/day averaged over 1 week at baseline, week 12, week 52). DXA body composition in a subset. Measurement frequency: Weight, waist/hip, hunger VAS, SSFFQ monthly; fasting bloods and DXA in subsets; activity at baseline, week 12, week 52. COVID-19 adaptations: Reduced delivery frequency with larger quantities, online sessions (Zoom), self-reported weight/measurements via standardized scales and tape measures mailed to participants; photos of empty packaging; some visits conducted via online questionnaires; some blood pressure/samples not collected. Analysis: Equivalence of NNS to water defined as two-sided p > 0.05 at week 52; power calculation assumed 27% attrition, n=316 (158/group) to provide 90% power to detect ±1.5 kg difference at week 52 (minimum n=248 excluding attrition). Primary analyses via ANCOVA with treatment group as predictor and baseline measure as covariate. Primary analysis on complete cases (participants with baseline and week 52 data). Missing data handled in sensitivity analyses via multiple imputation (predictive mean matching, 50 imputations) and last observation carried forward (LOCF). Additional sensitivity analyses included covariates: age, sex, measurement location (self vs clinic), and NNS naïveté.

Participants: 493 randomized and initiated treatment (water n=246; NNS n=247) between July 2016 and December 2021; 24.1% NNS-naïve. Week 12 completed by 383; week 52 completed by 262 (53.1% of starters: water n=137; NNS n=125). Among week-52 completers, 93.6% attended monthly sessions. Adherence high: at week 52, full compliance (two 330 mL servings/day) 98.2% (NNS) and 98.6% (water); partial 1.1% vs 0.6%; non-compliance 0.7% vs 0.8%. Primary outcome (complete cases): Mean weight change from baseline to week 52: water −6.1 ± 5.8 kg vs NNS −7.5 ± 5.9 kg; between-group difference 1.4 kg (90% CI 0.2, 2.6); p < 0.05; groups non-equivalent (i.e., significant difference). Sensitivity/imputation: Multiple imputation and LOCF showed no statistically significant between-group differences. ANCOVA sensitivity: Baseline weight significantly predicted week-52 weight; beverage group remained significant in complete cases and LOCF but not in multiple imputation. Measurement location (self vs clinic), sex, and NNS naïveté had no effect. Weight trajectory: Greatest loss in first 12 weeks; maximum loss at week 44 (water) and week 36 (NNS); subsequent regain slower with NNS. Anthropometrics: Waist and hip circumferences decreased significantly in both groups. Hip reduction greater with NNS (between-group difference in change 1.6 cm; 95% CI 0.2, 3.0); waist difference not significant. DXA subset (n=57): Significant reductions in fat mass, fat-free mass, and android/gynoid fat in both groups; no between-group differences. Biomarkers: Most improved from baseline in both groups. HDL cholesterol increased modestly with NNS (0.1 mmol/L) vs no change with water, yielding a significant between-group difference; other lipid, glycemic, and liver markers showed no significant between-group differences. Appetite and intake: Hunger VAS unchanged. Sugar consumption decreased similarly in both groups. Sweetener consumption decreased substantially in water group and was unchanged/slightly increased in NNS group, leading to a significant between-group difference as expected. Activity: Steps/day decreased slightly with water (−3.6 steps) and increased with NNS (+868.9 steps; p < 0.05 within-group), but between-group difference was not significant. Clinical significance: Although statistically significant, the 1.4-kg advantage for NNS did not meet the predefined 1.5-kg threshold for clinical significance.

Both water and NNS beverages, when integrated into a structured behavioral weight management program, supported significant weight loss and favorable changes in anthropometrics and biomarkers over 52 weeks. The NNS group achieved greater weight loss than the water group in complete-case analyses, indicating statistical non-equivalence driven by a modest advantage for NNS beverages. However, the between-group difference did not reach the predefined threshold for clinical significance (1.5 kg). Results are generally consistent with prior RCTs showing neutral to beneficial effects of NNS beverages on weight, including a similar 52-week trial reporting clinically significant superiority of NNS over water. Differences in observed effect sizes and significance may reflect program specifics, participant characteristics, dropout, and pandemic-related adaptations. Improvements in HDL cholesterol and hip circumference with NNS beverages suggest potential cardiometabolic and body composition-related advantages, though overall biomarker changes were similar between groups. The findings reinforce that both water and NNS beverages can be compatible with weight loss and maintenance strategies, irrespective of prior NNS exposure, while underscoring the importance of long-term randomized evidence to inform guidelines.

After 52 weeks (12 weeks of active weight loss plus 40 weeks of assisted maintenance), NNS beverages produced statistically greater weight loss than water, but the difference was not clinically significant. Both beverage strategies effectively supported weight loss and improvements in anthropometric and cardiometabolic measures within a behavioral program. Future research, including the planned 52-week unassisted maintenance extension of this trial, will clarify the durability of effects without routine support and further inform guidance on NNS beverage use in long-term weight management.

- Single-site study in England with no collection of race/ethnicity data, limiting generalizability to diverse populations. - Lower completion rate at week 52 (53%) compared to a similar trial (73%), potentially influenced by the COVID-19 pandemic and initial 2-year commitment. - COVID-19 necessitated protocol adaptations (self-reported measures, online sessions), although sensitivity analyses suggested no effect of measurement location on outcomes. - Some secondary assessments (e.g., blood samples, blood pressure) were missing for participants due to pandemic restrictions, reducing sample sizes for biomarker analyses.