Study Overview
Objective. To compare the efficacy of non-nutritive sweetened beverages (NNS) or water for weight loss during a 12-week behavioral weight loss treatment program.
Study design. 2-arm equivalence randomized clinical trial.
Setting and participants . Participants were recruited at the University of Colorado and Temple University. A total of 506 participants were screened and 308 were enrolled in the study. Inclusion criteria included being weight stable within 10 pounds in the 6 months prior to the trial, engaging in fewer than 300 min of physical activity per week and consuming at least 3 NNS beverages per week. Exclusion criteria included pregnancy, diabetes, cardiovascular disease, uncontrolled hypertension, and the use of medications affecting metabolism or weight. Participants also had physician approval stating they were in good health and could handle the nutrition and exercise requirements of the trial. Participants were randomly assigned to a NNS or water treatment arm using a computer-generated randomization that equally distributed men and women between the 2 groups. Participants had to be willing to discontinue consumption of NNS beverages for the duration of the 1-year study if they were randomized to the water-only group.
Intervention. The study was designed to include a 12-week weight loss phase followed by a 9-month maintenance phase. All participants received a cognitive-behavioral weight loss intervention called The Colorado Weigh. The program involved weekly hour-long group meetings led by registered dieticians or clinical psychologists. Groups were split by research arm and participants were taught about different weight loss strategies including self-monitoring, portion sizes, and physical activity. Participants were weighed at each meeting. The group curriculum was the same for both arms of the study except in the type of beverage they were encouraged to consume.
Participants were given individual energy targets based on their estimated resting metabolic rate (RMR), determined by using a Tanita Model TBF-300A bioelectrical impedance device that assesses body composition. Group leaders adjusted these targets as needed for participants in order to achieve a goal weight loss of 1 to 2 pounds per week. Physical activity targets were set to increase each participant’s typical physical activity by 10 minutes a week with a final target of 60 minutes a day, 6 days a week. Participants filled out daily exercise logs. Additionally, physical activity was assessed by the use of a Body Media armband that participants wore weeks 1 and 12.
Participants in the NNS group were asked to consume at least 24 fluid ounces of NNS beverage per day. Their water consumption was not limited. A beverage was considered NNS if it had less than 5 kcal per 8-ounce serving, was pre-mixed, and contained non-nutritive sweeteners. Participants in the water-only group were asked to drink at least 24 ounces of water a day and not drink any NNS beverages. They were allowed to eat foods that contained NNS but could not intentionally add NNS to beverages such as coffee. Participants in both groups were asked to record their beverage intake daily. Participants were given manufacturers’ coupons for bottled water or NNS beverages.
Main outcome measures. The primary outcomes were weight loss at 12 weeks (weight loss period) and at 1 year (weight loss maintenance). All assessments were conducted at baseline and after 12 weeks. This was designed as an equivalence trial, and the authors’ hypothesis was that there would be no clinically meaningful difference in weight change between the 2 groups. The authors pre-specified that the bounds of equivalence would be 1.7 kg. Waist circumference was recorded in addition to height and weight. Participant’s blood pressure was also recorded and blood samples were collected to measure lipids and glucose. Urine samples were collected to measure urine osmolality. Participants completed questionnaires at baseline and 12 weeks to assess changes in perceived hunger.
Results. A total of 308 patients were randomized following baseline assessment but 5 did not begin treatment. 279 of the remaining 303 participants completed the full 12-week weight loss phase of the study. The dropout rate in the water group was 10% compared to 5.8% in the NNS group, but this was not statistically significant. 80% of participants were female, 68% were white, and 27% African American. There were no significant differences at baseline in age, gender, race/ethnicity or other measures between the water-only and NNS groups. There was no significant difference in adherence to the beverage requirements between the 2 groups (96.6% in the NNS group and 95.7% in the water-only group), and similarly group attendance did not differ between the 2 groups (90.8% for NNS and 89.7% for water-only).
The mean weight loss difference between the water and NNS groups was –1.85 kg (90% confidence interval [CI], –1.12 to –2.58 kg). Because the lower confidence limit of –2.58 kg was outside the equivalence limit set in the hypothesis, the 2 treatments were not considered equivalent and paired comparisons were carried out. Analysis done using an intention-to-treat scheme indicated that the weight loss in the NNS group (5.95 kg ± 3.94 kg) was significantly higher than the weight loss in the water-only group (4.09 ± 3.74 kg, P < 0.001). 43.0% of participants in the water-only group lost > 5% of their body weight and 64.3% of participants in the NNS group lost > 5% of their body weight ( P < 0.001).
After 12 weeks of treatment there was no significant difference between the 2 groups in changes in waist circumference, blood pressure, HDL, triglycerides, or urine osmolality. Reductions in total cholesterol and LDL were significantly greater in the NNS group than the water group. There were no significant changes in physical activity between the 2 groups as measured by the exercise logs or the Body Media armbands. There was a statistically significant difference in hunger between the 2 groups ( P = 0.013): participants in the water group reported increased hunger, while participants in the NNS group reported a slight decrease in hunger.
Conclusion. Participants who drank at least 3 servings of NNS beverages a day at baseline lost more weight during a behavioral weight loss program when they continued to drink NNS beverages than participants who were asked to cut NNS beverages and drink only water. The study was designed as an equivalence trial but paired comparisons showed a significant difference in weight loss between the 2 groups.
Commentary
Obesity is a major public health concern in the United States and drinking sugar-sweetened beverages has been indicated as a significant contributing factor. Consumption of sugar-sweetened beverages increased considerably from 1994 to 2004 [1]. Fortunately, there is strong evidence that decreasing the consumption of sugar sweetened beverages can lead to weight loss [2]. Most studies look at the effect of replacing sugar-sweetened beverages with water [3] and, in fact, increased consumption of water has been shown to aid weight loss [4]. The relationship between diet drinks and obesity, however, has been a source of controversy. Since NNS beverages contain little to no calories they are a logical replacement for sugar-sweetened beverages, but observational studies have shown a positive correlation between diet drinks and obesity [5,6] as well as type 2 diabetes [7]. Additionally, a recent study by Suez et al [8] found that consumption of artificial sweeteners affects the gut microbiota and increases glucose intolerance. However this correlation may not be causal; NNS beverage consumption may be higher in overweight individuals. A study by Tate et al [9,10] looked at replacing sugar-sweetened beverages with water or artificially sweetened beverages and found no significant difference in weight loss between the 2 groups. However, the Tate et al study used beverage replacement as the primary intervention. This experiment by Peters et al is unique because it tested the hypothesis that NNS is equivalent to water alone when combined with a structured weight loss program. Their results reject the equivalence hypothesis and suggest that NNS beverages facilitate weight loss for patients already consuming them.
Strengths of this study included the use of a randomized, equivalence design. The study also examined secondary outcomes (eg, waist circumference, lipids, and urine osmolality) that helped reinforce that participants consuming NNS were able to lose weight without compromising their health. Further, they measured hunger and found that participants in the NNS beverage group had decreased hunger while those in the water group had increased hunger, which points to a potential mechanism for their findings.
However, the potential for bias in this study is concerning. One major weakness is that all the participants were initially regular drinkers of NNS beverages. The authors never explain why consuming 3 NNS beverages per week was an inclusion criteria. Participants in the water group had to change their behavior to abstain from NNS beverages and this may have impacted results. More concerning, this study was fully funded by the American Beverage Association, who has an obvious interest in promoting NNS beverage consumption. Finally, the authors mention that 5 participants dropped out after randomization but before the start of treatment and were excluded from the study after baseline assessment. The authors do not provide information about group allocation or if the participants knew which group they were assigned to, calling into question the integrity of the intention-to-treat design.
Applications for Clinical Practice
For patients who already drink NNS beverages and are motivated to lose weight, these results support continued use. However, it is unclear how NNS beverages impact weight loss efforts for patients who do not currently drink them. Further, since other studies have shown potential harm of NNS beverages [6–8], more studies are needed to better elucidate their health effects.
—Susan Creighton and Melanie Jay, MD, MS