Health Effects of Sugar-Sweetened and Artificially Sweetened Beverages: Umbrella Review and Evidence-Based Consensus Statement of the Korean Diabetes Association and the Korean Nutrition Society

Evidence review and statement development process

This consensus statement was developed on the basis of a structured evidence review that combined an updated meta-analysis of RCTs with an umbrella review of systematic reviews (SRs) of prospective cohort studies. The review protocol was prospectively registered in the International Prospective Register of Systematic Reviews (PROSPERO; Registration No. CRD42025 1029808). The key question guiding this process was whether replacing SSBs with ASBs improves health outcomes across children, adolescents, and adults. To address this question, we evaluated short- to intermediate-term effects using evidence from RCTs and long-term associations using evidence from prospective cohort studies. The framework of population, intervention, comparison, and outcomes (PICO) is summarized in Supplementary Table 1. A comprehensive literature search was conducted in MEDLINE (Ovid), Embase, and the Cochrane Library, applying predefined strategies that combined indexing terms (MeSH/Entree), and free-text terms related to NSSs and beverage types. The search covered studies published from 2010 onwards, with the final search completed on May 13, 2025. Detailed strategies and results are presented in Supplementary Table 2. Two independent reviewers (J.H.C. and S.J.S.) screened all titles and abstracts, assessed full texts, and determined study eligibility; discrepancies were resolved by consensus. The study selection process was documented according to Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) 2020 guidelines and is illustrated in Supplementary Fig. 1.

For the updated meta-analysis, eligible RCTs were identified based on the methodological foundation using McGlynn et al. [19], with additional recent trials incorporated. For the umbrella review, we included SRs and meta-analyses of prospective cohort studies reporting associations of SSBs and ASBs consumption with clinical or metabolic outcomes. Data extraction was independently conducted by two reviewers (J.H.C. and S.J.S.) using a standardized form, capturing study characteristics, intervention and comparator details, outcomes, follow-up duration, and effect estimates. Risk of bias (RoB) in RCTs was assessed using the Cochrane RoB 2.0 tool, while the methodological quality of included SRs was appraised using A MeaSurement Tool to Assess systematic Reviews, version 2 (AMSTAR-2). Results of these assessments are provided in Fig. 1, Supplementary Fig. 2 for RCTs and Supplementary Table 3 for SRs. Evidence synthesis was performed using the web-based version of RevMan. For RCTs, effect estimates were expressed as mean differences (MDs) with 95% confidence intervals (CIs) and were analyzed using a fixed-effects model; results are presented in Table 1, Fig. 1, and Supplementary Fig. 2. For cohort evidence, risk ratios (RRs) with 95% CIs were extracted from published meta-analyses and summarized at the review level; results are shown in Table 2, Fig. 2, and Supplementary Fig. 3, with detailed study-level estimates provided in Supplementary Table 4.

Fig. 1

Forest plots from the updated meta-analyses of randomized controlled trials comparing replacement of sugar-sweetened beverages (SSBs) with artificially sweetened beverages (ASBs). Results are shown for (A) body weight, (B) body fat, (C) glycosylated hemoglobin, and (D) fasting blood glucose. Effect sizes are presented as mean differences (MD) with 95% confidence intervals (CIs), using a fixed-effects model. Negative values indicate greater reductions in the intervention group (ASBs replacement) compared with the control group (continued SSBs consumption). Study identifiers are labeled as follows: ‘A’ for adult, ‘Ad’ for adolescent, and ‘C’ for child participants, followed by the intervention duration in weeks, the first author’s surname, and the year of publication (e.g., A12-Higgins 2019=adult participants, 12-week intervention) [26]. Risk of bias was assessed according to the Cochrane tool, including: A, random sequence generation; B, allocation concealment; C, blinding of participants and personnel; D, blinding of outcome assessment; E, incomplete outcome data; F, selective reporting. Symbols represent low risk (green), unclear risk (yellow), and high risk (red). SE, standard error; IV, inverse variance.

Table 1

Summary of findings: effect of replacing sugar-sweetened beverages with ASBs on body composition and metabolic outcomes (evidence from randomized controlled trials)

Table 2

Summary of findings: association between SSBs or ASBs consumption and mortality and cardiometabolic outcomes (evidence from prospective cohort studies)

Fig. 2

Forest plots from the umbrella review of systematic reviews and meta-analyses of prospective cohort studies evaluating the association between sugar-sweetened beverage (SSBs) and artificially sweetened beverage (ASBs) consumption and health outcomes. Results are shown for (A) SSBs and all-cause mortality, (B) ASBs and all-cause mortality, (C) SSBs and cardiovascular disease (CVD) mortality, (D) ASBs and CVD mortality, (E) SSBs and type 2 diabetes mellitus (T2DM), and (F) ASBs and T2DM. Effect estimates are presented as risk ratios (RRs) with 95% confidence intervals (CIs), using a fixed-effects model. Values >1 indicate increased risk associated with higher intake of SSBs or ASBs compared with lower intake or non-consumption. Each data point represents a systematic review or meta-analysis, not to individual cohort studies. This figure was designed for visual comparison of published meta-analyses, not for conducting a new pooled meta-analysis. All included reviews were based on prospective observational studies. The forest plots therefore summarize the overall evidence base at the review level, and not primary individual studies. Study identifiers are labeled as follows: ‘H’ for highest versus lowest categories and ‘S’ for per serving increment, followed by the first author’s surname, the initial of the first name, and the year of publication (e.g., H-Li B 2023=highest vs. lowest intake [18]). Panels (A, C, E) correspond to SSBs, while panels (B, D, F) correspond to ASBs, providing clear visual separation between the two beverage types. SE, standard error; IV, inverse variance.

The certainty of evidence was evaluated according to the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) framework. Summary assessments for RCTs and cohort studies are presented in Supplementary Tables 5–7. Publication bias was examined through visual inspection of funnel plots (Supplementary Fig. 4).

Evidence from RCT: results of updated meta-analysis

RCTs provide direct evidence on the short- to intermediate-term effects of replacing SSBs with ASBs. The updated meta-analysis included 14 RCTs with intervention durations ranging from a minimum of 3 weeks to a maximum of 76 weeks (Table 3). Most of these trials were conducted in Western countries such as the United States, the United Kingdom, and Europe, with a few recent studies from Asia. Participants were generally young to middle-aged adults, with the oldest mean age around 45 years, and represented a broad range of body types, from healthy lean individuals to those with overweight or obesity. Sample sizes varied, with both small, tightly controlled trials and larger pragmatic studies. The interventions also differed in design: some provided fixed daily volumes of ASBs ranging from 355 mL to 1,750 mL/day, while others used home delivery of ASBs every 2 weeks, thereby reflecting more realistic, free-living conditions. Various sweeteners were tested, including aspartame, sucralose, saccharin, stevia, and combinations thereof.

Table 3

Characteristics of randomized controlled trials included in the updated meta-analysis

Across 12 trials including 1,779 participants with follow-up durations ranging from 3 to 76 weeks, body weight was reduced by a mean of −0.73 kg (95% CI, −0.95 to −0.50) (Fig. 1A), body mass index (BMI) by −0.31 kg/m² (95% CI, −0.42 to −0.19) (Supplementary Fig. 2A), and body fat percentage by −0.72% (95% CI, −1.07 to −0.37) (Fig. 1B). These effects were rated as moderate certainty according to GRADE, downgraded mainly due to RoB in allocation concealment and blinding (Table 1, Supplementary Table 5). In three RCTs evaluating glycosylated hemoglobin, the pooled MD was −0.03% (95% CI, −0.11 to 0.05) (Fig. 1C), indicating no significant improvement with ASBs substitution. The certainty of evidence was rated low due to imprecision and the small number of available trials (Supplementary Table 5). Fasting blood glucose showed a modest reduction (−0.13 mmol/L; 95% CI, −0.21 to −0.04) (Fig. 1D), with moderate certainty (Supplementary Table 5). Insulin resistance, assessed by homeostatic model assessment of insulin resistance, did not show consistent benefit (MD, −0.25; 95% CI, −0.61 to 0.11) (Supplementary Fig. 2B), and the certainty was very low (Supplementary Table 5). The impact of ASBs on blood lipids was generally neutral. Across six RCTs (n=800 participants, follow-up 8 to 52 weeks), no meaningful changes were observed in low-density lipoprotein cholesterol (0.02 mmol/L; 95% CI, −0.07 to 0.11) (Supplementary Fig. 2C) or high-density lipoprotein cholesterol (0.03 mmol/L; 95% CI, −0.00 to 0.06) (Supplementary Fig. 2D). Triglycerides showed a small reduction (−0.09 mmol/L; 95% CI, −0.18 to −0.00) (Supplementary Fig. 2E), but the certainty of evidence for lipid outcomes was low due to small sample sizes and imprecision. Blood pressure outcomes were also inconclusive. In six RCTs with 8–52 weeks of follow-up, systolic blood pressure showed no significant change (−1.21 mm Hg; 95% CI, −3.20 to 0.78) (Supplementary Fig. 2F), and diastolic blood pressure showed a borderline but nonsignificant reduction (−1.45 mm Hg; 95% CI, −2.98 to 0.09) (Supplementary Fig. 2G). Both outcomes were rated as low certainty. Taken together, RCTs evidence supports that replacing SSBs with ASBs yields modest and consistent benefits for weight reduction, with uncertain and less consistent effects on glycemic control, lipid parameters, and blood pressure.

Evidence from prospective cohort studies: results of umbrella review

Unlike RCTs, observational studies did not directly compare the substitution of ASBs for SSBs. Instead, most prospective cohort studies evaluated the health impact of each beverage type separately, typically by comparing the highest versus lowest categories of intake or by examining risks associated with incremental increases of one or more daily servings. Thus, comparability between ASBs and SSBs is indirect, and interpretation requires caution. In conducting the umbrella review, we found that the meta-analyses from individual SRs of prospective cohort studies yielded remarkably consistent results across different populations and outcomes. Because of high degree of consistency among the included meta-analyses and their coverage of diverse health outcomes, we did not re-extract individual cohort data to perform a new meta-analysis of non-randomized studies. Instead, we synthesized the findings at the review level, thereby minimizing duplication and ensuring comprehensive integration of evidence across all reported outcomes. As summarized in Table 4, we identified 20 SRs and meta-analyses of prospective observational studies that satisfied our inclusion and exclusion criteria [17,18,20–51]. The included SRs encompassed large-scale cohorts, mostly conducted in Western populations of generally healthy adults at baseline. Participant mean ages ranged broadly from the early 20s to the mid-70s, with most cohorts clustered in middle-aged groups. Follow-up durations extended from approximately 3 years to more than 34 years, and sample sizes often reached hundreds of thousands, providing substantial power to detect associations with mortality and major cardiometabolic outcomes.

Table 4

Characteristics of systematic reviews of prospective cohort studies included in the umbrella review

For all-cause mortality, higher SSBs intake was consistently associated with increased risk, with RRs ranging from 1.03 to 1.14 when comparing highest versus lowest categories of intake, and from 1.04 to 1.10 per serving increment (Fig. 2A). Cardiovascular (CV) mortality demonstrated a similar pattern, with RRs between 1.13 and 1.31 for the highest versus lowest intake, and from 1.08 to 1.13 per serving increment (Fig. 2C). By contrast, cancer mortality showed no significant associations with SSBs consumption, with estimates close to null (RR, 0.96 to 1.02) (Supplementary Fig. 3A). For ASBs, findings were weaker and less consistent compared with SSBs. Reported RRs for all-cause mortality ranged from 1.09 to 1.15 for the highest versus lowest intake, and from 1.04 to 1.08 per serving increment (Fig. 2B). CV mortality demonstrated a similar pattern, with RRs between 1.05 and 1.26 for the highest versus lowest intake, and from 1.04 to 1.10 per serving increment (Fig. 2D). By contrast, cancer mortality estimates were generally close to null (RR, 1.01 to 1.04) (Supplementary Fig. 3B). The certainty of evidence for ASBs was rated low or very low due to heterogeneity, confounding, and possible reverse causality.

The association between SSBs and T2DM was particularly strong. Individuals with the highest SSBs intake had a 15%–29% greater risk of developing T2DM when compared with the lowest intake, and a 19%–27% greater risk per serving increment (RR, 1.15 to 1.29) (Fig. 2E). The associations between ASBs and T2DM risk were also strong, with individual in the highest intake group having a 15%–32% greater risk of developing T2DM when compared with the lowest intake, and a 13%–22% greater risk per serving increment (RR, 1.13 to 1.22) (Fig. 2F). All-cause mortality, CVD mortality and T2DM showed consistent evidence of a dose-response relationship, which further strengthened this conclusion, leading to an upgraded certainty rating (moderate, Table 2).

Regarding CVD incidence, SSBs were positively associated with increased risk (RR, 1.08 to 1.17), with consistent findings across cohorts. For ASBs, risk estimates for CVD incidence ranged from 1.07 to 1.21, but results were more heterogeneous across outcome definitions and generally of low certainty (Table 2). Other metabolic outcomes showed similar patterns. High SSBs intake was associated with increased risk of hypertension (RR, 1.10 to 1.27), metabolic syndrome (RR, 1.19 to 1.56), and obesity (RR, 1.12) (Supplementary Fig. 3). Although the certainty of evidence was low, the direction of associations was consistent across studies. For ASBs, results suggested modestly increased risks for hypertension (RR, 1.08 to 1.14), metabolic syndrome (RR, 1.31 to 1.44), and obesity (RR, 1.21), but all with low certainty due to residual confounding, publication bias, and the potential for reverse causality (Table 2, Supplementary Fig. 3).

Methodological quality of evidence from prospective cohort studies was assesses using the AMSTAR-2 tool (Supplementary Table 3). Among the 20 included meta-analyses, five were assessed as high quality, three as moderate quality, nine as low quality, and three as critically low quality. The quality criteria most typically lacking were not reporting funding of included primary studies (n=15), not specifying registered protocol (n= 10), not specifying duplicate study selection (n=4), and not specifying duplicate data extraction (n=4). All meta-analyses included in this study satisfied the following criteria: (1) following the PICO components to describe the aims and methods, (2) including detailed descriptions of primary studies, (3) evaluating RoB for included primary studies, (4) using appropriate methods for statistical analyses of the meta-analysis, (5) employing RoB assessments in sensitivity analyses, (6) including comments in the discussion considering results of RoB assessments, (7) discussing the potential sources of statistical heterogeneity, and (8) reporting conflicts of interest related to the study.

In summary, observational evidence strongly and consistently indicates that SSBs consumption is associated with higher risks of all-cause and CV mortality, incident CVD, T2DM, and adverse metabolic outcomes. For ASBs, observational studies suggest broadly similar associations across these outcomes, although uncertainties remain due to heterogeneity, potential confounding, and reverse causality.

Balance of benefits and harms

Our evidence synthesis highlights both the benefits and the limitations of using ASBs as substitutes for SSBs. High SSBs intake is strongly and consistently associated with obesity, T2DM, CVD, and premature mortality. Short-term RCTs show modest but consistent reductions in body weight when SSBs are replaced with ASBs, although improvements in glycemic control have not been demonstrated consistently. However, observational evidence regarding ASBs demonstrates similar associations with increased risks of mortality, CVD, and T2DM, with risk estimates generally comparable to those of SSBs.

Limitations of the current evidence include substantial variation in how intake is assessed, as both the dose per serving and the frequency of consumption differ across individuals and across studies. Most cohort studies rely on food frequency questionnaires, which are prone to misclassification of beverage type and portion size, further limiting the precision of risk estimates and their applicability to dietary recommendations. Moreover, real-world dietary patterns are more complex; people may consume SSBs and ASBs together, or intake may fluctuate from day to day, making it difficult to derive a single, clear recommendation for SSBs and ASBs. The potential for reverse causation between ASBs and metabolic outcomes should be taken into account when interpreting the study findings. Additionally, the lack of data from Korea and other Asian populations also represents a limitation of the current evidence base. In addition, recent studies have reported that NNS can activate sweet taste receptors not only in the oral cavity but also in gut endocrine cells and pancreatic islets, suggesting potential effects on appetite regulation, gut microbiota, and glucose metabolism [3,52]. These effects vary by the type of NNS and across individuals [6], limiting the generalizability of current evidence.

The balance of benefits and harms therefore suggests that ASBs can provide a net benefit for individuals with high SSBs consumption and elevated metabolic risk when used as a short-term substitution strategy. However, long-term habitual consumption of ASBs is not recommended due to the uncertainty of their safety profile and the possibility of reinforcing preference for sweetness. Importantly, the RCTs evidence is limited to child, adolescent or relatively young adults, short follow-up durations, and trials susceptible to bias from incomplete blinding. Observational studies lack head-to-head comparisons of SSBs and ASBs intake, are vulnerable to confounding, and may overestimate risks since ASBs consumers often already have higher baseline BMI or cardiometabolic risk [53]. These limitations highlight the need for careful interpretation of the evidence. ASBs are not recommended for initiation among individuals with little or no prior SSBs or ASBs intake, nor should they be relied upon long term by heavy SSBs consumers. Rather, ASBs may serve as a bridging strategy: in individuals with high SSBs intake, temporary substitution with ASBs can reduce immediate risks of obesity, T2DM, and CVD, while gradual transition to non-caloric, non-sweetened beverages such as water or unsweetened tea should be reinforced as the ultimate goal.

Consideration in the use of the recommendation

In considering this recommendation, regular use of ASBs, even as a replacement for SSBs, is not advised from a prevention and public health perspective. Reducing the intake of SSBs and ASBs requires both individual-level interventions and supportive environments. Clear labeling of sweetener type and content is particularly important for products marketed to children and adolescents. Schools, workplaces, and communities should limit the availability of SSBs and provide healthier beverage options. According to recent national statistics of Korea, energy and sugar intake from beverages was relatively higher among children, adolescents, and young adults, and approximately 64% of adolescents reported consuming SSBs three or more times per week [54,55]. Therefore, reducing SSBs intake in children, adolescents, and young adults deserves strong emphasis to prevent early exposure, which may increase later-life risk of adverse health outcomes. It is essential to develop tailored dietary guidelines for Koreans, along with the implementation of underpinning studies. Further research—including long-term cohort studies in the Korean population and RCTs in real-world settings—is also required to clarify the health effects and safety of ASBs consumption.

ASBs may be used as a substitution strategy to reduce added sugar intake in individuals with T2DM. However, given the potentially distinct physiological effects of different NNS, ASBs should be limited to a transition period and not promoted as regular consumption in place of SSBs. Our analysis highlights a significant discrepancy between experimental and observational evidence regarding the risk of developing T2DM. Short-term RCTs report that replacing SSBs with ASBs yields modest or inconsistent improvements in glycemic control, whereas long-term cohort studies link high ASBs consumption to an increased risk of T2DM, comparable to SSBs. Methodological limitations, including reverse causation and residual confounding, likely explain part of this gap [56,57]; however, emerging mechanistic data also suggest direct physiological effects. Several NNS activate sweet taste receptors in enteroendocrine cells and pancreatic islets, modulating incretin release, glucose transport, and insulin secretion [58–62]. The gut microbiome and appetite regulation may mediate heterogeneous glycemic responses [6,63,64]. These interactions are more complex in people with T2DM. Impaired incretin effect, insulin resistance, heterogeneous β-cell function, and concomitant therapies may modify the glycemic impact of ASBs. For instance, metformin’s effects on the gut microbiome interact with ASBs exposure [65]. ASBs consumed with carbohydrates can rapidly decrease metabolic and neural responses, suggesting dysregulated gut-brain control of glucose metabolism [66]. Some NNS have been associated with atherosclerotic risk, warranting caution in those at high CV risk or with established CVD [67,68].

As a major source of NNS, reducing ASBs intake is an important target for public health. However, future studies should move beyond ASBs alone and focus on individual NNS, using mechanism-informed stratification by sweetener type, metabolic status, and concomitant medications is needed [6]. Incorporating exposure biomarkers and phenotyping will help identify subgroups for whom ASBs substitution is neutral or beneficial versus unfavorable [69,70].

In conclusion, based on the current evidence, replacing SSBs with ASBs or water can lead to short-term improvements in body weight and fat mass. Dose-response analyses indicate that higher SSBs intake is consistently associated with increased risks of mortality, T2DM, and adverse metabolic outcomes, while ASBs show similar associations in observational studies. Therefore, the Food and Nutrition Committee of the Korean Diabetes Association and the Dietary Sugar Committee of the Korean Nutrition Society agreed that ASBs may be used as a short-term substitution for SSBs, particularly in individuals at high cardiometabolic risk, but their regular use is not advised for the prevention and management of long-term health outcome.