Effects of GLP-1 and Dual GIP/GLP-1 Medicines on Weight and Health in Children and Teens with Obesity, Including the Role of Diet and Behavior Support

This systematic review was conducted in accordance with the PRISMA 2020 statement [21]. A structured literature search was carried out between April and December 2025 using the following databases: PubMed, Scopus, ClinicalTrials.gov and Embase.

The review protocol was registered in the PROSPERO database (CRD420261347002).

The aim was to identify original clinical studies, including randomized controlled trials (RCTs) and observational studies, investigating the use of GLP-1 RAs for the treatment of overweight and obesity in children and adolescents aged 6–19 years. Additionally, we manually inspected the reference lists of relevant studies and checked for any publications of interest. Data extraction was performed independently by two reviewers using predefined, standardized data extraction forms developed for the purposes of this review.

A systematic literature search was conducted using the following Boolean combination of keywords and Medical Subject Headings (MeSH): (“GLP-1 receptor agonists” OR “glucagon-like peptide-1 receptor agonists” OR liraglutide OR semaglutide OR exenatide OR dulaglutide OR lixisenatide OR tirzepatide) AND (“Children and Adolescents” OR children OR adolescents OR youth OR “young people” OR pediatric OR pediatrics) AND (“Obesity” OR overweight OR “weight control” OR “weight loss”).

Only articles published in English were included. No study-design filters were applied in the search strategy to allow for the retrieval of both randomized and observational evidence. Study type eligibility was assessed during screening. The last search update was performed on 12 December 2025. The full search strategy for each database is shown in. Supplementary File S1

Inclusion Criteria

Exclusion Criteria

After duplicate removal, titles and abstracts were screened independently by two reviewers (D.L., K.G.). Full-text articles were then assessed independently for eligibility by the same two reviewers. Any discrepancies were resolved through discussion, and when necessary, consultation with a third reviewer (A.W.). Extracted data included study authorship and year of publication, participant characteristics (age, sex, sample size), type and dosage of GLP-1 RAs used, study duration, primary and secondary outcomes (including changes in BMI, body weight, HbA1c, and metabolic markers), adverse events and tolerability, and descriptions of lifestyle or dietary interventions, where applicable. Data were extracted according to predefined categories. Trial registry numbers and regulatory documents (e.g., FDA approvals) were used to verify the published results where applicable.

Risk of bias was independently assessed by two reviewers using the RoB 2 tool for randomized controlled trials and the ROBINS-I tool for non-randomized studies. Risk-of-bias visualizations were created using the robvis tool (R package and Shiny web app) [22]. Studies were assessed across all relevant bias domains, and overall risk-of-bias judgments were assigned according to the recommended approach for each instrument. Any discrepancies were resolved through discussion, and when necessary, consultation with a third reviewer.

The overall certainty of evidence was assessed using the GRADE framework. With the inclusion of both randomized controlled trials (initially graded as high) and observational studies (initially graded as low), we assessed five domains—risk of bias, inconsistency, indirectness, imprecision, and publication bias—to determine the final strength of evidence.

In addition to pharmacological interventions, we also reviewed the scope and structure of nutritional and behavioral support provided in each eligible study. Particular attention was paid to whether participants received: (1) Standardized dietary counseling, (2) Structured physical activity recommendations, (3) Behavioral or psychological support, (4) Family or caregiver education.

Data were categorized according to predefined components of lifestyle intervention. This assessment aimed to determine whether GLP-1 RAs were evaluated as stand-alone interventions or in combination with lifestyle interventions. Descriptions of nutritional protocols were extracted directly from the methods sections of the published trials and their supplementary materials. Studies that lacked clear reporting on dietary components were noted as limited in this regard.

Lifestyle co-intervention intensity was classified a priori using predefined, transparent criteria based on the number and structure of reported components. This classification approach was developed specifically for the purposes of this review based on available reporting across studies. Studies were categorized as low intensity when they included only general advice without structured follow-up (≤1 component); moderate intensity when they included two structured components (e.g., dietary counseling and physical activity guidance); and high intensity when they incorporated three or more components, particularly when including behavioral/psychological support and family or caregiver involvement. Where reported, the frequency and regularity of contact (e.g., scheduled visits or counseling sessions) were considered to refine classification. Intensity categorization was performed independently by two reviewers; discrepancies were resolved through discussion, and when necessary, consultation with a third reviewer. It should be emphasized, however, that these criteria were developed pragmatically for this review and have not been formally validated against an external standard or established behavioral-intervention reporting framework (such as the Template for Intervention Description and Replication [TIDieR] or the Workgroup for Intervention Development and Evaluation Research [WIDER] recommendations). The components and thresholds therefore reflect a degree of subjective, expert-based judgment about what constitutes “structured” lifestyle support, and the resulting categorization may not be fully reproducible by independent reviewers applying different rules. For this reason, the intensity grading is presented strictly as a descriptive, hypothesis-generating heuristic rather than as a quantitative or psychometrically validated measure, and all findings derived from it must be interpreted with corresponding caution.

These differences were recorded and discussed in the synthesis of clinical outcomes, acknowledging that observed weight loss effects were not solely attributable to pharmacotherapy, but rather to a combination of interventions.

The findings were summarized in tabular form and narratively discussed to highlight patterns in clinical efficacy and implications for integration with nutritional treatment strategies. It should be noted that this classification approach is inherently limited by variability and incomplete reporting of lifestyle interventions across studies, and should therefore be interpreted as a descriptive, hypothesis-generating approach rather than as a quantitative measure.

Quantitative synthesis (meta-analysis) was planned if outcomes were sufficiently comparable across studies. However, due to clinical and methodological heterogeneity in populations, intervention protocols, outcome metrics, and reporting of lifestyle co-interventions, meta-analysis was not feasible. Therefore, the findings were summarized using narrative synthesis and presented in tabular form. Although several trials reported nominally similar endpoints (e.g., BMI, body weight, HbA1c), the underlying metrics and effect-size formats were not directly poolable. Specifically, BMI-related outcomes were expressed inconsistently as absolute BMI (kg/m²), BMI standard deviation score (BMI-SDS/z-score), or percentage change from baseline, with different reference populations used for z-score calculation. Effect estimates were variably reported as between-group treatment differences (placebo-controlled RCTs), within-group pre–post changes (single-arm and observational studies), or proportions of responders (≥5% or ≥10% BMI reduction), and the dispersion measures provided (SD, SE, 95% CI, IQR) were not uniform. Populations differed substantially in baseline indication (obesity vs. T2D with obesity), age range (6–19 years, with one trial restricted to 6–<12 years), and pubertal status. Intervention duration ranged from 6 to 68 weeks, and follow-up time-points were not aligned. The pharmacological exposures themselves were heterogeneous, comprising six different molecules (liraglutide, exenatide, semaglutide, dulaglutide, lixisenatide, tirzepatide) at different doses, frequencies, and dose-escalation schemes and lifestyle co-interventions varied widely in scope, structure, and reporting completeness, precluding any meaningful adjustment for stratification by behavioral context. Pooling under such conditions would have produced statistically combinable but clinically uninterpretable estimates and would have masked rather than clarified between study variability. The decision to abandon quantitative pooling and to rely on a structured narrative and tabular synthesis was therefore made a priori once the extent of these incompatibilities became evident at full text screening, in line with PRISMA 2020. For the same reasons, formal sensitivity and subgroup analyses could not be performed quantitatively. Instead, these elements were addressed descriptively across the narrative synthesis and risk-of-bias assessment.

A total of 15 studies met the inclusion criteria, comprising 12 interventional studies (placebo-controlled and non–placebo-controlled trials) [23,24,25,26,27,28,29,30,31,32,33,34] and 3 observational studies [35,36,37]. The study selection process is summarized in Figure 1.

Most studies were published from 2020 onward and differed in study design, population characteristics, intervention protocols, and duration. Although some trials included adolescents with type 2 diabetes (T2D), all participants met the criteria for overweight or obesity. Overall, substantial heterogeneity was observed across studies in terms of design, population characteristics, intervention protocols, and outcome definitions.

The included studies involved 1448 participants aged 6 to 19 years. Evaluated agents included liraglutide (6 studies) [32,33,34,35,36,37], exenatide (5 studies) [27,28,29,30,31], semaglutide (1 study) [26], dulaglutide (1 study) [24], lixisenatide (1 study) [25], and tirzepatide (1 study) [23]. Across the controlled interventional studies, 972 participants received active treatment and 476 were assigned to placebo groups.

Study durations ranged from 6 to 68 weeks. Included populations comprised children and adolescents with obesity [26,27,28,30,31,32,34,37] and T2D with coexisting obesity [23,24,25,29,33,35]. Details of the included studies, including population characteristics and primary therapeutic context (weight management vs. glycemic control), are presented in Table 1.

Risk of bias was assessed using established methodological tools selected a priori according to study design. Randomized controlled trials (RCTs) were evaluated using the revised Cochrane Risk of Bias tool (RoB 2), whereas non-randomized studies were assessed using the ROBINS-I instrument.

Each study was independently examined across all relevant bias domains, after which an overall judgment of risk of bias was assigned. Any discrepancies were resolved through discussion.

Efficacy trials evaluated GLP-1 RAs in conjunction with lifestyle or behavioral components. However, the intensity, structure, and content of these interventions varied substantially, as summarized in Table 2.

Two high-quality studies, SCALE Teens [32] and STEP Teens [26], employed high-intensity multidisciplinary programs including individualized dietary counseling, caloric restriction (~500 kcal/day deficit), physical activity planning, motivational support, and consistent family involvement. Notably, these trials also reported the largest improvements in adiposity-related outcomes and metabolic parameters including high proportions of participants achieving ≥5% and ≥10% BMI reduction.

Moderate-intensity interventions typically provided structured but less frequent lifestyle support, such as standardized nutritional advice, moderate exercise guidance, and variable family participation. Examples include four studies combining basic diet and activity counseling with limited behavioral components [28,30,31,34].

In contrast, low-intensity studies offered minimal dietary advice or generalized diabetes education, with no formal behavior change support or structured follow-up [23,25,29]. These trials generally reported more variable or less consistent changes in BMI and metabolic outcomes.

Considerable heterogeneity was observed across studies with respect to both reported treatment outcomes and the structure and reporting of lifestyle co-interventions.

However, it should be emphasized that the intensity of these interventions was neither randomized nor consistently reported across studies; therefore, these observations should be interpreted as exploratory and hypothesis-generating rather than indicative of causal relationships. Moreover, the observed differences may also reflect the influence of confounding factors including variations in study design, population characteristics, follow-up duration, and pharmacotherapy regimens (see Table 2 and Table 3).

Most trials reported clinically meaningful reductions in BMI or BMI z-score (SDS), with the largest effect observed in the STEP Teens trial [26], where semaglutide resulted in a 16.1% BMI reduction. Similarly, liraglutide and exenatide trials demonstrated moderate improvements: for example, BMI decreased by approximately −0.22 SDS [32] and by −1.18 SDS [28]. The proportion of participants achieving at least 5% BMI reduction ranged from 4.92% [31] to 73% [26] depending on the intensity and structure of co-interventions.

A qualitative overview of lifestyle co-intervention intensity and the direction of reported adiposity and metabolic outcomes across the included studies is presented in Figure 3.

Substantial heterogeneity was observed across studies with respect to both reported treatment outcomes and the structure, intensity, and reporting of lifestyle co-interventions. Several studies reporting clinically meaningful BMI reductions also incorporated relatively detailed multidisciplinary lifestyle programs, although the independent contribution of these components could not be determined [26,30,32]. Studies incorporating less detailed or minimally reported lifestyle support also demonstrated substantial variability in weight-related outcomes across different study designs and populations [25,29]. However, because lifestyle intensity was not randomized and was variably reported across trials, these comparisons should be interpreted as hypothesis-generating rather than causal.

Among the studies that reported absolute weight change (kg), the findings also reflected this trend. In liraglutide trials, weight loss in treatment groups ranged from −2.7 to −2.3 kg, while the placebo groups varied more widely (−1.91 kg to +2.8 kg) [35]. Exenatide trials showed greater variability, with active treatment groups showing changes from −2.93 kg to −6.1 kg, and placebo groups ranging from +0.01 to +12.4 kg [28]. In the lixisenatide trial, weight increased in both groups, albeit more so in the placebo arm [25].

Reported weight-change outcomes varied substantially across studies, including those with limited or minimally described nutritional support, and should not be directly compared because of major differences in the study design, duration, and therapeutic context [25]. These observations further highlight the potential role of structured lifestyle support; however, causal relationships cannot be established based on the available data.

Unfortunately, weight change data were not reported in some key studies, including those evaluating semaglutide and dulaglutide. Nonetheless, the magnitude of the BMI and SDS reductions in these trials suggests clinically significant weight effects (see Table 3).

Beyond the anthropometric effects, several trials reported improvements in selected metabolic indices [Table 3]. Reductions in fasting insulin were reported in at least seven studies [24,26,28,31,32,33,37]. Improvements in HOMA-IR and ALT were also described in the selected trials; however, these outcomes were not consistently reported across studies and were therefore not included in the tabular summary (Table 3). Two studies measured adipokines, documenting decreased leptin and increased adiponectin, reflecting favorable changes in adipose tissue function [26,32].

However, the reporting of dietary behavior and metabolic biomarkers remained inconsistent. Recent evidence-gap work has shown that dietary intake and diet quality are rarely assessed or reported as outcomes in incretin-based trials, limiting mechanistic interpretation and the ability to link pharmacotherapy with nutrition-related behavioral change [38]. Our pediatric-focused synthesis complements this by describing how nutritional and behavioral co-interventions are implemented and reported alongside clinical outcomes in children and adolescents. Some studies reported data on lipid profiles [23,26,27,28,30,31,32,36,37], and none provided standardized measures of dietary intake or changes in food preferences. This limits the interpretation of GLP-1’s broader metabolic impact, especially in relation to nutrition-mediated mechanisms.

Table 3 summarizes the effectiveness of GLP-1 RAs in adolescents based on reported BMI outcomes, lipid changes (if available), and insulin. BMI reduction is reported either as a change in BMI (kg/m²), BMI z-score (SDS), or percentage change depending on available data.

Considerable heterogeneity was observed in outcome measures and reporting standards across included studies. Baseline BMI was inconsistently expressed (absolute values vs. BMI-SDS), and behavioral parameters such as dietary adherence or appetite-related outcomes were rarely assessed or reported. This variability limits direct comparability between studies and obscures the nutritional context in which GLP-1-related effects occur.

The observed clinical benefits may reflect interactions between pharmacological effects and lifestyle-related factors, particularly in studies incorporating more structured interventions; however, this relationship cannot be confirmed due to heterogeneity in study design and incomplete reporting.

Furthermore, most studies did not provide sufficiently detailed data to enable the recalculation or standardization of outcomes across a uniform analytical framework. The largest gaps in reporting were identified for metabolic and hormonal parameters, including HOMA-IR, leptin, and adiponectin, which were reported in only a minority of studies.

Given their potential relevance to the underlying mechanisms of action, these parameters were retained in the review; however, they were interpreted descriptively and not used to support definitive conclusions.