Systematic review of influence of ethnicity on efficacy and safety of pharmacotherapy for childhood and adolescent obesity

1. Protocol and registration

This systematic review was prospectively registered in the International Prospective Register of Systematic Reviews (PROSPERO), which is accessible at: http://www.crd.york.ac.uk/PROSPERO/. The full protocol can be accessed at: https://www.crd.york.ac.uk/PROSPERO/view/CRD420251176301 (registration number CRD420251176301).

2. Focus question

A structured PICO (Patient/Population, Intervention, Comparison, Outcome) framework was used to investigate the effect of ethnicity on the pharmacological treatment of childhood and adolescent obesity. The focus question is: In children and adolescents (≤18 years) with overweight or obesity (Population), do pharmacologic interventions for weight loss (Intervention), compared to placebo, standard care, lifestyle interventions, or other pharmacotherapies (Comparison), show differences in efficacy and/or safety outcomes (Outcome) across ethnic or racial subgroups (Study focus)?

3. Information sources

Eligible studies were identified via a thorough literature search of the PubMed, Embase, Cochrane Library, Web of Science, and Scopus databases. There is no relevant peer-reviewed literature on pharmacotherapy for obesity treatment in children and adolescents, with an emphasis on ethnic or racial subgroup analyses.

Only articles published in English and indexed from 2000 onward were considered. Full search terms and strategies for each database are detailed in the supplementary document entitled “Keywords and Search Strategy Details.”

4. Search

The search strategy was developed using the PICO framework and adapted across different databases. For transparency, the full electronic search strategy used in PubMed is detailed below, including all applied keywords, MeSH (medical subject headings) terms, and Boolean operators:

("Obesity"[Mesh] OR "Overweight"[Mesh] OR obesity [tiab] OR overweight [tiab]) AND ("Paediatrics" [Mesh] OR "Adolescent" [Mesh] OR "Child" [Mesh] OR children [tiab] OR adolescents [tiab] OR pediatric [tiab]) AND ("Pharmacological Treatment" [Mesh] OR "Drug Therapy" [Mesh] OR pharmacotherapy [tiab] OR medication [tiab] OR drug [tiab] OR "anti-obesity drugs" [tiab] OR liraglutide [tiab] OR semaglutide [tiab] OR metformin [tiab] OR orlistat [tiab] OR phentermine [tiab] OR topiramate [tiab]) AND ("Ethnic Groups" [Mesh] OR race [tiab] OR ethnicity[tiab] OR cultural [tiab] OR minority [tiab] OR disparities [tiab]) AND (efficacy [tiab] OR safety [tiab] OR adverse events [tiab] OR treatment outcome [tiab])

Filters applied: English language, human studies, year of publication 2000 onward. This strategy was modified appropriately for the Embase, Cochrane Library, Web of Science, and Scopus databases using specific indexing terms and syntax. The complete keyword and search term list is documented in the supplementary file entitled “Keywords and Search Strategy Details.”

5. Selection of studies

Records identified from the databases were uploaded to Rayyan and deduplicated. A prepilot screening checklist according to the inclusion and exclusion criteria was used to screen abstracts by 2 reviewers independently. Additional full-text articles meeting the initial search criteria were retrieved and reviewed in detail. The full-text screening was performed independently by 2 reviewers. Any discrepancies at any level were resolved by consensus and, if required, a third reviewer. A uniform eligibility list was used to direct both stages of the screening process, promote continuity across reviewers, and restrict bias. The screening process; number of studies identified, screened, and assessed for eligibility; and included in the review with reasons for exclusion at each stage are presented in the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) flow diagram [25].

6. Types of publications

This review included only articles published in peer-reviewed journals. Letters to the editor, commentaries, editorials, review articles, conference abstracts, and Ph.D. dissertations were excluded to maintain methodological rigor and ensure peer-reviewed reliability.

7. Types of studies

The review included randomized controlled trials (RCTs), nonrandomized comparative studies, and observational cohort studies published since 2000. All studies reported data on pharmacotherapy for childhood or adolescent obesity, with ethnicity-specific outcomes or subgroup analyses.

8. Types of participants/population

Eligible studies included children and adolescents aged ≤18 years who were diagnosed with overweight or obesity. Obesity was defined according to standardized criteria such as the WHO, Centers for Disease Control and Prevention, or International Obesity Task Force classifications. Participants from diverse ethnic or racial backgrounds were included if the study stratified or discussed the results by ethnicity.

9. Disease definition

Childhood and adolescent obesity was defined as excess body fat accumulation that presents health risks measured using BMI-for-age z scores, percentiles, or other accepted international growth standards.

10. Inclusion and exclusion criteria

11. Sequential search strategy

The following stepwise search strategy was used:

1. Comprehensive literature search across multiple databases;

2. Initial screening of titles and abstracts against the inclusion and exclusion criteria;

3. Full-text review of eligible articles; and

4. Final inclusion criteria included complete eligibility and relevance to the research objectives.

12. Data extraction

Data were collected using a pilot and standardized data abstraction form. Two authors independently extracted information about the study characteristics, participant demographics, intervention descriptions, comparator(s), ethnicity-related outcomes, and safety data. Differences were resolved through consensus or third-party consultation.

13. Data items

The following data were extracted:

Study design, year, and country

· Sample size and participant age

· Participants' ethnicity/racial composition

· Pharmacotherapy types used

· Comparator (e.g., placebo, lifestyle intervention)

· Efficacy outcomes (changes in BMI, BMI z score, and weight)

· Safety outcomes (adverse events and discontinuation rates)

· Subgroup analysis according to ethnicity (if available)

The results were organized in summary tables and grouped by pharmacotherapy class and ethnic subgroups to facilitate a systematic synthesis and comparison across study reports.

14. Risk of intrastudy bias

The risk of bias in the included studies was evaluated using different tools according to study type. The risk of bias of the RCTs was assessed using the Cochrane Collaboration tool [26,27]. Bias was assessed by 2 reviewers, and any disagreements were resolved by consensus or third-party adjudication.

15. Risk of interstudy bias

Since only 4 studies met the inclusion criteria, a formal assessment of publication bias with a funnel plot was not applicable according to the recommendations that consider at least 10 analyses necessary to perform a meaningful analysis. Therefore, we did not create a funnel plot. Instead, we qualitatively assessed the risk of bias across studies. The notable concerns include the following:

· Underrepresentation of minority and non-White participants across trials limits generalizability.

· Lack of prespecified ethnicity-based subgroup analyses in most studies leads to potential selective reporting bias.

· Post hoc nature of the subgroup analyses carries the inherent risks of analytical flexibility and an increased risk of false-positive or false-negative findings.

· Variations in how race and ethnicity were categorized, with some studies collapsing categories into broad groups such as “White” vs. “non-White” to reduce interpretive clarity.

· These factors were considered when interpreting the overall findings and confidence of the evidence.

16. Statistical analysis

A meta-analysis was initially planned; however, the results were ultimately not pooled due to substantial heterogeneity in the:

· Pharmacological agents (liraglutide, metformin, PHEN/TPM, and sibutramine)

· Outcome definitions (BMI, BMI z score, BMI percentage reduction, and metabolic indices)

· Duration of treatment and follow-up (range, 24–68 weeks)

· Analytical approaches to ethnicity (model adjustment, post hoc subgrouping, and categorical collapse)

The small number of included studies (n=4) further limited the feasibility of the quantitative synthesis. Instead, the findings were synthesized descriptively with focus on the direction and magnitude of the effects and consistency across studies. No pooled effect estimates, heterogeneity statistics (I²), or forest plots were generated.

17. Additional analyses

Due to the heterogeneity and limited number of included studies, no sensitivity, meta-regression, or statistical subgroup analyses were performed. Instead, intrastudy analyses reported by the trial authors were summarized narratively, including:

· Post hoc subgroup analyses for liraglutide evaluated whether sex, race, ethnicity, pubertal stage, and early treatment response predicted weight loss;

· Multivariable modeling for PHEN/TPM assessment of predictors such as age, sex, cognitive function, and baseline BMI;

· Adjusted models of metformin controlled for demographic factors; and

· Race-stratified comparisons of sibutramine in African American and Caucasian adolescents

However, no cross-study comparative analyses were feasible because of methodological variability, insufficient subgroup sample sizes, and a lack of harmonized race/ethnicity reporting.

1. Study selection

A total of 3,979 records were identified through the electronic database searches: 2,050 from PubMed, 423 from the Cochrane Library, 214 from Embase, and 1,292 from Scopus. After the removal of 395 duplicates, 3,584 records were subjected to title and abstract screenings.

Of these, 54 studies were subjected to full-text review. Among them, 22 studies were excluded after full-text assessment for reasons such as a lack of an ethnicity-stratified analysis, absence of comparator groups, and inapplicability of the study populations (Table 1) [16,28-48].

Table 1.

Summary of excluded studies examining the influence of ethnicity on pharmacotherapy efficacy and safety for childhood and adolescent obesity

Additionally, 28 studies could not be retrieved in full text despite repeated efforts. Ultimately, 4 studies met all inclusion criteria and were included in the qualitative synthesis (Table 2) [49-52].

Table 2.

Summary of studies examining the influence of ethnicity on pharmacotherapy efficacy and safety for childhood and adolescent obesity

The study selection process is illustrated in the PRISMA flow diagram (Fig. 1) [25].

Fig. 1.

PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) 2020 flow diagram detailing the identification, screening, and inclusion process of studies in this systematic review. *PubMed, Embase, Cochrane Library, Web of Science, and Scopus databases. **After title and abstract screening.

2. Study characteristics

The characteristics of the 4 studies included in this systematic review (Table 2) highlight a growing body of evidence evaluating the efficacy and safety of pharmacotherapy across diverse adolescent populations with obesity. In a post hoc analysis of the SCALE Teens RCT, Bensignor et al. [49] assessed the efficacy of liraglutide among 251 adolescents aged 12–17 years with obesity, of whom approximately 84% were White and 22% identified as Hispanic/Latino. The trial showed marked reductions in BMI with liraglutide versus placebo; although subgroup estimates did not show statistically significant differences, the study was too underpowered to determine whether true ethnic differences exist. Similarly, Wilson et al. [50] conducted a 48-week double-blind placebo-controlled multicenter trial of extended-release metformin in 77 adolescents (21% African American, 8% Asian). They observed a small but statistically significant BMI decrease in the metformin group. However, the absence of observed interaction effects should be interpreted cautiously given the limited subgroup sample sizes.

Another study by Bensignor et al. [51] reported the BMI responses of 222 adolescents (32% Black, 30% Hispanic) to PHEN/TPM in a phase IV RCT. The results showed that both mid- and top-dose regimens significantly reduced the BMI across all subgroups, with no statistically significant moderation by race/ethnicity, although the trial was not designed or powered to test ethnic differences.

Finally, Budd et al. [52] examined differential responses to sibutramine plus behavioral therapy in a racially stratified sample of 79 adolescents (45 Caucasians, 34 African Americans) and found greater weight and BMI reductions in the Caucasians, although the study lacked power for a definitive subgroup analysis and sibutramine has since been withdrawn owing to cardiovascular risks.

Collectively, these 4 studies incorporated racially and ethnically diverse samples and reported pharmacotherapy-associated BMI reductions across the study groups. However, none were adequately powered to determine whether ethnicity meaningfully modifies treatment effects, underscoring the need for intentionally designed trials.

3. Intrastudy risk of bias

The risk of bias across the included RCTs was assessed using the Cochrane RoB 2 tool (Cochrane Risk of Bias Methods Group, 2019), and the findings are presented in Fig. 2 [49-53]. Overall, the methodological quality was moderate with some variability in outcome- and study-level bias.

Fig. 2.

Risk of bias assessment using the Cochrane Risk of Bias 2 “traffic light” visualization for the 4 included randomized controlled trials [49-52].

Astrup et al. [54] conducted a randomized double-blind placebo-controlled trial assessing tesofensine in obese adolescents. that demonstrated a low risk of bias across all domains, including the randomization process, deviations from intended interventions, missing outcome data, outcome measurements, and the selection of reported results, yielding an overall low risk of bias.

Wilson et al. [50] investigated metformin XR in adolescents and showed a low risk of bias in most domains. However, based on moderate attrition (approximately 23%), the certainty of evidence was rated low for missing outcome data, and the risk of bias was labeled “some concerns.”

Bensignor et al. [51] conducted a secondary analysis of the PHEN/TPM trial in obese subjects. The management of missing outcome data and the choice of reported results (because they were very exploratory and non-per-specified) were, however, sources of some concern with respect to both internal study validity and generalizability. Therefore, the overall risk of bias was rated as “some concerns.”

Budd et al. [52] reported a secondary analysis of a sibutramine trial stratifying outcomes by race. Although the original trial maintained adequate randomization and blinding, the post hoc nature of the analysis introduced some concerns regarding deviations from intended interventions and outcome measurements. Moreover, the absence of a prespecified analytic plan and the likelihood of selective reporting led to a high risk of bias in the domain of selective reporting and an overall judgment of “some concerns” for the study. Overall, the risk of bias limitations further restrict confidence in ethnicity-specific findings, contributing to uncertainty about potential true subgroup differences.

4. Results of individual studies

Across the 4 included RCTs and their secondary or post hoc analyses, outcomes pertaining to weight reduction, BMI change, metabolic markers, and adverse events were extracted for each intervention and comparator group. Because ethnicity-specific efficacy estimates are limited and underpowered, the results are summarized primarily using overall treatment effects, with subgroup findings highlighted when available. The findings should be interpreted as reflecting insufficient evidence and not as evidence of equivalence across groups.

In the SCALE Teens post hoc analysis, adolescents treated with liraglutide 3.0 mg demonstrated significantly greater weight loss versus those treated with placebo, with higher proportions achieving ≥5% and ≥10% reductions in BMI; confidence intervals (CIs) around effect estimates favored liraglutide consistently across groups [49]. Importantly, subgroup analyses indicated no significant effect modification by sex, race, or ethnicity, suggesting similar treatment responses across demographic groups. Early responders (≥4% BMI reduction at week 16) showed a markedly greater probability of achieving clinically meaningful weight loss at week 56. However, the study was not powered to detect such differences, and the absence of statistical significance cannot be interpreted as evidence of a lack of an effect of ethnicity.

The Metformin XR RCT reported a mean BMI reduction of -0.9 units (standard error [SE], 0.5) in the metformin group compared with a BMI increase of 0.2 units (SE, 0.5) in the placebo group at 48 weeks, with a significant intergroup difference (P=0.03), although CIs overlapped for some metabolic outcomes [50]. Ethnicity did not significantly interact with treatment response (P>0.20), although limited subgroup sample sizes precluded a meaningful inference. Secondary outcomes such as dual-energy x-ray absorptiometry-measured fat mass and computed tomography–derived visceral fat showed no significant intervention effects.

For PHEN/TPM, both middose (7.5/46 mg) and topdose (15/92 mg) regimens produced significantly greater BMI reductions than placebo at 56 weeks, with ≥5% BMI reductions achieved in 38.9% and 46.9% of the treated adolescents, respectively, compared with 5.4% in the placebo group [51]. The effect sizes were large, and CIs did not cross zero for BMI change. Race and ethnicity were not significantly predictive of treatment response; however, this null finding reflects the limited analytical power and broad racial categorization, reducing interpretability. The subgroup power was low, but the analysis found no evidence of differences in the effectiveness of PHEN/ TPM across racial and ethnic groups.

For example, the sibutramine trial secondary data analysis comparing African American and White adolescents found that Whites on sibutramine lost a significantly greater amount of weight (-9.0 kg) than their placebo-treated peers (-3.0 kg), as did the African Americans, but the treatment effect was not statistically significant (-6.9 kg vs. -3.4 kg) and lower CIs for subgroup comparisons made it difficult to draw inferences from the trial findings [52]. The CIs were wide, and the subgroup sample sizes were insufficient to reliably evaluate ethnic differences.

Reductions in triglycerides, insulin, and HOMA-IR were noted within each ethnic group, and side effects were mild for both drugs, although the increases in pulse rate and blood pressure were higher with sibutramine. Because the outcomes, interventions, and analytic methods varied (and the effect size data were classified by race/ethnicity strata), a formal meta-analysis was not possible. However, across all 4 studies, treatment with pharmacotherapy (liraglutide, metformin, PHEN/TPM, and sibutramine) consistently demonstrated greater reductions in BMI than treatment with placebo, and no study found statistically significant differences in treatment response across racial or ethnic subgroups. Because the outcomes, pharmacologic agents, and statistical approaches were heterogeneous and ethnicity-stratified effect estimates were inconsistently reported, a formal meta-analysis was not feasible. The quantitative outcomes of each study are summarized in Table 3 to enhance transparency. These outcomes included intervention-placebo differences in BMI change, proportions achieving clinically meaningful weight loss thresholds, and ethnicity-specific effects, where numerical values were available. While subgroup analyses were conducted of all 4 RCTs, only the sibutramine trial reported ethnicity-stratified numerical outcomes (African American: -6.9 kg vs. -3.4 kg for placebo; Caucasian: -9.0 kg vs. -3.0 kg for placebo). The remaining trials reported no statistically significant interaction effects but did not provide stratified quantitative estimates, limiting comparative interpretations.

Table 3.

Summary of key quantitative outcomes across included randomized controlled trials

5. Synthesis of results

The 4 included RCTs collectively suggested that pharmacotherapy produces modest to substantial reductions in BMI among adolescents with obesity, with generally consistent effects across racial and ethnic subgroups where these were examined [49-52]. However, the available evidence is insufficient to determine whether meaningful racial or ethnic differences exist in treatment responses.

In the post hoc analysis of liraglutide, adolescents receiving 3.0 mg daily alongside lifestyle intervention were more likely to achieve clinically meaningful BMI reductions (≥5% and ≥10%) than those receiving placebo, and no significant effect modification by race or ethnicity was detected, indicating comparable efficacy across White, non-White, and Hispanic/Latino participants [49]. However, all such analyses were underpowered, exploratory, and based on heterogeneous racial categories, making it inappropriate to draw conclusions about equivalence.

Similarly, metformin extended-release produced a small but statistically significant reduction in BMI compared with placebo over 48 weeks, and an adjustment for race and ethnicity revealed no treatment-ethnicity interactions, although the overall effect size was modest and the trial was underpowered for subgroup analyses [50]. In the secondary analysis of a phase IV trial of PHEN/TPM, the mid- and top-dose regimens yielded substantially greater proportions of adolescents achieving a ≥5% BMI reduction versus placebo, and ethnicity was not a significant predictor of treatment response when considered alongside age, sex, pubertal status, and metabolic or psychological factors [51].

The sibutramine trial, the only study explicitly designed to compare outcomes of African American and Caucasian adolescents, suggested a potential differential pattern: Caucasian adolescents receiving sibutramine plus behavioral therapy had significantly greater weight and BMI reductions than those receiving behavioral therapy plus placebo, whereas African American adolescents showed numerically greater weight loss with sibutramine than placebo, but the difference did not reach statistical significance, likely reflecting limited power in the smaller African American subgroup [52]. Across all 4 trials, pharmacotherapy was generally well tolerated, although gastrointestinal adverse events were common with liraglutide and metformin, while sibutramine was associated with increases in pulse and blood pressure, consistent with its later withdrawal from clinical use [49,50,52].

Because the included studies evaluated different pharmacological agents (liraglutide, metformin, PHEN/TPM, and sibutramine), varied outcome definitions (absolute BMI change, percentage BMI reduction, and BMI z score), and differences in follow-up duration and analytical approaches to ethnicity (including post hoc and secondary analyses), a quantitative meta-analysis was not performed. Instead, a narrative synthesis was conducted. Taken together, the available evidence suggests that: (1) antiobesity pharmacotherapies combined with lifestyle interventions are more effective than placebo at reducing BMI in adolescents with obesity; (2) none of the trials demonstrated a statistically significant moderation of treatment effect by race or ethnicity; and (3) conclusions regarding ethnic differences remain constrained by small subgroup sizes, underpowered analyses, heterogeneous drug classes, and limited reporting of ethnicity-stratified outcomes [49-52]. Because these pharmacotherapies represent distinct mechanisms of action with different metabolic pathways and safety profiles, the absence of statistically significant ethnicity-related differences must be interpreted in the context of substantial pharmacologic heterogeneity.

6. Risk of bias across studies

The risk of bias across the included RCTs was assessed using the Cochrane Risk of Bias 2 tool, which evaluates 5 domains of potential bias in randomized studies [55]. Across the 4 trials, the overall risk of bias ranged from low to some concerns. Common limitations include reliance on post hoc subgroup analyses, moderate attrition in some trials, the absence of prespecified ethnicity analytic plans, and a risk of selective reporting. These factors reduce the certainty of the evidence regarding whether ethnicity moderates the effects of pharmacotherapy.

Funnel plotting could not be used to formally assess publication bias, as this meta-analysis was not possible because of the heterogeneity between pharmacological agents, outcome definitions, and analytical methods for ethnicity. Furthermore, considering the 4 included trials (each of which tested a different drug), the funnel plot was statistically inappropriate and unreliable. Therefore, small-study effects or publication bias could not be quantitatively evaluated.

However, a qualitative evaluation showed a possible risk of reporting bias within the studies (since all 4 trials reported the initial distribution of race/ethnicity and only one trial directly compared treatment effects with respect to ethnicity) [52]. The remaining studies reported either ethnicity-adjusted or exploratory results but not fully stratified efficacy or safety results. This practice may have led to the underreporting of subgroup effects, thereby introducing uncertainty regarding whether ethnic differences exist in pharmacotherapy responses.

The cumulative evidence is limited by: (1) underpowered subgroup analyses, (2) inconsistent presentation of ethnicity-specific outcomes, and (3) methodological shortcomings attributed to secondary analyses. Together, these issues compromise the assessment of the effect of ethnicity on the efficacy and safety of pharmacotherapy for childhood obesity.

7. Additional analysis

Subgroup and predictor analyses were the main additional analyses in the included trials, as no sensitivity analyses or meta-regressions could be performed owing to the small number of studies and diversity of interventions. In a post hoc analysis of the SCALE Teens study, Bensignor et al. [49] assessed whether baseline characteristics such as sex, race, ethnicity, Tanner stage, glycemic status, and obesity class modulated the treatment response to liraglutide. There was no significant treatment-by-race or treatment-by-ethnicity interaction, indicating a similar effectiveness of liraglutide across all race/ethnicity groups; however, an early response in BMI at week 16 (≥4% reduction) emerged as a strong predictor for weight loss at week 56. In the phase IV PHEN/TPM trial, Bensignor et al. [51] performed multivariable analyses to identify predictors of a BMI reduction, again finding that race/ethnicity did not significantly predict treatment response when evaluated alongside age, sex, pubertal status, glycemic status, cognitive performance, and quality of life, although ethnic categories were collapsed and the study was not powered for a detailed subgroup analysis. Wilson et al. [50] adjusted metformin treatment effects for race and ethnicity in their primary models and found no significant treatment-ethnicity interaction, but they did not present fully stratified subgroup results. Budd et al. [52] conducted explicit racial subgroup comparisons between African American and Caucasian adolescents receiving sibutramine plus behavioral therapy versus placebo, showing a statistically significant drug–placebo difference in BMI reduction among Caucasians and a non-significant but medium-sized effect among African Americans, likely reflecting limited power in the smaller subgroup. No meta-regressions or formal sensitivity analyses (e.g., exclusion of high risk of bias studies) were performed across trials due to the small number of studies, diverse drug classes, and variability in outcome definitions and follow-up durations [49-52].

1. Summary of evidence

This systematic review synthesized evidence from 4 RCTs and secondary analyses to evaluate whether ethnicity influences the efficacy or safety of pharmacotherapy in obese children and adolescents. In summary, these findings suggest that, although pharmacological treatments (liraglutide, metformin, PHEN/TPM, and sibutramine) are associated with clinically meaningful decreases in BMI in youth, the available evidence is of poor quality and inconsistent in supporting ethnic differences in treatment responses. Race and ethnicity did not significantly moderate the treatment effects across the 3 included studies. Post hoc analyses of the SCALE Teens study found that the reduction in BMI with liraglutide was similar across racial and ethnic groups; moreover, no treatment-ethnicity interactions were observed [49]. In addition, among the secondary analyses of a phase IV RCT of PHEN/TPM, race and ethnic descent failed to predict BMI despite multivariable adjustment, although the non-Caucasian ethnic groups were quite broad [51]. Similarly, the metformin extended release trial found no treatment-ethnicity interaction after accounting for demographic covariates, although the subgroup analyses were limited by small sample sizes and diverse populations [50].

The latter, although explored in a secondary analysis of the sibutramine trial, detected possible ethnic differences: compared with Caucasian adolescents, sibutramine-treated African American adolescents achieved an absolute difference (sibutramine-placebo) in weight (and BMI only slightly smaller but not significantly so; on average, these children had very mild increases or trivial decreases), although the power in this subgroup was likely inadequate to produce statistical significance [52]. This analysis is first of its kind to directly compare race-specific pharmacotherapeutic effects across trials.

There is low to moderate evidence taken together and across ethnic groups for or against the effect of ethnicity on the pharmacological treatment of obesity in youth, in which methodological limitations played a significant role. Between trials, non-White populations were underrepresented, none of the studies were adequately powered to test for effect modification by race or ethnicity, and subgroup analyses were post hoc rather than prespecified. This reduces confidence in the conclusions but also highlights major uncertainties for health staff and policymakers. However, taken together, the trials demonstrated that Food and Drug Administration (FDA)–approved drugs for pediatric obesity can be efficacious in a variety of populations, and the initial treatment response may be a better indicator than patient-specific factors. For providers, these results support the appropriateness of providing pharmacotherapy to eligible youths in all racial and ethnic groups, recognizing larger structural and access-related disparities that may affect treatment uptake. For policymakers, the continued underrepresentation of minority populations in obesity trials furthers the importance of fair research designs, mandatory reporting of race/ethnicity, and proactive recruitment in a way that will ultimately be generalized to the larger population. Because several trials exhibited concerns related to selective reporting, incomplete outcome data, and a lack of prespecified subgroup analyses, the overall certainty of evidence for ethnicity-related differences remains low and the current findings should be interpreted cautiously.

2. Interpretation of findings in the context of existing studies

The results of this review are largely consistent with those of other studies indicating that obesity in childhood and adolescence is disproportionately the highest among Black, Hispanic, American Indian, and some Asian subpopulations, but trials of pharmacotherapy for youth rarely offer a rigorous analysis by ethnicity. Previous studies confirmed that ethnic disparities in obesity are due to the interactions among socioeconomic disadvantages, the food environment, cultural norms, and genetic susceptibility (e.g., variation in obesity-related alleles and pharmacogenomic variants). However, despite these previously documented differences, the trials analyzed herein indicated that drug effectiveness does not meaningfully differ by ethnicity insofar as it can be determined from the existing evidence. A new multicenter analysis found no evidence of an interaction between ethnicity and response to hepatitis C virus treatment. This failure to detect an ethnic effect on treatment response may be due to true biological similarities among groups but is more likely to be the result of methodological constraints (limited subgroup sizes, post hoc testing) and inadequate representation of minority populations.

Interestingly, in adult obesity pharmacotherapy research, a few ethnic differences have emerged: studies are mixed on orlistat efficacy in some groups, there is variability in metformin response, and previous studies suggested differences in GLP-1 receptor agonist tolerability. The lack of such a pattern in pediatric trials indicates inadequate statistical power rather than true similarities. Only one study performed a secondary analysis directly along ethnic lines; it indicated a trend in which Caucasian versus African American adolescents responded differentially to sibutramine, indicating the potential for ethnicity-related variations when sufficiently explored [52]. Sibutramine is no longer approved because of the risk of cardiovascular disease, which reduces its modern applicability.

Overall, these results reinforce a central theme that pharmacotherapy might not only be equally effective across ethnic groups but that we may lack sufficient evidence to draw strong conclusions, and structural inequities in who participates (or does not) in research still stand as an obstacle to the true understanding of population-level differences.

The interpretation of ethnicity-related outcomes was further limited by the heterogeneity of the pharmacological agents evaluated. Liraglutide, metformin, PHEN/TPM, and sibutramine differ substantially in their mechanisms of action, expected weight loss, and side-effect profiles. As a result, the observed similarities in subgroup outcomes cannot be assumed to reflect true equivalence across ethnic groups but rather the inability of small, mechanistically diverse studies to detect differences. This pharmacological variability also limits the feasibility of the meta-analysis and underscores the need for future research evaluating the effects of ethnicity on specific drug classes.

3. Strengths and limitations of the evidence

The evidence provided herein has several strengths, including the implementation of an RCT design, which is considered the best method for assessing pharmacotherapy effectiveness and safety. All 4 studies were randomized double-blind placebo-controlled trials or strong secondary analyses of such studies and, thus, had high internal validity. Moreover, all studies included common outcome measures, including BMI, BMI z score, percent reduction in BMI, and metabolic outcomes, which provided for comparability across interventions. Multicenter enrollment in studies such as SCALE Teens and the metformin ER study improves generalization and the recruitment of diverse populations, especially for trials that have 20%–30% representation of African American or Hispanic adolescents that offer valuable, albeit limited, inferences regarding ethnicity-related effects.

However, several limitations weaken the robustness and certainty of this evidence. First, none of the trials were prospectively planned or powered to examine differences in treatment effects among racial subgroups; thus, there is an important risk of type II errors with any interpretation of subgroup findings. Three of the 4 studies were post hoc or secondary analyses, which are also at a higher risk of bias due to analytical flexibility, missing data, and selective reporting. Ethnic subgroups are often collapsed into broad categories (e.g., non-Hispanic White compared to “Hispanic and/or not White”), which obscured heterogeneity while blurring interpretation. In addition, the recurring underrepresentation of racial and ethnic minority groups, often at less than 15%–20% of total participants in trials, restricts the generalizability of findings for populations most burdened by childhood obesity.

An additional limitation was the wide heterogeneity of medications among the studies, including liraglutide, metformin, PHEN/TPM, and sibutramine. This heterogeneity, in addition to variations in the duration of follow-up, intensity of lifestyle-program interactions, and definitions of outcomes, precluded a formal meta-analysis and diminished the capacity to aggregate findings across classes of pharmacological agents. Furthermore, none of the studies included pharmacogenomic analyses, although ethnic variability in drug metabolism and GLP-1 receptor signaling has been recognized, restricting our understanding of the potential mechanistic differences in treatment response.

Finally, the growing obsolescence of these findings can be ascribed to the history of sibutramine discontinuation in clinical practice. These collective constraints highlight the urgent need for sufficiently powered, prospectively conducted, and ethnically diverse RCTs to inform equitable pediatric obesity pharmacotherapy.

4. Implications for practice and policy

The implications of this review are significant for clinical practice, health equity initiatives, and policy development regarding pediatric obesity treatment. The above evidence in the field of obesity medicine summarized, perhaps not surprisingly, that pharmacotherapy, including liraglutide, metformin, and PHEN/TPM, is safe and effective in a variety of racial and ethnic groups without clear evidence of diminished efficacy or increased harm among minority populations. For providers, this promotes the fair offering of pharmacological treatments for obesity to all eligible youths, including those with severe obesity or who do not respond well to lifestyle intervention efforts. An early naltrexone response, as opposed to patient demographic characteristics, is a better clinical predictive marker of long-term benefit and highlights the importance of close monitoring in the early stages and tailoring individual treatment.

However, structural inequalities remain critical. Racial and ethnic minority groups, including Black, Hispanic, American Indian/Alaska Native, and some Asian subgroups, are disproportionately affected by obesity but underrepresented in clinical trials, raising concerns that the evidence on which we rely to improve patient care does not reflect the populations most impacted. Healthcare systems and policymakers must focus more on the issue of access to obesity care, which includes insurance coverage for FDA-approved antiobesity medications, culturally sensitive behavioral approaches, and language support services. This is particularly concerning, as there is evidence that interpreters may increase medication prescriptions and reduce disparities in access to care for non-English-speaking families.

Policy-based frameworks should also require and incentivize diversity in pediatric clinical trial populations, including trial designs that pre-specify ethnicity-based subgroup analyses such as those planned here and have sufficient statistical power to detect intergroup differences. These efforts are in line with the long-term strategy of precision medicine and national equity efforts led by agencies such as the National Institutes of Health, FDA, and WHO.

Finally, addressing these disparities in pediatric obesity outcomes will necessitate a concerted effort to improve clinical care, research infrastructure, and public policies. The representation of diverse populations in research and the ability to access interventions that work are vital for the treatment of obesity in youth.

5. Future research directions

Substantial evidence gaps remain to be addressed in response to this review to further ensure the equitable evidence-based pharmacological management of pediatric obesity. First, there is an urgent need for large prospectively designed RCTs that are specifically powered to assess ethnic variations. Subgroups in Black, Hispanic, Indigenous, and Asian populations are not well represented; there are an inadequate number of cases and controls to allow for meaningful subgroup analyses in these groups.

Second, future trials should include pre-planned standardized race and ethnicity analyses with transparent reporting of subgroup sample sizes, interaction testing, stratified efficacy, and safety endpoints. If these analyses are limited to post hoc or collapsed categories, important heterogeneity will be masked and firm conclusions prevented. Standardized outcome measures (e.g., consistent outcomes for BMI-related measurements, metabolic biomarkers, and prolamin antibodies) also allow data to be pooled across and compared among studies.

Third, there is a high demand for research into pharmacogenomic, metabolic, and behavioral mediating mechanisms that could account for variations in the response to treatment. Pharmacogenetic variants that influence GLP-1 signaling, appetite control, insulin sensitivity, or drug metabolism may be ethnic group-specific and contribute to precision dosing or personalized therapy. Integrating genomic, metabolomic, and behavioral adherence would allow deeper insight into the observed response variability.

Fourth, future investigations must assess the real-world effectiveness of antiobesity medications across diverse population settings and relevant safety surveillance as well as treatment discontinuation/persistence and access barriers. Observational cohort studies in diverse, large, and representative health systems would complement RCTs and more closely reflect clinical heterogeneity. Trials should also evaluate the efficacy of combining pharmacotherapy with culturally adapted lifestyle interventions to determine the synergistic effects in minority populations.

Lastly, policy-relevant research is warranted on the additional role that insurance coverage, language access, socioeconomic status, and healthcare infrastructure play in disparities in obesity treatment utilization and outcomes. Incorporating clinical, biological, and sociostructural lenses into future research may provide a more holistic and equitable approach to pediatric obesity pharmacotherapy.