Effects of different exercise modalities and intensities on body composition in overweight and obese children and adolescents: a systematic review and network meta-analysis

Introduction: Childhood and adolescent overweight and obesity are global public health issues. Previous studies on exercise and overweight and obese children have produced inconsistent findings and lacked comparisons between different exercise modalities and intensities. Therefore, a network meta-analysis is necessary to provide evidence-based intervention programs. This study aims to identify the effects of different exercise modalities and intensities on changes in body composition in overweight and obese children and adolescents. Methods: A search for randomized controlled trials was conducted on Web of Science, PubMed, Scopus, and Embase involving exercise interventions aimed at improving body composition (body fat percentage, BMI, fat mass, fat-free mass, body weight) in overweight and obese children and adolescents. A random effects network meta-analysis was performed using STATA 14.0 software within a frequentist framework. The literature quality was assessed using the Cochrane Risk of Bias Tool 2.0. Results: Thirty-two papers involving 1,452 participants were included. There were six types of intervention involved in the study, including moderate intensity aerobic exercise, high-intensity aerobic exercise, moderate intensity resistance exercise, high-intensity resistance exercise, moderate intensity combined exercise, and high-intensity combined exercise. The network meta-analysis results revealed that high-intensity combined exercise was the best exercise mode for improving BMI [mean difference in kg/m2 = −1.65, 95% CI (−3.27, −0.02)] and reducing fat mass [mean difference in kg = −2.87, 95% CI (−4.84, −0.91)]. Moderate intensity combined exercise was the best mode for weight loss [mean difference in kg = −4.58, 95% CI (−5.94, −3.22)] and improvement in body fat percentage [mean difference in% = −2.52, 95% CI (−3.83, −1.20)]. High-intensity resistance exercise had the optimal effect in increasing fat-free mass [mean difference in kg = 1.10, 95% CI (0.22, 1.99)]. Conclusion: In conclusion, the study found that combined exercise, whether moderate or high intensity, was more effective than any other exercise modality in improving body fat percentage and BMI, reducing fat mass and weight. Resistance exercise was the most effective in increasing fat-free mass.


Introduction
Over the last five decades, overweight and obesity have evolved from a minor public health issue affecting developed countries to a major public health concern with a global reach. According to the Global Burden of Disease Study 2013, the prevalence of overweight and obesity in children and adolescents was over 23% in developed countries and over 13% in developing countries. In 2019, the World Obesity Federation estimated that by 2025, 206 million children and adolescents aged 5-19 will be obese, with that number increasing to 254 million in 2030 (Lobstein and Brinsden, 2019). The total direct and lifetime costs resulting from obesity and overweight are estimated at 144,200 Euros, posing a significant economic challenge to children, adolescents, their parents, society, and the state (Hamilton et al., 2018).
Overweight and obese children and adolescents face both immediate and long-term health risks. They are more likely to have cardiovascular and metabolic disease risk factors, such as high triglycerides, hypertension, and high cholesterol, with 70% having at least one risk factor (Freedman et al., 2007). Obesity is often accompanied by various metabolic syndrome manifestations such as hyperinsulinemia, insulin resistance, and abnormal glucose tolerance (Sinha et al., 2002). It is a strong predictor of the development of type 2 diabetes and can cause effects that extend into adulthood, especially in children and adolescents with severe obesity (Daniels et al., 2005;Kelly et al., 2013). Additionally, being overweight and obese has been the most common cause of social exclusion and bullying, leading to poor mental health, impaired social development and education, and high rates of self-harming behavior and suicide among overweight and obese children and adolescents (Pont et al., 2017;Puhl and Lessard, 2020;Jebeile et al., 2022).
Given the enormous medical and health costs associated with overweight and obesity in children and adolescents, identifying effective palliative and intervention measures is a crucial priority for public health agencies. Currently, exercise and dietary interventions, in addition to surgical and pharmacological interventions, are proving to be effective ways to prevent and treat overweight and obesity (Peirson, Fitzpatrick-Lewis, Morrison, Ciliska, et al., 2015;Peirson, Fitzpatrick-Lewis, Morrison, Warren, et al., 2015). Treating obesity and promoting health through exercise is one of the most significant health and fitness trends worldwide (Stewart and Sutton, 2012). Body composition, which includes fat mass, fat-free mass, body fat percentage, and body mass index, is a crucial factor in obesity. Adverse changes in body composition and weight gain are key features of obesity. Multiple and inconsistent findings exist on how exercise can affect the body composition of overweight and obese children and adolescents. Previous studies have used traditional pairwise comparative metaanalyses to determine the effects of different exercise modalities (such as aerobic, resistance, and combined exercise) on body composition (Schranz et al., 2013;Kelley et al., 2015;Méndez-Hernández et al., 2022). However, direct comparisons between these modalities still need to be made. Additionally, differences in the effects of exercise intensities on body composition in overweight and obese children and adolescents have been observed (Gutin et al., 2002;Lazzer et al., 2011). Therefore, the optimal exercise pattern (format and intensity) for improving body composition in this population remains unclear.
To address this research gap, we propose to use Network Meta-Analysis (NMA), which is an extension of the traditional head-to-head meta-analysis. NMA allows for the simultaneous comparison of the efficacy of three or more interventions (Watt and Del Giovane, 2022) and can analyze both direct and indirect comparisons (Li et al., 2011). When direct comparisons are lacking, NMA can compare the effects of two treatment measures indirectly through a common control. When direct comparisons exist, NMA combines the results of direct and indirect comparisons, thereby increasing the accuracy of the results (Jansen et al., 2012).
Therefore, in this study, we aim to use NMA to explore the effects of different exercise patterns on body composition in overweight and obese children and adolescents. Our goal is to provide an evidence-based basis for selecting effective intervention programs for this population.
2 Materials and methods

Protocol and registration
This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement and was registered in the PROSPERO database under the registration number CRD42023403600. The results are reported in compliance with the extended items of the network meta-analysis as recommended by PRISMA (Hutton et al., 2015). Additionally, specific aspects of the network meta-analysis were reported in accordance with the guidelines set forth by Chaimani et al. (2017) (as shown in Supplementary Table S1).

Eligibility criteria
Inclusion criteria for this systematic review were established based on the PICOS principles (population, intervention, comparator, outcomes, and study type).
Frontiers in Physiology frontiersin.org 02 Population (P): Participants aged 5-18 years, without any musculoskeletal disorders or clinical contraindications to exercise, who were overweight or obese were eligible for inclusion.
Intervention (I): Any exercise intervention trials were included. Comparators (C): Participants in the control group were required to maintain their current level of physical activity or perform a different intensity and/or form of exercise.
Outcomes indicators (O): Outcome measures related to body composition indicators, such as fat mass, body fat percentage, body mass index, fat-free mass, and weight, were considered.
Study type (S): Only randomized controlled trials (RCTs) published in English were eligible for inclusion.
1. Studies involving mixed samples of individuals with overweight/obesity and other non-communicable diseases. 2. Studies that included people aged >18 and <5 years. 3. Studies in which the effects of exercise interventions could not be isolated because they were part of a multi-component intervention (e.g., diet and exercise interventions). 4. Studies that did not provide details of the intensity of the exercise intervention. 5. Studies that did not evaluate the outcomes of interest of the article. 6. Studies that did not provide mean and standard deviation or data that could be translated into mean and standard deviation form. 7. Non-peer-reviewed English, abstracts, reviews, conference reports. 8. Studies for which the full text was not available.

Study selection
Endnote X9 literature management software was used to manage the literature search records. After removing duplicates, the remaining studies were screened independently by two authors (ZH and YL) using predefined inclusion and exclusion criteria, with any disagreements resolved by consensus. During the initial screening, the authors analyzed the titles and abstracts of all literature to identify potentially relevant literature. Subsequently, the full texts of potentially relevant studies were obtained and carefully read to identify the studies that met the inclusion and exclusion criteria. Finally, two authors discussed the final list of included studies, and a third author (YZ) was consulted to determine which studies could ultimately be included in this systematic review.

Data extraction
Two investigators extracted relevant data from the eligible literature. Extracted data included first author, year of publication, country, intervention duration, sample size, mean age, exercise intervention details (frequency, intensity, time, type), and outcome indicators. The classification of exercise intensity categories was based on the ACSM Guidelines for Exercise Testing and Exercise Prescription 9th edition for aerobic and resistance exercise intensity estimates (Medicine, 2013). High-intensity aerobic exercise was defined as >65% VO2max or >65% HRR or >75% HRmax or PRE of more than 5, and moderate intensity as 45%-65% VO2max or 50%-65% HRR or >65-75% HRmax or PRE of 3-4. High-intensity resistance training was defined as >75% of a one maximum load (1RM) or less than 10 repetition maximum, and moderate intensity as 50%-75% of a 1RM or 10-15 repetition maximum. The intensity of each exercise is described in detail in Table 1.

Risk of bias assessment
Two authors (ZH and JL) independently assessed the risk of bias. The Cochrane tool for assessing the risk of bias in randomized trials Version 2 (RoB2) was used to perform the risk of bias assessment (Sterne et al., 2019). RoB2 comprises five evaluation domains: randomization process, deviation from intended interventions, missing outcome data, outcome measurement, and selection of reported result. Each domain has several signal questions, and the risk of bias is classified as "low risk," "some concerns," and "high risk" based on the assessor's responses to the signal questions. The overall risk was determined based on the risk of bias rating for each of the five domains. If the risk of bias in all domains is evaluated as "low risk," then the overall risk of bias is "low risk." If the risk of bias in a domain is "some concerns," and there is no domain with "high risk," then the overall risk of bias is "some concerns." If a domain with a risk of bias is rated as "high risk," then the overall risk of bias is "high risk."

Evidence assessment
The CINeMA (Confidence In Network Meta-Analysis) model was used to comprehensively assess the quality of evidence for the results of the entire network analysis . The following six domains were considered: within-study bias (risk of bias), between-study bias (publication bias or reporting bias), indirectness, imprecision, heterogeneity, and incoherence. Each domain was graded as "no concern," "some concern," or "major concern" according to its severity, and the final NMA evidence quality rating was consistent with the GRADE system and classified as high, moderate, low, or very low .

Statistical analyses
The raw data were pre-processed using Microsoft Office Excel. The required data were converted to the mean and standard deviation of the before and after differences according to the formula provided in the Cochrane Handbook for Systematic Review (Higgins et al., 2019). The formulas are as follows: The mean difference and its 95% confidence interval (CI) were used for merged statistics. A pairwise meta-analysis of all outcome indicators was performed using a random-effects model to explore Frontiers in Physiology frontiersin.org 03 the relative effects of different exercise interventions. A Q test and I 2 statistic were used to quantify heterogeneity between included studies, with a significance level of p < 0.1. I 2 values of 25%, 50%, and 75% represent low, moderate, and high heterogeneity, respectively (Higgins and Thompson, 2002).
Network meta-analysis was performed with STATA 14.0 in a frequentist framework. Global inconsistency between interventions was assessed using the Wald test, and local inconsistency between direct and indirect comparisons between each intervention was assessed using the nodesplitting method. The consistency model was used if both global and local consistency assumptions were satisfied. Network evidence was plotted for interventions with different intensity exercise modalities. When there was a closed-loop structure, the Loop Inconsistency Test was performed to check the inconsistency of the closed-loop structure, and the inconsistency factors, 95% CI, and p values were calculated. Intervention efficiency was ranked using the surface under the cumulative ranking curve (SUCRA), with SUCRA values ranging from 0% to 100%. The closer the SUCRA value is to 100%, the better the intervention will be.

Study selection
The search resulted in 15,076 studies. After removing duplicate literature, 5,509 studies remained. After screening the titles and abstracts, 108 studies were read in full text for refinement. Then, 30 studies met the inclusion and exclusion criteria. After searching the reference list from relevant papers and reviews, two studies were obtained. The final meta-analysis included 32 studies (Figure 1).

Risk of bias assessment results
The risk of bias was assessed for the included literature. According to the risk of bias assessment, 26 studies (81.3%) were classified as having "some concerns," four studies (12.5%) were classified as "high risk," and two studies (6.3%) were classified as "low risk." The results of the risk of bias assessment for all studies are shown in Supplementary  Table S3 and Supplementary Figure S1.

Network evidence
The network meta-analysis included five body composition indicators: body fat percentage, BMI, fat mass, fat-free mass, and weight. Figure 2 shows the network plots of different intensity exercise modalities with the five body composition indicators. In each intervention's relationship, the circles represent different

Moderate-intensity resistance exercise
Intensity: 50%-75% of 1RM or 10-15 repetition maximum before muscle fatigue sets in Type: Forms of resistance training (e.g., free strength training, weight machines, and resistance band training)

High-intensity resistance exercise
Intensity: >75% of 1RM or less than 10 repetition maximum before muscle fatigue sets in Type: Forms of resistance training (e.g., free strength training, weight machines, and resistance band training)

Moderate-intensity combined exercise
Moderate-intensity aerobic combined with moderateintensity resistance exercise

High-intensity combined exercise
High-intensity aerobic combined with high-intensity resistance exercise HR max , maximum heart rate; HRR, heart rate reserve; RM, repetition maximum; VO 2max , maximal oxygen uptake. PRE, perceived effort.
Frontiers in Physiology frontiersin.org 04 interventions, with their size indicating the sample size included in that intervention. The lines represent the existence of direct comparisons between interventions, and the thickness represents the number of studies between two interventions. Both global and local inconsistency test analyses showed significant p > 0.05 for each outcome indicator, indicating that the consistency assumption was met (as shown in Supplementary Table S5, S6).

Body fat percentage
Twenty-eight studies compared the effects of interventions with different intensity exercise modalities on body fat percentage. The results of the pairwise and network meta-analysis are shown in Table 3. All exercise interventions, except for RT-M (p < 0.05), improved body fat percentage in obese children and adolescents when compared with the blank control group. Direct comparisons between the interventions showed that high-intensity resistance exercise improved body fat percentage better than high-intensity aerobic exercise

Body mass index
Thirty studies compared the effects of interventions using different exercise modalities of varying intensities on BMI. The results of the pairwise and network meta-analysis are shown in Table 5.

Fat mass
Fifteen studies compared the effects of exercise interventions with different intensity modalities on fat mass. The results of pairwise and network meta-analysis are presented in Table 7.

CON
The gray area represents the results of the network meta-analysis, and the white area represents the results of the pairwise meta-analysis. The bold values represent statistical significance at a level of p < 0.05. Estimates are shown as columns versus rows for network meta-analysis and rows versus columns for pairwise meta-analysis. The results are presented as mean difference (MD). N, number of studies in the comparison; NA, not available; AE-M, moderate-intensity aerobic exercise; AE-HI, high-intensity aerobic exercise; RT-M, moderate-intensity resistance exercise; RT-HI, high-intensity resistance exercise; COM-M, moderate-intensity combined exercise; COM-HI, high-intensity combined exercise; CON, blank controls.

Fat-free mass
Fifteen studies examined the effect of different intensity exercise modalities on fat-free mass. The results of pairwise and network meta-analysis are presented in Table 9. Pairwise meta-analysis results showed that RT-M significantly increased fat-free mass compared to blank controls [mean difference in kg = 0.70, 95% CI (0.46, 0.94)]. Direct comparisons between interventions showed that RT-HI intervention was superior to AE-HI [mean difference in kg = 0.69, 95% CI (0.47, 0.91)] and RT-M was superior to AE-M [mean difference in kg = 0.88, 95% CI (0.65, 1.11)]. Pairwise metaanalysis forest plots are available in the Supplementary Figure S4. The results of the network meta-analysis showed that RT-HI [mean difference in kg = 1.10, 95% CI (0.22, 1.99)] was significantly better than the blank controls (p < 0.05). Table 10

Weight
In this study, thirty-one different research works compared the effects of various exercise modalities with different intensities on weight loss. The results of pairwise and network meta-analysis are presented in Table 11     SUCRA, surface under the cumulative ranking curve. *Rank probability ranges from 0 to 100, where higher value indicates best intervention and lower value indicates worst intervention. AE-M, moderate-intensity aerobic exercise; AE-HI, high-intensity aerobic exercise; RT-M, moderate-intensity resistance exercise; RT-HI, high-intensity resistance exercise; COM-M, moderateintensity combined exercise; COM-HI, high-intensity combined exercise; CON, blank controls.

Publication bias
To test the publication bias, "corrected-comparison" funnel plots were drawn for body fat percentage, BMI, body fat mass, fat-free mass, and weight. The results showed poor symmetry on both sides of the funnel plots, suggesting the presence of publication bias or small sample effect in this study, as shown in the Supplementary Figure S9.

Discussion
The Network meta-analysis results showed that combined exercise (moderate and high intensity) was the most effective in reducing body fat percentage, BMI, fat mass, and body weight compared to the blank controls. Among the six type interventions, high-intensity resistance exercise was the less effective in reducing body weight, BMI, and fat mass but the most effective in increasing fat-free mass. These findings are consistent with the results of pairwise meta-analyses. Based on the effect size and Surface Under the Cumulative Ranking values   SUCRA, surface under the cumulative ranking curve. *Rank probability ranges from 0 to 100, where higher value indicates best intervention and lower value indicates worst intervention. AE-M, moderate-intensity aerobic exercise; AE-HI, high-intensity aerobic exercise; RT-M, moderate-intensity resistance exercise; RT-HI, high-intensity resistance exercise; COM-M, moderateintensity combined exercise; COM-HI, high-intensity combined exercise; CON, blank controls.
Frontiers in Physiology frontiersin.org (SUCRA), combined exercise (moderate and high intensity) had the highest likelihood of being the optimal measure of the combined effect of current interventions on body composition in overweight and obese children and adolescents. Overall, in terms of fat reduction, the effectiveness of all exercise models was ordered as combined exercise-aerobic exercise-resistance exercise. Weight and body mass index (BMI) are overall indicators of overweight and obesity and are the primary targets for treating obesity. In the current meta-analysis, all exercise interventions were effective in reducing body weight and BMI, except for high-intensity resistance exercise. Meanwhile, it was found that the effectiveness of exercise interventions depends mainly on the type of exercise. Specifically, combined exercise, whether moderate or high intensity, was more effective than single aerobic or resistance exercise, and aerobic exercise was more effective than resistance exercise. These findings are consistent with the results of previous studies (Willis et al., 2012;O'Donoghue et al., 2021).
The mechanisms of combined exercise suggest that aerobic exercise decreases plasma insulin levels and increases the secretion of glucagon, catecholamines, and epinephrine, prompting an increase in the activity of rate-limiting enzymes of the fat hydrolysis process. This accelerates fat hydrolysis and promotes lipolysis (Zouhal et al., 2013;Nakhostin-Roohi and Havaskar, 2018). Resistance exercise reduces fat mass mainly by activating the musculoskeletal system, increasing de-fatted body weight, and increasing resting metabolic rate (Dinas et al., 2014). Thus, integrating aerobic and resistance exercise stimuli increases exercise energy expenditure and resting metabolic rate, inducing a cumulative effect of both exercise modalities (Gäbler et al., 2018). Based on these theories, combined training may be the best exercise modality to reduce body weight and BMI.
Increases in fat mass and body fat percentage, and decreases in fat-free mass, are important features of obesity in children and adolescents. In the current study, the effects of different intensity  SUCRA, surface under the cumulative ranking curve. *Rank probability ranges from 0 to 100, where higher value indicates best intervention and lower value indicates worst intervention. AE-M, moderate-intensity aerobic exercise; AE-HI, high-intensity aerobic exercise; RT-M, moderate-intensity resistance exercise; RT-HI, high-intensity resistance exercise; COM-M, moderateintensity combined exercise; COM-HI, high-intensity combined exercise; CON, blank controls.
Frontiers in Physiology frontiersin.org exercise modalities varied for different outcome indicators. Combined exercise of two intensities was the most effective intervention for fat-related outcome indicators, such as fat mass and body fat percentage. This is consistent with the findings of George and Schwingshackl et al. (2013); George et al., 2017). For individuals specifically seeking to reduce total body fat volume, the best option is a form of exercise that produces the greatest total energy expenditure (Wilhelm and Pinto, 2019). However, individuals may experience fatigue when consistently performing a single exercise modality for long periods. Therefore, combined training can be an alternative to increase total energy expenditure by additionally increasing training volume and diversifying training modalities (Donges et al., 2013). Furthermore, combined training may reduce the amount of fat not only directly but also further reduce the percentage of body fat by increasing the amount of fat-free mass. The study also found that resistance exercise was less effective than other exercise modalities in reducing fat but was optimally effective in increasing fat-free mass. These findings are consistent with previous studies (Wilson et al., 2012;Schranz et al., 2013;Grgic et al., 2019). This may be because resistance exercise can selectively hypertrophy type II muscle fibers, thereby increasing total lean body mass (Fry, 2004;Farup et al., 2012). In contrast, aerobic exercise leaves type II muscle fibers unchanged or reduces muscle fibers (Edström and Ekblom, 1972).
This study may have several biases that could affect its validity. First, the total volume of exercise could be a crucial factor in the effectiveness of the intervention. Evidence suggests that combined aerobic and resistance exercise requires twice the training volume of single aerobic or resistance exercise, resulting in greater energy expenditure and fat reduction (Martins et al., 2010;Willis et al., 2012). Additionally, Rhea et al. (2003) found that combining resistance and aerobic exercise led to higher total training volumes and superior improvements in body composition compared to single-mode exercise (Rhea et al., 2003). Second, the frequency and duration of exercise should also be carefully considered. Although most studies in this review used a threetimes-per-week intervention, some studies used more or fewer sessions per week or varied the duration of each session. Such variations could lead to an unbalanced intervention protocol and introduce bias to the results. Therefore, it is important to standardize these variables to minimize their confounding effects on the outcomes. However, the study's methodological limitations precluded an in-depth analysis of the effects of exercise frequency and single exercise session duration on the intervention's efficacy.
To the best of our knowledge, this study is the first to focus on the effects of different types of exercise intensity on changes in body composition in overweight and obese children and adolescents using a network meta-analysis. This study can provide a reference for choosing the optimal exercise program for overweight and obese children and adolescents. Additionally, this study contains a relatively comprehensive set of body composition indicators, including body fat percentage, BMI, body fat mass, fat-free mass, and weight. It compares the effects of different exercise patterns on different indicators, providing instructors, researchers, and policymakers with useful and up-to-date information on the effects of different exercise patterns on different body components.
However, there are several limitations to this study. First, despite a comprehensive search of the published literature, some studies that used keywords other than those searched for in this paper may not have been included, resulting in selection bias. Second, the study currently lacks evidence for interventions such as low-intensity aerobic exercise, resistance exercise, and combined exercise, limiting the study's comprehensiveness. Third, as other components of the exercise intervention, such as time and frequency, were not considered in this study, this may have influenced the study's findings. Fourth, the included literature primarily focuses on aerobic exercise, with fewer studies on resistance and combined exercise interventions, which limits the number of direct comparisons between different exercise interventions and thus lowers the reliability of the evidence. Fifth, according to the CINeMA assessment, most evidence was rated as low or very low in the different comparisons. Thus, this may weaken the strength of the current findings.

Conclusion
In conclusion, combined exercise, whether of moderate or high intensity, has been found to be the most effective intervention for promoting weight loss and improving BMI, fat mass, and body fat percentage in overweight and obese children and adolescents. Additionally, high-intensity resistance exercise has been found to be the best intervention for increasing fat-free mass. These findings are significant for developing effective exercise interventions for managing obesity in children and adolescents. However, while this study provides important insights into the effects of different types of exercise intensity on changes in body composition in overweight and obese children and adolescents, there are several limitations that need to be addressed in future research to improve the comprehensiveness and reliability of the evidence.

Data availability statement
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.