Overweight and obesity are major global public health challenges and are strongly associated with hypertension, dyslipidemia, type 2 diabetes, premature mortality, and increased healthcare costs (Chooi et al., 2019; Lobstein et al., 2022; Wong et al., 2020). Although genetic susceptibility contributes to obesity risk, lifestyle-related factors, particularly chronic positive energy balance, are regarded as the principal drivers of its development (Berenson and group, 2012). Effective long-term management therefore requires sustainable strategies, especially regular physical activity (Hall et al., 2022; Khera et al., 2016; Petridou et al., 2019; Verboven and Hansen, 2021).

Among exercise-based interventions, recreational football has attracted increasing attention as a potentially effective and engaging modality. In this review, recreational football refers broadly to football-based activities performed in structured or semi-structured formats for health-related purposes rather than competitive performance, including small-sided games and organized sessions in school or community settings. Unlike conventional exercise programs, recreational football combines intermittent high-intensity activity, multidirectional movement, and social interaction, features that may support sustained participation while also providing meaningful physiological stimuli (Andersen et al., 2014a; Bangsbo et al., 2014; 2015; Krustrup et al., 2016; 2018; Milanović et al., 2019). Previous studies have shown that football participation can improve physical fitness, body composition, and selected cardiovascular and metabolic outcomes across different age groups (Andersen et al., 2014a; Bangsbo et al., 2014; 2015; Hernandez-Martin et al., 2021; Krustrup et al., 2016; 2018; Milanović et al., 2019; Seabra et al., 2014). In children and adolescents, football programs have been associated with increased physical activity, improved bone-related outcomes, reduced adiposity, and enhanced cardiorespiratory fitness (Hernandez-Martin et al., 2021; Krustrup et al., 2016; Seabra et al., 2014). In adults, participation has been linked to favorable changes in fat mass, cardiovascular function, and metabolic health indicators (Andersen et al., 2014a; Bangsbo et al., 2014; 2015; Krustrup et al., 2018; Milanović et al., 2019).

However, the effects of recreational football specifically in individuals with overweight or obesity remain unclear. Existing studies vary considerably in participant characteristics, intervention duration, training frequency, comparator conditions, and reported outcomes, complicating interpretation of the overall evidence and limiting its clinical applicability (Clemente et al., 2022; Hernandez-Martin et al., 2021; Milanović et al., 2019). In particular, age, sex, training dose, and comparator type may influence both the magnitude and consistency of the observed effects. Although children, adolescents, and adults differ in growth, maturation, and baseline metabolic characteristics, they share clinically relevant obesity-related outcome domains, particularly body composition and cardiometabolic health. A pooled evaluation may therefore help estimate the overall direction of effect across overweight/obese populations while identifying important sources of heterogeneity. Moreover, although football has been recognized as a health-promoting activity in previous research, evidence addressing both body composition and cardiometabolic outcomes in overweight or obese populations remains limited and fragmented. A focused quantitative synthesis is therefore needed to clarify the potential health effects of recreational football in this population.

Accordingly, this systematic review and meta-analysis aimed to evaluate whether recreational football, compared with non-football control conditions, improves body composition and cardiometabolic outcomes in individuals with overweight or obesity. BMI, BFP, WC, and LBM were defined as primary outcomes, whereas TG, TC, LDL-C, HDL-C, SBP, and DBP were examined as secondary outcomes.

This systematic review and meta-analysis was conducted in accordance with the PRISMA 2020 guidelines and was prospectively registered in PROSPERO on 17 May 2025 (registration ID: CRD420251054341).

We systematically searched Web of Science, PubMed, Embase, Scopus, and SPORTDiscus on 23 December 2025 for studies published up to December 2025. The search strategy combined keywords and controlled vocabulary related to recreational football, obesity, body composition, metabolic health, and randomized controlled trials. Detailed database-specific search strategies, including Boolean operators and search fields, are provided in Supplementary Table S1.

Studies were included if they met the following criteria: (1) randomized controlled trials (RCTs); (2) participants were classified as overweight or obese according to BMI-based criteria reported in the included studies (e.g., WHO definitions or age-specific BMI percentiles); (3) reported at least one outcome related to body composition (e.g., BMI, BFP, WC, and LBM) or metabolic indicators (e.g., blood lipids and blood pressure); (4) used recreational football as the sole intervention for at least 3 months; (5) were published in English; (6) included a control group receiving no exercise intervention, placebo intervention, or usual physical activity; and (7) were available in full text.

Studies were excluded if they met any of the following criteria: (1) nonrandomized studies (e.g., retrospective studies, case reports, or observational studies); (2) combined interventions involving other forms of exercise or dietary regulation, making it difficult to isolate the effects of recreational football; (3) intervention duration of less than 3 months; (4) participants who were not overweight or obese; (5) incomplete outcome data or absence of necessary statistical information; and (6) non-English publications or studies published as conference abstracts, reviews, or other non-original articles.

Relevant data were extracted into a predesigned electronic data extraction sheet. The following specific data items were collected: (1) general study information (first author, publication year, journal, country, and study design); (2) participant characteristics (sample size per group, age, sex distribution, and baseline BMI classification); (3) intervention details (modality, duration, frequency, session length, and exercise intensity); (4) outcome measures, including body composition outcomes (BMI, body fat percentage [BFP], waist circumference [WC], and lean body mass [LBM]) and cardiometabolic outcomes (triglycerides [TG], total cholesterol [TC], low-density lipoprotein cholesterol [LDL-C], high-density lipoprotein cholesterol [HDL-C], systolic blood pressure [SBP], and diastolic blood pressure [DBP]), along with the assessment tools used for each outcome; and (5) additional contextual and clinical variables where available (training adherence/exposure, medication use, dietary control, pubertal status, and baseline physical activity).

When relevant data were missing, we attempted to contact the corresponding authors. If no response was received, numerical data were extracted from graphs using GetData Graph Digitizer and checked for accuracy by a second reviewer. When duplicate reports or overlapping cohorts were suspected, study characteristics and sample information were compared, and only the most complete dataset was retained for analysis. All retrieved records were imported into EndNote 20 for deduplication. Two reviewers independently screened titles and abstracts for initial eligibility. The full texts of potentially eligible studies were then assessed in detail. Any disagreements were resolved through discussion or, when necessary, by consultation with a third reviewer.

Risk of bias was independently assessed by two reviewers using the Cochrane Risk of Bias 2 (RoB 2) tool. Because RoB 2 is a result-level tool, judgments were assessed with reference to the outcomes included in the meta-analysis. This tool evaluates five domains: (1) bias arising from the randomization process, (2) bias due to deviations from intended interventions, (3) bias due to missing outcome data, (4) bias in measurement of the outcome, and (5) bias in selection of the reported result.

Each domain was judged as “low risk,” “some concerns,” or “high risk” according to RoB 2 guidance. An overall risk-of-bias judgment was assigned for each study–outcome result. Any disagreements were resolved through discussion, with consultation from a third reviewer when required.

All statistical analyses were performed using Review Manager (RevMan, version 5.4). When outcomes were reported using comparable units and measurement conventions across studies, pooled effects were expressed as mean differences (MDs) with 95% confidence intervals (CIs). When outcome measures differed in scale or measurement convention, standardized mean differences (SMDs) were used. In the present review, MDs were applied to anthropometric and blood pressure outcomes, whereas SMDs were used for lipid-related outcomes, including TG, TC, LDL-C, and HDL-C.

Pooled effect estimates were calculated using post-intervention values to compare outcomes between the intervention and control groups. When required, standard deviations were derived from reported data according to standard meta-analytic procedures.

Statistical heterogeneity was assessed using the I² statistic, with values of 25%, 50%, and 75% representing low, moderate, and high heterogeneity, respectively. A fixed-effect model was used when heterogeneity was low (I² ≤ 50%), whereas a random-effects model was applied when substantial heterogeneity was present (I² > 50%).

Subgroup analyses were conducted to explore potential sources of heterogeneity according to participant age and training frequency. The comparator type was also considered a potential source of heterogeneity. However, because the number of studies within individual comparator categories was limited and unevenly distributed across outcomes, a separate subgroup analysis by comparator type was not performed. Age was categorized as <20 years and ≥20 years, and training frequency as <3 sessions/week and ≥3 sessions/week. These subgroup categories were selected on the basis of the distribution of the available studies and their potential clinical relevance. Random-effects models were used for subgroup analyses, and formal tests for subgroup differences were performed in RevMan. Sensitivity analyses were conducted by sequentially removing individual studies to assess the robustness of the pooled results.

Publication bias was assessed visually using funnel plots. When at least 10 studies were available for a given outcome, Egger’s regression test was additionally performed using R (version 4.5.0).

The certainty of evidence for each outcome was assessed using the GRADE approach across five domains: risk of bias, inconsistency, indirectness, imprecision, and publication bias. Because all included studies were randomized controlled trials, certainty was initially rated as high and then downgraded according to predefined criteria. Risk of bias was judged based on the RoB 2 judgments of the studies contributing data to each outcome; one level was downgraded when a substantial proportion of evidence came from studies with some concerns or high risk of bias, and two levels were downgraded when most evidence came from high-risk studies. Inconsistency was downgraded by one level when I² was ≥50% and by two levels when I² was ≥75% without a plausible explanation. Indirectness was downgraded when important population, intervention, comparator, or outcome mismatches were present. Imprecision was downgraded when the total sample size was <400 participants or when the 95% CI crossed or approached the line of no effect. Publication bias was downgraded when funnel plot asymmetry was observed or when Egger’s regression test indicated potential small-study effects (P < 0.10), when applicable. Certainty was rated as high, moderate, low, or very low accordingly. A summary table of the certainty of evidence for each outcome is presented in the main text.

A total of 909 records were initially identified through database searches (Web of Science, PubMed, Embase, Scopus, and SPORTDiscus). After title screening, 492 records were excluded because of irrelevance or duplication, leaving 417 records for abstract screening. Subsequently, 366 records were excluded after abstract review for reasons such as non-randomized designs, interventions not involving football, professional football interventions, or insufficient data. The remaining 51 full-text articles were assessed for eligibility. After full-text evaluation, 35 articles were excluded because of intervention duration of less than 3 months, non-randomized designs, multiple uncontrolled intervention components, or irrelevant outcomes. Finally, 16 studies (Andersen et al., 2014b; 2016; Cvetković et al., 2018; Faude et al., 2010; Hansen et al., 2013; Knoepfli-Lenzin et al., 2010; Krustrup et al., 2009; Lousa et al., 2017; Randers et al., 2010; Seabra et al., 2014; 2016; Soares et al., 2023; Vasconcellos et al., 2016; 2021; Wang et al., 2023; Weintraub et al., 2008) were included in this systematic review and meta-analysis (Figure 1).

A total of 16 randomized controlled trials (RCTs) (Andersen et al., 2014b; 2016; Cvetković et al., 2018; Faude et al., 2010; Hansen et al., 2013; Knoepfli-Lenzin et al., 2010; Krustrup et al., 2009; Lousa et al., 2017; Randers et al., 2010; Seabra et al., 2014; 2016; Soares et al., 2023; Vasconcellos et al., 2016; 2021; Wang et al., 2023; Weintraub et al., 2008) published between 2008 and 2023 were included, comprising 387 participants (200 in football intervention groups and 187 in control groups). Participants ranged in age from 8 to 68.1 years. Most studies (n = 11) focused on children and adolescents aged 8-17 years, whereas the remaining studies included adults with mean ages ranging from 20 to 68.1 years.

All studies targeted overweight or obese populations, with inclusion criteria generally based on BMI at or above the 85th or 97th percentile for age and sex, BMI > 2 standard deviations above WHO reference values, or BMI ≥ 20.5 kg/m² for prepubertal children. Intervention duration ranged from 3 to 6 months. In addition to study and intervention characteristics, several studies reported contextual factors relevant to interpretation of the findings, including training adherence/exposure, medication-related eligibility or restrictions, dietary control, pubertal status, and baseline physical activity. The most commonly reported outcomes were BMI (Andersen et al., 2014b; 2016; Cvetković et al., 2018; Faude et al., 2010; Hansen et al., 2013; Krustrup et al., 2009; Lousa et al., 2017; Seabra et al., 2014; Soares et al., 2023; Vasconcellos et al., 2016; 2021; Wang et al., 2023; Weintraub et al., 2008) (n = 13), BFP (Andersen et al., 2014b; 2016; Cvetković et al., 2018; Knoepfli-Lenzin et al., 2010; Krustrup et al., 2009; Lousa et al., 2017; Randers et al., 2010; Seabra et al., 2014; 2016; Soares et al., 2023; Vasconcellos et al., 2016; 2021; Wang et al., 2023) (n = 13), and LBM (Andersen et al., 21014b; 2016; Cvetković et al., 2018; Knoepfli-Lenzin et al., 2010; Krustrup et al., 2009; Seabra et al., 2014; 2016; Soares et al., 2023; Vasconcellos et al., 2016; Wang et al., 2023) (n = 10). WC (Knoepfli-Lenzin et al., 2010; Lousa et al., 2017; Seabra et al., 2016; Soares et al., 2023; Vasconcellos et al., 2021) (n = 5), TG (Andersen et al., 2016; Randers et al., 2010; Seabra et al., 2016; Vasconcellos et al., 2016; 2021) (n = 5), TC (Andersen et al., 2016; Knoepfli-Lenzin et al., 2010; Krustrup et al., 2009; Randers et al., 2010; Seabra et al., 2016; Vasconcellos et al., 2016) (n = 6), LDL-C (Andersen et al., 2016; Knoepfli-Lenzin et al., 2010; Krustrup et al., 2009; Randers et al., 2010; Seabra et al., 2016; Vasconcellos et al., 2016) (n = 6), HDL-C (Andersen et al., 2016; Knoepfli-Lenzin et al., 2010; Krustrup et al., 2009; Randers et al., 2010; Seabra et al., 2016; Vasconcellos et al., 2016; 2021) (n = 7), SBP (Andersen et al., 2016; Cvetković et al., 2018; Hansen et al., 2013; Krustrup et al., 2009; Seabra et al., 2016; Vasconcellos et al., 2016) (n = 6), and DBP (Cvetković et al., 2018; Hansen et al., 2013; Krustrup et al., 2009; Randers et al., 2010; Seabra et al., 2016; Vasconcellos et al., 2016; 2021) (n = 7) were also reported across multiple studies.

Anthropometric measurements were obtained primarily using stadiometers and dual-energy X-ray absorptiometry (DXA), including models such as the Hologic QDR 4500A and Lunar GE systems. Metabolic parameters were generally assessed using enzymatic assay kits and automated analyzers. Blood pressure was measured using standard clinical devices. Several studies did not specify the instruments used (Table 1).

All 16 included studies used an RCT design, with the intervention group receiving recreational football training. Most control groups involved no-exercise or usual-activity comparators, whereas others involved sedentary behavior, health education, or other non-football comparison conditions.

Intervention duration ranged from 3 to 6 months. Three-month interventions were the most common (n = 8), followed by six-month interventions (n = 6), and one study used a four-month protocol. Training frequency varied from 2 to 4 sessions/week, with 3 sessions/week (n = 9) and 4 sessions/week (n = 4) being the most common. Several studies allowed flexible training frequencies (e.g., 2-3 sessions/week). Session duration was generally about 60 min (n = 12), with a few extending to 60-90 min (n = 2). Exercise intensity was quantified mainly by average heart rate or percentage of maximal heart rate (HRmax), ranging from 50% to 83% of HRmax. Most studies (n = 8) reported intensities between 70% and 83% HRmax. However, some studies did not report a specific intensity level.

There was also heterogeneity in control-group conditions. “Daily activity” was not consistently defined across studies, and comparators such as health education or sedentary lifestyle may have represented different counterfactual conditions. In addition, several studies did not provide detailed descriptions of session duration or training intensity (Table 2).

Risk of bias was assessed using the Cochrane RoB 2 tool. Overall, most study–outcome judgments were rated as low risk or some concerns, whereas high-risk judgments were uncommon.

Some concerns were primarily related to the randomization process and selection of the reported result, mainly because of insufficient reporting of allocation procedures or the absence of clearly pre-specified analysis plans. In contrast, most study–outcome judgments were rated as low risk in domains related to missing outcome data and outcome measurement, particularly when objective outcome measures were used. Detailed results of the risk-of-bias assessment are shown in Figure 2 and Figure 3. The detailed RoB 2 assessment by study, outcome, and domain is provided in Supplementary Table S2.

Effects on body composition

Compared with control conditions, recreational football was associated with significant improvements in several body-composition indicators.

Pooled analyses showed significant reductions in BMI (MD = -2.58 kg/m², 95% CI: -3.35 to -1.81, P < 0.00001; I² = 10%, 13 studies (Andersen et al., 2014b; 2016; Cvetković et al., 2018; Faude et al., 2010; Hansen et al., 2013; Krustrup et al., 2009; Lousa et al., 2017; Seabra et al., 2014; Soares et al., 2023; Vasconcellos et al., 2016; 2021; Wang et al., 2023; Weintraub et al., 2008), BFP (MD = -2.95%, 95% CI: -4.45 to -1.46, P = 0.0001; I² = 10%, 13 studies (Andersen et al., 2014b; 2016; Cvetković et al., 2018; Knoepfli-Lenzin et al., 2010; Krustrup et al., 2009; Lousa et al., 2017; Randers et al., 2010; Seabra et al., 2014, 2016; Soares et al., 2023; Vasconcellos et al., 2016; 2021; Wang et al., 2023), and WC (MD = -8.15 cm, 95% CI: -11.60 to -4.71, P < 0.00001; I² = 17%, 5 studies (Knoepfli-Lenzin et al., 2010; Lousa et al., 2017; Seabra et al., 2016; Soares et al., 2023; Vasconcellos et al., 2021). Heterogeneity for these outcomes was low (Figure 4).

For LBM, the primary between-group analysis favored the control group (MD = -3.00 kg, 95% CI: -5.87 to -0.14, P = 0.04), with substantial heterogeneity (I² = 74%, 10 studies (Andersen et al., 2014b; 2016; Cvetković et al., 2018; Knoepfli-Lenzin et al., 2010; Krustrup et al., 2009; Seabra et al., 2014; 2016; Soares et al., 2023; Vasconcellos et al., 2016; Wang et al., 2023). In addition to the primary comparison, a supplementary within-group pre-post meta-analysis indicated a significant increase in LBM following football intervention (MD = +1.83 kg, 95% CI: 0.71 to 2.96, P = 0.001) (Figure 5). This supplementary finding should be interpreted cautiously because the primary between-group comparison favored the control group and showed considerable heterogeneity.

Recreational football was associated with significant reductions in both SBP and DBP. For SBP, a random-effects model showed a significant reduction (MD = -6.62 mmHg, 95% CI: -10.16 to -3.09, P = 0.0002; I² = 55%, 6 studies (Andersen et al., 2016; Cvetković et al., 2018; Hansen et al., 2013; Krustrup et al., 2009; Seabra et al., 2016; Vasconcellos et al., 2016). For DBP, a fixed-effect model also showed a significant reduction (MD = -2.60 mmHg, 95% CI: -4.56 to -0.65, P = 0.009; I² = 0%, 7 studies (Cvetković et al., 2018; Hansen et al., 2013; Krustrup et al., 2009; Randers et al., 2010; Seabra et al., 2016; Vasconcellos et al., 2016; 2021) (Figure 6).

TG levels were also significantly reduced in the football group (SMD = -0.57, 95% CI: -0.93 to -0.20, P = 0.002; I² = 30%, 5 studies (Andersen et al., 2016; Randers et al., 2010; Seabra et al., 2016; Vasconcellos et al., 2016; 2021).

By contrast, recreational football did not significantly affect other lipid parameters, including TC (SMD = -0.31, 95% CI: -0.62 to 0.00, P = 0.05; I² = 29%, 6 studies (Andersen et al., 2016; Knoepfli-Lenzin et al., 2010; Krustrup et al., 2009; Randers et al., 2010; Seabra et al., 2016; Vasconcellos et al., 2016), LDL-C (SMD = -0.28, 95% CI: -0.59 to 0.03, P = 0.07; I² = 20%, 6 studies (Andersen et al., 2016; Knoepfli-Lenzin et al., 2010; Krustrup et al., 2009; Randers et al., 2010; Seabra et al., 2016; Vasconcellos et al., 2016), or HDL-C (SMD = 0.30, 95% CI: 0.00 to 0.60, P = 0.05; I² = 44%, 7 studies (Andersen et al., 2016; Knoepfli-Lenzin et al., 2010; Krustrup et al., 2009; Randers et al., 2010; Seabra et al., 2016; Vasconcellos et al., 2016; 2021) compared with control conditions. Heterogeneity across these outcomes was low to moderate.

Subgroup analyses of LBM showed that the pooled effect was not significant in participants aged <20 years (MD = -2.31 kg, 95% CI: -6.53 to 1.90; P = 0.28; I² = 79%), but was significant in those aged ≥20 years (MD = -4.28 kg, 95% CI: -6.52 to -2.05; P = 0.0002; I² = 0%). However, the test for subgroup differences was not significant (Chi² = 0.66, df = 1, P = 0.42), indicating no statistically significant difference between age subgroups. In the training-frequency subgroup analysis, a significant pooled effect was observed in studies with <3 sessions/week (MD = -5.42 kg, 95% CI: -7.83 to -3.01; P < 0.0001; I² = 0%), whereas no significant effect was found in studies with ≥3 sessions/week (MD = -1.19 kg, 95% CI: -4.65 to 2.27; P = 0.50; I² = 71%). The test for subgroup differences approached statistical significance (Chi² = 3.87, df = 1, P = 0.05), suggesting that training frequency may have contributed to the observed variability (Figure 7).

Sensitivity analyses supported the robustness of the primary findings. For LBM, exclusion of Wang et al. (2023) reduced heterogeneity to I² = 37% without changing the direction of the pooled effect. For SBP, exclusion of Seabra et al. (2016) reduced heterogeneity to I² = 0% while the result remained statistically significant. For all other outcomes, sequential exclusion of individual studies did not materially alter the pooled estimates.

Publication bias was assessed visually using funnel plots for all outcomes (Figure 8). Overall, most funnel plots appeared broadly symmetrical, suggesting no clear evidence of substantial publication bias for BMI, BFP, WC, SBP, DBP, TG, TC, LDL-C, and HDL-C. However, the funnel plot for LBM showed some asymmetry.

For outcomes with at least 10 studies, Egger’s regression test was additionally performed. The results indicated no significant small-study effects for BMI (t = 0.65, df = 12, P = 0.5253) or BFP (t = 1.66, df = 11, P = 0.1260). In contrast, Egger’s test for LBM was statistically significant (t = -2.61, df = 8, P = 0.0314), suggesting potential small-study effects or publication bias for this outcome. This finding should nevertheless be interpreted cautiously because the number of included studies remained limited (Figure 8).

The certainty of evidence for the main outcomes was assessed using the GRADE approach. Overall, the certainty of evidence ranged from moderate to low across outcomes (Table 3). For body-composition outcomes, certainty was rated as moderate for BMI, BFP, and WC, mainly because of concerns regarding risk of bias in several included trials. Certainty for LBM was rated as low because of risk-of-bias concerns, substantial heterogeneity (I² = 74%), and potential publication bias.

For cardiometabolic outcomes, certainty was moderate for DBP but low for TG and SBP. Certainty for TG was downgraded mainly because of risk-of-bias concerns and imprecision, whereas certainty for SBP was down-graded because of risk-of-bias concerns and inconsistency. Similarly, certainty for lipid outcomes (TC, LDL-C, and HDL-C) was considered low because of risk-of-bias concerns and imprecision.

This systematic review and meta-analysis indicates that recreational football can improve several body-composition outcomes in individuals with overweight or obesity, particularly BMI, body fat percentage, and waist circumference. These findings suggest that football-based exercise may be an effective strategy for reducing overall and central adiposity in this population. Given the close association between excess adiposity, especially abdominal fat accumulation, and cardiometabolic risk, the observed reductions in BMI, BFP, and WC may be clinically relevant.

Several features of recreational football may help explain these favorable changes. Football training typically involves repeated bouts of running, acceleration, deceleration, changes of direction, and intermittent high-intensity activity performed over a sustained session (Krustrup et al., 2010). These activity patterns may increase total energy expenditure and provide a stimulus conducive to reductions in adiposity. In addition, the game-based and socially interactive nature of football may promote continued participation and enhance the cumulative effect of training over time (Krustrup et al., 2016; Krustrup et al., 2018; Milanović et al., 2019).

By contrast, the effect on LBM remains uncertain. In the primary between-group meta-analysis, recreational football did not show a clear benefit for LBM, and the pooled estimate was accompanied by substantial heterogeneity. Subgroup analyses suggested that the effect may have differed by age and training frequency, with somewhat more favorable patterns in studies involving adults and in those using lower training frequencies; however, these subgroup findings were not robust and did not fully explain the observed heterogeneity. Additional variation in intervention design, measurement methods, baseline characteristics, developmental stage, and comparator conditions may also have contributed to the inconsistent results. Previous reviews have likewise suggested that football-based interventions may produce more consistent effects on adiposity-related outcomes than on lean mass (Wang et al., 2024). Because intensity data were incompletely reported and definitions of overweight and obesity varied across studies, further subgroup exploration was limited. Overall, the current evidence does not support a firm conclusion regarding the effect of recreational football on LBM.

The present findings suggest that recreational football may improve selected cardiometabolic outcomes in individuals with overweight or obesity, with the clearest effects observed for triglycerides and blood pressure. The reductions in SBP and DBP are particularly relevant because elevated blood pressure is among the most important cardiovascular risk factors associated with excess adiposity. The reduction in triglycerides likewise indicates that the benefits of recreational football may extend beyond body-composition changes alone.

These favorable findings may be related to physiological adaptations commonly associated with intermittent exercise training. Recreational football typically combines repeated high-intensity efforts with periods of lower-intensity recovery, a pattern that may be relevant to both vascular regulation and triglyceride metabolism (Hambrecht et al., 2000; Wang and Eckel, 2009; Wang, 2004). From a clinical perspective, this pattern is important because improvements in blood pressure and triglycerides may contribute meaningfully to cardiometabolic risk reduction in populations with overweight or obesity.

Interpretation of these lipid-related findings is also complicated by variation in intervention duration and training frequency. Previous evidence suggests that relatively large exercise volumes may be required to reduce LDL-C and increase HDL-C (Kraus et al., 2002; LEON and SANCHEZ, 2001). In addition, dietary control and medication-related procedures were not reported consistently across the included trials. Some studies excluded participants using relevant medications or recorded dietary intake, whereas others provided limited or no monitoring of these factors, which may have contributed to uncertainty in the lipid-related findings. Finally, the relatively small number of studies and modest sample sizes for individual lipid outcomes may have limited statistical power.

Taken together, the available evidence indicates a more consistent pattern of benefit for triglycerides and blood pressure than for TC, LDL-C, and HDL-C. Recreational football therefore appears to offer cardiometabolic benefits in individuals with overweight or obesity, although the strength of evidence differs across specific outcomes.

Several limitations of this review should be acknowledged. First, the total sample size was relatively small, and methodological concerns remained in several trials, including incomplete reporting and heterogeneity in intervention protocols and comparator conditions. Although subgroup analyses were conducted where feasible, they did not fully explain the observed variability across outcomes.

Second, the external validity of the review is constrained by the composition of the available evidence base. Most included data were derived from male participants and from children or adolescents, whereas substantially fewer data were available for adult women and broader clinical populations. This is important because developmental stage, baseline metabolic status, and intervention setting may influence both the magnitude and pattern of response. Accordingly, the present findings are most directly applicable to populations resembling those most frequently represented in the included studies.

Third, the long-term sustainability of the observed benefits could not be evaluated because follow-up data were rarely reported. The present review therefore reflects intervention effects over the study periods included rather than the durability of those effects after the intervention ended. Finally, the restriction to English-language publications may have introduced language bias.

Future research should address these limitations through large, well-designed multicenter randomized controlled trials with longer follow-up and more consistent reporting of participant characteristics, intervention dose, comparator conditions, adherence, baseline metabolic status, and relevant co-interventions. Subgroup analyses based on sex, age, developmental stage, baseline BMI category, and training exposure may help identify the populations most likely to benefit from football-based interventions. Further studies are also needed to determine whether the effects of recreational football differ according to baseline risk profile and clinical context.

Recreational football appears to improve several obesity-related health outcomes in individuals with overweight or obesity, with the most consistent benefits observed for adiposity-related measures, triglycerides, and blood pressure. In contrast, the evidence for lean body mass and for conventional lipid markers, including TC, LDL-C, and HDL-C, remains less consistent. Because the current evidence base is derived predominantly from male participants and children/adolescents, these findings should be generalized with caution to women, adults, and broader clinical populations. Overall, recreational football represents a promising exercise modality for obesity-related health improvement, but larger and better-designed trials are needed to clarify the magnitude, consistency, and applicability of its effects.