Effectiveness of Lifestyle Nutrition and Physical Activity Interventions for Childhood Obesity and Associated Comorbidities among Children from Minority Ethnic Groups: A Systematic Review and Meta-Analysis

Lifestyle physical activity (PA) and nutrition are known to be effective interventions in preventing and managing obesity-related comorbidities among adult populations but less so among children and adolescents. We examined the effectiveness of lifestyle interventions in children from minority ethnic populations in Western high-income countries (HICs). Our systematic review included 53 studies, involving 26,045 children from minority ethnic populations who followed lifestyle intervention programmes lasting between 8 weeks and 5 years with the aim of preventing and/or managing childhood obesity and associated comorbidities, including adiposity and cardiometabolic risks. The studies were heterogenous in terms of lifestyle intervention components (nutrition, PA, behavioural counselling) and settings (community vs. schools and after-school settings). Our meta-analysis included 31 eligible studies and showed no significant effects of lifestyle interventions when they focused on body mass index (BMI) outcomes (pooled BMI mean change = −0.09 (95% CI = −0.19, 0.01); p = 0.09). This was irrespective of the intervention programme duration (<6 months vs. ≥6 months), type (PA vs. nutrition/combined intervention) and weight status (overweight or obese vs. normal weight) as all showed nonsignificant effects in the sensitivity analysis. Nonetheless, 19 of the 53 studies reported reductions in BMI, BMI z-score and body fat percentage. However, the majority of lifestyle interventions adopting a quasi-design with combined primary and secondary obesity measures (11 out of 15 studies) were effective in reducing the obesity comorbidities of cardiometabolic risks, including metabolic syndrome, insulin sensitivity and blood pressure, in overweight and obese children. Preventing childhood obesity in high-risk ethnic minority groups is best achieved using combined PA and nutrition intervention approaches, which jointly target preventing obesity and its comorbidities, especially the outcomes of diabetes, hypertension and cardiovascular disease. Therefore, public health stakeholders should integrate cultural and lifestyle factors and contextualise obesity prevention strategies among minority ethnic groups in Western HICs.


Introduction
Preventing the persistent rises in noncommunicable diseases (NCDs), such as cardiovascular disease, diabetes and cancer, is an immediate public health priority [1][2][3][4]. Obesity remains the main modifiable NCD risk factor, which has seen an alarming increase globally and is now associated with reduced life expectancy [5]. Prevalence estimates have shown concurrent increases in obesity, physical inactivity and poor dietary quality and patterns across all age groups [6][7][8][9]. Recent post-COVID-19 reports have estimated that over 380 million children are currently overweight or obese worldwide [10,11]. Alarmingly, this age group is also at risk of imminent rises in childhood obesity-related comorbidi-2 of 36 ties, including hypertension, insulin insensitivity, fatty liver, type-2 diabetes (T2D) and cardiovascular disease (CVD) [12].
There is established evidence in adult populations of the benefits of lifestyle interventions in preventing obesity-related cardiometabolic diseases, such as diabetes [13]. Although the effectiveness of lifestyle interventions in children is not well established [14,15], recent reviews and meta-analyses have concluded that overall, lifestyle interventions combining physical activity and nutritional modifications represent the most promising means for preventing childhood obesity [16,17]. However, these reviews have also highlighted that high-risk population groups, including those from ethnic minorities, who have increased risks of obesity and associated diseases are not targeted effectively with such interventions, especially at the community level [17]. For example, children from ethnic minority groups with low socioeconomic status have higher risks of obesity and healthcare disparity [18,19]. Consequently, the likelihood of obese children from high-risk minority groups developing comorbidities, such as fatty liver, hypertension, T2D and CVD, is increased. It has recently been shown that children from ethnic minority groups who have higher rates of obesity are more likely to be exposed to both NCDs and worse COVID-19 outcomes [10,20]. We previously reported a disparity in the prevalence of childhood obesity and related comorbidities, especially due to ethnicity and social inequality determinants [21]. Yet the available knowledgebase of effective intervention approaches to guide the development of childhood obesity and NCD prevention interventions targeted towards those at greatest risk remains limited [22][23][24].
The effective lifestyle interventions of improving dietary quality, physical activity (PA) levels and sedentary behaviour are known to ameliorate obesity and associated NCDs in children [25][26][27]. However, most studies have been conducted in predominantly white populations [27,28]. It has often been assumed that lifestyle interventions found to be effective in the general population are likely to be effective among ethnic minority populations, if appropriately adapted [28]. However, the discrepant effectiveness of behavioural interventions among different population groups has been reported [29,30]. Furthermore, there is currently limited evidence to prove or disprove the effectiveness of adapted behavioural interventions in preventing childhood obesity among minority ethnic groups [31]. Therefore, it has been suggested that targeted interventions are likely to be more effective than universal approaches in this circumstance because of the unique barriers and inequities faced by minority ethnic groups [32]. Moreover, risk stratification and targeted interventions have been shown to play a role when high-risk groups face unique barriers [33].
Therefore, it is important to identify lifestyle intervention approaches that are likely to be effective among children from minority ethnic groups in Western high-income countries (HICs), given the reported disparity in prevalence between HICs and low-and middleincome countries (LMICs) [21]. There are currently no reviews or analyses on whether or how lifestyle interventions are effective in preventing childhood obesity-related comorbidities among high-risk minority ethnic groups. Therefore, this systematic review and meta-analysis aimed to assess the effectiveness of lifestyle interventions among minority ethnic groups living in Western HICs and describe the salient features of effective interventions for preventing childhood obesity.
For the purpose of this review, the term minority ethnic groups is used to describe people of non-White descent living in Western HICs, in accordance with the common terminology applied in the UK [34].

Materials and Methods
The protocol for this review was registered with the International Prospective Register of Systematic Reviews (PROSPERO: CRD42022369557) and followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [35].

Search Strategy
We used the Patient/population Intervention Comparator and Outcome (PICO) framework to develop the search strategy [36] and conducted systematic searches of different electronic databases using a combination of free text and medical subheadings (MeSH) to locate published studies. We searched the MEDLINE, EMBASE, CINAHL, PsycInfo, SCOPUS, SPORTDiscus and Cochrane Controlled Trials Register databases and conducted manual searches using the lists of references from relevant studies. Furthermore, we searched the registers of controlled trials in progress, conference proceedings and the general Internet using Google Scholar. Initially a specific search strategy for MEDLINE (see Supplementary Materials Table S1) was developed with the help of a Teesside University librarian. This was then adapted to the other databases.
The following search terms were used: [(children or adolescents or paediatric or students or school pupils or youth or boys or girls or school age or juvenile or preteens or teens) AND (BME or BAME or Black and minority ethnic group or Black African or Indian or Pakistani or Bangladeshi or Chinese or mixed race or Hispanic)] AND [(physical activity or exercise or sport or cycling or walking or physical education or aerobics or fitness class/regime/programme or dance therapy or intervention for or sedentary lifestyle) OR (diet therapy/diets/dieting or fasting or healthy eating or fruit or vegetable or formula) OR (behaviour therapy or social support or psychotherapy, group or family therapy or counselling or social support or peer support or health education/health promotion or media intervention or community intervention school programme or health policy on food or nutrition)] AND [(obesity or body weight or adiposity or body mass index or waist circumference or neck circumference) OR (type 2 diabetes mellitus or hypertension or high blood pressure or cardiovascular disease or CVD or metabolic syndrome or nonalcoholic fatty liver disease or NAFLD or depression or psychological problem or anxiety or self-esteem or sleep apnoea or asthma or respiratory problem or dyslipidaemia or musculoskeletal problem)]. The searches were filtered for randomised controlled trials (RCTs) and quasi-RCT study designs.

Inclusion and Exclusion Criteria for Studies
As the review examined lifestyle interventions for the prevention of overweight/obesity and related NCDs among children from minority ethnic groups living in Western HICs, studies were included if (i) they were quasi-randomised studies or RCTs that compared lifestyle interventions with no intervention or other interventions with the primary aim of preventing or managing obesity and associated NCDs as these are the most appropriate study designs for determining effectiveness [37], (ii) they included 0 to 18 year old children, with minority ethnic groups constituting the majority (at least 60%) of the study participants, (iii) they used lifestyle interventions, such as physical activities, diet and reductions in sedentary activity, to prevent obesity and associated NCDs, (iv) they included the outcomes of interest of adiposity measures and metabolic risk factors of NCDs, (v) they were set in Western HICs and (vi) there were no restrictions on timing or language, provided the studies could be translated into English using Google Translate.
Studies were excluded if (i) they were of other designs (such as cohort studies, casecontrol studies, cross-sectional studies, case series or case reports), (ii) the participants were adults or from the general population without minority ethnic children being the main target group, (iii) they used lifestyle interventions for other outcomes and not obesity or NCD prevention or management or (iv) they were conducted in regions other than Western HICs (see Supplementary Materials).

Study Selection, Quality Assessment and Data Extraction
EndNote reference management software version 9, 2019, was used to upload and remove duplicates and share articles identified from the searches between reviewers.  Two reviewers independently screened all of the title/abstracts of the identified studies  against the inclusion and exclusion criteria (see Supplementary Materials Table S2). One reviewer then retrieved the full papers of the selected articles and two reviewers reviewed them in detail. The reasons for excluding any full text studies were documented. The search results were reported according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) flow diagram (Figure 1) [38].

Selection Process
Two reviewers independently screened all of the title/abstracts of the identified studies against the inclusion and exclusion criteria (see Supplementary Material Table S2). One reviewer then retrieved the full papers of the selected articles and two reviewers reviewed them in detail. The reasons for excluding any full text studies were documented. The search results were reported according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) flow diagram (Figure 1) [38].

Risk of Bias in Individual Studies
In this review, the risk of bias for all selected studies was assessed using the Cochrane risk of bias tool [39]. Unlike most tools used for assessing the quality of studies in the context of systematic reviews, the Cochrane tool is not a scale or a checklist [40]. It is a domain-based evaluation that allows for the critical assessment of different domains in RCTs [39]. Given that it is impossible to quantify bias in a given study, this tool allows for qualitative and quantitative judgments and, therefore, is more realistic than checklist-or scale-based tools [41].
Two reviewers independently determined the risk of bias in the individual studies using the Cochrane risk of bias tool (see Supplementary Materials Table S3). Six domains of the study designs and reporting were assessed: random sequence generation for randomisation, allocation concealment, the blinding of participants and personnel, the blinding of outcome assessment and selective reporting. With this tool, studies were classified

Risk of Bias in Individual Studies
In this review, the risk of bias for all selected studies was assessed using the Cochrane risk of bias tool [39]. Unlike most tools used for assessing the quality of studies in the context of systematic reviews, the Cochrane tool is not a scale or a checklist [40]. It is a domain-based evaluation that allows for the critical assessment of different domains in RCTs [39]. Given that it is impossible to quantify bias in a given study, this tool allows for qualitative and quantitative judgments and, therefore, is more realistic than checklist-or scale-based tools [41].
Two reviewers independently determined the risk of bias in the individual studies using the Cochrane risk of bias tool (see Supplementary Materials Table S3). Six domains of the study designs and reporting were assessed: random sequence generation for randomisation, allocation concealment, the blinding of participants and personnel, the blinding of outcome assessment and selective reporting. With this tool, studies were classified as having a high, unclear (where the domains were not clearly described) or low risk of bias [39]. One reviewer extracted data from the studies included in the review using a data extraction checklist and a second reviewer examined the extracted data (see Supplementary Materials Table S4). Any dispute was resolved through discussion. Data regarding study population, methodology, intervention/comparator details, outcome and main results were extracted and tabulated for analysis.

Data Synthesis
We carried out a narrative synthesis of the results of both RCTs and quasi-experimental studies to compile data and identify common patterns. We compared the results of studies that used direct or indirect lifestyle interventions, such as counselling, to prevent or manage obesity and associated NCDs among children. We also compared the outcome, type of intervention, intervention duration and setting of intervention. Ultimately, the data synthesis teased out intervention approaches that were effective among children from minority ethnic groups and those that were not effective. Both tables and text were used to summarise these findings.

Statistical Analysis
We conducted the meta-analysis using Review Manager (RevMan) 5.4 [42], where appropriate. For example, when complete pre-post measures, such mean BMI/BMI z-score, sample size, standard deviation or standard error, were available for the intervention and control groups, the RCT was included in the meta-analysis. The first step in the metaanalysis was to assess the mean differences (MDs) in the outcomes for both the intervention and control groups by comparing changes in the means as the difference between the post-intervention and baseline measures. To calculate the MDs, the available adjusted or unadjusted means, as reported in the included studies, were used. The corresponding changes in standard deviation (SD) were not directly reported in most studies; therefore, these were estimated using the formula suggested by the Cochrane handbook for the systematic review of interventions [43]. Where standard error (SE) was reported, this was converted into SD using the following formula: SD ( = SE * √ n (where SD = standard deviation, SE = standard error, n = sample size) [43]. The second step involved estimating the pooled effect for outcomes, where at least two RCTs reported the same outcome variables. The gain in the intervention group against the change in the comparator group was reported as the pooled effect, which was estimated with 95% CIs. The study weights were equal to the inverse of the variance of the estimated effect of each study, as suggested by DerSimonian and Laird [44]. The overall effect was interpreted as statistically significant if the 95% CI did not include the null value of 0 (no difference) in its range. Sensitivity analyses were performed to assess whether the correlations of 0.5 or 0.8 affected the interpretation of the pooled effect. Heterogeneity, i.e., variations in the intervention effects observed in the included studies, was quantified using the I 2 statistic. Results are to be interpreted with caution where there was significant heterogeneity (I 2 > 50%).

Study Selection
A total of 5751 unique articles were identified through the search process after removing duplicates. Following the title and abstract screening, we examined 246 full texts for eligibility. Of the full texts examined, 53 articles met our inclusion criteria (Figure 1).

Characteristics of Included Studies
We list the details of each of the included studies in Table 1. All but one of the included studies were conducted in the USA. The study that was not conducted in the USA was carried out in the United Kingdom (UK) [45]. Of the 53 included studies, 44 were RCTs and 9 were quasi-experimental pre-post designs. There were 19 studies conducted in school settings, 12 in community and home settings, 11 in more than one setting (such as school-based activities with homework or parental involvement and clinic and home settings), 9 in healthcare settings (such hospitals and primary healthcare and paediatric clinics) and 3 were web/online-based.
A total of about 26045 participants took part in the studies, with an average sample size per study of 477 and a range of 17 to 4044. About 62% of the included studies had participants who were predominantly Hispanic Americans, 13% of the studies had predominantly African American participants, 9% of the studies had participants of mixed ethnic minority groups and four studies had Asian Americans as their main participants. The age of the participants ranged from 0 (new-born infants to mothers who participated in interventions) [46] to 18-year-old adolescents. Female participants comprised 44% of the total sample and 45% of studies targeted participants who were either overweight or obese (BMI ≥ 85th United States CDC BMI percentile for age and sex).
Most intervention programmes (77%) were a combination of nutrition, physical activity and behavioural interventions. PA alone comprised 13% of interventions, while nutrition alone and general behavioural interventions comprised 5% each. The majority (64%) of the interventions were implemented for 6 or more months, whereas 36% were implemented for less than 6 months. The duration of implementation varied from 8 weeks to 5 years. There were three RCTs [47][48][49] that had more than one intervention group. There were 18 of the 53 included studies that reported theoretical frameworks or models for the interventions. These included social cognitive theory (n = 8), a socioecological approach (n = 3), a transtheoretical model (n = 3), the Chronic Care Model (CCM) (n = 1), behaviour change theory (n = 1), social contextual change (n = 1), self-efficiency (n = 1) and a health belief model (n = 1).
The type of control or comparison group varied across the studies (Table 1). Overall, 52% of the RCTs compared interventions using 'standard intervention'/'usual care', whereas 48% compared interventions with relatively more active comparisons, such as school-readiness programmes, self-esteem programmes, health and safety programmes, general health programmes and self-help programmes.
Most of the included studies (n = 49) targeted obesity prevention (primary and secondary) as the main outcome. There were 18 studies that targeted cardiometabolic NCD risk factors, such as insulin resistance, hyperglycaemia, hyperlipidaemia and high blood pressure, either as primary or secondary outcomes. The most used measures of adiposity were zBMI and BMI. Only 12 studies only used BMI as the measure of adiposity, whereas 9 studies only used zBMI as the measure of adiposity and 6 studies used a combination of BMI and zBMI as the measures of adiposity. Seven studies used BMI in combination with other measures of adiposity, such as percentage body fat, waist circumference (WC) and waist-hip ratio. Twelve studies combined measures of adiposity and cardiometabolic outcomes, whereas four studies only used measures of cardiometabolic risk factors as the primary outcomes.  After-school intervention programme had some effects on BMI, body fat and lipid profile in Black communities, but not statically significant. The interventions were difficult to adhere to in home and community settings because they lacked parental involvement.

Control online programme
There were no significant intervention effects on BMI but there was a significant effect of the group intervention on light physical activity among the parents at 16 weeks (B = 33.017, SE = 13.115, p = 0.012) and a similar trend for adolescents.
The online programme was not effective for BMI but had a useful impact on physical activity. Actual data were not shown on BMI mean differences between intervention and comparators for both children and parents.
Aerobic exercise in a controlled environment reduced hepatic fat, visceral fat and insulin resistance in obese participants. The sample was small and involved selected individuals with severe adiposity; therefore, the results may not be generalisable.   Non-exercising control group.
Changes in insulin sensitivity (×10 −4 min −1 mL −1 , using insulin and glucose sensitivity during OGTT): The intervention group significantly increased insulin sensitivity compared to the comparator group (p < 0.05).
Resistance training alone significantly reduced the metabolic risk factor of insulin sensitivity within 3 months in overweight/obese children. However, its effect on adiposity was not reported. There was no difference between high-intensity and moderate-intensity PA over 6 months in the group of obese African American adolescent girls. However, both groups showed some improvements in adiposity. There was no significant difference in the combined incidence of overweight and obesity between intervention schools (11.7%) and control schools ( The intervention led to significant decreases in mean 2-h glucose level (baseline: 144 mg/dL; 6 months: 132 mg/dL; p = 0.002) and increases in mean insulin sensitivity (baseline: 1.9 (0.2); 6 months: 2.6 (0.3); p = 0.001).
The 1-year education and structured PA intervention was effective in decreasing NCD metabolic risk in a high-risk group. However, there was no information on its effect on overweight/obesity. The intervention was shown to be effective in reducing BMI among boys but not girls. However, the intervention was tested in the age group of 11 to 13 years, a period of growth spurts in girls.    Did not receive any nutrition, cooking or gardening information from investigators.
Intervention group had a 1.2 cm (1.7%) reduction in WC, while the control group had a 0.1 cm (0.1%) increase after the intervention (p < 0.001). Fewer children had metabolic syndrome (n = 1) after the intervention than before (n = 7), while the number of children in the control group with metabolic syndrome remained essentially the same between pre-(n = 3) and post-intervention (n = 4).     There were no significant intervention effects on insulin sensitivity, body composition or most glucose/insulin indices, with the exception of glucose incremental area under the curve (p = 0.05), which decreased in the nutrition and nutrition + strength training groups by 18 and 6.3%, respectively, compared to a 32% increase in the control group.
The short duration and small size nutrition education and PA programme had no effects on adiposity or metabolic risk. Participants in the control group were asked to maintain their regular lifestyles and complete the yearly measurements.
No changes in any weight measures were statistically significant. Children in all groups increased their overall mean BMI z-score over the course of the study.
Despite the long duration, the nutrition education and support programme in family, school and combined family and school settings was not effective in reducing BMI. However, the effect of the intervention on other metabolic parameters was not assessed.  RCT; participants were identified from primary care clinics, radio advertising and local churches.
72 mostly Hispanic parent-child dyads; Children aged 8 to 11 years, with BMI ≥85% for age and sex, according to CDC growth chart; 54% females. Families in the control group received standard of care counselling from physicians trained using American Academy of Pediatrics guidelines, addressing both nutrition and activity.
Participants that had a higher baseline BMI were more likely to decrease their absolute BMI (Kg/m 2 ) (β= −0.22, p< 0.0001).
The counselling and education were effective for obese children but less so for normal weight or overweight children. Barkin et al., 2012, USA [95].
RCT, conducted in a community setting.
A programme: (i) Weekly 90-min skills-building sessions for parents and preschool-aged children designed to improve family nutritional habits, increase weekly PA and reduce sedentary activity.
A brief school readiness programme was conducted as an alternative to the active intervention because there was no standard care condition for comparison.
The effect of the treatment on post-intervention absolute BMI (Kg/m 2 ) was B = -0.59 (p = 0.001).
This skills-building intervention programme targeting both parents and children with obesity had a small but significant effect on BMI.
RCT, conducted in a school setting.
189 Hispanic adolescent students in grades 6 through 12, with BMI ≥85% for age and sex, according to CDC growth chart; 47% females.

6-month programme:
Trained peer-led discussion on the selected topic with their group of middle-school students during PE classes, e.g., what they were going to eat for lunch that day or their favourite vegetables.
Usual PE classes.
Significant differences were found between groups across time (F = 4.58, p = 0.01). After the 6-month intervention, the intervention group had a larger decrease in zBMI (F = 6.94, p = 0.01) than the control.
Adding peer-led nutritional education to PE classes reduced adiposity in high-risk Hispanic children. Control was physical education (PE) class as traditionally taught in the district.
The addition of the PA component to PE classes reduced overweight/obesity after 12 months; however, it was uncertain whether this was sustainable.
RCT, conducted in primary schools.
1392 non-White multi-ethnic children; Aged 5 to 6 years in year 1 of primary school; 51% females.
A 12-month programme: (i) Encouraged healthy eating and PA; (ii) Additional 30 min of school time every day for PA opportunities; (iii) A 6-week interactive skills-based programme in conjunction with a football club; (iv) Signposting of families to local PA places; (v) School-led family workshops on healthy cooking skills.
Ongoing health-related activities for pupils in year 2 study. In addition, citizenship education resources, excluding topics related to healthy eating and physical activity, were provided.
No significant effects of intervention on adiposity were observed in either the short or longer term. Although there were improvements in BMI, the differences were smaller the longer the duration of intervention.

Risk of Bias within Studies
All 53 studies were assessed for quality using the Cochrane risk of bias tool (random sequence generation for randomisation, allocation concealment, the blinding of participants and personnel, the blinding of outcome assessment and selective reporting) (see Supplementary Table S3, which contains full details of all studies). In total, 14 RCTs were deemed to have been conducted in a relatively unbiased manner, based on the Cochrane tool, and 15 RCTs were considered moderately biased, mostly because of a lack of a description of the blinding and allocation concealment processes.

Effectiveness of Interventions
There were 44 RCT/controlled studies and 9 quasi-experimental pre-and post-intervention studies (Table 1).

Quasi-Experimental Pre-and Post-Intervention Studies
The results from the quasi-experimental studies were mostly effective intervention (seven of the nine studies) ( Table 2). Eight studies measured BMI or other adiposity measures as the main outcome. Five studies reported significant effectiveness in reducing or maintaining BMI/zBMI/WC/%BF. Three of the studies that were shown to be effective in improving weight/obesity outcomes were conducted over a duration of more than 6 months and two were conducted for less than 6 months. In four of the studies, the interventions targeted children only, while one study targeted children and their families. Four of the studies were implemented in schools, two in community settings, one in a healthcare setting and one in combined healthcare and community settings. Four studies included measures of cardiometabolic risk factors, such as fasting glucose, blood pressure, insulin resistance and hepatic fats. Three [50,55,98] of the four studies showed significant effects in improving cardiometabolic risks.

Randomised Control Studies
There were 44 RCTs that used PA, nutrition and behavioural interventions for the prevention of obesity and NCDs. In total, 40 of these studies (Table 1) included obesity measures, such as BMI, BMI z-score, BMI percentile, percentage body fat and waist circumference, as primary outcome measures, whereas 4 studies did not include obesity outcome measures. Of the 40 studies that examined obesity outcomes, 14 (35%) studies reported significant improvements in the outcomes, 8 of which were implemented among overweight/obese children (BMI ≥85th percentile for age and sex). Nine of the effective interventions targeted children only, while three targeted children and parents or families and two targeted parents only. In 8 of the 14 effective studies, the participants were more than 50% female and the settings were predominantly multiple settings and schools (6 of 14 studies).
There were 15 RCTs that examined the effects of interventions on cardiometabolic outcomes. Of the 15 effective RCTs, 11 (73%) showed effectiveness in improving cardiometabolic risk outcomes. In nine studies, participants were either overweight or obese and four studies had more than 50% female participants. Nine of the effective RCTs used combined PA, nutrition and behavioural interventions, six of which were implemented for more than 6 months.
Overall, 15 RCTs were deemed suitable for the meta-analysis of the BMI z-score outcome and 16 for the BMI outcome (Figures 2 and 3). The pooled effect of the 15 studies did not show any significant differences between intervention and comparator groups in terms of change in mean BMI z-score (−0.03 (95% CI = −0.06, 0.01)) ( Figure 2). Similarly, the pooled effect of the 16 studies in the meta-analysis for BMI did not show any significant differences in terms of change in mean BMI (kg/m 2 ) (−0.09 (95%CI = −0.19, 0.01)) ( Figure 3). Although not significant, the pooled results were slightly in favour of intervention. However, according to our sensitivity analysis, none of the results changed significantly depending on the duration of implementation (<6 months vs. ≥6 months), the type of intervention (PA vs. nutrition/combined intervention) or weight status (overweight/obese vs. normal weight). The studies included in the meta-analyses for BMI z-score and BMI outcomes did not show significant heterogeneity.
outcomes. Of the 15 effective RCTs, 11 (73%) showed effectiveness in improving cardiometabolic risk outcomes. In nine studies, participants were either overweight or obese and four studies had more than 50% female participants. Nine of the effective RCTs used combined PA, nutrition and behavioural interventions, six of which were implemented for more than 6 months.
Overall, 15 RCTs were deemed suitable for the meta-analysis of the BMI z-score outcome and 16 for the BMI outcome (Figures 2 and 3). The pooled effect of the 15 studies did not show any significant differences between intervention and comparator groups in terms of change in mean BMI z-score (−0.03 (95% CI = −0.06, 0.01)) ( Figure 2). Similarly, the pooled effect of the 16 studies in the meta-analysis for BMI did not show any significant differences in terms of change in mean BMI (kg/m 2 ) (−0.09 (95%CI = −0.19, 0.01)) (Figure 3). Although not significant, the pooled results were slightly in favour of intervention. However, according to our sensitivity analysis, none of the results changed significantly depending on the duration of implementation (<6 months vs. ≥6 months), the type of intervention (PA vs. nutrition/combined intervention) or weight status (overweight/obese vs. normal weight). The studies included in the meta-analyses for BMI z-score and BMI outcomes did not show significant heterogeneity.

Findings
This systematic review and meta-analysis examined data from 53 studies, involving 26045 children from minority ethnic groups in Western HICs who followed lifestyle preventative interventions for obesity and related comorbidities. The main finding of our meta-analysis was that lifestyle interventions were not significantly effective when they focused on BMI outcomes, which are commonly used as primary prevention outcomes. However, of the 53 RCTs included, there were 19 studies that reported reductions in BMI, BMI z-score and percentage body fat following lifestyle interventions in minority ethnic children. Furthermore, our analysis showed that constant adherence to supervised physical activity, diet and lifestyle interventions that adopted a direct approach using a quasiexperimental pre-and post-design were most effective, especially when they combined primary obesity outcomes with secondary comorbidity measures, including metabolic

Findings
This systematic review and meta-analysis examined data from 53 studies, involving 26045 children from minority ethnic groups in Western HICs who followed lifestyle preventative interventions for obesity and related comorbidities. The main finding of our meta-analysis was that lifestyle interventions were not significantly effective when they focused on BMI outcomes, which are commonly used as primary prevention outcomes. However, of the 53 RCTs included, there were 19 studies that reported reductions in BMI, BMI z-score and percentage body fat following lifestyle interventions in minority ethnic children. Furthermore, our analysis showed that constant adherence to supervised physical activity, diet and lifestyle interventions that adopted a direct approach using a quasiexperimental pre-and post-design were most effective, especially when they combined primary obesity outcomes with secondary comorbidity measures, including metabolic syndrome, insulin sensitivity and blood pressure. Therefore, incorporating obesity-related comorbidity NCD outcomes alongside BMI within lifestyle interventions could be essential for preventing the obesity-related complications of T2D, hypertension and CVD in high-risk overweight and obese ethnic minority children.
The interventions that were effective in ameliorating obesity and related comorbidities were mostly RCTs with low attrition, conducted under controlled settings (home, school, under parental supervision) and combined both PA and nutritional components [63,73,81,83,84,99]. For example, there was 7% reduction in BMI in Hispanic children when a 12-week instructorled PA, nutrition and parental guidance programme was followed [100]. On the contrary, when interventions were remote, less supervised or carried out as compensation for a lack of PA, less or sometimes reverse effects were found. For example, BMI was either unchanged or increased with an intervention that provided handouts on recommended eating patterns (Aloha cookbook), farmers' market locations and PA locations/maps [101]. This suggests that when targeting ethnic minorities, it is essential to consider supervised activities, such as instructor-guided PA, guided nutritional education and environmental changes. There is already evidence suggesting that culturally tailored and directly monitored interventions work better in minority ethnic communities [28,102]. Therefore, culturally acceptable nutrition and PA interventions that are embedded within usual supervised activities are feasible and more likely to be effective in preventing overweight and obese children among minority ethnic groups.
There are various potential explanations for the apparently weak evidence for the effectiveness of lifestyle interventions among children from minority ethnic groups living in Western HICs (Table 1). Firstly, most of the interventions were adapted from existing interventions to address cultural and socioeconomic barriers encountered by minority ethnic communities. However, for most minority ethnic groups, traditions from original homelands have a strong influence on physical activity and dietary practices while living in Western HICs [103]. Therefore, culturally tailored approaches are required. Previous reviews that examined diet and physical activity behaviours among adults from minority ethnic groups in HICs have similarly not found any concrete evidence for the significant effectiveness of diet and PA for the prevention of overweight and obesity people among minority ethnic groups [103]. For example, a review of the effects of diet and physical activity interventions on weight, BMI and waist circumference among South Asian migrants, including 29 studies, observed no significant differences in adiposity parameters, except for a significant improvement in weight (mean difference: 1.8 kg; 95% CI: 2.5 to 1.2 kg) [28].
Therefore, the development of effective interventions may require qualitative and quantitative research on the knowledge, attitudes, behaviours, perceptions and differential effects of lifestyle interventions among different ethnicities. Evidence for the effectiveness of such approaches was demonstrated in a study that developed interventions in collaboration with the Pakistani community using a social cognitive theory framework and taking into consideration the values, behaviours and perceptions of the targeted community [104]. Another successful approach, used in Chinese adolescents, was a culturally specific, familybased, interactive programme delivered in clinics (supervised) and online. The intervention entailed the early involvement of key community stakeholders and adolescents in the design and implementation of the intervention [105]. Therefore, simply adapting interventions developed for the general population may not be effective among specific minority ethnic groups, unlike those developed in partnership or that are culturally appropriate and implemented with the involvement of the entire family.
Variations in approaches adopted to supervise lifestyle interventions across different populations provide another possible explanation for the observed lack of effectiveness in children from some minority ethnic groups (Figures 2 and 3). For example, African American children did not adhere well to an unstructured moderate-intensity school-based PA intervention combined with healthy snack education [51], despite the design and setting being effective when applied with other ethnic groups, such as American Indians and Hispanic groups [83,84]. The majority of the studies did not report the codesign of interventions with ethnic minorities groups, which is recommended to reduce disparities [20,23,24], thus limited their influence on intervention approaches. Therefore, there is an opportunity to improve interventions targeting minority ethnic groups by involving them in the design. The intervention that showed positive preventative effects involved a family setting and a combination of practical PA and nutrition interventions with adequate follow-up arrangements. Intervention type, setting, duration, frequency and follow-up varied among the studies included in this review. In terms of the intensity of PA, adolescent groups seem to benefit and adhere to more intense forms of vigorous PA compared to lower intensity forms of PA (Table 2).
Despite the weak evidence for the benefits of lifestyle interventions on adiposity and BMI outcomes shown in our meta-analysis (pooled BMI mean change = −0.09 (95% CI = −0.19, 0.01, p = 0.09) (Figures 2 and 3), lifestyle interventions were shown to be effective in reducing cardiometabolic NCD risk factors, such insulin sensitivity, metabolic syndrome and blood pressure, in overweight and obese children ( Table 2). Over 70% of the studies in this review that examined cardiometabolic risk factors as secondary outcomes showed effectiveness (Table 1). Of the 11 studies that examined both BMI and metabolic risk factors as outcomes, only 5 (45%) showed impacts on overweight/obesity, whereas 9 (81%) showed effectiveness on cardiometabolic risks. Most of the effective interventions had multiple components with PA as the prominent part, whereas those that were not effective were mostly counselling and telephone follow-up interventions [52,72,87]. These data suggest that multicomponent lifestyle interventions may be effective in reducing metabolic risk faster than adiposity in the short term. Similar conclusions about the impacts of multicomponent programmes on lipid profile were made by Ho et al., 2012 [106]. They conducted a meta-analysis on the effects of lifestyle interventions on cardiometabolic outcomes, including 33 studies conducted among overweight and obese children, and found significant improvements in low-density lipoprotein cholesterol (−0.30 mmol/L, 95% CI = −0.45 to −0.15), triglycerides (−0.15 mmol/L, 95% CI = −0.24 to −0.07), fasting insulin (−55.1 pmol/L, 95% CI = −71.2 to −39.1) and blood pressure up to 1 year from baseline. Similarly, a meta-regression by El-Medany et al., 2019, showed significant improvements in SBP, LDL, TG and HDL (p < 0.05) with lifestyle interventions among 4-to 19-year-olds, with minimal changes in BMI SDS [107]. The quasi-experimental intervention designs we presented (Table 2) also showed improved lipid profile, blood pressure and cardiometabolic risk when multicomponent PA and nutrition interventions were adopted in children aged 11-18 years [50,52,55,108]. It was also noted that the most effective of those studies in targeting multiple conditions [52,109] were those with direct PA approaches that were able to tolerate higher intensity levels. For example, in one study [62], intense forms of exercise were well tolerated among obese adolescents, who improved their insulin sensitivity from 0.8 to 2.2 (p < 0.01) following 1 year of instructor-led structured and unstructured PA activities. It is likely that children with obesity-related comorbidities can tolerate and adhere to more intense forms of PA, especially under supervision, which can be readily provided throughout their lived environment by parents, coaches and teachers. Therefore, supervised games, sports and group activities are recommended for this high-risk group, which is in line with recent recommendations for the use of lifestyle interventions for the prevention of multiple long-term conditions [110].
There was paucity of studies that specifically targeted children from minority ethnic groups in other Western HICs except the USA. In this review, only one study was conducted in another Western HIC (the UK) [45]. This indicated a lack of targeted responses to the obesity and NCD scourge among children from minority ethnic groups in most Western HICs [111], even though it is known that obesity and NCD in children and adolescents is most prevalent among minority ethnic groups in HICs [112]. Therefore, considering that ethnic minorities have an increased risk of developing childhood obesity [113,114], it is critical that researchers develop effective interventions that can minimise such disparities.

Implications for Practice and Research
We recommend that more obesity intervention studies targeting minority ethnic populations are conducted. These interventions should (1) be developed and implemented in partnership with minority ethnic communities, (2) be underpinned by appropriate and sound behavioural change techniques and theories and (3) be implemented with the involvement of the entire family. We also recommend that obesity-related comorbidity NCD risks, such as diabetes, hypertension and CVD, should be set as outcomes in addition to BMI and other adiposity measures in lifestyle intervention studies in high-risk children with obesity.
These recommendations would enable reviewers to assess how behavioural change techniques and theories moderate effectiveness, assess the equity impacts of interventions and examine explanations for heterogeneity between interventions. More research into the differential effects of lifestyle interventions for specific ethnic minority groups compared to others is also required.

Limitations
This review captured and analysed an extensive number of intervention studies and included a systematic analysis of many RCTs, on which a meta-analysis was conducted. While the included studies addressed interventions among minority ethnicities within Western HICs, emerging socioeconomic disparities within LMICs also require further research. While we attempted to highlight some of the emerging disparities elsewhere [21,115], further research is still needed on whether and how lifestyle interventions can be effective in reducing obesity-related comorbidities and health inequalities.

Conclusions
Lifestyle interventions are essential for the prevention and management of obesity and associated comorbidities among children. Currently designed interventions focusing on reducing adiposity are not significantly effective in children from minority ethnic groups in Western HICs. Effective interventions in high-risk ethnic minority groups require jointly targeting obesity and its comorbidities as outcomes, especially diabetes, hypertension and cardiovascular disease, as well as actively engaging minority ethnic populations in the design and implementation. Public health policy makers and obesity prevention stakeholders should integrate cultural and lifestyle factors and contextualise obesity prevention strategies among minority ethnic groups in Western HICs.