Introduction

The rising prevalence of childhood obesity has created a worldwide public health crisis.1 According to the World Health Organization (WHO), in 2014, approximately 41 million (6%) young children under the age of five years across the world were overweight or obese.1-3 While global prevalence data available for obesity in older children are currently being verified by WHO,2 the International Obesity Task Force (IOTF) (2000) estimated that approximately 155 million children aged five to 17 years were overweight (10%) or obese (3%).4,5 The Australian Health Survey (2011-12) showed that one in four Australian children were overweight (18%) or obese (7%), placing these children at increased risk of chronic disease from a young age.6 The Australia Burden of Disease Study (2011) indicated that high body mass index (BMI) related to overweight and obesity was the second highest contributor to disease burden.7 In Australia, overweight and obesity accounted for 5.5% of the total disease burden in 2011, including 49% of endocrine disease and 21% of cardiovascular disease.7 Early intervention for weight management is essential for disease prevention, as obesity tracks from childhood to adulthood.

Extensive research has been conducted in child obesity. This has included several systematic reviews (SRs)8-12 in both obesity prevention and treatment in children and adolescents, with evidence suggesting that parental involvement (mainly for primary school aged children) has increased intervention effectiveness in relation to improved weight outcomes and lifestyle behaviors. Systematic reviews of childhood obesity show that family-focused behavioral lifestyle interventions, often with direct parental involvement, can lead to positive outcomes in weight, BMI and other measures of body fat composition of the children.9,13-18 Behavioral interventions are classified as those that aim to change parents' and/or children's weight-related thinking patterns and actions - including dietary intake, physical activity and sedentary behaviors - which go on to determine a family's food and physical environment.19

Parents' attitudes, beliefs and behaviors have an effect on their child's risk of being overweight.20 Parental characteristics such as increased BMI, high alcohol intake, regular smoking, low socioeconomic status and low education level have been linked to greater possibility of their children being overweight.20 Moreover, parents are the key mediator of the obesogenic environment within the family home, particularly for young children who consume most meals at home. Parents usually control decision making about the types of food available in the home and how food is prepared for family meals. Parental decisions can have an impact on the development of child food preferences and eating habits. Family meal times, if they occur, provide a potential opportunity for parents to model healthy food choices and food-related behaviors, while promoting a positive atmosphere around healthy eating for better diet quality. For these reasons, parents are often targeted in interventions for child weight management.

Despite increasing research on obesity, the prevalence of overweight and obesity has risen globally, in both developed and developing countries, over the last decade.1 It remains a challenge for healthcare professionals to work effectively with the complex dynamics of family systems to improve child health outcomes, noting that this can require the active engagement of both parents to achieve effective behavioral change.21-23 There is an abundance of literature on childhood obesity interventions with parental involvement.24-26 However, the effectiveness of interventions to reduce a child's weight and/or change their weight-related lifestyle behaviors has been inconsistent, due in part to the lack of high quality, effective programs27,28 which have included an array of diverse strategies.29 A Cochrane review28 acknowledged that the heterogeneity of current literature in the area of childhood obesity treatment makes it difficult to conclude that one intervention component is more effective than the other. As parental influences are closely associated with the child's weight or weight-related behavior, especially in young children, the parental role in child obesity treatment is likely to be an essential element for effective interventions.24,30 However, there is limited evidence to inform how parents should be involved or targeted in interventions aiming to achieve behavioral changes in their children.9,18,24

Given a number of SRs have already been completed in the area of parental involvement in childhood obesity intervention, a comprehensive review of these SRs is sensible to map and analyze the available evidence. This umbrella review summarized current strategies effective in supporting parents with an overweight child to better manage their child's weight and/or weight-related behavioral change. To the authors' knowledge, this is the first systematic review of SRs on obesity interventions involving parents with overweight children.

Review question

What is the effectiveness of family-based behavioral or lifestyle weight management interventions for children with overweight and obesity? What are the strategies or characteristics of effective interventions in combating child obesity?

Inclusion criteria

Participants

Participants of interest were children aged 18 years and under who were classified as overweight or obese, based on WHO Child Growth Standards, Centers for Disease Control and Prevention (CDC) Growth Charts, or the International Obesity Task Force (IOTF).31-33 Systematic reviews were excluded where study participants included children of all weight status, and/or results were not reported separately for overweight children.

Interventions

The umbrella review included SRs which had a focus on behavioral and/or lifestyle interventions for child weight management. Interventions of interest are those that target weight loss as a primary outcome through changes to behavioral or lifestyle habits, including, but not limited to, dietary intake, physical activity, sedentary behavior, mealtime patterns and sleep. Interventions were included if they were family-based, which was defined as the direct involvement (i.e. attendance or participation in intervention sessions) of first- or second-degree relatives or caregivers cohabiting under one roof in interventions adapted from McLean et al.34 The interventions must have included a comparator group, such as a control group not receiving an intervention (usual care), or a control group receiving an alternative intervention. There were no limitations regarding frequency, duration, intensity and setting of interventions.

Outcomes

Published systematic reviews that reported a synthesis of child weight outcomes were considered for inclusion in this review. Primary outcomes of interest included change in body weight or BMI of the index child, measured from baseline to intervention-end and/or post-intervention follow-up. Where available, "behavior change" such as dietary intake or physical activity were included as secondary outcomes of interest.

Types of studies

Systematic reviews and meta-analyses of quantitative studies (randomized controlled trials (RCTs), quasi-experimental, and pre-post design) were included in the umbrella review. Mixed-method studies (i.e. both quantitative and qualitative) were included if the quantitative component could be extracted clearly. Systematic reviews of solely qualitative studies or studies that did not include an active intervention (e.g. cohort study, case study and cross-sectional study) were excluded as these studies were unlikely to report quantitative results, which were the outcomes of interest. An eligible SR must have a protocol describing the review question/s, search strategy, and inclusion criteria, often referred to as PICO (Participants, Interventions, Comparisons, and Outcomes).35 Therefore, narrative literature reviews were excluded. For SRs that did not explicitly limit inclusion criteria to intervention study designs, only results from relevant intervention trials were extracted for inclusion in the umbrella review. If results were not reported or not separable between intervention and non-intervention studies, the SR was excluded.

Methods

The umbrella review was conducted according to the protocol which was developed based on the methodology for Joanna Briggs Institute (JBI) umbrella reviews36 and published in September 2016 (doi: 10.11124/JBISRIR-2016-003082).37

Search strategy

Database searches were completed in May 2016 by an experienced academic medical librarian. Seven databases were searched, including MEDLINE, Embase, CINAHL, PsycINFO, Scopus, Database of Abstracts of Reviews of Effects and the Cochrane Database of Systematic Reviews, using keywords and index terms (Appendix I) identified by several experienced authors (LKC, TB, CC). Searches were limited to English, and publications between 1990 and May 2016. As there were very few SRs published prior to 1990,36 the search period was deemed appropriate to capture existing SRs on family-based childhood obesity treatment, given SRs only began to emerge from the year 2000.16 Reference lists of included SRs and additional databases including PROSPERO and JBI Database of Systematic Reviews and Implementation Reports were searched to identify any existing SRs on the same topic. The authors believe that it is unlikely that a comprehensive SR in this area of research will have been undertaken and not be published. Therefore, the umbrella review did not search for unpublished/gray literature consistent with the previously published SR protocol,37 as opposed to the JBI Umbrella Review Methodology chapter.36 All references were managed using EndNote X8 (Clarivate Analytics, Philadephia, PA, USA).

Study selection

Two reviewers (LKC and one of either TB, CM, KB, DWS, CC) independently reviewed the titles and/or abstracts of all records retrieved from the search. All potentially relevant full texts were retrieved and assessed independently by two reviewers (LKC and one of either TB, CM, CC). Any discrepancies were resolved through consensus or a third reviewer (TB, CC).

Assessment of methodological quality

All included SRs were critically appraised by two reviewers (LKC, TB) independently using the standard JBI Critical Appraisal Instrument for Systematic Reviews and Research Syntheses.36 Conflicts were resolved through discussions to reach consensus. All eligible SRs (based on PICO inclusion criteria) were included, regardless of methodological quality in order to summarize the current literature and quality of existing studies within SRs to date.

Data extraction

The JBI Data Extraction Form for Review for Systematic Reviews and Research Syntheses was used for extracting information including characteristics of included systematic reviews and weight-related outcomes.36 Relevant information on characteristics of included SRs was extracted and presented in line with the study protocol which has been published previously.37 As per the protocol, primary weight outcomes and weight-related anthropometric indicators, such as BMI, BMI percentile, and BMI z-score (zBMI is defined as the measure of relative weight adjusted for child age and sex) were extracted. In addition, changes in child/parental weight outcomes or weight-related behaviors, such as dietary intake, physical activity and sedentary behavior, were also extracted when they were reported as these were deemed important secondary outcomes in the context of family-based interventions with parental involvement. When results reported within the SRs were not clear (e.g. values reported in narrative synthesis were different from results tables), the original primary studies were sourced to extract the correct data in order to enhance the accuracy of umbrella review synthesis. Adverse consequences that arose as a result of interventions were also documented if reported in SRs. In cases where SRs included more detailed outcomes, such as population groups (e.g. children, adults), intervention contexts (e.g. family-, school-, clinical-based), and intervention components (e.g. behavioral, pharmacological, surgery), only that subset of relevant studies (e.g. children, family-based, behavioral) were extracted for synthesis, provided that the results of the subset of studies were reported separately in the SRs. In cases where an original research study was included in multiple SRs, the number of overlapping studies included in SRs was described in the report - full details of these are presented in Appendix II. For primary studies that were included in multiple reviews, results related to the primary study were cross-checked across multiple reviews for accuracy (when same outcomes were reported) and consolidated for reporting in the current umbrella review (when different outcomes were reported) to avoid duplicates of results.

Data summary

The effectiveness of interventions was extracted as results of meta-analyses conducted within the included SRs, or as reported in the results of included SRs. Quantitative findings were categorized by authors into seven intervention categories and presented in tables describing effect estimates within groups, and between groups, at the end of intervention and at the longest follow-up time.

The seven intervention categories were:

i. Parent-child interventions versus waitlist/no intervention control

ii. Parent-child interventions versus usual care

iii. Parent-only interventions versus waitlist/no intervention control

iv. Parent-only interventions versus usual care

v. Parent-only interventions versus parent-child interventions

vi. Parent-only interventions versus child-only interventions

vii. Parent-child interventions versus child-only interventions

The quality of evidence for each intervention category against weight-related outcomes was assessed using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach.38 The GRADE framework includes evaluation of the following five criteria: i) quality of primary studies (e.g. risk of bias and methodological limitations); ii) inconsistency (e.g. direction of intervention effects, magnitude of statistical heterogeneity measured by I²; low (I² <40%), moderate (I² 40-60%), high (I² >60%); iii) indirectness (e.g. direct comparisons with populations, interventions and outcomes relevant to context); iv) imprecision (e.g. magnitude of the number of included studies: large: >10 studies, moderate: 5-10 studies, small: <5 studies; and median sample size: high >300 participants, intermediate 100-300 participants, low <100 participants); and v) publication bias.38,39

The strengths of overall intervention effectiveness are presented in a table using a "stop-light" indicator, where green indicates an effective or beneficial intervention; amber indicates no intervention effect or no difference when compared to the comparator, or unclear effect due to insufficient information; and red indicates a detrimental or less-effective intervention when compared to the comparator.

Results

Study inclusion

The process of study selection is presented as an adapted PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram (Figure 1). The database searches identified 15,755 records; 697 potentially relevant full texts were retrieved and assessed after excluding 15,058 records following the examination of title and abstract against inclusion criteria. Of the 697 full texts, 14 SRs9-12,15,16,18,40-46 met the inclusion criteria and were included. The majority of the excluded articles were primary studies and/or SRs with irrelevant study designs, such as cohort study, cross-sectional study, or intervention trials without family involvement. A list of excluded studies with reasons is summarized in Appendix III.

Figure 1. PRISMA flowchart of study selection and inclusion process

Methodological quality

Of the 11 quality appraisal criteria listed in the JBI Critical Appraisal Instrument for Systematic Reviews and Research Syntheses, seven criteria (64%) were met by all 14 included SRs (Table 1). The remaining four criteria (5, 6, 7, 9) were not met or rated as unclear due to the lack of reporting in SRs. Six SRs16,40-43,46 did not provide information on whether risk of bias was assessed by more than one reviewer independently, whereas one SR12 was conducted by only one author with no second reviewer. Four SRs12,42,44,46 did not mention risk of bias assessment tools used nor the results of the quality appraisal. One SR15 mentioned that included studies had methodological weaknesses but did not specify the use of an appraisal instrument for formal quality assessment. Three SRs11,15,43 did not provide information about the data extraction tool used or specify the pre-determined study characteristics extracted. Three SRs16,40,42 did not mention whether two or more independent reviewers performed extraction or additional examinations. Only two SRs9,16 reported assessment for the likelihood of publication bias against weight outcomes, which found a low probability of publication bias as indicated by fail-safe N exceeding Rosenthal's recommendation (5k+10; with k = n of included studies).16

Table 1 Critical appraisal results for included systematic reviews

In general, trials included in the SRs were rated as being of low quality with a large proportion rated as unclear or at high risk of bias based on individual risk of bias criteria due to absence of reporting or under-reporting within primary intervention trials. The high risk of bias was for incomplete outcome data due to a higher dropout in parent-only interventions, failure to conduct intent-to-treat analysis, while most studies reported limited information about allocation concealment and randomization procedures. It was uncommon for trials to report power calculations. In an SR of eight trials, only three trials reported sample size calculations and of these trials, two did not meet the target sample size.18 This limited the power and sensitivity to detect significant differences between groups resulting from the interventions. A GRADE assessment of the outcomes pooled in this umbrella review led to trials being downgraded for risk of bias and imprecision owing to the small number of trials and small sample sizes (n = 8 to 80). Therefore, the overall interpretation of the data was synthesized more cautiously. Further details on appraisal instruments used and methodological quality are presented in Table 2.

Table 2 Findings of included systematic reviews and meta-analysis

Table 2 (Continued) Findings of included systematic reviews and meta-analysis

Table 2 (Continued) Findings of included systematic reviews and meta-analysis

Table 2 (Continued) Findings of included systematic reviews and meta-analysis

Table 2 (Continued) Findings of included systematic reviews and meta-analysis

Table 2 (Continued) Findings of included systematic reviews and meta-analysis

Characteristics of included studies

The 14 included SRs9-12,15,16,18,40-46 were published between 2004 and 2015; four included meta-analyses.9,16,41,46 The majority (n = 13 SRs) had searched at least three databases, with the databases most commonly searched being MEDLINE, CINAHL, PsycINFO and PubMed, and with publications retrieved from 1967 to May 2015. While all SRs included intervention studies, eight SRs9,10,12,15,18,41,44,45 specifically included RCTs only, with two SRs15,18 including only RCTs with at least six months follow-up. Two SRs also incorporated specific inclusion criteria for countries including UK11 and USA40 or ethnicity such as African-American girls.40 While the study populations were predominantly primary school age children, the most commonly included child age range was six to 13 years (n = 13 SRs),9-12,15,16,40-46 with some also reporting on children aged under six years (n = 9 SRs)9-11,16,42-46 or over 13 years (n = 10 SRs).11,12,16,18,40-45 One SR18 specifically described results of children aged eight to 11 years only. The included SRs9-12,15,16,18,40-46 had included 47 independent trials that were relevant for the umbrella review. Of the 47 trials, which were conducted in over 16 countries and published between 1975 and 2015, 22 trials (47%) were included in two or more SRs included in the umbrella review. Two trials were included in seven SRs; one trial was included in five SRs; two trials were included in four SRs; three trials were included in three SRs; and 14 trials were included in two SRs. All four trials included in the meta-analysis conducted by Jull et al.41 were also included in the SR by Loveman et al.9 which included 20 trials (see Appendix II). The intervention duration ranged from one month to two years, with the longest post-intervention follow-up time points ranging between three months and seven years. The most common primary outcome measures reported were BMI z-scores (zBMI) (n = 13 SRs),9-12,16,18,40-46 BMI (n = 8 SRs),9,11,12,15,18,40,45,46 percentage overweight (n = 7 SRs),12,16,18,42-44,46 BMI percentile (n = 6 SRs),10,16,18,42-44 and body weight (n = 6 SRs).9,12,15,40,45,46 Several SRs also reported on secondary outcomes related to behavioral changes such as diet (n = 7 SRs),9,12,16,40,42-44 physical activity (n = 5 SRs),9,12,16,40,44 sedentary behavior (n = 2 SRs),16,44 and/or parental outcomes (n = 4 SRs).9,12,43,44

The majority of SRs evaluated family-based studies which targeted parents and children in the interventions and compared with a waitlist or no intervention control group10-12,16,18,40,42-45 and/or usual care.10,12,42,44 Six SRs evaluated parent-only interventions in comparison with a waitlist or no intervention control group9,10,18,43,45 and/or usual care.9 Seven SRs9,12,16,18,41,43,44 examined intervention studies which compared parent-only conditions with parent-child conditions. Six SRs assessed child-only interventions and compared them with parent-only intervention arms15,18,43,44 and/or parent-child intervention arms.15,18,40,43 Two SRs12,42 provided a summary of the effectiveness of parent-child interventions based on different settings, family-, school- and clinic-based interventions, in the treatment of childhood obesity. Overall, interventions had aimed to change behavior of both the index child and their parents and/or family members through targeted intervention components including dietary change, physical activity and behavioral modification or cognitive behavioral therapy;9,11,18 and through intervention techniques, such as nutrition and physical activity education, and goal setting.11,44 Dietary interventions focused on increasing healthy food consumption through the use of traffic light dietary approaches (e.g. the Stoplight Food Guide) or similar strategies.16,41,44 Physical activity interventions aimed to increase physical activity and reduce sedentary behaviors, either through specified targets, or through individualized goal setting.41,44 Detailed characteristics of included SRs are summarized in Appendix IV.

Review findings

All reviews, except one,40 found that family-based lifestyle interventions were effective, as indicated by a decrease in weight or weight-related outcomes (e.g. zBMI, percentage overweight) from baseline. The one review40 which targeted African-American girls only was unable to draw clear conclusions due to most included studies being pilot trials with small sample sizes (n participants<50) and of a short duration (12 weeks or less). Overall, no studies reported adverse events. Detailed findings and interventions included in each SRs are presented in Table 2. Key findings for each pre-defined intervention of interest are described below. Detailed results and GRADE quality of evidence (QOE) are presented in the Summary of Findings 1 through 7 for each intervention of interest respectively.

Parent-child interventions versus waitlist/no intervention control

Eight SRs11,12,18,40,42-45 (eight trials; 581 children) and one meta-analysis16 (three trials; 274 children) provided evidence supporting the effectiveness of parent-child interventions in reducing zBMI compared to waitlist controls after interventions ranged between three and 12 months (moderate QOE). Results from systematic reviews11,12,18,40,42-45 found greater zBMI reduction in the active intervention groups for all but one trial and was consistent with the findings of the meta-analysis16 of three trials. At post-intervention follow-up (10 months to two years; four trials; 288 children; low QOE), the zBMI reduction was maintained.10,11,18,42-44 Therefore, the strength of overall intervention effectiveness was awarded the color "green"; indicating beneficial/positive intervention effects (Table 3). The overall QOE was rated as low to moderate.

Table 3 Summary of evidence

Body mass index percentile was reported in an SR45 (one trial; 105 children; moderate QOE) and a meta-analysis16 (four trials; 230 children; moderate QOE). There was a greater reduction of BMI percentile by -0.5% in intervention groups (three trials), while one trial found no significant difference between groups.16,45 At post-intervention follow-up (three to six months; five trials; 328 children; low QOE), the BMI percentile reduction was maintained.10,16 The outcome overall was beneficial and consistent in the SR and the meta-analysis, hence, resulting in award of the color "green". The overall QOE was rated as low to moderate.

Percentage overweight was reported in an SR45 (one trial; 40 children; moderate QOE) and a meta-analysis16 (three trials; 167 children; moderate QOE). All four trials observed a greater reduction of percentage overweight by -0.3% in intervention groups.16,45 The strength of overall intervention effectiveness was awarded the color "green". The overall QOE was rated as moderate. Waist circumference was reported in four SRs11,40,42,44 (three trials; 324 children; moderate QOE). Two trials found that the intervention group had lower waist circumference at six months, and 12 months, respectively, while another trial found no difference between groups at one month.11,40,42,44 At post-intervention follow-up (12 months; one trial; 116 children; low QOE), the waist circumference remained significantly lower.11,42 The strength of overall intervention effectiveness was awarded the color "green". The overall QOE was rated as low to moderate.

There were no meta-analyses that evaluated dietary changes or physical activity levels as a result of an intervention. Three SRs40,42,44 (four trials; 210 children; low QOE) found that interventions improved diet quality, however meta-analysis was not conducted due to the heterogeneity of the study methods as well as the dietary outcome measures used in reporting results (e.g. energy intake, nutrient intake, food groups servings). Two SRs40,44 (four trials; 253 children; low QOE) found physical activity levels and screen time were not different between groups (three trials) while one trial found that the intervention increased physical activity levels. The overall QOE was rated as low. Detailed results and quality of evidence are presented in the Summary of Findings 1.

Summary of Findings 1

Parent-child interventions versus usual care

Additionally, when compared to a usual care control group (six trials from four SRs; 308 children [two trials did not report sample size]; low QOE), for which interventions were mailed information, a workbook or minimal sessions, the parent-child interventions achieved a greater reduction in the child's BMI, BMI percentile, percentage overweight, and/or weight.10,12,42,44 However, the overall QOE was rated as low quality as each outcome was informed by only one trial with a small sample size (n = 16 to 192). Mixed effects on zBMI were found between intervention and usual care control groups, where one trial indicated that the intervention was effective, while another trial found no difference between the groups.42,44 Detailed results and quality of evidence are presented in the Summary of Findings 2.

Summary of Findings 2

Parent-only interventions versus waitlist/no intervention control

Four SRs (seven trials; 393 children) and one meta-analysis9 (two trials; 153 children) provided evidence supporting the effectiveness of parent-only interventions in improving child weight outcomes. Overall, when compared to a waitlist control group, parent-only interventions reduced zBMI (three trials; 224 children; moderate QOE),9,18,43 BMI (three trials; 55 children; low QOE)9,43 and BMI percentile (one trial; 98 children; low QOE),9,10 while mixed results were reported for parental BMI (two trials; 169 parents; low QOE).9 Apart from zBMI (moderate QOE), evidence for these listed outcomes was rated as low due to small sample sizes, the small number of studies and/or inconsistent results. Detailed results and quality of evidence are presented in Summary of Findings 3.

Meta-analysis9 (two trials; 153 children; moderate QOE) which reported outcomes of parent-only interventions presented results for zBMI only, and indicated that parent-only interventions had significantly lower zBMI by -0.12 following interventions that ranged between three and four months, and the changes remained significant at six to 12 months post intervention. Results from two SRs18,43 (one trial; 71 children; low QOE) supported the meta-analysis of two trials where zBMI reduced by -0.13 in the intervention group after four months intervention, and remained lower by -0.14 than control groups at 10 months. The outcome overall was beneficial and consistent in the SRs and meta-analysis, hence resulting in award of the color "green". The overall QOE was rated as low to moderate. Detailed results and quality of evidence are presented in Summary of Findings 3.

Summary of Findings 3

Parent-only interventions versus usual care

There was a smaller number of SRs that compared parent-only interventions with a usual care control group (seven trials from one SR; 925 children; moderate QOE) for which interventions were mailed information, a workbook or minimal sessions. Only one trial (170 children) reported zBMI and found no significant difference between groups after a three- to six-month intervention (low QOE).9 Five trials from one SR (648 children; moderate QOE) assessed BMI percentile while only one trial (107 children; low QOE) assessed BMI, and all reported a greater reduction in intervention groups.9 Overall, no trial reported negative effects (ineffective) on weight-related outcomes for parent-only interventions. At post-intervention follow-up (six to 24 months), parent-only interventions had greater reduction in BMI (two trials; 614 children; moderate QOE) and BMI percentile (one trial; 60 children; moderate QOE), and no differences in zBMI (one trial; 165 children; low QOE) compared to usual care control groups.9,18,43 The overall QOE was rated as low to moderate. Detailed results and quality of evidence are presented in Summary of Findings 4.

Summary of Findings 4

Parent-only versus parent-child interventions

Three SRs9,12,18 (three trials; 164 children) and three meta-analyses9,16,41 (five trials; 402 children) reported zBMI at the end of the interventions (10 weeks to six months). Results from the meta-analyses9,16,41 showed no significant difference in zBMI (moderate QOE) between the two interventions. Systematic reviews9,12,18 also reported consistent zBMI reduction in both groups (low QOE). Given there was no significant difference between parent-only interventions and parent-child interventions, the color "amber" was awarded, suggesting that both interventions were equally beneficial. The overall QOE was rated as low to moderate.

Body mass index percentile was reported in an SR,18 parental BMI was reported in an SR,9 and percentage of children who were overweight was reported in four SRs.9,18,43,44 Overall, there was no significant difference in child BMI percentile (one trial from one SR; 80 children; low QOE)18 and parental BMI (three trials from one SR; 207 parents; low QOE)9 between parent-only interventions and parent-child interventions. Mixed findings were reported for the percentage of children who were overweight (two trials from four SRs; 88 children; low QOE);9,18,43,44 with one trial reporting a greater reduction in parent-only intervention groups while the other trial found no difference between groups (percentage of children who were overweight reduced in both groups). No trial reported that parent-only interventions were less effective in comparison to parent-child interventions on the above outcomes. Detailed results and quality of evidence are presented in the Summary of Findings 5.

Summary of Findings 5

Child-only versus parent-only or parent-child interventions

For the purpose of comparing interventions with parental involvement to those without parental involvement, this section presents results of the two remaining intervention categories specified in the data summary section: parent-only interventions versus child-only interventions, and parent-child interventions versus child-only interventions. Detailed results and quality of evidence are presented in the Summary of Findings 6 and Summary of Findings 7. There was limited evidence that compared parent-child and child-only interventions (10 trials from five SRs; 546 children; moderate to low QOE),15,18,40,43,45 and even fewer studies that compared parent-only and child-only interventions (three trials from four SRs; 181 children; low QOE).15,18,43,44 Overall, no trial reported that child-only interventions were more effective than interventions with parental involvement. Parent-only and/or parent-child interventions demonstrated positive improvement on weight (three trials from one SR; 91 children; low QOE),15 zBMI (two trials from one SR; 236 children; low QOE),18 BMI (one trial from one SR; 36 children; moderate QOE),40 percentage overweight (six trials from five SRs; 288 children; moderate QOE),15,18,43-45 parental weight (one trial from one SR; 76 parents; low QOE)15 during follow-up at one to seven year/s. The overall QOE was rated as low.

Summary of Findings 6

Summary of Findings 7

Summary of evidence

The strength of overall intervention effectiveness immediately post intervention is presented in the Summary of Evidence (Table 3) using a traffic-light visual indicator. Parent-child and parent-only interventions were awarded the color "green" on most outcomes, indicating that interventions were effective or beneficial in improving weight-related outcomes. No intervention received a "red" indicator, meaning no intervention had a detrimental or less-effective impact when compared to the comparator group.

In summary, family-based behavioral lifestyle interventions targeting parents, with or without child involvement, can be effective in achieving successful weight change outcomes in children aged two to 18 years. When compared to a waitlist control group, parent-child interventions10,12,15,16,18,40-44 (one month to two years follow-up) and parent-only interventions9,10,12,15 (10 weeks to 10 months follow-up) were both effective in improving weight-related outcomes, such as a reduction in zBMI, BMI and BMI percentile. However, these interventions did not result in an impact on parent outcomes, including parents' BMI, waist circumference, and/or weight.9,41 A smaller number of studies compared parent-child interventions12,18,41 or parent-only interventions9 to a usual care control group (mailed information or a workbook or minimal sessions). The studies found that interventions with greater parental involvement and multiple treatment components, such as intense dietary monitoring, physical activity and behavioral modification, had greater overall effectiveness.9,12,18,41 Both parent-child interventions and parent-only interventions showed greater effectiveness when compared to child-only interventions, despite the limited number of studies reporting such comparisons.10,11,15,16,41 These overall findings are supported by evidence showing multi-component interventions with higher intensity or greater parental involvement were usually more effective in improving child weight outcomes.16,46

While interventions for children often require parents' involvement, SRs and meta-analyses suggest that interventions with parents only were equally effective when compared to interventions with parents and children.9,12,18,41,44 Five SRs indicated that parent-only interventions had similar (four SRs)9,12,18,41 or greater (one SR)44 effectiveness compared to parent-child interventions. However, all 14 SRs included interventions with parental involvement, but did not specify clearly whether mothers, fathers or both parents participated in the interventions. This has prevented this umbrella review from further synthesizing the results by sub-category to compare intervention effectiveness by different parental roles (e.g. mother-child versus father-child interventions). There was an insufficient number of SRs reporting behavioral outcomes (secondary outcomes), such as dietary intake and physical activity, from which any conclusions could be drawn regarding such parameters. Hence, intervention effectiveness in the present review mainly refers to improvement in weight, body composition and weight-related anthropometric indicators.

Discussion

The current umbrella review systematically identified, synthesized and graded a wide range of evidence on the effectiveness of targeting parents within individual-level treatment interventions for relative weight loss or weight maintenance in children aged 18 years and under who were overweight or obese. Results indicate that family-based behavioral interventions appear to be an effective strategy for weight management in children aged between two and 18 years, as indicated by a reduction in weight or weight-related outcomes (e.g. zBMI, percentage overweight) from baseline. The findings of the current umbrella review are similar to a previous umbrella review which assessed only RCTs with longer term intervention duration (>=6 months) in child weight management.39 The SRs found that a comprehensive multi-component intervention was effective in improving child metabolic and anthropometric measures, and appears to have the best overall outcomes when compared to single component interventions focused on physical activity, diet, education, and the pharmacological or surgical approach.39 The effectiveness of a multi-component intervention combining dietary advice, physical activity and behavioral modifications was also frequently mentioned in the SRs included in the current umbrella review.12,46 Evidence consistently supports the effectiveness of childhood obesity interventions that set goals for behavioral change, such as consuming five servings of fruits and vegetables each day and replacing sugar sweetened beverages with sugar-free beverages.44 Studies to date have recommended interventions that engage children in 60 minutes of moderate to vigorously intense physical activity on most days of the week, and limit screen time (leisure television and computer use) to no more than two hours per day.44 These findings are consistent with the Australia's Physical Activity and Sedentary Behavior Guidelines for Children (five to 12 years).47 The recent Australian 24-Hour Movement Guidelines for the Early Years recommend that preschoolers aged two to five years spend at least 60 minutes throughout the day in energetic play, including running, jumping, kicking, and throwing, and to limit screen time to no more than one hour per day.47 However, within the included SRs, there was a lack of reporting on behavioral change such as dietary intake and physical activity as a result of the interventions.

Parents, as the gatekeeper of the family food supply and as nutrition role models for their children, have a major influence on their children's eating habits.24,30,48,49 It is acknowledged that parents may play different roles as children age; however, the involvement of parents in interventions is essential, and this is supported by evidence showing that the parent's weight and lifestyle behavior are related to those of their children.20,50,51 Family-based interventions included in the current umbrella review directly involved one or both parents,12,45 and/or included family members or siblings15,16 in the treatment, and these interventions demonstrated greater effectiveness compared to control groups without parental or family involvement. Although the existing SRs suggested that including parents in weight management interventions enhanced outcomes, they did not provide clear insights into which of the many possible aspects of parental influence were modified in the interventions and were key to achieving the desired weight outcomes (e.g. feeding practices, food parenting).46 An SR of nine trials reported that no clear pattern emerged in terms of physical activity intervention effectiveness related to family member involvement (whole family, parents and the index child, or child only), goal of the family member, format of the intervention delivery (parents and child together or in separate groups), or age of the child.40

Few weight management intervention trials had similar intervention characteristics and, together with the mixed outcomes assessed and reported results, it was difficult to establish whether there was any particular intervention type (parent-only versus parent-child) that was more likely to lead to a successful outcome in terms of change in child weight outcomes.9 Nevertheless, the current umbrella review found that no intervention had a detrimental or ineffective impact on child weight-related outcomes when compared to the comparator control group. Studies suggest that if parents recognize the importance of their child's weight, they will be motivated to influence their children in terms of lifestyle behaviors related to weight control.42 Encouraging participating family members to change their own behaviors and reduce their own body weight may be an effective strategy for overweight children in terms of reducing excess weight or preventing further weight gain.40

Evidence also indicates that low parental self-confidence predicts dropout rates in family-based behavioral treatment,42 with one SR10 of seven trials indicating potential predictors of program success (greater reduction in child BMI) including higher parental motivation, lower baseline BMI percentile in children, higher parental attendance, younger children and lower socioeconomic status. Future interventions could include strategies targeting parents' self-confidence to actively engage them in interventions, and to motivate and encourage them to be good role models for their children by improving their lifestyle behaviors.

There was emerging evidence indicating that parent-only interventions are as effective, if not more effective, in improving child weight and/or weight-related behavior as parent-child interventions.9,12,18,41,44 The primary modality of intervention delivered to parents was through face-to-face educational sessions.44 Key strategies targeting parents included providing education on healthy eating and physical activity, fostering the development of parenting skills to promote positive health behaviors in children, and coping with difficult situations.10,41 Interventions targeting nutrition and/or physical activity education along with parenting skills showed larger and more significant changes compared to interventions with educational plus behavioral control components.16 Effectiveness was demonstrated in child weight management interventions that targeted parents as the agent of change through educational sessions on nutrition and/or physical activity, authoritative parenting styles (setting boundaries, providing a nurturing environment), positive parenting skills (self-monitoring, reinforcement, role modelling), and child behavior management strategies to encourage positive behaviors in weight management programs for overweight children.43 Interventions targeting parents to improve self-efficacy and confidence in managing health behavior also assisted in forming positive lifestyle habits within the family.50,52,53 It is therefore important to note that interventions that involved parents only were likely to be less costly than interventions that involved the whole family, especially when parents and children were in separate groups.11 However, the most commonly involved populations within the included SRs were children aged six to 13 years when parents were usually the gatekeeper of the family food supply. Parents' roles usually evolve as their children grow into adolescence and begin to gain more control and independence in making decisions including food preferences, such as lunchbox meals and snack choices, when eating at home or eating out with peers. Therefore, parent-only approaches for families with adolescents may need to be different from those with younger children. Nevertheless, there are numerous issues to consider due to the lack of high-quality evidence and high attrition rates in parent-only interventions. Further investigations are warranted to explore whether parent-only interventions are more cost-effective and sustainable,9,18 and to examine the barriers to participation and other complexities behind higher attrition rates in parent-only interventions through qualitative research.18

While previous research supports effective interventions that involve greater parental involvement as a whole, the majority of interventions targeting parents did not clearly specify whether mothers or fathers were involved.10 Whenever mentioned, studies commonly referred only to maternal involvement, with the paternal role generally overlooked.54-57 A recent systematic review seeking to assess father involvement in pediatric obesity prevention trials found that only 6% of parents in studies limited to one parent participation were fathers (N = 123).23 While only 2% of included studies identified a lack of paternal participation as a potential limitation, 99% of included studies did not explicitly attempt to engage with fathers.23 However, evidence shows that fathers are involved in child feeding, cooking, shopping and food choices,58 as well as other aspects of child health and wellbeing.56 Paternal BMI has been reported to be more strongly linked to childhood obesity than maternal BMI.59 This suggests that the beliefs and behaviors of fathers need to be taken into account when implementing weight related lifestyle intervention within the family.60 Future research should consider actively engaging both mothers and fathers in parent-targeted interventions for child weight management.

Limitations of the review

The current umbrella review had a number of limitations, as with any SR, one of which was that potentially relevant studies may have been omitted as the review only included published SRs in English. The JBI Reviewer's Manual recommends the inclusion of gray literature searches, and this approach is often followed in standard SRs. Therefore, unpublished gray literature would have been reported in the included SRs in the current umbrella review. There is the possibility that inherent bias was present in the reporting of this review where errors may have arisen in the initial appraisal and data extraction of the included SR or meta-analysis, and these may have been carried though in the current umbrella review.36 In some of the included SRs, it was unclear whether there was more than one independent reviewer for study selection (n = 7 SRs) and/or data extraction (n = 4 SRs), which quality appraisal or risk of bias instrument was used (n = 6 SRs), and the assessment of the presence of publication bias (n = 12 SRs) was unclear. There were a few occasions where results reported within SRs (narrative synthesis and results tables) were ambiguous. To address this, the original primary studies included in the SRs were examined to obtain information to enhance the accuracy of umbrella review synthesis.

This umbrella review was also dependent on the reporting of the included research syntheses, which may limit reporting of desirable details of interventions in the present report. For example, a limited number of SRs reported dietary and physical activity outcomes, which impeded further synthesis of the intervention effectiveness on these behavioral outcomes of interests in the current umbrella review. Positive behavioral change outcomes as a result of an intervention will provide an indication that an intervention is effective in modifying health behavior, which is likely to lead to weight loss in the longer term. Better reporting of behavior outcomes would help in evaluating the effectiveness of obesity interventions on promoting health behavior. It is particularly essential to report health behavior change as a result of obesity interventions when weight change is usually not observed or not statistically significant within a short study duration (e.g. generally between three and six months). As the majority of the included SRs did not adequately report on statistical significance (p-values) of the intervention trials, the umbrella review has not been able to synthesize a precise summary of intervention types that were significantly more effective than the other intervention types on various outcomes of interest. However, using a systematic approach, the umbrella review was able to provide recommendations after grading the quality of evidence on a range of interventions and the strength of intervention effectiveness against numerous weight-related outcomes in children aged 18 years and under who are overweight or obese.

Conclusions

Lifestyle behavior interventions targeting parents only, or parents with their child, are effective in achieving successful weight management outcomes in children aged two to 18 years. Multi-component family-based interventions combining dietary, physical activity and behavioral modifications have consistently demonstrated effectiveness. Effective interventions employ parent-targeted strategies, including nutrition and physical activity educational sessions, positive parenting skills, role modelling, and child behavior management.

Recommendations for practice

Health professionals can work with parents as the key agents of change for their children to encourage healthy eating and lifestyle behavioral change across the family. It is, however, not possible to recommend one intervention over another. Therefore, the recommendations for practice include a summary of strategies and interventions related to parental involvement within interventions. Parents can be provided with education on healthy eating and physical activity, not only to increase knowledge but to enhance self-efficacy and confidence in managing health behaviors within the home. Parent-targeted consultations can focus on fostering positive parenting skills to promote positive health behaviors in children and to cope with difficult situations related to health behavioral change (e.g. challenges at family mealtime). Positive parenting skills, such as monitoring, reinforcement, role modelling and providing nurturing environment, are relevant to supporting parents in facilitating healthy lifestyle changes in the family.

Recommendations for research

Future interventions need to examine whether engaging both parents within the parental component of interventions, especially fathers, can further enhance intervention effects. It is recommended for researchers to explicitly describe the role of parents (e.g. mothers, fathers) involved in the interventions as opposed to using the term "parents" when referring to participants who are often predominantly mothers. Future research should include larger and more diverse population groups, and examine the impact of interventions of a longer duration and follow-up. There is a need for more comprehensive reporting of health behavior outcomes (e.g. dietary intake, physical activity levels) in order to assess which intervention components contribute to effectiveness and their relationship with change in health risk factors that are also associated with overweight and obesity.

Funding

LKC is supported by the University of Newcastle International Postgraduate Research Scholarships, the Barker PhD Award Top-up Scholarship, and Emlyn and Jennie Thomas Postgraduate Medical Research Scholarship through the Hunter Medical Research Institute (HMRI). CEC is an NHMRC Senior Research Fellow and a University of Newcastle, Faculty of Health and Medicine, Gladys M Brawn Senior Research Fellow. KB is supported by the Australian Postgraduate Award and HMRI Top-up Scholarship. TB is supported by a University of Newcastle, Faculty of Health and Medicine, Early Career Brawn Fellowship.

Acknowledgments

The authors would like to thank Mrs. Debbie Booth (University of Newcastle, Faculty of Health and Medicine librarian) for assisting in the development of the search strategy. This study forms part of the thesis work for PhD candidate LKC at the University of Newcastle.

Appendix I: Search strategy

MEDLINE 1946 to present with daily update (searched on 2^nd May 2016)

Embase 1974 to 2016 week 18 (searched on 2nd May 2016)

PsycINFO 1806 to April week 4 2016 (searched on 2nd May 2016)

CINAHL complete (searched on 2nd May 2016)

Cochrane Library - DARE and CDSR (searched on 2^nd May 2016)

Scopus (searched on 2^nd May 2016)

Appendix II: List of relevant primary studies included in systematic reviews

Appendix III: List of excluded studies and reasons for exclusion

Appendix IV: Characteristics of included studies (systematic reviews)

References