Preventing and treating childhood overweight and obesity in children up to 5 years old: A systematic review by intervention setting

Abstract The prevalence of childhood obesity is increasing worldwide with long‐term health consequences. Effective strategies to stem the rising childhood obesity rates are needed but systematic reviews of interventions have reported inconsistent effects. Evaluation of interventions could provide more practically relevant information when considered in the context of the setting in which the intervention was delivered. This systematic review has evaluated diet and physical activity interventions aimed at reducing obesity in children, from birth to 5 years old, by intervention setting. A systematic review of the literature, consistent with Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) guidelines, was performed. Three electronic databases were searched from 2010 up to December 2020 for randomised controlled trials aiming to prevent or treat childhood obesity in children up to 5 years old. The studies were stratified according to the setting in which the intervention was conducted. Twenty‐eight studies were identified and included interventions in childcare/school (n = 11), home (n = 5), community (n = 5), hospital (n = 4), e‐health (n = 2) and mixed (n = 1) settings. Thirteen (46%) interventions led to improvements in childhood obesity measures, including body mass index z‐score and body fat percentage, 12 of which included both parental/family‐based interventions in conjunction with modifying the child's diet and physical activity behaviours. Home‐based interventions were identified as the most effective setting as four out of five studies reported significant changes in the child's weight outcomes. Interventions conducted in the home setting and those which included parents/families were effective in preventing childhood obesity. These findings should be considered when developing optimal strategies for the prevention of childhood obesity.


| INTRODUCTION
Over the last two decades, the global prevalence of overweight or obesity in children under the age of 5 has risen from 32 to 42 million (World Health Organisation, 2014). Once obesity is established in early life it may extend into adulthood, therefore creating a lifelong condition that is difficult to resolve (Geserick et al., 2018;Simmonds et al., 2016). Predictive modelling suggests, given the current levels of childhood obesity, that 60% of children today will have obesity by 35 years of age (Ward et al., 2017).
In the short term, children with obesity are at greater risk of adverse physical and psychological comorbidities (Pulgarón, 2013) as well as musculoskeletal difficulties, asthma and obstructive sleep apnoea (Narang & Mathew, 2012). In the longer term, there is an increased risk of morbidity, including type 2 diabetes, cardiovascular disease, cancer and premature death (Owen et al., 2009;Prospective Studies Collaboration, 2009). The substantial cost of treating the associated health implications has led to an intense focus on reducing rates of childhood obesity worldwide (Sonntag et al., 2016;World Health Organisation, 2016).
Among the potential causative factors, nutritional exposures in early life make a major contribution to the development of childhood obesity (Fogel et al., 2020), including the development of suboptimal eating habits and behaviours (Dalrymple et al., 2019), high intake of energy-dense foods and a higher rate of food consumption (Fogel et al., 2017). Low levels of physical activity and the hours of TV viewing time are also implicated (Janssen et al., 2005). Longitudinal studies suggest that these dietary habits (Mikkilä et al., 2005;van Jaarsveld et al., 2014) and sedentary behaviours (Jago et al., 2005) established in early childhood may track into adult life (Rovio et al., 2018). Interventions that modify dietary intake and/or physical activity early in life, therefore, have the potential to improve lifelong health.
A recent Cochrane Review of childhood obesity interventions identified 153 randomised controlled trials (RCTs) targeting children from 0 to 18 years of age (Brown et al., 2019). Outcomes were stratified by age group and the authors reported that for children aged 0-5 years, dietary and physical activity interventions resulted in a modest reduction in childhood body mass index (BMI) and BMI z-score (zBMI), and that the effect of interventions differed between settings. However, due to the scope and design of the Cochrane Review, it was not possible to identify which intervention was effective for whom and in which setting (Brown et al., 2019).
Previous systematic reviews have summarised the evidence of intervention to reduce childhood obesity within a specific setting, such as home (Pamungkas & Chamroonsawasdi, 2019) or school/ childcare (van de Kolk et al., 2019). Findings to date have been mixed, and the optimal setting to deliver intervention is not yet clear. As children spend a significant proportion of their day in either the home or school/childcare environment, interventions in these settings may potentially be more effective than those undertaken in a hospital/ community-based setting. To support obesity prevention strategies, it is important to evaluate interventions by setting to identify the optimal location of delivery. The aim of the present review was to systematically evaluate diet and physical activity interventions aimed at reducing obesity in children up to 5 years of age by intervention setting.  Studies meeting the following criteria were excluded: (1) nonrandomised and observational studies; (2) antenatal interventions that aimed to reduce the risk of childhood obesity or studies that aimed to assess breastfeeding initiation or duration, as these areas of research have recently been reviewed (Dalrymple et al., 2018;Rito et al., 2019);

T A B L E 2 Study characteristics
(3) conference abstracts and protocols; (4) interventions that did not mention the intervention setting; (5) absent or unpublished body composition data and (6) studies not reported in English.

| Literature search
In January 2021, the following databases were searched from 1st January 2010 up to 31st December 2020: Medline, Embase and the Cochrane Central Register of Controlled Trials. The search strategies are outlined in Supporting Information 1.

| Primary and secondary outcomes
The primary outcomes of this review were measures of childhood body composition, including BMI (z-scores and percentiles), SKM, weight and height. The secondary outcomes included behavioural outcomes associated with eating habits and physical activity. Tables 3-8 only report outcomes where a significant result was found. We deemed a study effective if at least one reported measure of obesity differed significantly between the intervention and control arms of the study.

| Study quality
The methodological quality of each study was assessed by two independent reviewers (H.W.-A. and F.S.) using The Cochrane Handbook for Systematic Reviews of Interventions (Higgins & Green, 2011 Cochrane criteria for judging risk of bias and were ranked using high, low or unclear for each domain and overall risk of bias was assigned to each study. Inconsistent assessments were discussed, and a consensus was reached among the authors.
Parent group clinic sessions of 90 min (dietary education and PA) in Weeks 8, 10 and 12.

Control:
Enhanced standard care, presented the same diet and activity recommendations, but was delivered in a one-time Paediatrician Counselling session.

Duration: 6 months
Paediatricians and researchers in weekly and biweekly sessions Significant decrease in BMI z-score for the intervention group versus control at 6 months, which was maintained at 12 months for each measurement: The sample sizes across studies ranged from 18 (Stark et al., 2011) to 1211 (Natale et al., 2017) participants and age at baseline ranged from birth (Hodgkinson et al., 2019) to 5 years (Bocca et al., 2012).

| School/childcare-based interventions (n = 11)
One study used a diet-only intervention, which was adopted from a nutrition-focused US National Aeronautics and Space Administration  (Bellows et al., 2013). Four of the interventions implemented a health and wellbeing policy in the school or childcare setting, which aimed to influence a child's physical environment and sociocultural factors to create a setting that favours obesity prevention (Hodgkinson et al., 2019;Natale et al., 2014Natale et al., , 2017Stookey et al., 2017). Dietary advice was provided by classroom/daycare centre teachers and staff (Bellows et al., 2013;Fitzgibbon et al., 2011;Hodgkinson et al., 2019;Natale et al., 2014Natale et al., , 2017Salazar et al., 2014). Three of these studies and teachers (Kim et al., 2019;Natale et al., 2014) and through guidelines and tips presented on posters as well as tailored feedback on activity and diet-related behaviours (Verbestel et al., 2014).
The intensity of the interventions varied from twice weekly educational sessions (Fitzgibbon et al., 2011) to up to 6 months of educational sessions in the first year with four booster sessions delivered in Years 2 and 3 of the intervention (Natale et al., 2017). All interventions except for one (Verbestel et al., 2014) involved group-based sessions for either children or parents or children and parents, with only one intervention also offering one-to-one sessions (Hodgkinson et al., 2019). Four interventions included family/parent group sessions (Lumeng et al., 2017;Natale et al., 2017;Salazar et al., 2014;Walton et al., 2015).
Five of the eleven studies reported a significant impact on childhood obesity measures with decreases in intervention group zBMI (Hodgkinson et al., 2019;Verbestel et al., 2014), BMI percentile (Natale et al., 2017;Stookey et al., 2017) and a significant change in fat mass, fat-free mass, skinfold thicknesses and percentage body fat in the intervention group compared to controls (Salazar et al., 2014).
Six of the eleven studies did not report a significant difference in obesity. However, four of these studies observed significant differences in dietary behaviours (Fitzgibbon et al., 2011;Kim et al., 2019;Lumeng et al., 2017;Salazar et al., 2014), of which two also reported a significant difference in physical activity (Fitzgibbon et al., 2011;Salazar et al., 2014) (Table 3).

| Home-based interventions (n = 5)
One dietary only intervention assigned participants to growing up infant milk, 'GUMli' (reduced protein with synbiotics and micronutrients) or whole, pasteurised cows' milk, both in powder form (Wall et al., 2019). The four combined interventions used a variety of strategies to modify dietary intake and physical activity behaviours, including incorporating culturally appropriate programmes (de la Haye et al., 2019;Tomayko et al., 2016), family-based interventions to increase intake of fruit and vegetables (de la Haye et al., 2019;Tomayko et al., 2016), limiting consumption of foods high in saturated fat and SSBs (Haines et al., 2018) or SSBs and fruit juice while providing education on healthy portion sizes (de la Haye et al., 2019). In one study, families were given a paper family routine tracker to record their health behaviours and possible barriers to change (Haines et al., 2018).
One intervention recruited children at risk of overweight or obesity (Sherwood et al., 2015). This intervention provided home counselling visits to raise awareness of obesity risk, obesity prevention tips, while a phone coaching programme for parents aimed to reduce screen time and increase physical activity (Sherwood et al., 2015). Interventions were delivered by trained home visitors who were matched to each family based on race, ethnicity and language preference (de la Haye et al., 2019), researchers (Wall et al., 2019), health educators (Haines et al., 2018), trained home mentors who were tribal members with long-standing employment in the community (Tomayko et al., 2016) and paediatric care providers (Sherwood et al., 2015).
Four of the five studies reported a significant difference in the following measures of child obesity, including a reduction in percentage body fat at 12 months (Wall et al., 2019), a reduction in fat mass (Haines et al., 2018), BMI percentile and BMI z-score (Sherwood et al., 2015;Tomayko et al., 2016). All five reported some significant differences in diet and physical activity behaviours, including a decrease in SSB intake (de la Haye et al., 2019), an increase in fruit and/or vegetable intake (Haines et al., 2018;Tomayko et al., 2016; Table 4).
One study used a family-based behavioural model aimed at modifying dietary and physical activity behaviours through activities, such as creative play and skill development, to promote fruit and vegetable acceptance (Skouteris et al., 2016). Another consisted of two intervention groups: (1) responsive parenting intervention (Tot-TOPS) focused on toddler diet, physical activity and behaviours, such as soothing without relying on food, appropriate portion sizes for toddlers and recreational physical activity guide for toddlers, and (2) a maternal lifestyle intervention (Mom-TOPS), which focused on maternal diet and physical activity (Black et al., 2021). A further study used interactive group sessions informed by theoretical models (Daniels et al., 2013). Two interventions implemented family-based programmes targeting healthy nutrition and activity, including Zumba classes and programmes that focused on healthy nutrition choices and increased activity (Berry et al., 2011;Campbell et al., 2013).
Dietary advice was provided by members of the community or community allied health professionals (e.g., maternal and child health nurse, childcare worker) (Skouteris et al., 2016), health educators (Berry et al., 2011;Black et al., 2021), dietitians (Campbell et al., 2013;Daniels et al., 2013) and psychologists (Daniels et al., 2013). The intensity of the interventions varied from quarterly sessions delivered during parents' regular group meetings (Campbell et al., 2013) to once a week for 3 months (Berry et al., 2011). The duration of the interventions varied from 10 weeks (Skouteris et al., 2016) to 15 months (Campbell et al., 2013). The community-based interventions did not report a significant difference in obesity measures, except for one which reported a reduction in BMI percentiles (Berry et al., 2011).
Four studies reported significant differences in dietary behaviours (Black et al., 2021;Campbell et al., 2013;Daniels et al., 2013;Skouteris et al., 2016). No studies reported an effect on offspring physical activity, except for one study (Black et al., 2021 3.4 | Hospital/clinic-based interventions (n = 4) The hospital/clinic-based interventions were conducted in primary care clinics (Martínez-Andrade et al., 2014;Quattrin et al., 2014), a university-based clinic (Fisher et al., 2019) or an outpatient clinic (Bocca et al., 2012). The only diet-only study utilised a food fun and families parenting intervention to reduce children's intake of 'empty calories' from solid fat and added sugar (SoFAS) through group sessions for mothers with low-income levels (Fisher et al., 2019). For the three diet and activity interventions, two recruited children with overweight and obesity (Bocca et al., 2012;Quattrin et al., 2014).
These three studies aimed to modify dietary behaviour using a variety of approaches, including sessions with parents and children on improving diet, increasing physical activity and decreasing sedentary activity (Quattrin et al., 2014). One focused on eating breakfast while having an active lifestyle, consistent with elementary school exercises. Parents received behavioural therapy on how to become healthy role models for their children, changing family attitudes, removing unhealthy food triggers and knowing the difference between hunger and food cravings (Bocca et al., 2012). An obesity awareness and prevention curriculum was used by another study and included diet, healthy growth and physical activity workshops (Martínez-Andrade et al., 2014). The duration of the interventions varied from 6 weeks (Martínez-Andrade et al., 2014) to 12 months (Quattrin et al., 2014). For the diet-only study, the intervention was provided by a graduate-level interventionist (Fisher et al., 2019). The diet and physical activity interventions were delivered by dietitians (Bocca et al., 2012;Quattrin et al., 2014), nurses and nutritionists (Martínez-Andrade et al., 2014) and practice enhancement assistants (Quattrin et al., 2014). The intensity of the interventions ranged from more than once per week (Bocca et al., 2012) to monthly (Quattrin et al., 2014). Three of the studies reported significant differences in dietary intake (Bocca et al., 2012;Fisher et al., 2019;Martínez-Andrade et al., 2014) (Table 6), of which, two reported a significant reduction in body composition outcomes postintervention and at subsequent follow-ups (Bocca et al., 2012;Quattrin et al., 2014).

| eHealth-based interventions (n = 2)
One study recruited families via social media and delivered guidance to parents in monthly video clips on feeding-related topics, including the evolution of taste preferences in children, appropriate food types and textures and responsive feeding practices in addition to providing cooking and recipe information (Helle et al., 2019). The other aimed to change diet and physical activity behaviours via a freely accessible smartphone app for parents, which was compatible with both iOS (version 6.1.3 or higher) and Android (version 2.3.5 or higher) smartphones. The content was underpinned by social cognitive theory and behaviour change techniques and was based on guidelines for healthy eating and physical activity in preschool-aged children (Nyström et al., 2017). The interventions were delivered by monthly video clips via email (Helle et al., 2019) or bi-weekly via a smartphone application (Nyström et al., 2017). Neither study reported a significant difference in measures of childhood obesity.
However, one study reported a significant improvement in dietary intake and eating behaviours (Helle et al., 2019; Table 7).

| Mixed setting interventions (n = 1)
The mixed setting intervention was conducted in a clinic and at home (Stark et al., 2011) and recruited children with a BMI percentile ≥95.
The dietary advice centred on meals, snacks and beverages and included a calorie target per day for each child. Parents were instructed to keep a 7-day food diary and both parents and children were provided with pedometers and a daily step goal. Children also attended group sessions. Home sessions were delivered by psychology postdoctoral fellows while the group sessions were delivered by psychologists and research coordinators. The intervention was delivered in two phases: Phase 1 was delivered on a weekly basis for 12 weeks, alternating between group-based clinic sessions and individual home visits and phase 2, which consisted of sessions every other week for 12 weeks, alternating between the clinic and home.
The study reported a significant decrease in BMI percentile and zBMI at 6 and 12 months and an improvement in dietary intake (Stark et al., 2011; Table 8).

| Quality of included studies
The overall quality of the included studies varied (Table S1). Nine studies were assessed to be at 'high risk of bias', 10 studies classified as 'low risk of bias' and the remaining 9 studies classified as 'moderate'. Common sources of bias included no information on randomisation or high rates of attrition. Of the effective interventions, five were found to be at low risk of bias (Haines et al., 2018;Quattrin et al., 2014;Stark et al., 2011;Stookey et al., 2017;Wall et al., 2019), four were moderate (Bocca et al., 2012;Hodgkinson et al., 2019;Sherwood et al., 2015;Verbestel et al., 2014) and four were classified as high risk (Berry et al., 2011;Natale et al., 2017;Salazar et al., 2014;Tomayko et al., 2016).

| DISCUSSION
This comprehensive and contemporary evaluation of interventions to prevent and reduce childhood obesity in young children has demonstrated differential effects of interventions on childhood obesity outcomes by intervention setting. Out of 13 studies that reported a significant difference in obesity measures, 5 were performed in a school/childcare setting, 4 in a home-based setting, 2 in a clinic/hospital-based setting, 1 in the community and 1 in a mixed intervention setting. The effective interventions all encompassed a diet and physical intervention, except for one, and they all recruited children over the age of 1 year. The duration of these studies ranged from 4 months to 2/3 years, with the majority of the successful trials intervening for at least 6-12 months. Five of these studies were found to be at low risk of bias, 4 were moderate and 4 were classified as high risk. No differences in obesity measures were reported in studies that utilised an eHealth intervention.

| School/childcare-based interventions
The
Parental involvement has been shown to be necessary when implementing dietary changes in young children (Alderman & Headey, 2017). Parents have control over the dietary intake of their child, particularly during the early years; therefore, interventions that target the family as a whole may have a more positive impact on encouraging the development of healthy eating patterns in offspring (Savage et al., 2007). Our finding is consistent with a systematic review and meta-analysis that assessed the influence of parental practices on child promotive and preventive food consumption behaviours and found that alongside food availability, parental modelling showed a strong association with both healthy and unhealthy food consumption (Yee et al., 2017).
In the present review, three of the four studies that reported a significant difference in childhood body composition outcomes also reported improvements in dietary behaviour, of which two studies reported positive changes in physical activity. Previous research examining the role parents have in shaping their children's physical activity and sedentary behaviours has suggested that parental involvement may increase daily physical activity and contribute to the prevention of weight gain (Pyper et al., 2016). One of the interventions utilised a different dietary strategy to reduce childhood obesity, evaluating standard cow's milk against lower protein milk for 12 months in 1-year-old children in New Zealand. Percentage body fat was modestly lower in the children assigned to the lower protein milk arm (Wall et al., 2019), which aligns with previous studies reporting an association between high protein intake in infancy and a higher BMI in later childhood (Günther et al., 2007;Michaelsen & Greer, 2014). Larger RCTs are required to examine whether this approach is an effective intervention to reduce childhood obesity.

| Community-based interventions
The community-based interventions were directed at individuals, rather than populations. One of the five interventions conducted in a community setting reported an improvement in childhood obesity measures. Several studies reported greater dropout rates from parents of lower socioeconomic status, with one noting that dropout/missed sessions were due to the cost of travel, childcare and work commitments (Skouteris et al., 2016). This may have influenced the level of engagement and outcomes of the intervention, with possible barriers of travelling to a community setting if it is not a frequent destination, in comparison to interventions set within participants' own home or a school/ childcare setting.

| Clinic/hospital-based interventions
Our review found that two of the four clinic/hospital-based interventions led to differences in childhood obesity measures between treatment groups (Bocca et al., 2012;Quattrin et al., 2014).
Both studies recruited either children with overweight or obesity and utilised an intense dietary and physical activity intervention, which led to some improvements in dietary intake. Interventions set in this environment targeted both children and parents; however, this setting may be less amenable to addressing other social and physical environment influences on behaviour, which may limit effectiveness, compared to other settings.

| eHealth
Two studies in this review evaluated the use of parent-focused eHealth interventions in reducing obesity in young children, neither of which reported a difference in childhood obesity measures. Some eHealth interventions have been shown to be successful in adults (Hutchesson et al., 2015); however, there is a paucity of data in children. Hammersley et al. (2016) assessed parent-focused eHealth interventions in children and adolescents and out of eight included studies, no significant differences in obesity measures were reported. The use of eHealth interventions may provide the opportunity to overcome barriers that arise from in-person sessions, such as providing greater scheduling flexibility and reaching at-risk families (Jacobs et al., 2016;Reinwand et al., 2015). More research is needed to establish whether parent-focused eHealth interventions could play a role in preventing obesity in young children.

| Future implications
This systematic review has identified that interventions conducted in the home setting and those which included parents/ families are the most effective in reducing obesity in children from birth to 5 years. School/childcare interventions which included policy-based interventions were also effective. Furthermore, studies that solely recruited children with overweight or obesity all reported significant differences in the measure of BMI between the trial arms (Bocca et al., 2012;Quattrin et al., 2014;Stark et al., 2011). This suggests that interventions in children with overweight or obesity have the potential to have a greater influence on body composition outcomes. Future reviews of childhood obesity interventions could therefore stratify outcomes by BMI category to explore this observation further. We know that a child's food choices and lifestyle decisions are controlled by parents/caregivers at home; this may also extend to childcare settings where 76% of UK children under 5 years old attend a form of part-time or full-time childcare (Scaglioni et al., 2018). With evidence that one-third of a child's daily calories are consumed within schools; interventions that target both parents and policies within the childcare/school settings should enable continuity in modelling healthy food patterns, therefore preventing mixed messages and compromising learnt behaviours (Liu et al., 2019), while optimising opportunities to promote better health outcomes.

| Strengths and limitations
Strengths of this systematic review include the use of a structured process and framework guided by the Centre for Reviews and Dissemination (2009)

| CONCLUSION
This contemporary and comprehensive review highlights the differential effect of interventions on measures of child obesity by setting.
Interventions conducted in a home setting and those that included parent/family involvement more frequently led to improvements in childhood obesity measures than those in the community or hospital settings or involving eHealth coaching. We also found that studies that recruited children with overweight or obesity were also effective. The successful interventions used a variety of strategies to modify diet and physical activity behaviours and the implementation of policy to influence the early-years environment appeared to be effective. The success of interventions may be attributed to the involvement of adults who are key to the child's care, for example, the entire family/parents and/or teachers, known to play an important role in obesity prevention and treatment in this age group.
In particular, in childcare settings children are often influenced heavily by their peers or in the case of the home environment, by siblings/family members. These findings should be considered when developing optimal strategies for the prevention or treatment of childhood obesity.

AUTHOR CONTRIBUTIONS
The research question and design study were formulated by Kathryn