Obesity: a systematic review on parental involvement in long-term European childhood weight control interventions with a nutritional focus

In Europe, about 20% of children are overweight. Focus on parental responsibility is an effective method in weight control interventions in children. In this systematic review we describe the intensity of parental involvement and behaviour change aimed at parents in long-term European childhood weight control interventions. We include European Union studies targeting parents in order to improve children's weight status in multi-component (parental, behaviour change and nutrition) health promotion or lifestyle interventions. The included studies have at least one objectively measured anthropometric outcome in the weight status of the child. Parental involvement was described and categorized based on the intensity of parental involvement and coded using a validated behaviour change taxonomy specific to childhood obesity. Twenty-four studies were analysed. In effective long-term treatment studies, medium and high intensity parental involvement were identified most frequently; whereas in prevention studies low intensity parental involvement was identified most frequently. Parenting skills, generic and specific to lifestyle behaviour, scored frequently in effective weight control interventions. To list parental skills in generic and specific to lifestyle, descriptions of the included studies were summarized. We conclude that intensity of parental involvement and behaviour change techniques are important issues in the effectiveness of long-term childhood weight control interventions.


Introduction
Childhood overweight and obesity are a serious public health problem in Europe. In Europe, about 20% of children (aged 0-16) is currently overweight, of which a third is obese. This percentage represents 15 million school children. Prevalence among infants and preschoolers (aged 0-5) is high and of epidemic proportion, ranging from 12% in eastern European countries to 33% in Mediterranean countries. Over 60% of children who are overweight before puberty will be overweight in early adulthood (1,2). Childhood obesity has adverse psychological, social and health consequences in childhood and later in life (3). Children have 10 times higher risk for obesity when both their parents are obese (4). Additionally, it is known that energy balance related behaviour such as dietary behaviour is established and set before the age of 5 (5,6). Experts in the area of childhood obesity recommend that prevention and treatment of obesity in the formative pre-and primary school years should focus on parents (7).
Parental beliefs, attitudes, perceptions and behaviour appear to have significant impact on the development of early overweight (4,(7)(8)(9)(10)(11). Greater parental involvement and making parents responsible for participation in and implementation of lifestyle changes are identified as effective techniques in the prevention and treatment of obesity and being overweight in children (9,12,13). To study the effectiveness of interventions based on behaviour change processes and techniques in the programme contents, a behaviour change taxonomy was developed and validated (14). Golley et al. specified this taxonomy further for childhood obesity and studied short-term effectiveness of parental involvement in lifestyle interventions including a nutrition or activity component and a behavioural change component in worldwide studies that included children aged 1-18 years. Intervention effectiveness was favoured when behaviour change techniques (BCTs) spanned the spectrum of behaviour change process (9). However, the long-term effectiveness of parental involvement in these interventions remains unclear. Therefore, in this review, we aimed to describe the intensity of parental involvement and behaviour change processes and techniques aimed at parents in long-term childhood weight control lifestyle interventions, with a focus on nutritional components, in children in the age of 0-12 years in the European Union.

Inclusion criteria
Published and unpublished (non-) randomized controlled trials (RCTs), clinical controlled trials, pilot studies and observational trials and reports were eligible for inclusion in this systematic review. Restriction to RCT designs might compromise the types of parental involvement implemented in present interventions.
Intervention participation involved at least one parent or caregiver, with their child(ren). Interventions were included if they had a parental component, a behavioural change component and a nutritional component. For the multiple components, we formulated the following definitions. A parental component is an intervention with the parents or caregivers as key participants. Parents were considered key participants if the reviewers were able to identify direct parental exposure to intervention; identify active parental participation and identify active use of parenting skills in lifestyle behaviours including dietary and physical behaviour. We defined a behavioural change component as a theory-based behavioural change or well-reported method of behaviour change. A nutritional component was defined as targeting a single or multiple dietary change through education in a group or by individual counselling. The programme settings were (pre)school-based, communitybased, day-care based or clinic outpatient setting.
Only trials performed in the European Union were considered for review. Trials with interventions aimed at the primary, secondary, tertiary prevention and treatment of weight control (stabilization of weight and weight loss) were included. Since we were specifically interested in parental involvement in intensive lifestyle interventions, trials were included if the duration of the interventions was 10 weeks minimum. Trials with a primary prevention focus needed to have a follow-up of at least 1 year from the start of the trial. Trials of weight control interventions needed to have a follow-up of at least 6 months after the end of the intervention was performed. Studies with at least one objectively measured (not self reported) anthropometric outcome in the weight status of the child -body mass index (BMI), BMI-standard deviation scores (BMI z-scores) or percentage overweight of the child -at end of intervention or interim, and/or Յ1 year post-intervention and/or Ն2 years post-intervention were included.

Exclusion criteria
Studies focusing on a single intervention component and studies in which weight status was not reported as an outcome variable were excluded. Furthermore, studies on obesity resulting from eating disorders, and studies in which pharmaceuticals were used as an intervention for the treatment of obesity were excluded. Trials on obesity in a specific subgroup with non-obesity related comorbidity were excluded as well.

Search methods for identification of studies
The search was carried out in April 2011. Interventions published between January 1996 and April 2011 were searched. Search strategies and keywords for the different electronic databases were developed and assessed by both researchers and specified with the help of an information specialist. The search was performed using the following electronic databases in a systematic, structured and repro- For included studies, if parental involvement was reported only briefly, we searched for publications of the same study that provided more information on parental involvement or requested more information from the authors. In addition, reference lists of all retrieved articles and review articles were screened for potentially eligible articles. Furthermore, a number of websites of research groups that conduct and publish systematic reviews, websites and contents of programme details were systematically searched, with a key focus on Dutch websites and programmes. The International Standard Randomised Controlled Trial Number register, the register of the National Institute of Health (http:// www.controlled-trials.com) and the Dutch Trial register (http://www.trialregister.nl) were searched for ongoing trials. All authors of ongoing trials were contacted for details on unpublished results. The 2010 report of the European network for public health, health promotion and disease prevention EuroHealthNet was screened for obesity prevention and treatment programmes (15).

Selection of studies
After performing the search strategy described previously, the second author (FK) assessed the records retrieved by the database search and selected full texts for eligibility, based on title and abstract. Review criteria were further specified based on the records retrieved. Two authors (JvdK, FK) independently assessed these full texts to identify studies meeting the inclusion criteria, using a self-developed form. Discrepancies were resolved between reviewers by reaching consensus. In case of discrepancies between the authors, final resolution by a third party arbitrator (CL) was made. In Fig. 1 the search results are visualized.

Data extraction and synthesis
Two reviewers (JvdK, FK) performed data extraction independently, using piloted and standardized coding forms. The data collection included a description of the parents' involvement, a categorizing of the intensity of parental involvement, a list of reported BCTs and a summary of results of primary outcomes. Moreover, only those aspects of the interventions which were aimed at the parent(s) and child were coded. Only the intervention under study was coded. Studies categorized with unclear parental involvement were not coded. The methodological quality of included studies was assessed by two reviewers independently (JvdK, FK) using the CONSORT statement with the extension for non-pharmacologic treatment (NPT) interventions (17). The risk of bias was assessed as instructed in the Handbook of the Cochrane review group (18). Discrep-ancies were resolved between the reviewers by consensus. Studies with a low quality rating or high risk of bias were excluded from the review.
Methods of parental involvement were described and categorized based on the intensity of parental involvement: high involvement, medium involvement, low involvement or unclear involvement (19). The intensity of parental involvement was measured scoring the time the healthcare provider spent in direct contact with the parents and the frequency of contact moments. High involvement was defined as parents are directly involved in multiple activities or in structural behaviour change methods of the lifestyle intervention, delivered by multiple sessions, home visits or individual counselling over an extended period of time. There are opportunities for the parent to contact the caregiver at all times. Medium involvement was defined as parents are directly involved in at least two activities within the intervention, delivered in at least four sessions over a period of time of at least 3 months. There are opportunities to consult the intervention team at a set time point in the week. Low involvement was defined as parents are directly involved in at least one occasion/session and are approached in an indirect way during a period of at least 3 months. Additionally, unclear involvement was defined as based on the report of parental involvement in the full text; not enough detail is available to code the behaviour aspect of parent involvement, and no further details were available or have been provided by the authors on request.
In the behaviour change taxonomy of Golley et al., five behaviour changes processes and 32 BCTs were identified (9). The five processes underpinning behaviour change process are identify and motivate readiness to change, facilitate motivation to change, provide relevant information and advice/behaviour change strategies, build self-efficacy (and independence) and prevent and manage relapse. The 32 BCTs are summarized in Table 3. An instruction manual was available on how to identify the processes and the BCTs. To increase interrater reliability, a pilot test was performed, in which four trials not included in the final review were reviewed. Both in the pilot test and in the final review, two reviewers (JvdK and FK) performed the coding independently. Different outcomes in the coding were discussed and proposed to an expert (FL) in coding the BCTs of Golley et al.
To determine which type of intensity of parent involvement and BCTs were effective, studies were categorized as effective or ineffective based on child weight status and cross tabulated with the intensity of parental involvement and BCTs used in the intervention. A study was classified as supporting effectiveness of the intervention, if it showed a significant change in an objectively measured (not self-reported) variable of obesity (e.g. BMI z-score or percent overweight). Due to heterogeneity in both the multi-component interventions and study design, study data were not pooled and results are presented in a narrative form. Table 1 summarizes 24 studies on health promotion or lifestyle interventions, which aimed to reduce or control body weight and change dietary behaviour in young children in Europe. Eight studies targeted the primary prevention and 16 studies targeted the treatment of overweight and obese children. Individual counselling, group sessions with child and parent, or written materials were the top three modes of intervention delivery. Seven studies primarily investigated obese children only and nine studies primarily investigated overweight and obese children. Longterm treatment effects are available from most intervention trials, with follow-up ranging from 6x months up to 5 years post-tertiary prevention.      Mend (

Quality assessment
The results of methodological quality and risk of bias are reported in Table 1. The three studies with a low risk of bias had a randomized controlled design (28,39,58). In seven studies, follow-up was reported in one or more publications. Quality assessment of these studies was based on all publications (21,24,35,42,46,53,54). In 17 studies, the risk of bias was assessed as unclear or unclear to high. In 8 of these 17 studies, study quality was impaired due to poor study design (32,43,44,54,56,64,66,67). Other common reasons for unclear to high risk of bias in reviewed studies are non-randomization, small sample size and high dropout. Loss to follow-up increased over the course of the time of post-intervention measurements (43,50,53).

Parental involvement
All 24 trials were classified according to the criteria of high, medium or low involvement. Table 1 describes how parents were involved in the intervention. In primary prevention programmes, parents were involved in meetings which were organized by school and by provision of written information materials. Weight loss interventions were mostly delivered in group sessions, in which child and parent were together, or were in separate sessions, occasionally interacting. Table 2 shows the intensity of parental involvement and effectiveness of the studies at the end of the intervention or at interim and 2 years post-intervention. In prevention studies, only low intensity parental involvement was identified, whereas in effective treatment studies medium and high parental involvement were identified most frequently.

Behavioural change techniques
Using the taxonomy of BCTs, 22 studies were coded (7). The interventions of EPODE and Utrecht Overvecht Gezond Gewicht were not coded, due to the complexity of these multiple interventions and the inability to describe clearly parental involvement (30,69). Table 3 lists the five behaviour change processes and the 32 techniques used to code intervention description. We scored the results of effective and ineffective interventions at the end of the intervention or at interim (n = 22), Յ1-year post-intervention (n = 16) and Ն2 years post-intervention (n = 10).
Sixteen studies reported all five behaviour change process steps, four studies reported four out of five and two studies reported three out of five. All treatment studies included a minimum four behaviour change process steps. In total, 310 techniques were reported in all studies (n = 24). Primary prevention studies scored 23 BCTs and the treatment studies scored 287 BCTs. Treatment studies used more BCTs than the primary prevention studies. Table 4 shows the descriptions found in the included studies about the BCTs parenting skills generic and parenting skills specific to lifestyle. Parenting skills generic and specific are reported in Ն50% in effective studies at the end of the intervention or at interim, Յ1-year post-intervention and Ն2 years post-intervention.

Behavioural change techniques in long-term effective studies
From the 32 BCTs in the studies, 10 techniques scored Ն 11 in effective studies at the end of intervention or interim (n = 22) ( Table 3). More than 2 years postintervention, the most frequently reported techniques in effective studies were provide general information on behaviour-health link (six out of six studies), provide information consequences (five out of six studies), prompt intention information (five out of six studies), provide instruction (five out of six studies), tailored or personalized delivery (five out of six studies), feeding practices (five out of six studies) and plan social support social change (six out of six studies).

Discussion
We systematically reviewed the intensity of parental involvement and BCTs aimed at parents in long-term   obesity reviews childhood weight control intervention using a taxonomy to describe studies in terms of intervention content. This review showed that in all prevention studies, the intensity of parental involvement was identified as low, whereas in treatment studies the intensity of parental involvement was identified as low, medium or high. No conclusions could be drawn concerning the low intensity of parental involvement in prevention programmes, due to the small number of the included primary effective prevention studies at the end of the intervention or at interim (n = 3) and >2 years post-intervention (n = 1). In the included treatment studies measured at the end of intervention or at interim (n = 13) and measured >2 years post-intervention (n = 5), medium and high intensity of parental involvement were identified most frequently. This finding suggests that the level of intensity of parental involvement is an important issue in weight control interventions. We identified less reported BCTs in primary prevention studies compared with treatment studies. The BCTs most frequently identified differed per time of follow-up. The most frequently identified BCTs Ն2 years post-intervention were provide general information on behaviour-health link, provide information consequences, prompt intention formation, provide instruction, tailored or personalized delivery and plan social support/social change. In contrast to Յ1 year post-intervention, parenting skills generic was infrequently identified Ն2 years post-intervention.
The results of this systematic review are in line with the findings of the study by Hingle et al., in which direct approaches to engage parents were more likely to result in positive outcomes than indirect methods (70). Within the reported techniques of parental barriers in the treatment of childhood obesity, we found similarities with the study by Pocock et al. (11). The most common theme related to parental perception is lack of time, which acts as a barrier to child exercise and healthy diet (11). In our review, however, the BCT time management was reported only twice in all included studies. For that reason, we recommended attention to time management in developing or implementing weight control interventions.
The strength of this review is the scope of reviewing long-term multi-component lifestyle interventions with a parental, a behavioural change and a nutritional component in childhood obesity by identifying the role of parental involvement in the European Union. We specifically studied sustainable effects of parental involvement on weight control. Effectiveness of interventions was identified both using a classification system of intensity of parental involvement and a taxonomy to describe studies in terms of intervention content. Also, parenting skills general and parenting skills specific to lifestyle were identified explicitly.
However, this review has several limitations. Due to the multi-component design of the interventions under study, change in energy-related behaviour or weight can be inferred as being caused or influenced by many factors such as physical activity and sedentary behaviour, and not by parental involvement only. In our opinion, interventions should focus on both energy intake and energy expenditure. However, the taxonomy by Golley et al. does not Table 4 Parenting skills generic and specific to lifestyle Parenting skills generic: Parenting skills specific to lifestyle: Skills to use praise and adequate reward to reinforce the children, how to apply positive enforcement, importance of using praise with children Enforcing dinner table rules in a positive way: taking the time to eat, no eating in front of the TV, any food on the table is offered to all members of the family, keeping stress away from the family table Increasing the quality of authority and control of the parent over the child, instruction on how to set boundaries for the child Giving the good example of modelling physical activity Stimulating and supporting the child to deal with bullying Providing structure in frequency of meals Skills training to support adequately the child, skills to help the child deal with negative emotions Controlling portion sizes of meals in a positive way, helping the child to differentiate between hunger and craving Learning to provide an open environment for communication, listening to each other, sharing ideas and opinions, consulting model Undertaking activities together (fun and play) Teaching parents how to deal with different children in one family Create awareness of who is responsible for achieving good physical activity and good food habits Changing interaction patterns between parents and child by teaching them how to support the child instead of controlling them Providing a supportive environment in terms of food availability and accessibility Addressing important topics with parents/carers to help them implement these topics at home Practices used to control the child's dietary intake and to monitor the child's food intake Family rules family rewards, nurturing our families nurturing ourselves Skills to implement anti-obesogenic strategies Skills to implement and maintain behaviour change Developing family rules that will support the development of healthy lifestyle behaviours within the home Training in assertiveness Enabling mothers to cope with stigmatization of obesity in their offspring Skills to modify behaviour step by step obesity reviews Parental involvement in childhood obesity J. J. van der Kruk et al. 757 identify physical activity as a specific BCT. After finishing the process of coding, a refined version of the manual of the CALO-RE taxonomy was published (71). The specific focus of the CALO-RE taxonomy is changing physical activity and healthy eating behaviours. This taxonomy may offer additional value to prevent childhood obesity since both aspects in energy balance (intake and expenditure) are included. Another limitation is the lack of identifying intensity of the various BCTs. Unfortunately, the taxonomy of Golley et al. identifies underpinning processes and BCTs in behaviour change only. In addition to the use of the taxonomy of Golley et al., we assessed intensity of parental involvement by classifying this involvement into low, medium or high parental involvement. However, it is unknown whether the value of each separate BCT in the context of the multi-component intervention is equal in outcome of effectiveness. Therefore, insight in the intensity of BCTs is needed. Furthermore, more clinical relevant cut-off points may be needed to determine whether an intervention is effective. Our process of coding behaviour change was based on manuals and description of the intervention, and limited by the quality of reporting in the publications of the studies. In the publications, we found no details on differences between description and implemented intervention in practice. Therefore, the assumption was made that the intervention was implemented in the exact same way as stated by the description. However, discrepancies in coding might be possible for encodings from different publications. Further, we included only studies published in English, Dutch or German. Many interventions in national or regional settings are published in other languages such as French or Spanish. These language restrictions limited the extent of the publications reviewed and had consequences for viewing all of the literature in the cultural diversity of Europe.

Conclusion
This systematic review provides a detailed overview of the intensity of parental involvement and BCTs in childhood weight control interventions in children (aged 0-12 years) in the European Union. Low parental involvement was identified in prevention studies, whereas medium and high parental involvements were frequently reported in longterm effective treatment studies. In treatment studies, BCTs were identified more frequently compared to prevention studies. Furthermore, the BCTs most frequently identified differed per time of follow-up. The analysis of parenting skills, generic and specific to lifestyle behaviour offers additional content information and was identified as being high in effective weight control interventions specific during the intervention until Յ1 year post-intervention.

Conflicts of interest statement
The authors declare there is no conflict of interest.