Effects of Parent and Child Behaviours on Overweight and Obesity in Infants and Young Children From Disadvantaged Backgrounds

Systematic Review With Narrative Synthesis

Catherine Georgina Russell; Sarah Taki; Rachel Laws; Leva Azadi; Karen J. Campbell; Rosalind Elliott; John Lynch; Kylie Ball; Rachael Taylor; Elizabeth Denney-Wilson


BMC Public Health. 2016;16(151) 

In This Article


Study Selection Criteria

Studies were selected for inclusion in the review if they addressed one of the following pathways, nominated due to previously reported associations with overweight or obesity in a variety of populations (see Fig. 1): (A) between parenting behaviours and child eating, (B) between parenting behaviours and child activity (physical activity or sedentary behaviours), (C) between children's eating and children's weight, (D) between child activity levels (including sedentary behaviours and physical activity) and child weight and (E) between parenting behaviours and child weight. Child eating was defined as dietary intake (including breast milk or formula), diet patterns, intakes of specific foods or beverages, food choices, food preferences, eating styles and eating behaviours. The age at which children started consuming solid foods was also included as a 'child eating' variable.[41,42] Parenting behaviours included specific feeding behaviours (e.g. using food as a reward, modelling) and general parenting behaviours. All studies assessing pathways C, D and E had to provide measures of anthropometric status.

Figure 1.

Studies were included in the review if they assessed any one or more of the selected pathways through which parents and children may affect children's overweight or obesity

To be included, studies needed to focus on low socio-economic or Indigenous groups or with the overall results stratified by socioeconomic or Indigenous group, or to report on interactions between socioeconomic or Indigenous group and the pathway variables. Socioeconomic disadvantage was defined based on families being described as having low income, low level of education or occupation, and/or living in an area defined as disadvantaged using aggregate indicators. Indigenous was defined using Cobo's criteria.[43] As our focus was on early life, only studies of children aged 0–5 years were included. Studies focusing on weight loss or with children with underlying medical conditions were excluded.

No limitations were placed on publication year, although studies needed to be published in English and use human participants. However, we limited our search to OECD countries as our focus was on the effects in high-income countries.[44,45] Studies included had to be primary studies or papers presenting secondary data analyses from these studies, and be published in a peer-reviewed journal or edited book. The full search strategy is available from the authors.

The Search Strategy

Between June 2013 and November 2014, we conducted a systematic literature search of ten electronic databases (Academic Search Complete, PsycINFO, CINAHL, Medline, EMBASE, Health Collection, Google Scholar, Joanna Briggs Institute, Scopus, Proquest), guided by the PRISMA statement.[46] The search terms child* OR infant* OR famil* OR parent* OR mother OR maternal OR father/Indigenous* OR socioeconom* OR socio-econom* OR poverty OR disadvant* OR unemploy*/obes* OR overweight OR weight/exercis* OR play OR activ* OR sedent* diet* OR nutrition OR Eating behavi* OR food/breastfe* OR "breast fe*" OR feeding OR Parent*/Empirical research OR clinical trial* OR randomi* OR qualitative OR cohort study OR quantitative were used to identify relevant literature. We also searched via subject headings in PsycINFO, CINAHL and Medline. Limiters were set to exclude literature reviews, systematic reviews, meta-analyses, letters, and editorials. We also visually scanned reference lists from relevant studies and undertook citation searching.

The Study Selection Process

Articles retrieved through the electronic search process were entered into an EndNote bibliographic database. A process of electronic elimination of duplicates subsequently took place. Titles and abstracts were then screened by two authors (CGR and ST) and papers were classified as either (A) appearing to meet the selection criteria, (B) meeting selection criteria difficult to determine, or (C) excluded articles (did not meet selection criteria or duplicate). Full document texts from the potentially eligible groups (A and B) were examined. Any discrepancies between the two authors were resolved by discussion. A third researcher (EDW) checked 10 % of the titles and abstracts classified as excluded studies, to check reliability of the screening process.

Data Extraction and Synthesis

Two researchers (CGR and ST) extracted key data from each of the papers using a developed template. Extracted data covered bibliographic information, study background and aims, setting, inclusion and exclusion criteria, recruitment strategy, when and how data were collected, sample size and response rate, characteristics of parents (e.g. age, gender, education level, employment status, income, ethnicity, BMI), child/infant characteristics (e.g. age, gender, weight, whether breast- or bottle-fed), outcomes measured and relationships tested, covariates, definitions and measures of key variables, results, author's interpretation and conclusions. In instances where publications reported multiple dependent variables only those results fitting the inclusion criteria were extracted. A cross check of 10 % of the extracted studies was undertaken by a third researcher (EDW) to ensure accuracy of data extraction. Differences in data interpretation and extraction were resolved through discussion.

Quality Assessment

The quality of the selected studies was appraised using the Mixed Methods Appraisal Tool (MMAT)[47] independently by one of the authors (RE), with 10 % of the sample cross-checked by another author (EDW). The MMAT was selected in order to accommodate the various study methodologies and to enable us to meet the aims of the review. The MMAT comprises two screening questions (applied to all studies) and four questions for five broad study methodologies; qualitative, quantitative randomized controlled trials, quantitative non-randomized controlled, quantitative descriptive and mixed methods. Mixed methods studies are evaluated using the appropriate criteria for the study methodologies employed in the particular study as well as the mixed methods criteria and the methodology with the lowest score is the final quality rating. A quality rating is derived from the number of 'yes' responses to the criteria; this is either reported as a raw score or expressed as a percentage of the total number of criteria for that study methodology. Quality ratings range from a raw score of zero to four, (0–100 %) where zero indicates that none of the criteria were met and four (100 %) indicates that all criteria were met. The MMAT has been extensively tested for reliability and validity.[47] A quality rating was expressed as a percentage and derived by dividing the number of 'yes' responses by the number of applicable criteria multiplied by 100. To ensure accuracy of the quality assessment of the sub-studies in our review the main study was also located and the protocol thoroughly examined.