The influence of the maternal peer group (partner, friends, mothers’ group, family) on mothers’ attitudes to obesity-related behaviours of their children

This study uses a retrospective cohort study design, with data from a cohort of first-time parents involved in the Melbourne Infant Feeding, Activity and Nutrition Trial (InFANT) Program, a cluster-randomised controlled trial testing an intervention (six two-hour sessions delivered by a dietitian) that aimed to improve children’s diets, promote active play and reduce television viewing. Data for the current study was obtained from parent questionnaires administered at a two-year follow-up of this trial cohort (including parents from both treatment and control groups) when the children were aged 3.5 years [24].

The InFANT intervention protocol has been previously reported [25] but in brief, the program was delivered quarterly over a 15-month period from June 2008–February 2010 to 542 parents (equal numbers in control and intervention groups) who were members of 62 randomly selected first-time parent groups in Melbourne, Australia. Children were aged approximately 4 months at baseline and 20 months at the end of the intervention. Parents were eligible for inclusion if they were consenting, first-time parents, who were members of a participating new parents group’ and able to communicate in English [25].

Participants (n = 307) were followed up in 2012–13, approximately two years after the intervention ceased to assess long-term effects [24]. Ethics approval was granted by the Deakin University Human Research Ethics Committee (2007–175). Written informed consent to participate was obtained from all participants.

A questionnaire was mailed to participants who were asked to complete it prior to a home visit by a research assistant. Child height and weight were measured by research assistants at home visits. BMI z-score was calculated using the ‘zanthro’ add-on in Stata which uses WHO sex-specific growth standards [26]. Maternal height and weight measurements were also taken at home visits where possible or otherwise self-reported [24].

The results presented here relate to questions about external influences on child feeding, physical activity and TV viewing. These were purpose-designed questions (provided as Additional file 1) on which test-retest reliability and internal reliability analyses were performed. In relation to a) mothers’ group, b) friends, c) partner, and d) other family members (all potential influencers of behaviour), mothers were asked to rate on a four-point scale (where 1 = no influence, 4 = major influence) the following questions: 1) How much of an influence on your attitudes towards feeding your child have the following been? 2) How much of an influence on your attitudes towards your child’s television viewing have the following been? and 3) How much of an influence on your attitudes towards your child’s activity levels have the following been?

Maternal education was self-reported as: no formal education, year 10 or equivalent, year 12 or equivalent, trade/apprentice/certificate, certificate/diploma, University degree or higher degree.

For assessment of peer influence by education level, maternal age and maternal weight, dichotomous categories were created based on a) those with a University degree or higher, vs. those with lower levels of education, b) mothers less than 35 years of age vs. those over 35 years and c) healthy weight vs. overweight/obese (BMI ≥ 25 kg/m²) respectively.

To examine the association between peer influence and related behaviours, questions were asked about the following: 1) duration of breastfeeding, 2) age of introduction of solids, 3) time child spends outside on an average day over the last week, and 4) average number of serves of fruits (per day), vegetables (per day) and hot chips (per week, includes hot chips, French fries, wedges, fried potatoes) consumed by the child.

Demographic statistics were calculated as mean (standard deviation or min/max), percentage or n as appropriate. The level of influence of each of mothers’ group, friends, partner and other family members was calculated (percentage reporting no influence, little influence, some influence or a major influence respectively, with 95% confidence intervals) for each of child feeding, child television viewing and child physical activity. Stratified analyses were conducted according to whether mother/child dyads were in the control or intervention arm of the InFANT program, whether they had attended a mothers’ group meeting in the last four weeks, and based on maternal education, age and BMI. Chi square tests were used to test for differences in the level of influence reported for each group. One-way analysis of variance (ANOVA) was used to test for associations between peer influence ratings and behaviours expressed as continuous variables (breastfeeding duration, time of introduction of solids, time spent outside) and t-tests were used to assess associations with behaviours expressed as categorical variables (consumption of various foods). Analyses of peer influence on behaviours were conducted for the whole sample, and then for the control group only to account for the potential impact of the InFANT intervention. Stata version 14.0 (StataCorp, College Station, TX, USA) was used for analysis.

The self-reported influence of different relationships on mothers’ attitudes toward each of child feeding, television viewing and physical activity behaviours is presented in Table 2. Partners were the strongest influence on attitudes toward each behaviour. Other family members (i.e. mothers’ parents, in-laws, aunties/uncles, cousins and siblings) were considered to be the next strongest influence, followed by friends. Overall, participants rated their mothers’ group as the weakest influence on their attitudes toward their child’s obesity-related behaviours. Of all the behaviours examined, attitudes toward child feeding were most likely to be influenced by mothers’ group peers, followed by child physical activity. Attitudes toward child television viewing was reported as being least likely to be influenced by any relationships.

Significant differences in who mothers rated as the strongest influence on attitudes toward obesity-related behaviours were found according to maternal education (Table 3). University educated mothers were more likely to report that their mothers’ group, friends and partners were an influence on their attitudes toward child feeding and child activity behaviours, than were non-University educated mothers (all p < 0.02). University educated mothers were also more likely to rate partners (p = 0.001) and other family (p = 0.032) as strong influences on their attitudes toward their child’s TV viewing compared to non-University educated mothers.

When comparing treatment and control groups, mothers who took part in the InFANT intervention were more likely to rate their mothers’ group as having influenced their attitudes toward their child’s feeding than their control group counterparts (p = 0.045). Of those in the intervention group, 34.4% rated their mothers’ group as some or a major influence on attitudes to feeding compared to 23.2% of controls. Mothers in the treatment and control groups were similar in their rating of the influence of partners, friends and other family on their attitudes toward all behaviours (all p > 0.05).

The self-reported influence of others on attitudes toward obesity-related behaviours according to recent attendance at a mothers’ group meeting is presented in Table 4. Those who had attended a meeting in the past four weeks rated their group to be a greater influence on child behaviours than those who had not attended during this period (p < 0.001 for child feeding, child TV viewing and child physical activity).

The only behaviours to be associated with peer influence using data from the total sample were a) breast feeding, where partner influence on attitudes was significantly associated (p = 0.004) and b) consumption of hot chips, where influence of other family on attitudes was significantly associated (p = 0.012). Mothers who rated their partners as either some influence on attitudes toward child feeding (average breastfeeding duration = 9.0 months) or a major influence (8.1 months) had significantly longer breastfeeding duration than mothers who rated partners as either a little influence (6.2 months) or no influence (6.7 months). No association was seen between the influence of any peer groups (mothers’ group, friends, partner, other family) on maternal attitudes, and other behaviours examined, including age of introduction of solids, amount of time child spends outside or the consumption of fruits or vegetables (all p > 0.05). When analyses were conducted using participants from the control group only (in order to exclude any potential impact of the InFANT intervention on results), the influence of partners on breastfeeding duration remained (p = 0.019), while the influence of other family on hot chip consumption was no longer significant (p = 0.31). Using the control group participants only, the only other behaviour to be influenced by peers was consumption of vegetables, with partner influence on attitudes being associated with consumption (p = 0.002).

A key strength of this research is that it includes child feeding, physical activity and sedentary behaviours. Limitations include the self-reported nature of some data collected, and the educated sample which may reduce the generalisability of the findings. Collection of data on the influence of others was years after collection of data on breastfeeding and introduction of solids due to the retrospective nature of the study design, and responses may have differed if these data were collected simultaneously. It is possible that recall/recency bias may have been a factor in mothers from active mothers’ groups rating the influence of these groups as stronger than mothers who were not actively meeting with mothers’ groups. No similar research has been undertaken previously, with further refinement of this tool potentially also including questions on the direction of the social influence – whether it was seen as being positive or negative. This could be particularly important given that even though here we found that associations between partner influence and breastfeeding duration were positive, separate qualitative research among 26 mothers from the InFANT Program has found the influence of partners on later feeding practices was frequently negative [19]. How to identify mothers who are at risk due to either a lack of positive social influencers or the presence of negative influencers could be an important topic for further research.