Parents' beliefs about appropriate infant size, growth and feeding behaviour: implications for the prevention of childhood obesity

The relationship between breast feeding and obesity risk in childhood has been investigated in a number of studies, with contradictory findings[6]. A systematic review and meta analysis conducted in 2004, concluded that breast feeding has a small but consistent protective effect against the development of obesity during childhood[6]. There is emerging experimental evidence that the higher protein content of formula milk may be partially responsible for greater weight gain[7]. There is also observational evidence that early weaning onto solid foods (< 4 months) is significantly associated with overweight or obesity at 3 years[8]. A UK longitudinal cohort study found that infants fed with formula milk were introduced to solid foods earlier than breast fed infants and were less likely to have consumed fruit and vegetables[9]. These study results require cautious interpretation as the cohort consisted exclusively of white, singleton children, who were from more affluent backgrounds. There may be other differences between mothers who choose to breastfeed and those that do not (e.g. breastfeeding mothers may be more health conscious, more responsive to their infant and have a higher socio-economic status (SES)).

Parental response to infant temperament is a risk factor for childhood obesity[10]. A qualitative study of 14 low income mothers with overweight young children reported that food was used to soothe fussy infants, calm temper tantrums and shape behaviour[11]. In another study infants with a temperament dimension described as "fear" in relation to their acceptance or rejection of new objects or people were significantly slower to gain weight than those described as "distress to limitations" which is characterized by frustration, reduced sleep, excessive crying and fussiness. The authors speculate that parents who perceive their infants to be fussy or frustrated may use food as comfort which may account for the differences in weight gain between the two groups[12].

Parental perceptions about appropriate size, growth and feeding patterns play an important role in the development of childhood obesity. Baughcum and colleagues[11], in the study described above, reported that mothers had strong beliefs that heavy infants were desirable as weight was the best marker of health and successful parenting. In a later study, Baughcum and colleagues[13] assessed maternal feeding practices and beliefs using feeding questionnaires for infants (n = 453) and pre-school children (n = 634). They described different feeding behaviours between high and low income mothers. Low income mothers reported greater concern about their child's hunger, a lower tendency to use food to calm and greater difficulty feeding their children. They also reported pushing their child to eat more and engaging in more age-inappropriate feeding.

Parental beliefs and understanding are crucial determinants of infant feeding behaviour, therefore any intervention will need to take account of these views. This study aimed to explore UK parents' beliefs concerning their infant's size, growth and feeding behaviours and their receptiveness to early intervention aimed at reducing the risk of childhood obesity.

The objectives were:

Six study sites were selected with the intention of ensuring a higher probability of recruiting parents whose infants were at risk of developing obesity. The UK National Child Measurement Programme, which started in 2005, routinely measures children aged 4-5 years (Reception) and 10-11 years (Year 6),[14] and the intention was to use this data to select five study sites where the risk of childhood overweight and obesity was high and one site where the risk was low. The inclusion of a low risk site was to determine whether the views of these parents were vastly different from those parents who resided in a high risk site. Two urban sites with high rates of overweight and obesity were selected using this data (Table 1)[15]. Unfortunately National Child Measurement data was unavailable for the surrounding rural population. As social disadvantage is a risk factor for childhood obesity[8], the Index of Multiple Deprivation (IMD) (2007)[16] of the health centre location was used as a proxy measure for childhood obesity risk. Three further study sites were selected with high IMD scores and one with a low score (Table 1). Permission for the multi-centre study was obtained from the Nottingham Research Ethics Committee 1 and Nottinghamshire and Lincolnshire Primary Care Trusts (PCTs). The study was registered with the Primary Care Research Network (PCRN).

At each site, 8-10 parents attending child health (baby) clinics were approached by Health Visitors (UK specialist practitioners in public health) and asked to participate in the study. Inclusion criteria were that parents should be living in the study site and have an infant less than one year of age, during 2008-2009. One focus group took place in the local health or children's centre at each of the six study sites. Participating parents were offered a £20 store voucher and travel expenses.

Each focus group included between 4 and 9 parents, together with a facilitator and note taker. The focus group sessions followed the guidelines set out by Krueger and colleagues (1994)[17]. Participants were provided with a resume of the study's aims and objectives and given an opportunity to ask further questions prior to providing written informed consent. Participants were informed they could withdraw from the focus group at any time but, as their views would be audio-recorded along with other participants; their data could not be erased until transcription.

Participants were asked to complete a short, anonymous questionnaire detailing demographic details (by postcode) and ethnicity, information about family size, infant feeding history, parental weight and height. The facilitators offered parents help to complete the questionnaire, where appropriate. Each parent was provided with a set of ground rules prior to the focus group session. The facilitators were aided by a semi-structured topic guide (Table 2). The guide was revised following each focus group as issues important to the participants emerged. In addition to responding to the topics raised by the facilitators, participants were encouraged to express their own views and to lead the discussion in other directions within the remit of the research topic.

The focus groups were audio recorded and transcribed verbatim. Transcription was performed immediately after each focus group to enable identification of individual parents. Participants were provided with an identification number on the transcribed interviews. Notes were used to track each participant's contribution.

Data from the questionnaire was analysed using descriptive and frequency statistics facilitated by SPSS software[18]. The interview data were approached from an interpretative perspective. This involves the researchers looking beyond the actual spoken words in an attempt to understand the meaning behind the discussions[19]. All of the transcribed focus groups were entered into NVIVO 8.0[20]. Two researchers independently coded three of the focus group transcripts each using an inductive approach[19]. The codes were then arranged into related clusters to form a coding frame. The coding frame was then explored for linked and explanatory themes. A coding book was developed following the guidelines set out by Boyatzis (1998) which comprised codes, definitions and examples[21]. An inter-rater reliability check conducted on 10% of the coded data revealed 88% agreement. The coding frame was further revised between two coders and the main themes agreed. The data were then re-analysed as appropriate within the revised framework of themes and sub-themes.

Thirty-eight participants consented to take part in the study (n = 36 female, n = 2 male). Participants' ages ranged from 19-45 years (mean 30.1 years, SD 6.28). Body mass index (kg/m²) was calculated for each participant using self reported weight and height and (BMI < 25), with 12/38 classified as overweight (BMI 25-29.99) and 8/38 as obese (BMI >30) (Table 3).

The mean age of the infants was 5.51 months (range 1-11 months) and most (27/38) were breast fed for at least a month (mean 4.48 months, SD 2.58, range 1-10 months) (Table 4).

The descriptive data, including parents' height and weight, was collected by parental self-report. Self-reported weight measures have been reported as inaccurate with trends for under-reporting for weight and BMI and over-reporting for height[38]. This may mean that the participants in our study were more overweight or obese than reported. Caution is needed, however, since all the female participants had given birth within the previous year and the weight reported may not reflect their typical weight. Qualitative methods were used to understand parent's explanations for their infant's size, growth and diet. The results represent an underlying social reality for the participants and generate various hypotheses which contribute to our understanding of the theory around infant feeding practices[39]. An attempt was made to include study sites with high and low rates of childhood obesity in order to recruit more parent participants whose infants may be at high risk of rapid weight gain. However, local and national data on rates of childhood obesity were incomplete at the time the study was set up and the IMD of the health centre was used as a proxy measure[16] to select four of the six study sites. The median IMD for each of the six study sites was retrospectively re-calculated using the parent participant's individual postcodes (Table 1). This revealed that in some of the study sites the parents who participated in the focus groups were living in more affluent locations than predicted. We cannot, therefore, say with certainty that the areas selected were those where infants were at greatest risk, or in the case of one area lowest risk. Participants' views were more homogenous than expected which may be due to a group effect. It may also have reflected that, despite the wide sampling, personal commonalities between the participants outweighed the impact of their respective geographic locations. Participants were volunteers and the details of those who were approached but did not participate were not collected. Some of the views expressed may, therefore, be different from non-participants in some meaningful way.

Both the researchers, who undertook the field work for the study, are qualified health visitors. The 'health-care-professional-as-researcher' is uniquely placed as they are already immersed in the field and have important insights into patient issues[40]. However, they can add a significant power imbalance and raise difficulties for participants in how free they feel to be open and/or critical[41]. At the time of the study the researcher were not practising and this information was shared with the participants. This research was undertaken in areas where the researchers have never practised and, therefore, could not have had a professional relationship with the parents.