Mother–child feeding interactions in children with and without weight faltering; nested case control study

Abstract

Aims: To explore whether the Mellow Parenting assessment system can detect any difference in parent–child meal time interaction between children with weight faltering (failure to thrive) and normally growing children. Subjects and methods: Thirty mother–infant dyads with weight faltering and 29 healthy controls nested within the Gateshead Millennium prospective cohort study were assessed at mean age 15.6 months (range 13–20). Video-tapes of two standardized meals per child by a researcher blind to infant health status were analysed using a simplified version of the Mellow Parenting Coding System (MPCS), an all events measure of maternal–child interactivity. These were linked to questionnaire data on eating behaviour and growth held on the children. Results: The MPCS had good inter-rater reliability (0.82) and coherent inter-relationships between coding domains. During case meals there were significantly fewer positive interactions overall: cases median 81.5 (IQR 4–496); controls 169.5 (40–372) and within all the commonly observed domains (Anticipation (p = 0.013), autonomy (p = 0.003), responsiveness (p = 0.005) and cooperation (p = 0.016)). There were only low levels of distress and control or negative behaviours and no significant differences were found in these between the groups. The case infants had significantly lower reported appetite by the age of 4 months and higher reported avoidance of feeding at the age of 8 months than controls. Conclusions: Mothers of weight faltering infants generally showed fewer interactions with their infants at mealtimes. It is not clear whether this is causal or simply a maternal adaptive response to their child's eating behaviour.

Introduction

It is generally accepted that maternal child interaction is of high importance and it is in the process of feeding that this interaction is most readily evident. Chatoor argued that the feeding interaction is one which is central in establishing a parent's dyadic relationship with his or her child (Chatoor, 2000) and other authors have emphasized the importance of concepts such as “sensitivity” and “responsivity” (Ainsworth, Blehar, Waters, & Wall, 1978), “connectedness” and “mirroring” (Stern, 1985) in characterizing optimal parental interaction. Since feeding is such an important form of interaction, it has long been suggested that disruption of the maternal child interaction may play a key role in the causation of weight faltering or ‘failure to thrive’. This was particularly so early on, when studies in referred populations linked ‘failure to thrive’ closely to maternal disturbance or neglect (Oates and Hufton, 1977, O’Callaghan and Hull, 1978). A number of early studies also suggested that infants with ‘failure to thrive’ had higher rates of insecure attachment (Ainsworth et al., 1978, Gordon and Jameson, 1979) less physical and verbal interaction (Pollitt, Eichler, & Chan, 1975) and lower rates of maternal sensitivity, emotional expression, responsiveness, acceptance, and cooperation (Casey et al., 1984, Drotar et al., 1990). However these studies often used highly subjective definitions and most were of children referred for treatment, making selection bias likely.

In more recent years studies in unselected populations have found little evidence of a link between slow weight gain and maternal disorder (Boddy and Skuse, 1994, Drewett et al., 2004) or even poverty (Blair et al., 2004, Wright et al., 1994b). Few rigorous studies of maternal child interaction have been done in un-referred populations. One small study described mothers of cases as displaying more negative affect, and less likely to give instructions, communicate or socialize at mealtimes (Hepinstall et al., 1987), but another larger population based study found no difference in the level of interaction (Skuse, Wolke, & Reilly, 1992).

Observational research focusing on general parent–child interactions has been shown to distinguish problem dyads (Dowdney et al., 1985, Mills et al., 1985, Puckering et al., 1995) but it is time consuming and demanding. A number of studies have used various mealtime observational schedules, but none are in widespread use. Probably the most widely used are the infant Feeding Scale (Chatoor et al., 1997) and the Cognitive Growth Fostering subscale of Nursing Child Assessment and Teaching Scale (Barnard, 1978). The Parent–Child Early Relational Assessment has also been used in at least two studies (Hutcheson et al., 1993, Black and Nitz, 1996), but is itself unpublished. The Bob and Tom Method of Assessment (BATMAN) (Klesges et al., 1983) has also been used in at least two studies. However these all rely on a global observation of the whole meal, an approach which tends to be culturally specific and is difficult to standardise.

The Gateshead Millennium Baby Study (GMBS) is a population based birth cohort, set up to study various aspects of feeding and growth prospectively (Parkinson et al., 2010). Work in the cohort as a whole has already demonstrated that weight faltering is not clearly associated with either maternal disturbance or poverty (Wright, Parkinson, & Drewett, 2006b) and that children with slow weight gain have lower appetite (Wright, Parkinson, & Drewett, 2006a). A case control study nested within this cohort set out to directly observe infant behaviour in infants with weight faltering. This used an existing all events coding scheme which describes a child's feeding behaviour in terms of feeding and eating acts (e.g., accept, refuse, give, self-feed) with high reliability (>0.87) (Drewett et al., 2003, Parkinson and Drewett, 2001). An earlier version of this scale had shown some differences in a population of children referred with failure to thrive compared to controls (Drewett et al., 2003). On this occasion no difference was found in the number of feeding acts between children with weight faltering and controls, but they did consume less energy (Parkinson, Wright, & Drewett, 2004).

However, this coding scheme only counted child and parent feeding acts and not the process or style of interaction, which might be expected to be more important. In seeking a measure of interaction we became aware of the Mellow Parenting system, an all events observational tool designed to examine parent–child interaction, which had been developed by CP. This has been used for assessment in a group intervention for parenting problems (Puckering et al., 1996, Puckering et al., 1994) but not for children with growth or feeding problems.

We thus set out to develop a shortened version of the Mellow Parenting system suitable specifically for use at mealtimes and then to use it to reanalyse the video-tapes from this study, both to explore it utility in this context and to explore whether interactive behaviour differs in children with weight faltering compared to controls. We hypothesized that, at mealtimes, parents of children with weight faltering would show fewer positive interaction and more negative interactions compared to controls.