This review examined the evidence for effects of parental and child behaviours on overweight and obesity in disadvantaged groups. Overall, in disadvantaged (and particularly, Indigenous) populations, there remains a relatively scant body of evidence describing the influence on children’s weight of most of the behavioural variables assessed; and that where more evidence exists (e.g. for some categories like parental feeding practices) the variation in study designs, dietary outcomes, measures and other limitations make it difficult to draw strong conclusions. The review has shown that further exploration of causal pathways linking parenting and children’s eating and activity levels with weight in a range of disadvantaged groups is needed.
Associations between parenting behaviours and child eating and weight
Findings on associations between parenting and child or infant weight and dietary intakes were generally suggestive of maternal feeding behaviours being in reaction to children’s weight status [
44,
65]. Parents with heavier children used more restriction [
72,
73] and less pressure to eat [
44,
65,
72] reflecting findings in samples of mixed SEP and ethnicity [
24,
75‐
78]. Parental feeding pressure and restriction were also associated with children’s greater intakes of unhealthy foods and beverages in some [
29,
79], but not all [
44] of the included studies, as well as greater energy intake [
44]. Finally, lower responsive feeding scores were associated with toddler overweight [
73].
Associations between child eating and weight
Of the included studies assessing the importance of children’s diet, there was evidence that consumption of soft drinks [
24,
62], was associated with greater weight [
32]. Juice consumption was also predictive of greater weight in some [
24,
32,
60] studies. There were several contrasting findings, with obese children eating more fruit than non-obese children (this study did not report on vegetable consumption) in one study [
32] whilst another [
59] found that neither fruit nor vegetable intake was related to overweight. Similarly, obese children ate more bread and other carbohydrates as well as total calories in comparison to non-obese children in one study [
32] but in another [
60] grain intake was not linked with weight status. In another study [
59] the measures of diet were related to infant overweight.
Dietary intake (foods and beverages) is associated with overweight and obesity in samples of mixed ethnicity and SEP [
80,
81] although results are not always consistent [
82‐
84]. It is possible that differences in children’s initial risk of obesity may account for some of the observed discrepancies across studies [
24,
62]. Further, a number of the included studies recruited participants from WIC or other clinics where participants would be receiving information on nutrition, weight and health, which may have affected their behaviours or responses [
24,
60]. Parental education level is also found to vary widely, even within the low-income WIC populations, for instance [
65,
72] and may have confounded results.
Associations between breast- and bottle-feeding and child weight
Results on associations between breastfeeding and weight were mixed. Of the six studies reporting on the relationship between breastfeeding and weight status, only two reported an inverse dose–response relationship between breastfeeding and infant overweight [
66,
85] with a minimum duration of three or four months of breastfeeding required. Importantly, these studies had a very large sample size, a long duration of follow-up and were rated high in quality, potentially allowing for modest protective effects of breastfeeding to be detected than in the other smaller studies. However, the positive effect of breastfeeding found in both of these studies was only seen in non-Hispanic Whites and no other racial or ethnic group. This is significant given that in two of the studies reporting no relationship between breastfeeding and overweight [
59,
61] the study populations were entirely Hispanic; and in one of the other studies also finding no protective effect, [
58] just over half of the participants were Hispanic.
The reported effects of breastfeeding on infant weight were often sizeably reduced when a range of other variables were controlled for. For instance controlling for variables previously linked with child weight such as maternal smoking, weight gain during pregnancy or maternal obesity substantially weakened or negated the effects of breastfeeding on overweight or obesity [
66‐
68]. Moreover, breastfeeding has been associated with other health behaviours such as a more positive eating pattern and later introduction of solid foods [
86]; behaviours also independently associated with obesity in childhood [
35]. It is possible that the reported associations between breastfeeding and infant and child overweight seen in disadvantaged families could therefore at least be partly accounted for by other health behaviours associated with greater breastfeeding duration, several of which were not controlled for in the present studies[
87,
88]. A further complication was that only one study [
66] examined differences between exclusive breastfeeding and breastfeeding with concurrent formula feeding ([
59] measured it but did not report results) despite suggestions that a large proportion of low SES mothers breast- and formula- feed concurrently [
69].
It is unclear why the protective effect of breastfeeding in relation to weight status is found almost exclusively in populations of White European descent [
89]. Ethnic differences in the effects of breastfeeding on weight suggest a behavioural, rather than a biological mechanism. There is evidence, for example, that Hispanic, Black and White mothers feed their children differently in terms of breastfeeding, feeding children SSBs, but also restriction and indulgent feeding, for instance [
90,
91]. Other possible explanations are residual confounding [
92], biological or socioeconomic factors [
60]. Furthermore, small numbers of breastfeeding mothers in non-White groups, particularly after six months, mean that studies of breastfeeding in such groups may be underpowered. Nonetheless, studies of socioeconomically heterogeneous groups of children have shown that breastfeeding – initiation, longer duration or exclusivity – may exert a modest protective effect on child overweight [
89,
93,
94]. The reported associations between formula feeding and weight were also mixed, likely because formula feeding was rarely examined and when it was definitions and measures of formula feeding varied [
35,
54,
58,
61,
66,
69]. Despite this, formula feeding appeared to reduce the protective effects of breastfeeding on weight gain, whilst frequency of formula feeds was somewhat predictive of overweight [
69,
95].
Diversity in measures and definitions of formula feeding behaviours and indeed breastfeeding (e.g. whether ever breastfed, breastfed for six months or more, exclusively breastfed and so on) reported in these studies is a limitation. How mothers breastfeed, formula feed or feed from a bottle may be more important than simply whether mothers breast- or formula-feed [
69,
96]. The reasons for this may be that satiety responsiveness or calorific self-regulation, as well as maternal sensitivity to infant hunger and satiety cues are important mediators through which maternal feeding practices may influence weight gain [
69,
97]. However, it is difficult to determine whether these potential behavioural mediators existed in the other studies, as just one of the studies included in this review assessed these factors [
69]. In order to understand the relative importance of breastfeeding and formula feeding as predictors of overweight in disadvantaged infants and children higher quality studies are needed that use well-defined and detailed measurements of breastfeeding and formula- behaviours, including protein content and type, methods of preparation and amounts consumed, and the introduction and provision of solids.
Associations between age of introduction of solid foods and child weight
The age of introduction to solid foods was also rarely assessed, with only four studies examining this as a predictor of infant adiposity [
35,
61,
63,
69]. Just one of these studies, the highest in quality, reported significant associations [
35]. In socioeconomically diverse populations, introduction of solid foods before an infant is four months of age has been associated with greater weight gain [
98]. However, as noted earlier, early introduction of solid foods is also associated with other behaviours linked to obesity including earlier introduction of high fat foods and SSBs [
35,
61,
86]. As these obesity-promoting behaviours cluster and are more prevalent in disadvantaged families, isolating the independent impact of early solids introduction on weight gain was challenging within the available set of studies.
Associations between physical activity or sedentary behaviour and child weight
There was only one study [
33] examining associations between sedentary behaviour or physical activity and children’s weight and both reported significant positive results. The number of hours that children watched TV appeared to be an important correlate of excess weight [
33]. Children from disadvantaged backgrounds watch more hours of TV per day than children from more advantaged backgrounds [
99‐
101], however this relationship may be complicated by the influence of other demographic factors such as ethnicity [
33]. It is also possible that greater TV viewing by low SES children is more strongly associated with obesity in these children because TV viewing is associated with other obesity-promoting behaviours, such as consumption of energy-dense foods [
102]. However, due to the small number of studies included in the review examining the influence of sedentary behaviours or physical activity in disadvantaged groups, more research is required before any firm conclusions can be drawn.
Documenting the mechanisms that may explain the development of obesity-promoting habits amongst parents and young children from low socioeconomic or Indigenous backgrounds is fundamental to understanding how to prevent them. However a large number of the studies included in this review were of cross-sectional design, therefore limiting their ability to examine causality, including bi-directional effects. Bi-directionality is imperative to understand in parent–child interactions as parenting is at least partially reactive to the child’s extant characteristics such as weight status, temperament and eating behaviours [
103,
104]. Moreover, because cross-sectional studies are unable to distinguish between incidence and persistence of overweight in children, testing associations between, for instance, dietary intake and weight in a sample that includes children who are already overweight is difficult. Prospective studies utilising validated and culturally reliable measures of key variables (e.g. parent feeding behaviours) are needed if we are to further reveal causal relationships in parent–child feeding (and activity) relationships in disadvantaged families. Additionally, future studies utilising large population-based samples in socioeconomically and ethnically diverse groups, measuring the possible pathways of effect and testing for mediators, whilst controlling for confounders, including mediator-outcome confounding, are needed.
This review has also highlighted that research in this area is hindered by the availability of appropriate or adequate measurement tools, a challenge that has been highlighted previously [
74,
105,
106]. Many of the tools utilised in the included studies were either purpose-developed with few data on validity and reliability [
48,
64], or were established tools that were developed in different ethnic and socio-economic groups (primarily high advantage White families) than in which they were applied [
65,
107]. The appropriateness of these tools to collect data in disadvantaged ethnic minorities or Indigenous populations is unknown. Children in White, higher income families likely have very different feeding environments (e.g. foods available, parental feeding behaviours and beliefs) and both the types of relevant behaviours and their measurement may not be appropriate for other groups. Indeed the CFQ [
75], a widely used tool to measure parental feeding behaviours [
108] appears better able to detect associations between parental feeding and child weight in all-White mother-daughter samples [
109,
110] than in other samples [
111,
112]. Similarly, clear definitions of each of the concepts (e.g. early introduction of solids, breastfed, restriction) under study were often lacking and appeared to differ across studies. Future research aiming to develop and apply tools more specific to the target populations, different food groups and eating contexts may help tease out relationships.
Obesity prevention interventions may be less effective in disadvantaged populations [
113], likely because they have not been tailored towards the specific requirements of disadvantaged families. The present review has highlighted that there is only a small evidence base explaining causal relationships between parent and child behaviours and children’s weight status upon which interventions tailored to disadvantaged groups could be designed. Overall, only a small number of factors that could affect weight gain in disadvantaged families has been considered. The focus to date has been on the duration of breastfeeding, socio-demographics, dietary intakes and a selected few parental feeding behaviours such as restriction, control and pressure to eat [
108,
114]. Consideration of other factors that affect weight such as other parental feeding behaviours, or how children are breast- or formula-fed (e.g. feeding to appetite) has seldom been undertaken in disadvantaged groups. Additionally, because many of the parent and child behaviours associated with overweight co-occur [
115], studies that isolate or control for confounding are needed if we are to elucidate mechanisms of effect.
There are a number of limitations to this review. The search strategy was limited to parental feeding, child eating, physical activity and weight although there are other factors that influence children’s weight trajectories. We used a limited number of search engines, did not examine grey literature, and limited our search to English language and therefore may have missed some relevant studies. We did not examine all possible relationships that may affect infant or child weight, including relationships between child sedentary behaviour and child eating. Furthermore, we examined only two disadvantaged groups– low income and Indigenous and restricted our geographic search to the OECD. Finally, we were unable to perform a meta-analysis due to heterogeneity in population types, measures and outcomes.