Observational evidence

Although natural hunger-driven eating behaviours are known to regulate the EI of infants and toddlers, environmental cues such as large food PS, have been shown to have the ability to disrupt this innate self-regulation⁽ Reference Rolls, Engell and Birch ^{38 ,} Reference Adair ^{46 –} Reference Mrdjenovic and Levitsky ^{49 )}. In infants and young children (aged less than 11 months), Fox et al. (2006)⁽ Reference Fox, Devaney and Reidy ^{41 )} reported a negative relationship between ED and average PS z-scores, suggesting that as the ED of diet increased, there was a corresponding down-regulation of food intake. In contrast, no such association was shown in toddlers (aged 12–24 months)⁽ Reference Fox, Devaney and Reidy ^{41 )}. Analysis of data from large-scale dietary surveys⁽ Reference McConahy, Smiciklas-Wright and Birch ^{40 ,} Reference McConahy, Smiciklas-Wright and Mitchell ^{50 )} has also identified that large PS across many food groups are positively associated with obesity in young children. In UK adolescents, an increase in the PS of, and EI from snacks was also observed between 1997 and 2005, particularly for all drinks, crisps and savoury snacks and breakfast cereals⁽ Reference Kerr, Rennie and McCaffrey ^{51 )}. Further research has shown that the PS of meals is positively associated with BMI percentiles in boys aged 6–11 years and in children aged 12–19 years⁽ Reference Huang, Howarth and Lin ^{52 )}, and overall PS is consistently positively associated with both EI and body-weight in children⁽ Reference McConahy, Smiciklas-Wright and Birch ^{40 )}. However, these observational data, cannot establish causality.

Intervention studies

To date nine studies have been conducted in children to assess their responsiveness to increasing PS (Table 1). The majority of these studies have been conducted in the USA, and demonstrate that doubling the PS of a macaroni and cheese entree resulted in a 10–40 % increase in EI⁽ Reference Rolls, Engell and Birch ^{38 ,} Reference Fisher ^{42 ,} Reference Fisher, Rolls and Birch ^{53 –} Reference Fisher, Arreola and Birch ^{55 )}, whereas a 4-fold increase in entree PS increased the total meal EI by 61 %⁽ Reference Savage, Fisher and Marini ^{33 )}. These observations were first reported in 5-year-old preschool children, but not in 3-year-old children⁽ Reference Rolls, Engell and Birch ^{38 )}, supporting the self-regulation hypothesis in younger children⁽ Reference Adair ^{46 –} Reference Birch and Fisher ^{48 )}. Subsequent studies, however, have demonstrated significant positive effects of larger PS on EI in children even as young as 2 years⁽ Reference Savage, Fisher and Marini ^{33 ,} Reference Fisher ^{42 ,} Reference Fisher, Rolls and Birch ^{53 ,} Reference Fisher, Liu and Birch ^{54 ,} Reference Looney and Raynor ^{56 ,} Reference Smith, Conroy and Wen ^{57 )}. Moreover, there is evidence to suggest that these effects are sustained for up to 24 h⁽ Reference Fisher, Arreola and Birch ^{55 )}.

Table 1. Studies investigating children's energy intake (EI) response to a change in portion size (PS)

Δ, change; yr, year; B, boys; G, girls; AA, African-American; wk, week; ↑, increase; NS, not significant; H, Hispanics; NR, not reported ↓, decrease.

* Net effect on energy intake (i.e. over total eating occasion/study).

† Study also investigated the effect of self-serve portions on children's energy intake.

‡ Study also investigated the effect of energy density (5·4 and 7·5 kJ/g) on children's energy intake.

§ Study also investigated the effect of energy density (6·7 and 5·0 kJ/g) on children's energy intake.

∥ Study also investigated the effect of energy density (1·8 and 5·0 kJ/g) on children's energy intake.

¶ Study only reported change in food intake (g) which significantly ↓ in 4-yr-old children (179 and 183 v. 256 g), but significantly ↑ in 6-yr-old children (252 v. 325 v. 441 g).

In the only studies investigating the impact of ‘reduced’ PS in children, while no change in EI was observed when the PS of the entree decreased by 25 %⁽ Reference Leahy, Birch and Fisher ^{58 )}, a positive effect on EI was apparent in other studies when age-appropriate PS were assessed⁽ Reference Savage, Fisher and Marini ^{33 ,} Reference Smith, Conroy and Wen ^{57 )}. In the study by Smith et al. (2013)⁽ Reference Smith, Conroy and Wen ^{57 )}, it is also noteworthy that, unlike the 6-year-old children, the younger children (4-year olds) did not respond to the larger PS by eating more. In fact, they consumed significantly less food when served the large PS, in comparison with the smaller (reference) PS of the same meal (183 (sd 76) g v. 256 (sd 75) g; P≤0·01)⁽ Reference Smith, Conroy and Wen ^{57 )}.

These short-term studies, although limited in number, provide supportive evidence that from an early age children are susceptible to PS cues. In the short-term children will immediately respond to increasing PS by consuming more, but there is limited evidence to establish whether they will compensate for this at subsequent eating occasions⁽ Reference Fisher, Liu and Birch ^{54 ,} Reference Leahy, Birch and Fisher ^{58 )}. Although evidence from adults suggests larger PS result in a sustained increased in EI for up to 11 d⁽ Reference Rolls, Roe and Meengs ^{59 ,} Reference Kelly, Wallace and Robson ^{60 )}, no experimental or free-living studies to date have explored the longer-term effects of PS manipulation on the quantity of food consumed in children.

Observational evidence

A recent systematic review commissioned by the US Dietary Guidelines Advisory Committee (2010)^{( 61 )} has concluded that the available evidence consistently supports a positive relationship between ED and body-weight in children and adolescents as well as adults. The evidence for the association in young people was based on four methodologically rigorous longitudinal studies, whose key strengths were: (1) use of objective measures of adiposity (including dual-energy X-ray absorptiometry or doubly-labelled water) rather than reliance on proxy measures such as BMI; (2) mis-reporting of dietary EI was appropriately adjusted for; (3) ED was calculated by recommended methods that excluded all or most beverages, to avoid attenuation of results⁽ Reference Johnson, Mander and Jones ^{62 –} Reference McCaffrey, Rennie and Kerr ^{65 )}. The latter issue is of critical importance in evaluating associations between ED and adiposity⁽ Reference McCaffrey, Rennie and Kerr ^{65 )}. These authors convincingly demonstrated that when the calculation of ED included liquids (i.e. water, energy-free and energy-containing beverages), there was no association between ED and the change in adiposity between baseline and follow-up in their study cohort. In contrast, when the ED was calculated based on solid foods, including milk as a food, the ED of the diet at baseline did positively predict the change in adiposity over time. Furthermore, in this study it was the ED of the total diet rather than any particular part of the dietary pattern (e.g. the ED of snacks) that was associated with the change in adiposity⁽ Reference McCaffrey, Rennie and Kerr ^{65 )}.

Intervention studies

Studies investigating children's responsiveness to changes in ED are shown in Table 2. Although manipulating the ED of a single snack did not significantly affect children's EI at that eating occasion⁽ Reference Looney and Raynor ^{56 )}, reducing the ED of an entree has been shown to reduce children's total EI at that meal⁽ Reference Fisher, Liu and Birch ^{54 ,} Reference Leahy, Birch and Fisher ^{58 ,} Reference Leahy, Birch and Rolls ^{66 )}. These studies were similar in design to the PS interventions described earlier and the effect of decreasing ED on EI was also of a similar magnitude (17–18 % decrease). Further research has shown that this effect on EI can be sustained when the ED of multiple meals were manipulated over 2 d⁽ Reference Leahy, Birch and Rolls ^{67 )}.

Table 2. Studies investigating children's energy intake (EI) response to a change in energy density (ED)

Δ, change; yr, year; B, boys; G, girls; H, Hispanics; wk, week; ↑, increase; NR, not reported; ↓, decrease; AA, African-American.

* Net effect on energy intake (i.e. over total eating occasion/study).

† Study also investigated the effect of portion size (250 and 500 g) on children's energy intake.

‡ Study also investigated the effect of portion size (300 and 400 g) on children's energy intake.

§ Information only provided for twenty-five out of the twenty-six children.

∥ Study also investigated the effect of portion size (150 and 300 g) on children's energy intake.

Reductions in ED appear to have a positive effect on adiposity in the longer-term, particularly when children and parents are provided with more positively focused messages⁽ Reference Epstein, Paluch and Beecher ^{68 )}. Decreases in BMI z-scores were significantly greater in children advised to ‘increase their intake of healthy foods’ compared to those children advised to ‘reduce intake of high ED foods’ at both 12 (−0·30 v. −0.15 zBMI units; P=0·01) and 24 months (−0·36 v. −0·13 zBMI units; P=0·04)⁽ Reference Epstein, Paluch and Beecher ^{68 )}. From a public health perspective this study has revealed key insights about the importance of using positive messages to communicate dietary messages about weight control.

Additive effects of portion size and energy density on energy intakes

Three studies so far have simultaneously manipulated both the PS and ED of children's food intake⁽ Reference Fisher, Liu and Birch ^{54 ,} Reference Looney and Raynor ^{56 ,} Reference Leahy, Birch and Fisher ^{58 )} to investigate their independent and/or additive effects on EI (Tables 1 and 2).

In the study by Leahy et al. (2008)⁽ Reference Leahy, Birch and Fisher ^{58 )}, children's EI at a meal was affected by ED (but not PS). In contrast, Looney & Raynor⁽ Reference Looney and Raynor ^{56 )} demonstrated an effect of PS (but not ED) on children's EI at a single snacking occasion. Leahy et al.⁽ Reference Leahy, Birch and Fisher ^{58 )} showed that a decrease in ED of a meal (6·7 v. 5.0 kJ/g) resulted in a 17 % decrease in EI, irrespective of the PS served⁽ Reference Leahy, Birch and Fisher ^{58 )}. Conversely, Looney & Raynor⁽ Reference Looney and Raynor ^{56 )}, showed that an increase in the PS of a snack (150 v. 300 g) resulted in an 18 % increase in EI, irrespective of the ED of the snack provided. These findings suggest that different strategies to reduce children's EI may be required depending on the type of eating occasion. However, in the study by Fisher et al. ⁽ Reference Fisher, Liu and Birch ^{54 )}, PS and ED had both independent and additive effects on children's EI. Overall, an increase in both PS (250 v. 500 g) and ED (5·4 v. 7.5 kJ/g) resulted in a 34 % increased EI at that meal: an effect that was approximately double that reported when the main effects of each factor were analysed individually.

Together this evidence has prompted researchers to investigate novel strategies to reduce ED of children's meals, without compromising on the PS, so that children will compensate by increasing their intake of other foods.

Novel strategies manipulating portion size and/or energy density to modify children's energy intakes

Similar to the findings in studies conducted in adults⁽ Reference Rolls, Roe and Meengs ^{72 )}, offering children a large portion of a low-energy dense first course (e.g. vegetable soup)⁽ Reference Spill, Birch and Roe ^{73 ,} Reference Spill, Birch and Roe ^{74 )} or larger portions of fruit and vegetables with a meal⁽ Reference Kral, Kabay and Roe ^{75 ,} Reference Mathias, Rolls and Birch ^{76 )}, were effective strategies in both promoting fruit and vegetable intake and decreasing the ED of children's meals. For example, in the study by Kral et al. ⁽ Reference Kral, Kabay and Roe ^{75 )} when the PS of the broccoli and carrots side dish was doubled (75 g v. 150 g), children consumed significantly less of the pasta entree. Although the difference in overall EI at the meal was NS, the ED of the foods which children consumed at the meal did significantly decrease with the larger portions of fruit and vegetables (3·97 (sd 0.08) kJ/g v. 3.72 (sd 0.08) kJ/g; P=0·005). Moreover, others have shown that reducing the PS of high ED foods (e.g. French fries) and replacing with apple slices⁽ Reference Wansink and Hanks ^{77 )} or serving dessert alongside the main meal, as opposed to after⁽ Reference Huss, Laurentz and Fisher ^{78 )}, can attenuate EI irrespective of the PS or choice of main entree. These studies clearly demonstrate the potential to employ more novel strategies for decreasing children's EI, while at the same time not compromising on palatability.

Spill et al. ⁽ Reference Spill, Birch and Roe ^{79 )} also recently demonstrated that incorporating pureed vegetables into multiple meals, to achieve a 25 % decrease in ED, resulted in a down-regulation of preschool children's EI by 12 % (3–5-year-old boys and girls, n 40). Interestingly this reduction in EI persisted over a full day, and moreover, children did not compensate by consuming greater quantities of the un-manipulated snacks and side dishes offered throughout the day, as might have been hypothesised⁽ Reference Spill, Birch and Roe ^{79 )}.

Overall, these studies demonstrate that simply serving or covertly hiding more fruit and vegetables within children's meals can positively influence children's EI, at the same time as encouraging the consumption of these more healthy foods.

Parental influences

Dietary habits formed in early life are predictive of future eating patterns⁽ Reference Birch ^{80 )} and evidence suggests the earlier and broader the experience with food, the healthier the child's diet will be⁽ Reference Skinner, Carruth and Bounds ^{81 )}. Given that obese parents are more likely to have an overweight/obese child⁽ Reference Whitaker, Wright and Pepe ^{82 ,} Reference Lobstein, Baur and Uauy ^{83 )}, arguably this increased risk of childhood obesity is highly likely to be influenced by the epigenetic interactions within the shared family experience, related to both the food and activity environment⁽ Reference Faith, Scanlon and Birch ^{84 –} Reference McCaffrey, Rennie and Wallace ^{87 )}.

Research has shown that a coercive feeding approach, e.g. encouraging children to ‘clean the plate’, may actually have a counterproductive effect on food intake⁽ Reference Birch, McPheee and Shoba ^{88 –} Reference Wansink, Payne and Werle ^{91 )}. Disruption of children's innate ability to self-regulate their EI and therefore encouraging eating in the absence of hunger has indeed been prospectively linked to an increased risk of becoming overweight, albeit only in girls⁽ Reference Fisher and Birch ^{92 )}. Furthermore, this compensation ability has also been shown to decrease with age, particularly in obesity-prone children (i.e. those born to mothers with a pregnancy BMI >66th percentile)⁽ Reference Kral, Stunkard and Berkowitz ^{39 )}.

Further research has shown that the amount of food parents or caregivers serve their children is directly related to the amount of food served⁽ Reference Mrdjenovic and Levitsky ^{49 )} and is also predicted by the amount they serve themselves⁽ Reference Johnson, Hughes and Cui ^{93 )}. In a repeated-measures cross-sectional study (145 parents and their preschool children), Johnson et al. (2014)⁽ Reference Johnson, Hughes and Cui ^{93 )} reported a positive association between the amount of food parents served themselves and the amount of food served to and consumed by their children during three at-home evening meals. These findings add support to studies conducted in the laboratory setting, by showing that in a more natural family environment, children will also respond to larger PS by eating more.

Other environmental cues

Again, similar to findings in studies conducted in adults⁽ Reference Wansink and Cheney ^{94 –} Reference Raynor and Wing ^{98 )}, children's PS can also be influenced by food preferences and visual cues, such as self-served portions and the size of tableware.

The self-service of food or beverages requires some form of conscious effort and thought from an individual about how they are going to serve it, as well as the PS they wish to consume. For example, an individual may be perceptually driven to completely fill a cereal bowl, if they perceive an increased level of hunger at the time⁽ Reference van Ittersum and Wansink ^{99 )}. Analysis of large-scale survey data from 4966 fifth grade US students (aged 10–11 years) showed that children's liking for higher-energy dense foods (e.g. French fries, meats and potato chips) led to a preference for larger-than-recommended portions compared with that for lower-energy dense foods (e.g. vegetables)⁽ Reference Colapinto, Fitzgerald and Taper ^{100 )}. However, the results of experimental studies in this area are equivocal suggesting that self-serving may not provide a ‘one-size-fits-all’ approach to facilitate the avoidance of overeating in response to larger PS⁽ Reference Fisher ^{42 ,} Reference Fisher, Rolls and Birch ^{53 ,} Reference Savage, Haisfield and Fisher ^{101 )}.

The PS of self-served food is also partly influenced by the size of the plate, bowl or glass used⁽ Reference van Ittersum and Wansink ^{99 )}. Early research in this area suggests that children perceive taller containers to hold more of a food/beverage than shorter, wider containers⁽ Reference Piaget, Inhelder and Szeminska ^{102 )}. Wansink & van Ittersum⁽ Reference Wansink and van Ittersum ^{103 )} subsequently showed that this visual illusion caused children to pour and consume significantly more energy from fruit and soft drinks when they were given a short, wide glass compared with a taller narrow glass (of the same volume capacity). Similar findings have since been confirmed for foods served using different sizes of spoons⁽ Reference Fisher, Birch and Zhang ^{104 )} or onto larger adult-sized dinnerware⁽ Reference DiSantis, Birch and Davey ^{105 ,} Reference Wansink, van Ittersum and Payne ^{106 )}, with effects reported to be greater in extraverted compared with introverted children⁽ Reference van Ittersum and Wansink ^{107 )}.