Observational studies

Numerous studies have identified that EI increases with exposure to larger PS, showing a potential risk factor for overweight and obesity^{(Reference Rosenheck60)}. The increase in consumption rates due to larger PS is up to 25% during lunch, and up to 45% for snacking^{(Reference Young and Nestle24)}. Longitudinal and cross-sectional studies have proposed a positive association between the consumption or occurrence of eating outside the home and increased BMI or weight gain^{(Reference Binkley, Eales and Jekanowski61–Reference Kant and Graubard64)}. Representation of an appropriate amount to consume is structured by PS encountered in various places, such as supermarkets, restaurants, marketing images or the home^{(Reference Marteau, Hollands and Shemilt65)}. Nevertheless, the wide diversity in the characteristics of the food sector within the data leads to disagreement. Some associations were not identified^{(Reference Orfanos, Naska and Trichopoulos66,Reference Marin-Guerrero, Gutierrez-Fisac and Guallar-Castillon67)} , or they were identified only in women^{(Reference Kant and Graubard64)} or only in men^{(Reference Bezerra and Sichieri68,Reference Naska, Orfanos and Trichopoulou69)} . Furthermore, serving the same food amount at various junctures established a notable consistency in the food type and the amount consumed, especially in individuals who learn the PS they need to feel satiety (‘previous experience/expectation mechanism’). The choice of PS may be influenced by prior experiences^{(Reference Steenhuis and Poelman33)}. For instance, the PS chosen and consumed at a later time depends on prior experiences with the degree of fullness produced by a food, in both adults^{(Reference Steenhuis and Poelman33)} and children^{(Reference Brunstrom70)}.

Recent data from the Kantar Market Research Group revealed that during the coronavirus disease 2019 (COVID-19) pandemic eating patterns changed considerably globally, highlighting an increase in sugar, sweeteners, herbs, seasonings, olive oil and alcohol purchases for home consumption, whereas other purchases such as health and beauty products declined⁽⁷¹⁾. This speculates that the population worldwide, including the UK, was treating themselves to food. A recent scoping review indicated that the lockdown had positive (increased consumption of fresh produce and home cooking) and negative impacts (unhealthy snacking, mental health issues, physical inactivity and weight gain) on dietary practices globally^{(Reference Bennett, Young and Butler72)}. A study commissioned by the UK Food Standards Agency (FSA) identified the same beneficial changes in eating habits in households an increase in unhealthy snacking^{(Reference Lasko-Skinner and Sweetland73,74)} . The latter was indicated in another UK study^{(Reference Scott and Ensaff75)} and the European Institute of Innovation and Technology⁽⁷⁶⁾. Furthermore, it is indicated that the existing food security issues experienced in Brexit^{(Reference Barons and Aspinall77,Reference Ranta and Mulrooney78)} , post-Brexit^{(Reference Ranta and Mulrooney78)} and the COVID-19 pandemic^{(Reference Lasko-Skinner and Sweetland73,74,Reference Barons and Aspinall77,Reference Rivington, King and Duckett79)} have changed the consumption patterns, highlighting a change in food trends.

Experimental studies

Various experimental studies have shown a potential long-term association between fast-food or takeaway consumption, and high EI^{(Reference Rosenheck60,Reference Ello-Martin, Ledikwe and Rolls80,Reference Zlatevska, Dubelaar and Holden81)} , independent of individuals’ satiety levels^{(Reference Diliberti, Bordi and Conklin82)}. This suggests that adults ignore both hunger and satiety signs in the presence of external cues, such as large PS^{(Reference Kral, Roe and Rolls83,Reference Wansink, Painter and North84)} , especially when eating out with others^{(Reference Ello-Martin, Ledikwe and Rolls80,Reference Hermans, Larsen and Herman85)} . It is proposed that individuals learn to eat in the absence of hunger as children and continue to adulthood with this eating behaviour^{(Reference Fisher, Rolls and Birch39,Reference Rolls, Roe and Meengs86)} , indicating a major determinant of food consumption^{(Reference Nielsen and Popkin14,Reference Jeffery, Rydell and Dunn87)} . Indeed, continuous exposure to more than 6 months of high-energy lunch consumption led to a significant increase in EI and weight gain^{(Reference French, Mitchell and Wolfson88)}. A recent scoping review of randomised-controlled trials (RCT) and quasi-experimental studies highlighted a significant effect in lowering food consumption when offered a single smaller package compared with a larger one^{(Reference Liu, Tam and Rangan89)}. A systematic review and meta-analysis of RCT indicated a moderate to large reduction in daily EI when comparing smaller with larger PS^{(Reference Robinson, McFarland-Lesser and Patel90)}. A meta-analysis of RCT and cross-over trials (COT) indicated that the association between PS and food intake is not linear in population groups^{(Reference Zlatevska, Dubelaar and Holden10)}. Furthermore, a Cochrane review of seventy-two RCT indicated that doubling a PS leads to an increase of approximately 35% in energy consumption, mostly noticed in men and non-overweight individuals^{(Reference Hollands, Shemilt and Marteau11)}.

Studies have indicated that individuals determine the food amount consumed according to what they are accustomed to eating. This is related to frequent exposure to large portion consumption (unit size) over time^{(Reference Ello-Martin, Ledikwe and Rolls80,Reference Vandenbroele, Van Kerckhove and Zlatevska91)} . This is called the portion size effect (PSE) or portion size response^{(Reference Rolls, Roe and Meengs92)} (more food is offered, more is consumed)^{(Reference Almiron-Roig, Navas-Carretero and Emery12)} and is documented in naturalistic environments, such as offices and restaurants^{(Reference Diliberti, Bordi and Conklin82,Reference Geier, Rozin and Doros93)} . A Cochrane review of RCT identified a consistent PSE on EI in adults consuming more food when offered in larger-size versions^{(Reference Hollands, Shemilt and Marteau11)}. Furthermore, a systematic review and meta-analysis of RCT of PSE indicated that consuming larger PS was related to higher daily EI (295 kcal; 95% CI: 202, 388 kcal)^{(Reference Higgins, Hudson and Hayes94)}. A study has attempted to identify mechanisms of the PSE on food intake, indicating the importance of the dual-process theory^{(Reference Kahneman95)}. It has been proposed that the mind has two thinking systems: system 1 (intuition) and system 2 (reasoning). A research dialogue stated that system 1 is heuristic, with a preference for a dominant option that stands out based on appeal. However, when there is no dominant option, consumers may have difficulty making a decision. As a result, system 2 is activated, selecting the alternative according to their goals by comparing the attributes and values^{(Reference Dhar and Gorlin96)}. Consumers use the latter to make decisions based on time constraints, processing capacity, desired level of accuracy, and fatigue^{(Reference Dhar and Gorlin96)}. An independent relationship of other factors, such as nutritional status^{(Reference Rolls, Morris and Roe27,Reference Wansink, Painter and North84)} , sex^{(Reference Rolls, Morris and Roe27)} and posterior compensation^{(Reference Flood, Roe and Rolls29,Reference Rolls, Roe and Meengs97)} was observed. More research is required to address the reasons that individuals do not comply with the satiety cues in their eating environments^{(Reference Ello-Martin, Ledikwe and Rolls80)}.

Larger effects on PS consumption have been identified in men^{(Reference Rolls, Roe and Kral26,Reference Rolls, Roe and Meengs86,Reference Rolls, Roe and Kral98)} compared with women. Additionally, if larger PS was combined with higher energy density (ED) (the amount of energy in each weight of food), stronger effects were observed on total daily EI^{(Reference Marteau, Hollands and Shemilt65)}. Moreover, the effects of PS can persist for several days as evidenced at 2^{(Reference Rolls, Roe and Kral26,Reference Rolls, Roe and Meengs86)} , 4^{(Reference Kelly, Wallace and Robson99)} and 11 d^{(Reference Rolls, Roe and Meengs97)}. The larger PS of high-energy-dense foods has the greatest impact on EI, up to 279 kcal/d, especially in pre-packed foods. COT have suggested that ED directly influences ad libitum EI and provides an independent effect on the macronutrient composition of food^{(Reference Rolls100)}. Table 1 summarises studies’ evolution over time, indicating the relationship between PS and EI, including ED in adults.

Table 1.

Shows studies researching the relationship between portion sizes (PS) and energy intake (EI) and/or energy density (ED) in adults

PS, portion sizes; EI, energy intake; ED, energy density.

To summarise, studies with similar design and methodology indicate a hypothetical association between large PS and obesity^{(Reference Rolls, Roe and Kral26,Reference Rolls, Morris and Roe27,Reference Kral, Roe and Rolls83,Reference Wansink, Painter and North84,Reference Rolls, Roe and Meengs86,Reference Rolls, Roe and Meengs92,Reference Rolls, Roe and Meengs97–Reference Kelly, Wallace and Robson99)} . However, it is difficult to determine whether an independent risk factor exists, due to confounding factors. Further research is needed due to the controlled environment of the above studies, where social interactions and food-related reminders were non-existent and blinding bias was absent^{(Reference Benton101,Reference Gough, Haynes and Clarke102)} . Exploring the exact relationship between high EI, energy-dense foods and PS in real-life settings is needed to draw stronger conclusions^{(Reference Kral, Roe and Rolls83)}. Longer-duration studies are required in real-life environments to establish whether a cause-and-effect relationship exists between PS and daily EI^{(Reference French, Mitchell and Wolfson88,Reference Hetherington and Blundell-Birtill103)} .

Observational studies

Studies have indicated that large PS is positively associated with obesity in young children aged 1–5 years old^{(Reference McConahy, Smiciklas-Wright and Birch106,Reference McConahy, Smiciklas-Wright and Mitchell107)} . In the UK between 1997 and 2005, adolescents consumed an increased amount of PS and EI from snacks (drinks, crisps, savoury snacks) and breakfast cereals^{(Reference Kerr, Rennie and McCaffrey108)}. Moreover, PS is positively associated with BMI percentiles in boys from 6 to 11 years old and children from 12 to 19 years old^{(Reference Huang, Howarth and Lin109)}. For pre-schoolers, 4–6 years old, EI is regulated by natural hunger-driven eating behaviours. However, environmental cues, such as large PS, can disturb this self-regulation^{(Reference Mrdjenovic and Levitsky110)}. In infants (11 months and younger) the relationship between ED and average PS is negative, proposing that as ED decreases, food intake downregulates correspondingly. In contrast, no association has been noted in toddlers (1–2 years old)^{(Reference Fox, Devaney and Reidy111)}. Evidence from a systematic review in the USA supports the positive relationship between ED and weight gain throughout life⁽¹¹²⁾. Overall, PS is consistently positively associated with both EI and children’s weight^{(Reference Rolls100)}; nevertheless, these data cannot determine causality^{(Reference Pourshahidi, Kerr and McCaffrey31)}.

Experimental studies

Most experimental studies have taken place in the USA and have established that when children aged 3–6 years old doubled the PS of their main meal it resulted in an increase of EI by 40%^{(Reference Fisher, Rolls and Birch39,Reference Rolls, Engell and Birch113–Reference Fisher, Liu and Birch115)} , where the main food given was macaroni and cheese^{(Reference Fisher, Rolls and Birch39,Reference Rolls, Engell and Birch113,Reference Fisher, Liu and Birch115)} or a selection of foods, including macaroni and cheese^{(Reference Vandenbroele, Van Kerckhove and Zlatevska91)}. Vegetables, mainly carrots, were offered alongside the food^{(Reference Young and Nestle30,Reference Rolls, Engell and Birch113–Reference Fisher, Liu and Birch115)} . The average PS consumed by 2-year-olds appears to have remained stable over the last 20 years, while PS increased^{(Reference Ello-Martin, Ledikwe and Rolls80)}. This supports the hypothesis that 3-year-old children self-regulate their intake according to hunger and satiety rather than food cues^{(Reference Rolls, Engell and Birch113,Reference Birch and Fisher116)} . As children grow, internal cues, such as satiation, are less effective on food intake, while external factors are more influential, such as watching TV^{(Reference Ello-Martin, Ledikwe and Rolls80,Reference Boulos, Vikre and Oppenheimer117)} . Nevertheless, studies in a controlled environment have validated the significant positive effect of larger PS on EI in 2-year-old children^{(Reference Fisher40,Reference Fisher, Arreola and Birch114,Reference Fisher, Liu and Birch115)} . It is not known what the results would be in a free-living environment.

Few short-term studies have examined the impact of a reduction in children’s PS with a positive effect in reducing the EI in age-appropriate PS^{(Reference Smith, Conroy and Wen118,Reference Savage, Fisher and Marini119)} . No change in EI was noted when the PS of an entrée was decreased by 25%^{(Reference Leahy, Birch and Fisher120)}. While studies are limited, they provide evidence that children from an early age are vulnerable to PS cues. Children increase the PS by eating more, but the evidence is weak to determine if they compensate for this at following eating points^{(Reference Leahy, Birch and Fisher120)}, such as in adults^{(Reference Kelly, Wallace and Robson99,Reference Rolls100)} . No available data exist to investigate the long-term effects of PS in children^{(Reference Pourshahidi, Kerr and McCaffrey31)}. A study in a controlled environment indicated that computerised manual PS selection can be observed in children between 5 and 11 years old and the correspondence between manual portion selection and actual intake improves with age. This highlights that the relationship between children’s cognitive development and PS may help to develop age-appropriate PS^{(Reference Cox, Hinton and Sauchelli121)}.

COT have indicated that reducing the ED of an entrée reduces children’s total EI^{(Reference Leahy, Birch and Fisher120,Reference Leahy, Birch and Rolls122,Reference Leahy, Birch and Rolls123)} . However, manipulation of the ED of a single snack did not significantly affect children’s EI at a single eating juncture^{(Reference Looney and Raynor124)}. Additional research revealed that the EI effect could be continued when the ED of multiple meals were manipulated over 2 d^{(Reference Leahy, Birch and Fisher120)}. Also, reduction of ED has a positive effect on adiposity in the longer term, particularly when individuals, both children, and adults, are receiving positively focused messages about weight control^{(Reference Epstein, Paluch and Beecher125)}. Table 2 summarises studies’ evolution over time, indicating the relationship between PS and EI, including ED in children.

Table 2.

Shows studies researching the relationship between portion sizes (PS) and energy intake (EI) and/or energy density (ED) in children

PS, portion sizes; EI, energy intake; ED, energy density.

A great deal of evidence from COT positively associates ED, adiposity and PS in children^{(Reference Leahy, Birch and Fisher120,Reference Leahy, Birch and Rolls122–Reference Looney and Raynor124)} . Studies strongly support that 5-year-old children respond to increasing PS^{(Reference Rolls, Engell and Birch113–Reference Fisher, Liu and Birch115)}. While a direct causal link between obesity and PS has yet to be determined, consumption of large PS of ED foods promotes obesity-eating behaviours in children^{(Reference Pourshahidi, Kerr and McCaffrey31,Reference Rolls100)} . More research is required on influencing PS education and downsizing strategies for parents/carers, thus helping children to consume age-appropriate PS^{(Reference Small, Lane and Vaughan126,Reference Hetherington127)} . This could identify a possible EI reduction strategy that has still not been demonstrated^{(Reference Savage, Haisfield and Fisher128)}. Additional strategies are required to help children to recognise and respond appropriately to internal signals and resist environmental influences on PS^{(Reference Small, Lane and Vaughan126)}.

Labels on food packaging

Serving size guidance is voluntarily included on food packaging across the European Union (EU)⁽¹²⁹⁾, and is mandatory in the USA⁽¹³⁰⁾; however, consumers prefer household measure guidance^{(Reference Faulkner, Livingstone and Pourshahidi131)}, such as a portion control cup^{(Reference Logue, Doherty and McCann5)}, rather than referring to food labels for managing PS^{(Reference McConahy, Smiciklas-Wright and Birch106)}. Household measures result in positive behavioural changes, particularly in staple foods such as cereals, rice and pasta^{(Reference Logue, Doherty and McCann5)}. Consumers refer mostly to quality, quantity, brand, price and sell-by date, and less to ethical and sustainable food labels^{(Reference Ghvanidze, Velikova and Dodd132)}, therefore preferring visual impressions of packages and PS^{(Reference Ordabayeva and Chandon133)}. This leads to ‘awareness and estimation bias’^{(Reference Steenhuis and Poelman33)} as they fail to identify and understand the quantity information or representation of food^{(Reference Lennard, Mitchell and McGoldrick134)} in the packages, particularly if it is obtainable in non-metric units^{(Reference Viswanathan, Rosa and Harris135)}. A study that looked at the differences between suggested serving sizes in energy-dense foods indicated a lack of clarity in serving size guidance’ emphasises. This clearly indicates the need for effective and meaningful guidance on pre-packed foods^{(Reference Spence, Livingstone and Hollywood44)}. Contrarily, consumers with medical issues such as food allergies have reported problems with confusing and contradicting information on the food labels^{(Reference Cornelisse-Vermaat, Voordouw and Yiakoumaki136,Reference Hendriks, Frewer and van der Meulen137)} , readability of the label^{(Reference Cornelisse-Vermaat, Voordouw and Yiakoumaki136)}, lack of harmonisation between the different countries, and the position of the labels in the food package^{(Reference Voordouw, Cornelisse-Vermaat and Yiakoumaki138,Reference Sheth, Waserman and Kagan139)} . A literature review conducted by FSA and Food Standards Australia New Zealand highlighted the same issues⁽¹⁴⁰⁾. Men experience more difficulties in estimating appropriate PS^{(Reference Lewis, Ahern and Solis-Trapala141)}. However, other studies show that perceived healthiness of the food^{(Reference Faulkner, Pourshahidi and Wallace142)}, the ED^{(Reference Almiron-Roig and Solis-Trapala143)} and the BMI^{(Reference Lewis, Forwood and Ahern144)} may also contribute to shaping appropriate PS perceptions.

Different food labelling laws between countries add to consumers’ confusion. Clear information is provided in the consistent EU approach (EU Regulation No. 1169/2011) and retained regulations in the UK for the food labels on the pre-packed products⁽¹⁴⁵⁾; however, there is no harmonised approach for front-of-pack (FoP) labelling⁽¹⁴⁶⁾. A decade ago, the UK Government announced via a press release new consistency guidelines based partially on previous research^{(Reference Fisher, Liu and Birch115)} stating that different FoP labels could hinder consumer understanding and discourage use^{(Reference Draper, Adamson and Clegg147)}. Although nutrition labels on pre-packaged food products are cost-effective at the population level, governments need to maximise their potential, by identifying new layouts and different types of information content to ensure that nutrition information is available and comprehensible for consumers’ benefit^{(148,Reference Campos, Doxey and Hammond149)} . The food industry is pressured to reduce the PS and calories as part of the ongoing strategy to reduce childhood obesity⁽⁵⁰⁾. However, this leads to further issues, as consumers want value for money^{(Reference Rimmer150)}. Furthermore, a scoping review, including experimental and observational studies, indicated that food package labelling for serving size is unclear^{(Reference van der Horst, Bucher and Duncanson151)}. Specifically, consumers believe that the labelled serving size set by the manufacturer is the government recommendation^{(Reference van der Horst, Bucher and Duncanson151)}. Occasionally, the serving sizes are considered to be unrealistically small; resulting in consumers’ overestimating not only to the number of servings in one package but also the caloric content^{(Reference Young and Nestle19,Reference Bryant and Dundes34,Reference Marchiori, Corneille and Klein152)} . Most consumers interpret the package size as a single serving size when it contains multiple servings^{(Reference van der Horst, Bucher and Duncanson151)}. Likewise, using terms such as small, medium and large leads to confusion in interpretation and differences^{(Reference Young and Nestle153)}. Regrettably, visual perception is not reliable in indicating the package size or the PS because of biases, which are highlighted in Table 3 ^{(Reference Ordabayeva and Chandon133)}. Additionally, more emphasis towards dietary guidelines as well as the nutritional composition of the products sold may be warranted^{(Reference Culliford, Bradbury and Medici154)}. For instance, consumers widely know the number of PS of fruits and vegetables they need to consume every day. However, it remains unclear if they consider the consumption frequency and quantity of ultra-processed foods, such as confectionery and biscuits, when the EatWell Guide does not quantify how many portions of each food group should be consumed, unlike other food-based dietary guidelines^{(Reference Culliford, Bradbury and Medici154,Reference Miller, Eckberg and Decker155)} .

Table 3.

Shows the summary of package and portion size (PS) estimation biases

Ordabayeva N and Chandon P (2016) In the eye of the beholder: Visual biases in package and portion size perceptions. Appetite, 103, 450–457 [adapted]. Copyright © 2015 Elsevier Ltd. All rights reserved.

Unit bias

The unit bias model suggests that one serving is appropriate to consume at once, irrespective of the size^{(Reference Geier, Rozin and Doros93)}. However, it is argued that segmentation bias is more applicable to use as individuals consume less food when it is divided into smaller units^{(Reference Kerameas, Vartanian and Herman156)}. A study supports that the anchoring effect works as a reference point for PS, as consumers eat a specific amount of the food. This means that PS acts as an influential anchor for determining the amount to consume, and succeeding adjustment processes do not negate the effect of that anchor^{(Reference Marchiori, Papies and Klein9)}.

The package size or the unit creates a ‘consumption norm’, operating for most individuals^{(Reference Zlatevska, Dubelaar and Holden10)}, which arguably is not appropriate according to the nation’s food recommendations^{(Reference Geier, Rozin and Doros93,Reference Wansink157)} . The ‘clean your plate’ notion is encouraged, especially when the food is free; yet, in public settings, it promotes the bias to finish it, as the appropriateness mechanism is activated^{(Reference Birch and Fisher116)}. The PS sets a norm and dictates the amount consumed; therefore, PS and not hunger leads to food consumption^{(Reference Steenhuis and Poelman33,Reference Geier, Rozin and Doros93)} . This is associated with human evolution, involving psychological and physiological mechanisms protecting against low adiposity^{(Reference Berthoud158,Reference Zheng and Berthoud159)} . Consequently, consumers could be sensitive to detecting food inadequacy in terms of smaller than normal PS^{(Reference Haynes, Hardman and Halford160)}.