Dietary assessment in general is challenging, but especially in children and adolescents, due to lower literacy skills and limited memory compared with adults(Reference Livingstone, Robson and Wallace1,Reference Baranowski and Domel2) . Children’s food choices differ only slightly from those of adults, with the exception of age-appropriate adjustment of portion sizes(Reference Rothpletz-Puglia, Fredericks and Dreker3). Hence, dietary assessment methods similar to those used in adults, such as 24-h recall, dietary records or FFQ, are currently used to assess dietary intake in young populations(Reference Perez-Rodrigo, Artiach Escauriaza and Artiach Escauriaza4). In case of semi-quantitative tools like FFQ, researchers must apply information on portion sizes, the quantity of food consumed at one eating occasion, to quantify usual dietary intake. Unfortunately, there is a noticeable lack of reference values for portion sizes, especially tailored to different age groups in childhood and adolescence, highlighting a clear need for such data.
An FFQ is a commonly used survey method to retrospectively assess usual food intake over a past period of time, such as 6 or 12 months(Reference Willett5). Participants are asked to indicate the frequency of intake (e.g. several times per day, per week and per month) of commonly consumed foods and beverages(Reference Shim, Oh and Kim6). To quantify usual dietary intake and further calculate energy and nutrient intake, researchers need to multiply consumption frequencies by standard portion sizes. Although the list of foods included in the FFQ may be the same for children and adults, the portion sizes clearly differ. Therefore, knowledge of age-specific food portion sizes is crucial, especially in dietary assessment methods where participants are not encouraged to weigh their food and beverages, such as FFQ. For example, the FFQ used in the representative German National Health Interview and Examination Survey for Children and Adolescents (KiGGS) employed common household measures of forty-five food groups to estimate food intake. A validation study among 12- to 17-year-old adolescents concluded that the FFQ was suitable to rank children and adolescents based to their food intake(Reference Truthmann, Mensink and Richter7).
Age-appropriate food portion sizes are also required when dietary assessment is done by a 24-h recall. With this assessment method, participants retrospectively report the dietary intake over the past 24 h(Reference Shim, Oh and Kim6). Given that children and adolescents have difficulties to estimate accurately their food and beverage consumption, researchers might provide time-consuming and labour-intensive tools, such as food photographs, household measures or two-dimensional grids for more precise assessments(Reference Bouchoucha, Akrout and Bellali8). However, if detailed standard portion sizes were available for various age groups, it would greatly facilitate the post-processing of 24-h recalls in terms of both time and costs.
Up to now, empirically derived portion sizes for different age groups remain scarce for children and adolescents. Previously, only portion sizes of selected food groups for young children (6–36 months) were released as part of the German GRETA study(Reference Hilbig, Drossard and Kersting9). The authors used data from the DOrtmund Nutritional and Anthropometric Longitudinally Designed (DONALD) study, a dynamic cohort study collecting weighed dietary records since 1985 from infancy to young adulthood(Reference Kroke, Manz and Kersting10). With approximately 10 000 dietary records available, it is possible to calculate empirically age-specific portion sizes throughout childhood and adolescence. Therefore, the aim of the current analysis was to provide portion sizes for a wide range of food groups and age groups for children and adolescents aged 4–18 years.
Research design and methods
Study design
The DONALD study is an ongoing dynamic cohort study since 1985 in Dortmund, Germany. Its primary objective was to investigate the association between dietary intake, metabolism and growth from infancy into early adulthood. Details on recruitment, study design and methods have been published previously(Reference Perrar, Alexy and Nothlings11). In brief, each year, 30–40 infants from the region of Dortmund were unsystematically recruited and follow-up into adulthood. Among the examinations are 3-d weighed dietary records (3dWDR), which were conducted annually from infancy onwards. The study is conducted according to the Declaration of Helsinki and was approved by the Ethics Committee of the University of Bonn. All examinations were carried out with written informed consent from study participants themselves (≥16 years) or their parents. The study was registered in the German Register of Clinical Trials (DRKS-ID: DRKS00029092).
Study population
At the time of our data analysis, 10 880 3dWDR collected between 1985 and 2022 were available form 1,338 participants aged 4–18 years. Dietary records were categorised as underreported if the ratio of reported energy intake to estimated BMR according to the age- and sex-specific equations of Schofield(Reference Schofield12) was below the cut-off values proposed by Goldberg(Reference Goldberg, Black and Jebb13). After exclusion of 921 (8·5 %) potential underreported dietary records, the final dataset consisted of 9,828 3dWDR from 1,325 participants (49·3 % female). To facilitate our analysis, we divided the dietary records into five groups based on the age of the participants: 4–6 years, 7–9 years, 10–12 years, 13–15 years and 16–18 years. Due to the nature of our open and on-going cohort study, the number of dietary records as well as the number of participants gradually decreased with increasing age groups. It is important to note that participants in the first age group (4–6 years) may not have yet reached the age range for the next age group (7–9 years) and will therefore have the opportunity to submit dietary records within that age group in the coming years.
3-day weighed dietary records
Parents or older participants themselves collected detailed information on all foods and beverages consumed over three consecutive days using electronic food scale (±1 g). Since data on dietary intake in younger children (<7 years) are often considered less reliable(Reference Livingstone, Robson and Wallace1), but dietary recording already starts in infancy, it is carried out by parents. Since then, participating families conduct annual dietary records and with increasing age and cognitive abilities, children assist by recording guided by their families who have routine in weighing and recording of foods through years of practice. Findings from the DONALD study indicated that 80 % of participants aged 10–12 years assisted their parents in recording their dietary intake. By the time they reached the age of 15–18 years, 70 % of the boys and 90 % of the girls were able to independently maintain their dietary records without parental assistance(Reference Kersting, Sichert-Hellert and Lausen14). When participants consumed food away from home, such as at school or in a restaurant, and were unable to weigh all consumed items, they were permitted to provide semi-quantitative estimates such as household measurements or the number of portions. Participants were requested to document generic food items as accurately as possible, including details such as the fat content of milk or the type of grain (whole grain or refined grain) in dishes. For commercially packaged food items, participants collected both the packages and the accompanying food labels. Nutrient information and ingredient lists from each newly recorded food item were used to simulate recipes, enabling the addition of commercially available food items to the in-house food composition table LEBTAB, resulting in a continuously growing food database(Reference Sichert-Hellert, Kersting and Chahda15).
Food group classification
Classification of foods and beverages is a complex task, as there is no standardised procedure for classification into food groups available. To derive portion sizes, we have therefore used three steps: First, the 11 570 food items from the LEBTAB database of the DONALD study were categorised based on shared characteristics (e.g. milk: milk, chocolate milk and strawberry milk) or commonly culinary usage (e.g. cream: cream, sour cream, crème fraiche). Second, we aggregated food groups according to those usually used in FFQ and 24-h recall. In case of composite dishes, the large heterogeneity and diversity of the products made it necessary to aggregate them into joint groups. Third, we took the grouping of the German Food Book into account, to be as consistent as possible with the eating habits(Reference Rehlender16). Two researchers collaboratively defined the major food groups and corresponding underlying food groups. One scientist assigned all items to the previously defined food groups. In case of ambiguity, a second scientist was consulted for clarification. A total of 2,247 food items were excluded for the analysis as separate portion sizes for these groups were deemed unnecessary for estimating usual food intake. Those included commercial infant or toddler foods (e.g. baby mixed dishes or milk-cereal mixtures for infants, n 236), instant powder (e.g. for sauces, n 546), supplements (e.g. vitamins or minerals, n 334), flour and baking mixtures (e.g. baking mixes for bread or cakes, n 205), alcoholic beverages (e.g. beer, wine or whisky, n 49) and herbs and spices (e.g. salt, garlic granules or dried basil, n 35). It also included food items that could not be assigned to any broader food group, such as additives, colourants or preservatives (n 842). The remaining 9,323 items were then classified into sixteen major food groups: dairy, eggs, meat, fish, fat/oil, cereals, potatoes, vegetables, fruits, pulses, beverages, sweets, snacks, spreads, dishes and plant-based alternatives.
To ensure precision in portion size determination, these sixteen major food groups were further subdivided into a total of seventy-one underlying food groups. This subdivision allowed us to segregate foods that exhibited substantial differences in typical consumption quantities. For instance, the food group dairy was subdivided into milk, fermented dairy, cheese, curd cheese and cream. A comprehensive list of the major food groups, along with their respective underlying food groups and descriptions of associated food items, can be found in see online supplementary material, S1 Table.
Additional covariables
Trained nurses assessed participants’ weight (kg) and height (m) to calculate their BMI (kg/m2). Further, the LMS method was used to determine the BMI sd score (SDS), using age- and sex-specific values for median (M), CV (S) and skewness (L) from the German national reference(Reference Kromeyer-Hauschild, Wabitsch and Kunze17).
Statistical analysis
All statistical analyses were conducted using the statistical software SAS (Version 9·4). Participants’ characteristics were presented as median values (25th percentile; 75th percentile) for continuous variables and as relative frequencies (%) for categorical variables unless otherwise noted.
To derive median individual portion sizes based on actual consumption, we exclusively included participants who consumed quantities greater than 0 g within the respective food group. Since portion sizes are strongly dependent on age, it is useful to smooth the percentiles. By using a quantile regression (PROC HPQUANTSELECT), random fluctuations in the percentiles were eliminated effectively(Reference Le Cook and Manning18). This approach allowed us to represent portion sizes as a simple function of age. The rationale behind this procedure is that minor changes in the covariate age are likely to result in continuous alterations in portion size, thus changing uniformly across different age groups(Reference Cole and Green19). Subsequently, we calculated the median amount (in g) of weighed food group per eating occasion for each food group, which represents the empirically derived portion size for this respective age group. Differences in food portion sizes between boys and girls were tested with the Mann–Whitney U test to identify the need to report sex-specific portion sizes. We did observe sex differences in food portion sizes for a small number of underlying food groups, namely milk, vegetable juices and energy drinks (P < 0·05). However, these differences were not consistent across the age groups. Thus, portion sizes for children (4–12 years) were not stratified by sex. Due to significant differences in energy intake among adolescents, we decided to present sex-specific portion sizes for adolescents (13–18 years).
Results
Participant characteristics are described in Table 1. The number of participants as well as the number of 3dWDR decreased continuously between the youngest age group (4–6 years, n 2,847) and the oldest age group (16–18 years, n 1,113). In contrast, total energy intake increased by a median of 68 % from the youngest to the oldest age group (1,331 v. 2,239 kcal). With regard to the overall study population, the median age was 9·9 years (25th percentile: 6·2; 75th percentile: 13·0) and 7·5 dietary records (range: 1–16) were on average available per participant.
Characteristics* of 1,325 study participants of the DONALD study
3dWDR, 3-day weighed dietary records; SDS, Standard Deviation Score.
* Data shown as median (25th percentile; 75th percentile) or relative frequency (%).
† Calculation of total energy intake includes all foods and beverages consumed, not just those used for portion size derivation.
Reported frequencies within each food and age group (see online supplementary material, S2 Table) ranged from no consumption (e.g. wraps, 4–6 years) to 19 582 times (bread, 4–6 years). In case of no reported consumption, it was not possible to derive portion sizes for the respective age groups, which was the case for energy drinks in three age groups, instant dishes in one age group, ready-to-eat salad with sauce in two age groups, premade grain-based salad in four age groups and Traditional German dishes in one age group. Similar cases were identified for certain underlying food groups within the major food group plant-based alternatives. Across all food groups, the median of reported frequencies was 482 (25th percentile: 39; 75th percentile: 2,278). In terms of reported frequencies, 49·9 % of the underlying food groups had reported frequencies below the median, mostly covering plant-based alternatives and dishes.
The empirically derived portion sizes (Table 2) cover all food groups except for plant-based alternatives. Minimum portion size was 2 g (liquid oil or solid fat, 4–6 years) and maximum portion size was 500 g (energy drink, for 16–18-year-old boys). In general, portion sizes increased with age across all food groups, except for the food group milk and all food groups belonging to the major food group plant-based alternatives. When comparing the youngest with the oldest age group, it is noticeable that portion size increases between 22·2 % (processed meat: 18 g v. 22 g) and 286·7 % (savoury snacks: 15 g v. 58 g) for general food groups. Regarding dishes, portion sizes increased between 38·1 % (soup: 194 g v. 268 g) and 1,190·3 % (Traditional German Dishes: 31 g v. 400 g). Portion sizes for the major food group plant-based alternatives, including 12 underlying food groups, are provided in the appendix (see online supplementary material, S3 Table).
Empirically derived portion sizes in gram for 4–18-year-old children and adolescents from 9,828 3-d weighed records
n/a, not available; RTEC, ready-to-eat-cereals.
Consumption frequencies ≥482 (median of reported frequencies across all underlying food groups) were marked in light grey, and consumption frequencies <482 were marked in white. Of note, the database of vegetarian and vegan products covers a relatively small number of food items (n 237). In addition, the reported frequencies were below the median (482), and thus portion sizes might therefore not be considered as highly reliable.
Discussion
Using the comprehensive data from the DONALD study, we have empirically determined portion sizes for sixteen major food groups, encompassing a total of seventy-one underlying food groups. These determinations have been made across five distinct age groups spanning from 4 and 18 years and additionally for both sexes from the age of 13 years. As expected, our analyses have shown a consistent increase in portion weights with increasing age. It is important to note that the values represents actual portion sizes observed within our study sample and do not constitute evidence-based intake recommendations.
Our empirically derived portion sizes align closely with those reported by Wrieden and colleagues. Their portion sizes were drawn from the British National Diet and Nutrition Survey, where dietary intake was similarly assessed through the use of weighed food records(Reference Wrieden, Longbottom and Adamson20). They chose age ranges that are directly comparable to those in our analysis and derived portion sizes from an extensive dataset comprising 3,374 weighed dietary records encompassing 1,675 children and adolescents. Given that Wrieden and colleagues analysis is more than 15 years old, a new derivation of portion sizes might be beneficial, as portion sizes may have evolved over time or new food groups may have emerged. While some distinctions in food group categorization exist between the DONALD study and the British population, the resulting portion sizes are notably similar. For instance, in the case of pasta, for individuals aged 16–18 years, our study indicates a portion size of 161–200 g, whereas the British dataset suggests 185 g. Similarly, for milk consumption among 7- to 9-year-olds, our findings indicate 156 g, while the British dataset suggests 186 g. Of note, the British portion sizes are always slightly larger than the German portion sizes. Various factors affect portion sizes, including energy intake or energy requirement(Reference Rolls, Roe and Meengs21), dietary habits or cultural background(Reference Rozin, Kabnick and Pete22). Unfortunately, Wrieden and colleagues did not provide specific information on energy intake, thus precluding an assessment of its comparability to the DONALD participants. Likewise, no information was available concerning energy requirements based on the level of physical activity.
The German Nutrition Society (DGE) has established portion size guidelines as part of its quality standards for catering in kindergartens(23) and schools(24), ranging from 1 to 19 years. Although portion sizes derived from the DONALD study align closely with the DGE’s recommendations, our portion sizes often are below those recommended by the DGE. For rice or potatoes, the DGE recommends 90 g for children aged 4–6 years, whereas the portion sizes observed in the DONALD study were 60 and 70 g, respectively. We found similar results for legumes among 7–9-year-olds (DGE: 80 g v. DONALD: 50 g). Of note, portion sizes of the DGE serve as intake recommendations to reach the recommended nutrient intakes, whereas portion sizes of the DONALD study reflect actual consumption amounts. In addition, it is not clear on which data the DGE portion sizes are based or whether they were derived empirically at all.
While weighed dietary records are considered as the gold standard in dietary assessment and generally represent precise tools(Reference Carlsen, Lillegaard and Karlsen25), there are certain limitations, which potentially introduce inaccuracies in portion sizes determination. First, the prospective nature of these records can lead to underreporting of unhealthy foods and overreporting of healthy foods(Reference Schoch and Raynor26). Consequently, this can led to the reporting of a dietary pattern, which is socially acceptable but may not truly represent the usual diet. Second, portions sizes reported for unhealthy foods and beverages may be consciously or unconsciously underestimated due to social desirable bias(Reference Whybrow, Horgan and Macdiarmid27), although we excluded records with underreported energy intake for more reliable portion sizes. Third, accurately estimating food consumption away from home presents unique challenges, given the variability in portion sizes and ingredients(Reference Young and Nestle28). Notably, only one percent of the records was mainly based on semi-quantitative estimates, ensuring that our portion sizes reflect real-world consumption. Overall, these issues are particularly pertinent among younger children, where surrogate reporting is often necessary due to limited self-reporting capabilities(Reference Livingstone, Robson and Wallace1).
Research on sex-specific portion sizes for children and adolescents is scarce. Considering sex-specific portion sizes may increase the accuracy of consumption assessments for absolute nutrient intakes in adults(Reference Lee, Kang and Song29–Reference Kang, Park and Boushey31). Analyses in our sample suggest that sex-specific portion sizes are not necessary until adolescence, as consumption levels are very similar before puberty, although energy intake is higher in boys at all ages. Obviously, the higher energy intake is not reflected in the portion size of each food group. From late adolescence, when physiological differences as well as changes in body composition between girls and boys are more prominent, sex-specific portion sizes should be taken into account, which is why we calculated sex-specific portion sizes from 13 years onwards. Of note, due to sex stratification, the number of reported food items in some food groups is been considerably lower due to their infrequent consumption in our study sample and should be used with caution.
The major food group plant-based alternatives, which is a novel food group emerging due to the trend towards plant-based vegetarian and vegan diets(Reference Craig, Mangels and Fresan32), should be discussed in particular. The market for products, such as soya, rice, almond or oat drinks as milk alternatives, or plant-based alternatives for fish, meat and eggs, is steadily growing(Reference Bundesamt33). Hence, portion sizes for these products are also desirable, primarily to facilitate the development of dietary assessment instruments for vegetarians and vegans. Furthermore, given that omnivores increasingly incorporate these alternatives into their diets, the relevance of portion size information extends beyond plant-based eaters. With the predominant omnivore study sample of the DONALD study, these products have only been consumed in significant frequencies since 2016. This has resulted first in a limited number of available food items and second in consumption frequencies per food item below the median of reported frequencies within our study sample. However, even now, our data indicate lower portion sizes for alternative plant-based products compared with their animal source counterparts (16–18 years for milk v. milk alternatives: 150–218 g v. 96–127 g).
It is plausible that changes in dietary habits may not result in a proportionate increase in portion sizes in line with energy requirements. A reduction in milk consumption during late adolescence is well documented(Reference Baird, Syrette and Hendrie34,Reference Parker, Vivian and Oddy35) . We observed similar findings in our data. Despite statistical adaptation, consumption of milk was larger in girls aged 13–15 years than in girls aged between 16 and 18 years (152 g v. 150 g). One possible explanation for this shift could be that milk is no longer consumed in its pure form beyond the age of 16 but is instead used as an additive for coffee or ready-to-eat-cereals. Our data reveal an increase in vegetables with age. Nevertheless, this increase might primarily stem from a higher nutrient demand rather ran an increased energy demand, as vegetables typically contribute minimally to total energy intake due to their low energetic content. The reason for this age-related rise in vegetable consumption could be attributed to the enhanced sensory acceptance of vegetable taste, texture and smell(Reference Luszczki, Sobek and Bartosiewicz36).
In the German KiGGS study, food consumption data obtained from the validated FFQ were transformed into mean intakes as grams per day using predefined portion sizes(Reference Truthmann, Mensink and Richter7). Age ranges in the KiGGs study are considerably broader than in our study, making direct comparisons challenging due to differences in age grouping. However, when comparing specific portion sizes, such as desserts (DONALD 142–159 g v. KiGGS: 150 g), there is a notable degree of concordance. In contrast, portion sizes for chocolate (DONALD: 14–25 g v. KiGGS: 50 g) or nuts (DONALD: 7–14 g v. KiGGS: 25 g) show less agreement. In general, it appears that the portion sizes reported in the KiGGS study are slightly larger than those observed in the DONALD study. Of note, dietary intake in g of the 1,249 participants from the KiGGS study was roughly estimated, while in the DONALD study dietary intake was actually weighed.
Definition of age groups and categorisation of food groups might be essential for portion sizes. Our classification was based on subjective judgement, although we made diligent efforts to document our decisions (see online supplementary material, S1 Table) to ensure replicability. Food grouping is a complex task and during the process of coding, it might be necessary to revise initial decisions. For instance, we initially subdivided vegetable oils into three food groups. The first group encompassed rapeseed and sunflower oil, which are commonly used in Germany for frying. The second group included olive oil, primarily used for salad dressings and thus typically in smaller quantities. The last group included expensive oils, such as nut oils, linseed oil or soybean oil, used in smaller quantities due to their cost and culinary value. However, our statistical analyses showed that the consumption levels across all age groups were identical or differed only slightly (±1 g). Therefore, we combined all vegetable oils to increase the number of consumption occasions, resulting in more reliable portion size estimates.
Strengths and limitations of the used data should be mentioned. The study sample is characterised by a high socio-economic status, which limits the generalisability of our findings to the broader German population. Participants with a low socio-economic status are difficult to recruit due to the study’s long-term and time-consuming design(Reference Kroke, Manz and Kersting10). Consequently, participants with very unhealthy dietary habits might not be adequately represented in our dataset. In addition, the study is geographically limited to the region of Dortmund and thus regional differences in dietary intake cannot be represented. If more northern areas with seaside access were included, consumption of fish and fish products might have been more frequent, providing more robust estimates. Furthermore, we observed only a few food groups where portion sizes did not increase with age, such as rice dishes (10–12 years: 150 g v. 13- to 15-year-old girls: 122 g). This might be due to the small number of recorded rice dishes (n 7) among 13- to 15-year-old girls. Despite the high number of 9,828 available weighed dietary records, reported frequencies for certain food groups might be too low to calculate reliable portion sizes, necessitating caution when using these estimates. The strength of the study is the multiple collected dietary data with a mean of 7·5 3dWDR per participant. In addition, weighted dietary records often serve as reference instruments for dietary assessment methods. Nevertheless, when weighing is not possible (i.e. out of home consumption), participants are allowed to use household measurements. However, in only 5·8 % of the dietary records, between 50 and 75 % of all items consumed were reported using household measurements.
In conclusion, our current analysis provided empirically derived age-specific food portion sizes for a wide range of food groups spanning 4 and 18 years. They are also likely to be particularly useful for evaluating 24-h recalls or FFQ in young age groups. Nevertheless, there is a compelling need for further research to validate our portion sizes in other populations.
Acknowledgements
The authors would like to thank the staff of the DONALD study for collecting and coding the dietary records as well as all study participants and their families for the provision of their data.
Financial support
The DONALD study is financially supported by the Ministry of Science and Research of North Rhine Westphalia, Germany.
Conflict of interest
The authors declare no conflict of interest.
Authorship
Conceptualisation: M.S., U.A.; statistical analysis: M.S.; writing – original draft preparation: M.S.; writing – review and editing: U.A., U.N.; supervision: U.N.
Supplementary material
For supplementary material accompanying this paper visit https://doi.org/10.1017/S136898002400017X
Ethics of human subject participation
The study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving research study participants were approved by the Ethics Committee of the University of Bonn. Written informed consent was obtained from all subjects themselves (≥16 years) or their parents.
Open access
