Overweight and obesity continue to increase worldwide, with prevalence having risen markedly since 1975(1). Adult obesity is defined as having a BMI of 30 kg/m2 or above. According to the WHO estimates, 43 % of adults were living with overweight and 16 % with obesity in 2022, highlighting the ongoing global burden of excess adiposity(2). In England, Health Survey for England 2022 data indicate that 29 % of adults were living with obesity, with prevalence 28 % in men and 30 % in women(3). The UK also remains among the higher-burden countries in the WHO European Region for adult overweight/obesity(4). Marked inequalities persist, with adult obesity prevalence in England being 36 % in the most deprived areas compared with 22 % in the least deprived, demonstrating a substantial socio-economic gradient(3) that is consistently associated with disproportionately higher obesity-attributable morbidity and mortality in deprived populations(Reference Newton, Briggs and Murray5).
Food portion size (PS) is the food amount served for instant consumption in a single-eating event(Reference Almiron-Roig, Navas-Carretero and Emery6) and is an important determinant of energy intake (EI)(Reference Butland, Jebb and Kopelman7). Consistent consumption of larger-than-appropriate PS contributes to sedentary behaviour and high EI(Reference Berg, Lappas and Wolk8–Reference Hollands, Shemilt and Marteau11). Both have been linked to an increase in obesity prevalence(Reference Livingstone and Pourshahidi9) and have altered the food environment(Reference Marchiori, Papies and Klein12,Reference Roy, Kelly and Rangan13) , leading to non-communicable diseases such as CVD, type 2 diabetes and several cancer types(Reference Young and Nestle14), as well as psychosocial diseases such as depression and anxiety disorders(Reference Hruby, Manson and Qi15). Although there is no direct evidence linking PS to obesity, both the energy density of commonly consumed foods and their consumption frequency have risen over recent decades(Reference Livingstone and Pourshahidi9,Reference Rippin, Hutchinson and Jewell16) .
Since the 1970s, a fivefold increase in the ‘standard’ food PS has been noted in the USA(Reference Young and Nestle14), and this has also been shown worldwide over the last 40 years(Reference English, Lasschuijt and Keller17), via the introduction of larger-sized portions in popular and energy-dense foods, such as white bread products, macaroni(Reference Rolls, Morris and Roe18), pre-packed snacks(Reference Rolls, Roe and Kral19,Reference Steenhuis, Leeuwis and Vermeer20) , cakes, popcorn(Reference Wansink and Park21), soft drinks(Reference Steenhuis, Leeuwis and Vermeer20,Reference Flood, Roe and Rolls22) and alcoholic drinks(Reference Young and Nestle23).
Much of the previous research has used nationally representative surveys to investigate trends in the PS at the population level. In Australia, it was shown that the PS of discretionary food products such as pizza, cakes, ice cream, sausages and other processed meats have increased over 17 years, whereas products such as pastries, snack foods and chips have decreased(Reference Zheng, Rangan and Meertens24). An Irish survey of commonly consumed foods and beverages by adults aged 18–64 over 10 years indicated significant increases in the PS of five foods (milk, white slice bread, wholemeal/brown bread, all cooked meat and roasted poultry) and significant decreases in three foods (potatoes, chips/wedges and sliced ham). No significant changes occurred over time in five foods (yogurt, cheese, butter/spreads, processed potato products and ham/bacon)(Reference O’Brien, Livingstone and McNulty25).
The UK Government campaigns for healthy eating and calorie reduction via the Eatwell Guide (current food-based dietary guidelines)(26). However, guidance on the appropriate PS of each food group and key foods within food groups is limited(27) and arguably outdated(Reference Rippin, Hutchinson and Jewell16,Reference Clift28) , compounding consumers’ confusion on appropriate food PS(Reference Lewis, Ahern and Jebb29). Furthermore, over 20 years, the UK National Diet and Nutrition Survey (NDNS) has revealed inconsistent trends in PS(Reference Wrieden, Gregor and Barton30).
There are limited data on typical PS of commonly consumed foods among adults over the past decade, making it difficult to assess food over- or under-consumption and its impact on dietary adequacy and weight status. This is particularly relevant given the current prevalence of obesity rates in the UK. Further research on changing food consumption patterns is needed to guide current government dietary policies.
Using nationally representative data from the NDNS, this study primarily aims to identify trends in the PS of commonly consumed food groups over the last 10 years among the UK adult population. The key objectives are to examine differences in PS consumption according to sex and weight status, to identify changes in the frequency of consumption of key food groups and to examine associations between energy-dense foods and measures of adiposity. The hypothesis is that as obesity rates have increased over time, food PS has increased in the UK adult population, especially in the overweight/obese population.
Materials and methods
Survey design and sample size
Data from the NDNS were obtained from the UK Data Archive, University of Essex. The NDNS of adults aged 18–65 years is part of a Rolling Programme (RP), which is a continuous cross-sectional survey collecting detailed, quantitative information from a representative sample of the UK population on food consumption, nutrient intake and nutritional status of the general population aged from 1·5 years up and living in private households(Reference Ashwell, Barlow and Gibson31). Sample households were randomly selected from the Primary Sampling Units of the Postcode Address File, where one adult and one child were randomly chosen(32,33) .
For this analysis, two independent cross-sectional NDNS datasets were used to represent two different time points: Years 1–4 (2008/2009–2011/2012) and Years 9–11 (2016/2017–2018/2019). The sample sizes were 2506 adults in Years 1–4 (n 1250 males, n 1256 females) and 1880 adults in Years 9–11 (n 935 males, n 945 females). As the two samples are independent and unpaired, comparisons reflect population-level changes rather than individual-level trends. NDNS participants completed a 4-d food diary and an interview providing information on eating and drinking habits, anthropometric measurements, sociodemographic characteristics of the household, medication use and supplement intake.
Ethical approval for Years 1–4 (2008/2009–2011/2012) was obtained from the Oxfordshire A Research Ethics Committee (Reference 07/H0604/113)(32). For Years 9–11 (2016/2017–2018/2019), ethical approval was obtained from the Cambridge South NRES Committee (Reference 13/EE/0016)(34). Each laboratory obtained approval from the respected Research and Development Committee where required (NatCen Social Research, National Institute of Health Research Cambridge Biomedical Research Centre)(32,34) .
Dietary intake data
Food intake data were collected using estimated food diaries for four consecutive days, beginning at any point during the week, representing a daily average of the days assessed. Participants were asked to record all food and drink consumed inside and out-of-home, including the brand name and, if applicable, the preparation method, and to describe food PS using household measures and weights from packaging. Photographic aids were also used to estimate accurate recording(32,33) . Once completed, food diaries were checked with interviewers for completeness, missing values and improved recording and were collected no later than 3 d after the last day of recording(34,35) .
The food records were coded using Diet In Nutrients Out (DINO) software(Reference Fitt, Cole and Ziauddeen36), incorporating Public Health England’s NDNS Nutrient Databank, Food Standards Agency Portion Sizes and manufacturer data where applicable(34). Nutrient intakes were calculated for each participant based on the Lower Reference Nutrient Intake (LRNI) and Reference Nutrient Intake (RNI) dietary reference values for sex and age(37).
Food group analysis
Every food item in the dietary record was assigned to one of sixty individual food group codes, and each code was assigned to one of 155 subsidiary food groups(38). For the purpose of the current analysis, an internal protocol was used to create a working database from the main NDNS food file(39), where sixty food groups were recoded into twenty-four core food groups for analysis. During the recoding process, food items that were not considered to contribute meaningfully to habitual, everyday intake were excluded from the core food group to improve clarity of the aggregated groups. The data were subsequently aggregated and restructured to investigate the mean PS, EI and frequency of key food group consumption at both time points during the recording period. Total EI was divided by the number of days’ dietary intake recorded to calculate the average EI from foods. The contribution of dietary supplements and artificial sweeteners were excluded from the current analysis.
Anthropometric measurements
In NDNS at both time points, weight was measured to the nearest 0·1 kg on a variety of weighing scales: Soehnle, Seca 850, Seca 870 or Tanita THD-305. To measure height to the nearest 0·1 cm, a portable Leicester height stadiometer was used. At the midpoint between the iliac crest and the lower rib, waist circumference (WC) was measured to the nearest 0·1 cm(32,40) . BMI was calculated using Quetelet’s formula (kg/m2), weight status (defined as normal weight, ≥ 18·5–24·99 kg/m2; and overweight/obese, 18·5–≥ 40 kg/m2), WC (> 102 cm for men; > 88 cm for women, for substantially increased risk of metabolic complications and abdominal obesity) and waist:hip ratio (WHR) (≥ 0·90 cm for men; ≥ 0·85 cm for women, for substantially increased risk of metabolic complications, an indicator of serious health conditions) were described using WHO standards cut-offs(41).
Misreporting
For the current paper, misreporting was identified by examining the % difference between reported EI and estimated energy expenditure, known as the % estimated energy requirement (EER)(Reference Henry42–Reference Jessri, Lou and L’Abbé45), calculated via the following equation: (EI-EER)/EERx100(46). EER was calculated by multiplying the estimated BMR (according to sex and age (Oxford equations))(Reference Henry42), by an estimated physical activity level of 1·63, based on average UK adult energy requirements(46). Participants who reported ≤ 30 % of their total EER were considered to be ‘under-reporters’, while those who reported ≥ 30 % were considered to be ‘over-reporters’. Participants who reported between –30 % and +30 % of their EER were considered to be ‘accurate reporters’(Reference McCrory, Hajduk and Roberts47).
Statistical analysis
Data were statistically analysed using the Statistical Package for Social Sciences (SPSS) tool (version 27, manufacturer IBM). Study weights were used in analyses accounting for sampling and non-response errors. Both combined datasets were rescaled to produce representative results(48). The Kolmogorov–Smirnov test was used to determine the data’s normality. Normally distributed data are presented as means and standard deviations. Skewed data are presented as median and interquartile range. Independent t tests were used to assess the differences in energy contribution of each food group over the 10 years, accounting for sex. χ 2 tests were used to compare categorical data. %EER and EI:EER were compared between time points by independent t test. ANCOVA, to improve precision and enhance the statistical power, was conducted to test for associations between food PS intake and consumption frequency adjusting for sex, age and misreporting (%EER). Partial correlations adjusted for sex, age and misreporting were used to determine associations between PS of energy-dense foods, fruits and vegetables and markers of adiposity (BMI, WC and WHR). The statistical significance was set at the 5 % level (P < 0·05) and the CI at 95 %.
Results
General characteristics
Anthropometry
Table 1 summarises participant characteristics, indicating a statistically significant reduction in overall BMI (26·44 kg/m2 v. 26·17 kg/m2, P = 0·003) and between the sexes (males: 26·92 kg/m2 v. 26·88 kg/m2, P = 0·039; females: 25·76 kg/m2 v. 25·45 kg/m2, P = 0·028) in the 10-year period. The women in Years 9–11 were significantly taller (162·3 cm v. 163 cm, P = 0·004) than their counterparts in Years 1–4. WC significantly decreased over the decade overall (91·10 cm v. 89·74 cm, P = 0·001) and in females (85·05 cm v. 83·25 cm, P = 0·003). No other significant difference was identified. Values refer to Years 1–4 and 9–11, respectively.
General and dietary characteristics of National Diet and Nutrition Survey of Years 1–4 (2008/2009–2011/2012) and Years 9–11 (2016/2017–2018/2019)
Table 1. Long description
The table compares general and dietary characteristics of participants in the National Diet and Nutrition Survey of Years 1-4 (2008/2009-2011/2012) and Years 9-11 (2016/2017-2018/2019). It includes data for the total sample, males, and females, divided into two age groups: Years 1-4 and Years 9-11. The table has 15 rows and 24 columns, with column headers including General characteristics and Dietary characteristics. Row labels include Age, Height, Weight, BMI, WC, WHR, and various dietary metrics. Notable trends include a significant reduction in overall BMI and waist circumference (WC) over the decade, with specific significant differences in height and BMI between males and females. The table provides median and interquartile range (IQR) values for each characteristic, along with P-values indicating statistical significance.
IQR, interquartile range; WC, waist circumference; WHR, waist:hip ratio; EI, energy intake; CHO, carbohydrates; EER, estimated energy requirements.
General and dietary characteristics presented are median and IQR (presented in one value the difference between third and first quartile).
Misreporting is calculated by the % difference between reported EI and EER via the equation (EI-EER)/EERx100(46).
* General and dietary characteristics are obtained from independent t test; categories of BMI, WC and WHR and misreporting are obtained from χ 2 test; statistically significant values are presented in bold.
Dietary characteristics
The EI showed a significant decrease overall (1809 kcal v. 1763 kcal, P = 0·044) and in males (2075 kcal v. 1989 kcal, P = 0·011) (Table 1). Carbohydrate intake presented a significant decrease overall (60·9 g v. 60·4 g, P < 0·0001) and in both sexes (males: 60·5 % v. 60·1 %, P = 0·011; females, 61·2 % v. 60·5 %, P < 0·0001) over the decade. A significant reduction overall and in both sexes was observed in total (19·8 % v. 18·1 %, P < 0·0001; males: 19·3 % v. 17·5 %; P < 0·0001; females: 20·3 % v. 18·7 %, P < 0·0001) and free sugars (13·5 % v. 11·2 %, P < 0·0001; males: 16·9 % v. 13·1 %, P < 0·0001; females: 11·5 % v. 9·9 %, P < 0·0001). Fibre intake showed a significant increase overall (17·4 g v. 18·4 g, P < 0·0001) and in both sexes (males: 18·8 g v. 19·5 g, P < 0·0001; females: 16·4 g v. 17·0 g, P < 0·0001). Protein intake increased significantly overall (20·3 g v. 20·5 g, P < 0·0001) and in males (20·2 g v. 20·6 g, P = 0·036) while remaining the same in females (20·4 g v. 20·4 g, P = 0·014) over the decade. The fat consumption presented a significant increase overall (18·6 g v. 19·0 g, P < 0·0001) and in females (18·4 g v. 19·0 g, P < 0·0001). No other differences were observed in dietary characteristics over the 10 years.
Misreporting
No change over time was noted in the EER or %EER among the sample overall or in either males or females (Table 1). No difference in EI:EER between the time points in the overall sample or in females was observed. However, EI:EER was significantly higher among the males in Years 1–4 compared with Years 9–11 (0·73 v. 0·69, P = 0·001), suggesting greater misreporting of EI in Years 9–11 among males. The overall percentage of the sample size identified as under-reporters was similar between time points (48·1 % v. 51·3 %) and among females (50·9 % v. 51 %), while a higher % of males were considered to be accurate reporters in Years 1–4 compared with Years 9–11 (53·6 % v. 46·8 %; P = 0·007, respectively). Misreporters were included in the total sample to retain the sample size.
Comparison of food groups percentage consumption
Table 2 illustrates the percentage difference in food group consumption between both time points in the overall sample. The highest increase trend was noted in the consumption of nuts & seeds (20·5 %), condiments (14·7 %), eggs (11 %), low-calorie carbonated & soft drinks (8 %) and pasta, rice & pizza (7·6 %). In contrast, a notable decrease was noted in the consumption of carbonated & soft drinks (–12·9 %), breakfast cereals (–8·7 %), red meat (–8·6 %), processed meat (–7·1 %) and alcoholic drinks (–5·3 %). Milk and milk products showed no difference over the decade.
Percentage consumption of food groups in adults aged 18–65 years in Years 1–4 (2008/2009–2011/2012) and Years 9–11 (2016/2017–2018/2019) of the NDNS
Table 2. Long description
The table presents data on the percentage consumption of various food groups by adults aged 18-65 years during two time periods: Years 1-4 (2008/2009-2011/2012) and Years 9-11 (2016/2017-2018/2019). The table includes 23 food groups, with each row listing the food group, the number of participants (n), and the percentage consumption for both time periods, along with the percentage difference between the two periods. Row 1: Pasta, rice & pizza, Years 1-4: n 1853, 73.9 percent, Years 9-11: n 1533, 81.5 percent, Percentage difference: 7.6. Row 2: Bread, Years 1-4: n 2436, 97.2 percent, Years 9-11: n 1749, 93.1 percent, Percentage difference: -4.1. Row 3: Breakfast cereals, Years 1-4: n 1544, 61.6 percent, Years 9-11: n 994, 52.9 percent, Percentage difference: -8.7. Row 4: Buns, cakes, pastries & biscuits, Years 1-4: n 1965, 78.4 percent, Years 9-11: n 1419, 75.5 percent, Percentage difference: -2.9. Row 5: Milk & milk products, Years 1-4: n 2468, 98.5 percent, Years 9-11: n 1852, 98.5 percent, Percentage difference: 0. Row 6: Eggs, Years 1-4: n 1373, 54.8 percent, Years 9-11: n 1237, 65.8 percent, Percentage difference: 11. Row 7: Oils & spreads, Years 1-4: n 2368, 94.5 percent, Years 9-11: n 1830, 97.4 percent, Percentage difference: 2.9. Row 8: Processed meat, Years 1-4: n 1975, 78.8 percent, Years 9-11: n 1348, 71.7 percent, Percentage difference: -7.1. Row 9: Red meat, Years 1-4: n 1702, 67.9 percent, Years 9-11: n 1115, 59.3 percent, Percentage difference: -8.6. Row 10: White meat, Years 1-4: n 1812, 72.3 percent, Years 9-11: n 1420, 75.6 percent, Percentage difference: 3.3. Row 11: Fish, Years 1-4: n 1516, 60.5 percent, Years 9-11: n 1079, 57.4 percent, Percentage difference: -3.1. Row 12: Vegetables, Years 1-4: n 2448, 97.7 percent, Years 9-11: n 1853, 98.6 percent, Percentage difference: 0.9. Row 13: Potato & potato products, Years 1-4: n 2256, 90.0 percent, Years 9-11: n 1635, 87.0 percent, Percentage difference: -3.0. Row 14: Fruits, Years 1-4: n 1946, 77.6 percent, Years 9-11: n 1505, 80.0 percent, Percentage difference: 2.4. Row 15: Condiments, Years 1-4: n 1727, 68.9 percent, Years 9-11: n 1571, 83.6 percent, Percentage difference: 14.7. Row 16: Savoury snacks, Years 1-4: n 1225, 48.9 percent, Years 9-11: n 955, 50.8 percent, Percentage difference: 1.9. Row 17: Sugar & chocolate confectionery, Years 1-4: n 1433, 57.2 percent, Years 9-11: n 1098, 58.4 percent, Percentage difference: 1.2. Row 18: Fruit juice & smoothies, Years 1-4: n 1063, 42.4 percent, Years 9-11: n 733, 39.0 percent, Percentage difference: -3.4. Row 19: Alcoholic drinks, Years 1-4: n 1467, 58.5 percent, Years 9-11: n 1000, 53.2 percent, Percentage difference: -5.3. Row 20: Beverages, Years 1-4: n 2464, 98.3 percent, Years 9-11: n 1875, 99.8 percent, Percentage difference: 1.5. Row 21: Nuts & seeds, Years 1-4: n 536, 21.4 percent, Years 9-11: n 787, 41.9 percent, Percentage difference: 20.5. Row 22: Carbonated & soft drinks, Years 1-4: n 1306, 52.1 percent, Years 9-11: n 736, 39.2 percent, Percentage difference: -12.9. Row 23: Carbonated & soft drinks low calories, Years 1-4: n 875, 34.9 percent, Years 9-11: n 805, 42.9 percent, Percentage difference: 8. Row 24: Soups, sauces, other beverages & condiments, Years 1-4: n 2378, 94.9 percent, Years 9-11: n 1865, 99.2 percent, Percentage difference: 4.3.
NDNS, National Diet and Nutrition Survey.
* Bread types include white, wholemeal, brown granary, wheatgerm and other breads.
† Cereal types include high-fibre breakfast and other breakfast cereals.
‡ Oils and spreads types include butter, PUFA margarine/oils, low-fat spread margarine and non-PUFA-reduced-fat spread.
§ Processed meat types include burgers and kebabs, sausages, meat pies and pastries and other meat and meat products.
|| Beverages include tea, coffee and water.
Comparison of consumption frequency and portion size of food groups
When one-way ANCOVA adjusted for age, sex and %EER was applied, several statistically significant changes in both daily frequency of consumption and PS were observed between Years 1–4 and Years 9–11 (Table 3). Among the analysed food groups, significant increases were observed for pasta, rice and pizza (frequency: 0·51 (sd 0·31) v. 0·60 (sd 0·40) times/d, P < 0·001; PS: 79·1 g v. 87·5 g/d, P = 0·003), eggs (frequency: 0·44 (sd 0·29) v. 0·53 (sd 0·34) times/d, P = 0·001; PS: 28·5 g v. 31·0 g/d, P = 0·001), oils and spreads (frequency: 1·22 (sd 0·76) v. 1·77 (sd 1·07) times/d, P = 0·001; PS: 11·7 g v. 12·9 g/d, P = 0·001), vegetables (frequency: 3·12 (sd 2·15) to 4·05 (sd 2·88), P = 0·001; PS: 149·1 g to 168·6 g/d, P = 0·001), fruits (frequency: 1·39 (sd 1·13) to 1·77 (sd 1·59) times/d, P = 0·001; PS: 93·1 g to 105·9 g/d, P = 0·005), nuts and seeds (frequency: 0·64 (sd 0·67) to 1·10 (sd 1·43) times/d, P = 0·001; PS: 8·2 g v. 11·0 g/d, P = 0·001) and low-calorie carbonated/soft drinks (frequency: 1·40 (sd 1·50) v. 1·78 (sd 1·79) times/d, P = 0·001; PS: 200·0 g to 267·1 g/d, P = 0·001). In contrast, significant decreases in both frequency and PS were found for bread (frequency: 1·23 (sd 0·58) v. 1·10 (sd 0·57) times/d, P = 0·001; PS: 81·0 g v. 73·6 g/d, P = 0·001), potato & potato products (frequency: 0·65 (sd 0·30) v. 0·61 (sd 0·32) times/d, P = 0·001; PS: 90·0 g to 78·8 g/d, P = 0·001), condiments (frequency: 1·85 (sd 1·77) v. 1·58 (sd 1·47) times/d, P = 0·001; PS: 13·0 g v. 9·0 g/d, P = 0·001) and fruit juice & smoothies (frequency: 0·66 (sd 0·52) v. 0·62 (sd 0·49) times/d, P = 0·032; PS: 87·5 g to 62·5 g/d, P = 0·001).
Differences in frequency of consumption and portion size (g) per day among UK adults aged 18–65 years from Years 1–4 (2008/2009–2011/2012) v. Years 9–11 (2016/2017–2018/2019) of the NDNS
Table 3. Long description
The table presents data on the frequency of consumption and portion size (g/d) of various food groups among two age groups, Years 1-4 and Years 9-11. It includes mean and standard deviation (SD) for frequency consumption, and median and interquartile range (IQR) for portion size consumption. The food groups analyzed include pasta, rice and pizza, bread, breakfast cereals, buns, cakes, pastries and biscuits, milk and milk products, eggs, oils and spreads, processed meat, red meat, white meat, fish, vegetables, potato and potato products, fruits, condiments, savoury snacks, sugar and chocolate confectionery, fruit juice and smoothies, alcoholic drinks, beverages, nuts and seeds, carbonated and soft drinks, carbonated and soft drinks low calories, and soups, sauces, other beverages and condiments. Notable trends include significant increases in the consumption of pasta, rice and pizza, eggs, oils and spreads, vegetables, fruits, nuts and seeds, and low-calorie carbonated/soft drinks, and significant decreases in the consumption of bread, potato products, condiments, and fruit juice smoothies between the two age groups.
NDNS, National Diet and Nutrition Survey; PS, portion size; IQR, interquartile range.
Frequency consumption presented are mean and standard deviations, and PS consumption presented are median and IQR (presented in one value the difference between third and first quartile).
* P-value obtained from one-way ANCOVA adjusted for age, sex and misreporting; statistically significant values are presented in bold.
† Bread types include white, wholemeal, brown granary, wheatgerm and other breads.
‡ Breakfast cereal types include high-fibre breakfast and other breakfast cereals.
§ Oils and spreads types include butter, PUFA margarine/oils, low-fat spread margarine and non-PUFA-reduced-fat spread.
|| Processed meat types include burgers and kebabs, sausages, meat pies and pastries and other meat and meat products.
¶ Beverages include tea, coffee and water.
Several food groups displayed different trends, characterised by a significant increase in consumption frequency alongside a simultaneous reduction in PS. These included pasta, rice & pizza (frequency: 0·51 (sd 0·31) v. 0·60 (sd 0·40) times/d, P < 0·001; PS: 79·1 g v. 87·5 g, P = 0·003), red meat (frequency: 0·46 (sd 0·25) v. 0·49 (sd 0·25) times/d, P = 0·012; PS: 50·0 g v. 43·8 g, P = 0·001) and white meat (0·54 (sd 0·35) v. 0·61 (sd 0·38) times/d, P = 0·001; PS: 56·3 g v. 52·2 g, P = 0·144). A further group of foods demonstrated a significant increase in consumption frequency without a corresponding change in PS, including sugar & chocolate confectionery (0·62 (sd 0·52) v. 0·75 (sd 0·58) times/d, P = 0·001), savoury snacks (0·50 (sd 0·32) v. 0·55 (sd 0·37) times/d, P = 0·001), white meat (0·54 (sd 0·35) to 0·61 (sd 0·38), P = 0·001) and fish (0·49 (sd 0·37) v. 0·56 (sd 0·39), P = 0·001), and in PS only for milk & milk products (172·9 g v. 172·0 g/d, P = 0·042), processed meat (46·0 g v. 39·7 g/d, P = 0·001), alcoholic drinks (256·3 g v. 200·0 g/d, P = 0·001) and beverages (1022·3 g v. 1234·8 g/d, P = 0·001).
Differences in portion size of food groups according to weight status
Table 4 presents the differences in average daily PS of food groups in both time points stratified by weight status, using one-way ANCOVA adjusted for age, sex and %EER. Among normal-weight adults, a significant increase in PS over time was observed for pasta, rice & pizza (84·7 g v. 89·3 g, P = 0·020), vegetables (152·1 g v. 176·9 g, P = 0·001), beverages (986·7 g v. 1254·1 g, P = 0·001), nuts & seeds (7·7 g v. 12·0 g, P = 0·001) and oils & spreads (12·1 g v. 13·2, P = 0·019). Contrary, a significant decrease was observed for bread (80·1 g v. 74·7 g, P = 0·001), processed meat (43·3 g v. 35·6 g, P = 0·001), red meat (46·9 g v. 39·4 g, P = 0·001), potato & potato products (85·5 g v. 71·3 g, P = 0·001), condiments (15·0 g v. 10·0 g, P = 0·001), fruit juice & smoothies (100·0 g v. 50·0 g, P = 0·001) and alcoholic drinks (210·3 g v. 185·2 g, P = 0·001). No significant changes were observed in the other food groups.
Differences in portion size consumption (g) per day of food groups according to weight status between Years 1–4 (2008/2009–2011/2012) and Years 9–11 (2016/2017–2018/2019) of the NDNS
Table 4. Long description
The table presents the differences in average daily portion sizes (PS) of food groups between two time periods (Years 1-4 and Years 9-11) stratified by weight status (normal weight and overweight/obese). The table includes median and interquartile range (IQR) values for each food group and weight status category. It also indicates significant changes over time using P-values. The table has 25 rows and 12 columns. Column headers include Normal weight (Years 1-4 and Years 9-11) with Median and IQR values, Overweight/obese (Years 1-4 and Years 9-11) with Median and IQR values, and P-values for each food group. Row labels include various food groups such as Pasta, rice & pizza, Bread, Breakfast cereals, Buns, cakes, pastries & biscuits, Milk & milk products, Eggs, Oils & spreads, Processed meat, Red meat, White meat, Fish, Vegetables, Potato & potato products, Fruits, Condiments, Savoury snacks, Sugar & chocolate confectionery, Fruit juice & smoothies, Alcoholic drinks, Beverages, Nuts & seeds, Carbonated & soft drinks, Carbonated & soft drinks low calories, Soups, sauces, other beverages & condiments.
NDNS, National Diet and Nutrition Survey; PS, portion size; IQR, interquartile range.
PS consumption presented are median and IQR (presented in one value the difference between third and first quartile).
* P-value obtained from one-way ANCOVA adjusted for age and misreporting; statistically significant values are presented in bold.
† Bread types include white, wholemeal, brown granary, wheatgerm and other breads.
‡ Breakfast cereal types include high-fibre breakfast and other cereals.
§ Oils and spreads types include butter, PUFA margarine/oils, low-fat spread margarine and non-PUFA-reduced-fat spread.
|| Processed meat types include burgers and kebabs, sausages, meat pies and pastries and other meat and meat products.
¶ Beverages include tea, coffee and water.
Among adults with overweight/obesity, a significant increase was noted for pasta, rice & pizza (75·0 g v. 86·1 g, P = 0·001), eggs (29·3 g v. 33·5 g, P = 0·011), vegetables (146·2 g v. 168·0 g, P = 0·001), oils & spreads (11·5 g v. 12·9 g, P = 0·001), fruits (93·5 g v. 105·6 g, P = 0·009), sugar & chocolate confectionery (19·2 g v. 21·5 g, P = 0·005), beverages (1064·3 g v. 1225·6 g, P = 0·001), nuts & seeds (8·0 g v. 9·7 g, P = 0·009) and low-calorie carbonated & soft drinks (200·0 g v. 324·4 g, P = 0·001). A decrease in PS was observed for bread (81·7 g v. 72·5 g, P = 0·001), processed meat (48·8 g v. 43·1 g, P = 0·022), red meat (52·5 g v. 48·0 g, P = 0·016), potato & potato products (95·0 g v. 84·8 g, P = 0·001), fruit juice & smoothies (78·6 g v. 71·0 g, P = 0·045) and condiments (12·1 g v. 8·0 g, P = 0·001). No other significant changes were observed.
Associations between energy-dense, fruit and vegetable food groups and measures of adiposity over a 10-year period in the National Diet and Nutrition Survey
Partial correlations (r-values) were performed to examine the relationships between average intakes (g) of energy-dense and healthy food groups and measures of adiposity (BMI, WC and WHR), while controlling for sex, age and misreporting (Table 5). Although several correlations reached statistical significance (P < 0·05), all associations were small in magnitude, indicating weak relationships between food group intakes and adiposity outcomes. Among energy-dense food groups, carbonated and soft drinks showed the most consistent positive associations with adiposity, particularly for central measures. Positive correlations with WC were observed in both time points (Years 1–4: r 0·072, P = 0·020; Years 9–11: r 0·157, P < 0·001) and with WHR in Years 9–11 (r 0·153, P = 0·003), but not in Years 1–4 (r 0·033, P = 0·292). Sugar & chocolate confectionery intake was positively associated with BMI (r 0·125, P < 0·001) and WC (r 0·157, P < 0·001) only in Years 9–11, whereas no significant associations were observed in Years 1–4 (BMI: r 0·017, P = 0·507; WC: r 0·006, P = 0·823; WHR: r −0·045, P = 0·110). Buns, cakes and pastries & biscuits were positively associated with BMI (r 0·057, P = 0·008) and WC (r 0·073, P = 0·003) only in Years 1–4, with no significant associations in Years 9–11 (BMI: r 0·053, P = 0·081; WC: r 0·030, P = 0·400). Oils & spreads showed a small positive association with BMI in Years 1–4 (r 0·044, P = 0·026), but not in Years 9–11 (r 0·035, P = 0·203), and no significant associations with WC (Years 1–4: r 0·033, P = 0·135; Years 9–11: r 0·038, P = 0·240) or WHR (Years 1–4: r 0·024, P = 0·274; WHR Years 9–11: r 0·032, P = 0·313). No significant associations were observed for savoury snacks across any adiposity measures or time points. Using fruit and vegetable as diet quality markers, fruit intake was inversely associated with adiposity, particularly WHR. Fruit intake was negatively correlated with WHR in both Years 1–4 (r −0·069, P = 0·005) and Years 9–11 (r −0·100, P = 0·004), with additional inverse associations observed for BMI in Years 1–4 (r −0·056, P = 0·011) and WC in Years 9–11 (r −0·094, P = 0·007). Vegetable intake was not significantly associated with BMI (Years 1–4: r −0·008, P = 0·668; BMI Years 9–11: r 0·042, P = 0·120) or WC (Years 1–4: r −0·038, P = 0·124; WC Years 9–11: r −0·053, P = 0·099); however, it was inversely correlated with WHR in both time points (Years 1–4: r −0·071, P = 0·001; Years 9–11: r –0·064, P = 0·044).
Partial correlations between intakes (g) of energy-dense food groups and fruits/vegetables and measure of adiposity (BMI, WC and WHR) between Years 1–4 (2008/2009–2011/2012) and Years 9–11 (2016/2017–2018/2019) of the NDNS
Table 5. Long description
The table presents partial correlations (r-values) between average intakes of energy-dense and healthy food groups and measures of adiposity (BMI, WC, and WHR), controlling for sex, age, and misreporting. The table has 12 columns and 8 rows, including row labels and column headers. The columns are grouped into three sections: BMI (kg/m^2), WC (cm), and WHR (cm), each with data for Years 1-4 and Years 9-11, including R and P values. The row labels include different food groups such as savoury snacks, carbonated and soft drinks, buns, cakes, pastries and biscuits, sugar and chocolate confectionery, oils and spreads, fruits, and vegetables. Each row provides the correlation values and significance for the respective food groups with the measures of adiposity across the specified years. Notable trends include positive associations of carbonated and soft drinks with central measures of adiposity, particularly in Years 9-11, and inverse associations of fruit intake with WHR in both time points. Vegetable intake shows inverse correlations with WHR but not with BMI or WC.
WC, waist circumference; WHR, waist:hip ratio; NDNS, National Diet and Nutrition Survey.
R-value and P-value are obtained from partial correlation, adjusted for age, sex and misreporting; statistically significant values presented in bold.
Discussion
The current study examined decade-long changes in the prevalence, frequency of consumption and PS of food group consumption among UK adults aged 18–65 years using NDNS data. Overall, shifts in food patterns were heterogeneous across food groups, with notable divergence between the proportion of consumers, consumption frequency and PS. The proportion of consumers increased for pasta, rice & pizza, eggs, oils & spreads, white meat, vegetables, fruits, condiments, savoury snacks, sugar & chocolate confectionery, beverages, nuts & seeds, low-calorie carbonated & soft drinks and condiments. Whereas it decreased for bread, breakfast cereals, buns, cakes, pastries & biscuits, processed meat, red meat, fish, potato & potato products, fruit juice & smoothies, alcoholic drinks and carbonated & soft drinks.
When food groups were considered in relation to the Eatwell Guide, changes in consumption frequency across core food groups were mixed rather than uniform over time. Frequency increased for fruits and vegetables but decreased for bread and potato and potato products, while no clear change was evident for milk and milk products. Discretionary food groups also showed variable trends, with increased frequency for sugar & chocolate confectionery and savoury snacks, but decreases for condiments and fruit juice & smoothies. Changes in PS were similarly heterogeneous, with reductions observed for processed meat, condiments and alcoholic drinks, alongside increases for fruits, vegetables, beverages, nuts & seeds and low-calorie carbonated & soft drinks. Overall, these findings suggest inconsistent alignment with food-based dietary guidelines in the adult population(49).
These findings partly contrast with findings from other European populations. For example, a Norwegian study reported increased dairy products consumption alongside reductions in fruit and vegetable intake, trends not consistently observed in the present analysis(Reference Huseinovic, Hornell and Johansson50). Furthermore, a Danish cohort and cross-sectional study reported a decreased frequency of white bread and potato consumption(Reference Osler, Heitmann and Schroll51). Adults in the USA generally exhibit higher frequencies of consumption across most food groups compared with populations in Brazil and other countries, except for staple foods such as rice, beans and legumes(Reference Bezerra, Goldman and Rhodes52). These discrepancies likely reflect differences in food environments, cultural dietary norms and national nutrition policies.
In line with these findings, changes in PS were heterogeneous across food groups over the decade, with reductions observed for several energy-dense foods, including processed meat, red meat, potato and potato products and condiments. Only a limited number of food groups, including pasta, rice & pizza, eggs, vegetables, beverages, nuts & seeds, fruits and low-calorie carbonated & soft drinks, showed increases in PS. The increase in consumption of low-calorie carbonated & soft drinks may reflect reformulation and consumer response following the introduction of the UK Soft Drinks Industry Levy(Reference Bandy, Scarborough and Harrington53,Reference Scarborough, Adhikari and Harrington54) , suggesting that fiscal policies may influence PS and product choice beyond sugar-sweetened beverages alone.
An Australian study reported increased PS consumption of processed foods, which decreased in the present study. However, Australian participants were older overall, and the study methodology differed, as the dietary intake was measured with different methods(Reference Zheng, Rangan and Meertens24). An Irish study supports the findings of this research only regarding the decreased consumption of potato and potato products and processed meat. Notably, it did not account for under-reporting, and the PS was accounted for, regardless of the quantification method used(Reference O’Brien, Livingstone and McNulty25).
This study partially contradicts the evidence from the USA, where most foods’ PS consumption increased consistently over time(Reference Nielsen and Popkin55,Reference Smiciklas-Wright, Mitchell and Mickle56) . For instance, while the daily PS of foods such as French fries reportedly increased in the USA, the present study found a significant decrease in PS for potato & potato products. Beverage PS consumption has increased in both the USA and the present study. Such differences may partly reflect variation in dietary assessment methods and differences in eating location and food source patterns(Reference Nielsen and Popkin55). Moreover, while the USA study focused on widely consumed food patterns in a wider age range(Reference Smiciklas-Wright, Mitchell and Mickle56), this study explored all food groups in the UK adult population. Other USA studies have indicated decreased sugar-sweetened beverage consumption, which aligns with the current study findings(Reference Nielsen and Popkin57–Reference Vercammen, Moran and Soto60).
Between 1987 and 2006, PS increased in savoury foods, sugar and chocolate products and fast foods, while it decreased in others, such as potatoes/chips, desserts, yogurts, ice creams, breaded chicken and some fruits and vegetables(Reference Clift28,Reference Wrieden, Gregor and Barton30,Reference Church61) . Findings of 2000/1 and 2008/9 NDNS surveys support these trends, as PS varied with some foods increasing, some decreasing and others remaining constant(Reference Whitton, Nicholson and Roberts62). However, the 2008/9 NDNS data (Year 1, RP) had a smaller sample size than the 2000/1 survey. Increased consumption of cereal, grain and meat is indicated in the National Food Surveys (1949–2005)(Reference Benson63). However, it is not reflected in the present study, possibly due to evolving food trends. Data from the UK National Survey of Health and Development noted an increased PS of fruits and vegetables and decreased whole milk, butter and red meat intake in both sexes(Reference Prynne, Paul and Mishra64). The British Birth Study identified PS trends similar to the current study, with increases in bread (wholemeal and granary), fish, fruits, vegetables and semi-skimmed milk. Decreases were noted in white bread, whole milk, fats and oils, meat and meat products, coffee, sugar, preserves and alcoholic drinks(Reference Pot, Prynne and Almoosawi65). Some studies focused on PS changes based on manufacturer weights or examining food groups(Reference Young and Nestle14,Reference Osler, Heitmann and Schroll51,Reference Young and Nestle66,Reference Doody, Wreiden and Barton67) . In contrast, others examined specific food changes rather than those contributing most to EI using dietary data(Reference Nielsen and Popkin55,Reference Young and Nestle68) or combining both(Reference Benson63,Reference Prynne, Paul and Mishra64) . Three were cohort studies(Reference Osler, Heitmann and Schroll51,Reference Prynne, Paul and Mishra64,Reference Pot, Prynne and Almoosawi65) with repeated dietary measurements from the same individuals. Overall, the mixed direction of changes observed across studies, including the present one, might be due to the population’s ageing, food trends and changes in food availability over time.
In line with earlier studies(Reference Zheng, Rangan and Meertens24,Reference O’Brien, Livingstone and McNulty25) , this study discovered a 60·4 % decrease in carbohydrate consumption over the decade, yet consumption remains greater than the recommendations. Although total and free sugar intake has decreased over the decade(Reference Rauber, Steele and Louzada69), the population continues to consume higher amounts than is recommended primarily because of processed foods(70). Although fibre intake has increased, mainly due to veganism(Reference Selwyn71), it remains below the recommended 30 g/d(72). Fat intake has increased by 19 % but remained below the recommendations, similar to previous research(Reference Zheng, Rangan and Meertens24,Reference O’Brien, Livingstone and McNulty25) . Protein consumption has risen by 20·5 %, exceeding recommendations(73). Lifestyle changes(Reference Davis, Clarke and Coulter74), sustainability(Reference Hartmann and Siegrist75), increased need for quick protein sources(Reference Bauerle and Kuhn76) and alternative protein options(Reference Joseph, Searing and Watson77,Reference Karmaus and Jones78) have been attributed to this.
The current study identified modest associations between energy-dense food group intakes and measures of adiposity, with patterns differing between the two NDNS survey periods. While overweight and obesity remain highly prevalent among UK adults, diet–adiposity relationships were generally weak. Contrary to the study hypothesis that overweight/obese individuals would consume larger portions compared with their normal-weight counterparts, the observed differences were not consistent across food groups. These findings contrast with previous research showing obese individuals self-reporting larger PS(Reference Berg, Lappas and Wolk8,Reference Pereira, Mendes and Crispim79,Reference Rippin, Hutchinson and Jewell80) and that obesity is associated with increased consumption of various foods and food groups(Reference Joseph, Searing and Watson77,Reference Kelly, Rennie and Wallace81) . Previous research also indicates that high-energy and processed/ultra-processed foods are linked with adiposity and, thus, weight gain and obesity(Reference Rauber, Steele and Louzada69,Reference Du, van der A and Ginder82–Reference Livingstone, Sexton-Dhamu and Pendergast85) . Temporal variation in associations was evident, with sugar & chocolate confectionery intake positively associated with BMI in Years 9–11, and oils & spreads associated with BMI in Years 1–4. Stronger associations were observed for central adiposity, with carbonated & soft drink intake positively related to WC and WHR in Years 9–11. In contrast, fruit intake was inversely associated with several adiposity measures, while vegetable intake showed an inverse association with WHR only. Although statistically significant, the effect sizes were small, suggesting limited clinical relevance. As misreporters were retained, these associations might have been weakened, given that obese individuals tend to under-report their intake(Reference Goris, Westerterp-Plantenga and Westerterp86).
This study has several public health implications, particularly for developing strategies to combat obesity in the UK. Limiting multi-buy deals(87) and junk food advertising policies(88) were announced in 2020; however, while restrictions on multi-buy deals came into effect on 1 October 2025(87) in England, implementation of junk food advertising restrictions came into effect on 5 January 2026(89). Evidence supports the benefits of fat reduction content in foods, extending beyond the sugar levy(70), as observed successfully in international measures, such as energy-dense foods taxation(Reference Biro90–Reference Reyes, Smith Taillie and Popkin92). Introducing mandatory calorie labelling in UK restaurants(93,Reference Rincon-Gallardo, Zhou and da Silva Gomes94) promotes lower-calorie options(Reference Grummon, Petimar and Soto95). A UK study indicated that a 20 % fat reduction in manufactured and out-of-home foods could reduce overweight and obesity rates by 1·5 % and 5·3 %, respectively, and prevent type 2 diabetes over 20 years(Reference Alessandrini, He and Ma96).
Despite the obstacles presented by culturally different eating habits, governments and organisations worldwide must collaborate to provide transparent technical regulations and standards, including labelling, packaging and marketing(Reference Delhomme97). Front-of-pack nutrition labels help educate the public about healthy eating(Reference Kliemann, Kraemer and Scapin98) despite claims that they are a marketing tool rather than a public health strategy(Reference Brownell and Koplan99). Notably, nutrition labels are not required to include PS information(Reference Delhomme97,Reference van Camp, de Souza Monteiro and Hooker100–Reference Temple102) .
The NDNS is the only survey providing nationally representative data on food consumption and nutrient intake in the UK, monitoring and investigating dietary trends. Data collection through food diaries is a comprehensive method that helps minimise missing information(Reference Ashwell, Barlow and Gibson31,32) . However, misreporting in self-reported dietary methods remains an issue that requires further investigation(Reference Macdiarmid and Blundell103). EI may be underestimated due to PS estimates or food composition assumptions being measured in household measures(104). In addition, public health campaigns raising awareness of healthy eating may lead to under-reporting of high-fat foods(Reference Heitmann, Lissner and Osler105).
The study included all the participants for the statistical analysis, as misreporters’ exclusion may introduce selection and unknown bias(Reference Heitmann, Lissner and Osler105,Reference Poslusna, Ruprich and de Vries106) . Weekend days were overrepresented in the NDNS RP Years 1–4 combined data. This possibly weakened some expected relationships and introduced bias in reported food consumption and nutrition intake, as dietary patterns differ for sex and age groups(Reference Thane and Stephen107). While the data were weighted, including the non-responses, bias reduction is not guaranteed, as the weighting is equivalent to replacing the non-respondents from a sub-group with respondents from the same sub-group(104,108) .
This study recoded the NDNS food groups into core food groups, making direct comparisons with other studies difficult. When disaggregated composite food codes, fruit and vegetable intake can increase by up to 11 % and 29 %, respectively(Reference Krebs-Smith, Cook and Subar109,Reference O’Brien, Kiely and Galvin110) . While beans and pulses are included in the protein food group in the Eatwell Guide, they are classified in the vegetable food group in the NDNS(39,Reference Fitt, Mak and Stephen111) . This may explain the increase in vegetable consumption. The recoded food groups combined diverse items, such as different types of processed meat, milk & milk products or oils & spreads, which may mask important within-group variation in nutrient composition, processing level and associated health impacts. In particular, food groups such as alcoholic drinks can vary widely in energy content, and PS, and changes in their consumption may have a disproportionate impact on total EI. Finally, age, socio-economic status and geographic characteristics were not accounted for, which may have limited the representativeness of the results and the ability to identify specific trends.
These findings highlight changes in the UK adult population’s consumption patterns across several commonly consumed food groups over the past decade. Some reductions in PS, including those for energy-dense foods such as processed meat, may represent favourable trends; however, the overall pattern of change was mixed. The findings also provide scope for policy interventions to increase fibre intake and reduce sugar consumption, particularly among overweight/obese individuals. Monitoring trends through nationally representative surveys remains critical to inform targeted health promotion campaigns and support healthier dietary patterns.
Acknowledgements
This work has been conducted as part of the MSc in Food Regulatory Affairs course at the University of Ulster and University College Dublin.
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Maria Papagiannaki: Conceptualisation-Lead, Data curation-Lead, Formal analysis-Lead, Investigation-Lead and Methodology-Lead; Maeve A. Kerr: Supervision-Supporting and Writing – review & editing-Supporting.
This research was undertaken as part of the MSc Food Regulatory Affairs by M. P., who contributed to the development of the research question, conducted the literature search, interpreted the findings and drafted the manuscript. Dr M. A. K. contributed to the development of the research question and provided supervision and guidance throughout the research process.
No conflict of interest.
