What this is
- This research examines changes in eating patterns among Korean adults from 1998 to 2018.
- It focuses on the percentage of energy contributed by main meals and snacks, as well as the sources of these foods.
- Using data from the Korea National Health and Nutrition Examination Surveys, it analyzes trends in energy intake from minimally processed and ultra-processed foods.
Essence
- From 1998 to 2018, Korean adults reduced energy intake from breakfast while increasing it from snacks, with a notable rise in ultra-processed food consumption.
Key takeaways
- Energy from breakfast decreased from 25.0% in 1998 to 16.7% in 2018, while energy from snacks increased from 14.0% to 19.0%. This shift indicates a potential compensatory behavior in eating patterns.
- At all eating occasions, energy from minimally processed foods declined, while energy from ultra-processed foods increased significantly, particularly among younger adults.
- Younger adults (<50 years) exhibited greater changes in energy intake patterns compared to older adults, suggesting a shift towards less healthy eating habits.
Caveats
- Self-reported dietary assessments may lead to measurement errors, including underreporting of energy intake, particularly among specific demographics.
- The study's findings may not be generalizable beyond Korean adults aged 20-69 years, limiting broader applicability.
- The NOVA classification system used for food categorization may not fully capture the nuances of Korean cuisine, potentially affecting the accuracy of food source assessments.
Definitions
- NOVA system: A food classification system categorizing foods based on the extent and purpose of their processing.
Simplified
Introduction
Nutrition research has transitioned from individual nutrient- or food-based approaches to meal-based approaches to better understand the effect of the whole diet on health [1]. Although there is no accepted definition, the term “eating patterns” includes frequency of meals, regularity of meals, meal skipping, meal timing, meal food sequencing, nutrient composition, presence of others at a meal, and eating location [1]. Previous studies have shown a link between unfavorable eating patterns (e.g., skipping breakfast, lower meal frequency (≤ 2 meals/day), eating alone, late eating) and poor nutritional quality [2–6] and/or health outcomes, including obesity-related measures [3, 4, 6–13].
Among various eating patterns, characterizing changes in the distribution of eating throughout the day may be helpful to determine whether changes in one eating behavior lead to compensatory changes in other eating behaviors [14, 15]. For example, it was previously reported that increased energy consumption from snacking was compensated for by decreasing energy from main meals in US adults [16]. Additionally, consuming more energy early in the day than later in the day may provide physiological benefits, including decreased inflammation and improved circadian rhythm [17]. Above all, what and how much food is eaten at each eating occasion can be more important. For example, if someone consumes most of their energy from breakfast but mostly from unhealthy foods (e.g., ultra-processed foods), then we cannot conclude that this person follows a healthy eating pattern.
An increase in the frequency of eating occasions, particularly the frequency of snacking, and its subsequent contribution to energy intake have become evident in numerous countries. For instance, data indicate a rise in meal frequency from three to five times a day and an increase in the percentage of energy from snacks between 1971 and 2010 among US adults [14]. In the Chinese population, upward trends in the percentage of energy from snacks were observed between 1991 and 2018 [18]. A prospective cohort study has demonstrated a positive association between self-reported between-meal snacking and a risk of substantial weight gain [19]. Furthermore, in numerous instances, heightened snacking habits have been linked to increased consumption of ultra-processed foods, which, in turn, is associated with an elevated risk of developing obesity and various metabolic disorders [20, 21].
A few studies reported that Koreans have also changed their eating patterns during the last decade. Between 2011 and 2019, consumption of grains, vegetables, fruits, and alcoholic drinks decreased, while that of beverages and meat increased [22]. The proportion of skipping breakfast increased, while that of skipping lunch or dinner largely remained unchanged [22]. Between 2010 and 2018, there were increasing trends in ultra-processed food consumption [23]. However, there is no study characterizing the trends in energy distribution throughout the day including snacking and food sources in Korea.
We examined secular trends in the energy intake contribution of main meals and snacks and the four food groups categorized by NOVA in each main meal and snack among Korean adults aged 20–69 years using the Korea National Health and Nutrition Examination Survey (KNHANES) data from 1998, 2007–2009, and 2016–2018.
Methods
Study population
KNHANES is a cross-sectional and nationally representative survey administered by the Korea Disease Control and Prevention Agency to monitor the health and nutritional status of the Korean civilian, noninstitutionalized population using a complex and multistage probability sampling design [24]. KNHANES data are collected through health interviews, health examinations including the collection of biomarkers, and nutrition surveys. More details about KNHANES data collection methods have been published elsewhere [24]. The present study combines data from the 1st (1998), 4th (2007–2009), and 7th (2016–2018) KHANES.
Among 32,228 adults aged 20–69 years who completed both health interviews and nutrition surveys, we excluded those with the following conditions: implausible energy intake based on < 1st or > 99th percentile (< 500 or > 5000 kcal/day) (n = 1125); pregnancy or breastfeeding (n = 255); and severe chronic diseases, including heart diseases, cancer, liver cirrhosis, and renal failure (n = 1459). The final analytic sample included 29,389 adults (12,329 males and 17,060 females) (Supplemental Fig. 1).
Data collection
In mobile examination centers, trained interviewers conducted health interviews to collect detailed information on sociodemographic characteristics, health-related behaviors, and morbidity.
In participants’ homes, trained dieticians administered a 24-h dietary recall interview 1 week after the health interviews and examinations. Participants reported on all foods and beverages consumed in the previous 24 h, including the description, quantity (in units of volume), and time and place of eating. To minimize respondent burden and to obtain an accurate food recall, the multiple-pass approach was used with the assistance of tools such as two-dimensional measuring guides and/or standard measuring tools. For the home-cooked dishes, the person in charge of cooking reported the details of his or her home recipes.
Energy and nutrient intake assessment
Daily total energy and nutrient intakes were estimated by multiplying the intakes of all foods and beverages reported by participants and publicly available in the KNAHNES dietary data. When calculating total energy and nutrient intake, the Korean Food Composition Table of the Rural Development Administration was used (5th edition for 1998; 7th edition for 2007–2009, 9th edition for 2016–2018) [25]. Our main interest was total energy intake (kcal).
Assessment of outcome
Eating occasion
Each eating occasion was defined based on eating events named by respondents during the 24-h dietary recall interview. There were four eating occasion response options for respondents to choose from: breakfast, lunch, dinner, and snacks. The percentage of energy (% kcal) at each eating occasion was used as a main outcome. If the given eating occasion was skipped, the % kcal at that eating occasion was set to 0. For example, the % kcal of breakfast would be 0 for an individual who skipped breakfast.
Food sources for each eating occasion using the Korean NOVA system
To identify the major food sources of each main meal and snack, we used a 5-digit food code to categorize all food items by NOVA. NOVA enabled us to assess diets according to the extent and purpose of food processing [26]. In NOVA, there are four food groups, including unprocessed or minimally processed foods (Group 1), processed culinary ingredients (Group 2), processed foods (Group 3), and ultra-processed foods (Group 4) [26].
In general, the principle suggested by Monteiro and colleagues [26] was applied to the overall food classification in this study. We first identified the main ingredients of the food, and if processed, we categorized them into either Group 3 or Group 4, with a focus on the maintenance of the food matrix. However, the NOVA system developed in Brazil may not be directly applicable to Korean cuisine, which is characterized by dishes that combine many ingredients. Thus, we employed a Korean NOVA system that adheres to the rationale of the NOVA classification but specifically considers the preservation of the natural food matrix and traditional eating experiences for application to Korean cuisine. If a dish was not clearly distinguishable between Group 3 and Group 4 based on its ingredients, another criterion was used whether the dish was part of the traditional Korean diet before the mass production of processed foods [27]. Additionally, we used food codes in the KNHANES dataset to identify traditional ingredients, such as Korean fermented sauces like soybean paste. According to the original NOVA system [26], soybean paste can be classified as an ultra-processed food. In contrast, in the Korean NOVA system, it could be classified as either processed or ultra-processed depending on whether it is produced traditionally or through mass production. This distinction was made by checking the food codes in the KNHANES dataset, which uses separate codes for homemade and factory-made products. Detailed examples are described elsewhere [27].
Two investigators (S.J. and JY. K.) classified all consumed food and beverage items at the food or ingredient level into one of the four NOVA food groups. In cases of disagreements in classification, we (1) identified the raw materials by checking product names and manufacturer information, and (2) reviewed the nutrient database to ascertain the sodium or sugar contents of such items [28]. During the 24-h dietary recall interview conducted in each survey cycle, participants reported consuming 1453 items in 1998, 2725 items in 2007–2009; and 3894 items in 2016–2018, respectively. Unprocessed or minimally processed foods included raw vegetables, fruits, meats, grains (557 items in 1998; 798 items in 2007–2009; 966 in 2016–2018). Processed culinary ingredients comprised condiments and oils (34 items in 1998; 147 items in 2007–2009; 247 items in 2016–2018). Processed foods include soups, stew, kimchi, and grilled or marinated vegetables and meat (275 items in 1998; 334 items in 2007–2009; 504 items in 2016–2018). Ultra-processed foods included bread, snacks, distilled alcoholic beverages, soft drinks, and noodles (587 items in 1998; 1446 items in 2007–2009; 2177 items in 2016–2018). Supplemental Table 1 shows the most frequently consumed food or ingredient items for each NOVA group across the survey cycle for each eating occasion. For each eating occasion, the percentage of energy (% kcal) from each NOVA group was calculated and used for further analysis.
Assessment of covariates
Covariates included age (years), sex (males, females), residency (urban, rural), education level (elementary school, middle school, high school, college and higher), household income (quartiles), and marital status (married or not) for presenting the general characteristics of study participants. For stratification analysis, trends in the percentage of energy and the food sources from all eating occasions were further assessed among population subgroups by age: 20–29, 30–39, 40–49, 50–59, and 60–69 years old.
Statistical analysis
Sociodemographic characteristics of Korean adults by survey cycle were presented as weighted proportions with standard errors (SEs). After adjusting for age and sex, linear regression models were used to estimate (1) the percentage of energy at each eating occasion and (2) the percentage of energy from each NOVA group for each eating occasion across the survey cycle. Statistical significances of linear trends over time were examined with the survey cycle modeled as an orthogonal polynomial [29]. A potential interaction effect by age was checked using an interaction term between survey cycle and age in linear regression models, and age-stratified estimates were obtained.
Survey weights accounting for the complex sampling design of the KNHANES were applied to all analyses. All statistical analyses were conducted with SAS software (version 9.4, SAS Institute, Cary, NC, USA) using PROC SURVEY procedures, with significance determined at an α level of 0.05.
Results
Population characteristics
From 1998 to 2018, the mean age of respondents increased from 39.4 to 43.4 years old. The proportion of adults aged 20–39 years decreased from 56.2 to 40.4%, whereas those aged 40–69 years increased from 43.9 to 59.6%. The proportion of urban residency increased from 77.9 to 87.9%. The proportion of respondents with an education level of college graduate or higher increased from 27.5 to 48.0%. The proportion of married respondents decreased from 73.9 to 66.8% (Table 1).
| Overall (= 29,389)n | 1st KNHANES, 1998 (= 5748)n | 4th KNHANES, 2007–2009 (= 11,900)n | 7th KNHANES, 2016–2018 (= 11,741)n | |
|---|---|---|---|---|
| Age, years | 42.2 (0.1) | 39.4 (0.3) | 41.6 (0.2) | 43.4 (0.2) |
| Age group, years | ||||
| 20–29 | 21.2 (0.4) | 27.2 (0.9) | 21.3 (0.7) | 19.6 (0.6) |
| 30–39 | 23.5 (0.4) | 29.0 (1.1) | 25.0 (0.7) | 20.8 (0.6) |
| 40–49 | 24.1 (0.4) | 20.8 (0.8) | 25.2 (0.6) | 24.1 (0.6) |
| 50–59 | 19.6 (0.3) | 13.4 (0.6) | 17.8 (0.5) | 22.6 (0.5) |
| 60–69 | 11.6 (0.2) | 9.7 (0.6) | 10.6 (0.3) | 12.9 (0.4) |
| Sex | ||||
| Males | 51.0 (0.3) | 50.4 (0.5) | 51.1 (0.5) | 51.1 (0.5) |
| Females | 49.0 (0.3) | 49.6 (0.5) | 48.9 (0.5) | 48.9 (0.5) |
| Residency | ||||
| Urban | 84.7 (0.9) | 77.9 (1.3) | 82.9 (1.6) | 87.9 (1.3) |
| Rural | 15.3 (0.9) | 22.1 (1.3) | 17.1 (1.6) | 12.1 (1.3) |
| Education level | ||||
| Elementary school | 11.6 (0.3) | 18.1 (0.9) | 14.2 (0.4) | 7.6 (0.3) |
| Middle school | 9.5 (0.2) | 13.0 (0.6) | 10.6 (0.4) | 7.6 (0.3) |
| High school | 39.9 (0.5) | 41.3 (1.0) | 43.2 (0.7) | 36.7 (0.7) |
| College and higher | 39.1 (0.6) | 27.5 (1.3) | 32.0 (0.8) | 48.0 (0.9) |
| Household income | ||||
| Q1 | 11.6 (0.3) | 16.8 (1.0) | 11.9 (0.5) | 10.0 (0.5) |
| Q2 | 23.9 (0.5) | 21.6 (1.1) | 25.3 (0.7) | 23.4 (0.7) |
| Q3 | 31.1 (0.5) | 32.1 (1.2) | 30.7 (0.7) | 31.1 (0.7) |
| Q4 | 33.5 (0.6) | 29.5 (1.5) | 32.1 (1.0) | 35.5 (0.9) |
| Marriedb | 69.5 (0.5) | 73.9 (1.0) | 71.6 (0.8) | 66.8 (0.8) |
Trends in energy intake contribution of main meals and snacks
From 1998 to 2018, there was a significant upward trend in total energy intake; however, the estimates were not substantially different (1986 to 2048 kcal; difference, + 62 kcal [SE, 21]; P = 0.004). During the same period, the percentage of energy from breakfast decreased from 25.0 to 16.7% (difference, − 8.2% [SE, 0.3]; P = 0.004); that from lunch did not substantially change (29.9–30.5%; difference, + 0.6% [SE, 0.4]; P = 0.12); that from dinner increased from 31.1 to 33.8% (difference, + 2.7% [SE, 0.4]; P < 0.0001); and that from snacks increased from 14.0 to 19.0% (difference, + 5.0% [SE, 0.5]; P < 0.0001) (Table 2).
| Weighted mean (SE)a | Mean difference, 7th vs. 1st | % difference, 7th vs. 1st | trendlinearPb | |||
|---|---|---|---|---|---|---|
| 1998 (1st) | 2007–2009 (4th) | 2016–2018 (7th) | ||||
| Total energy intake (kcal) | 1986 (19) | 1932 (10) | 2048 (10) | 62 (21) | 3.1 | 0.004 |
| % Energy intake from breakfast | 25.0 (0.3) | 20.8 (0.2) | 16.7 (0.2) | − 8.2 (0.3) | − 32.8 | < 0.0001 |
| % Energy intake from lunch | 29.9 (0.3) | 30.7 (0.2) | 30.5 (0.2) | 0.6 (0.4) | 2 | 0.12 |
| % Energy intake from dinner | 31.1 (0.3) | 31.4 (0.2) | 33.8 (0.2) | 2.7 (0.4) | 8.7 | < 0.0001 |
| % Energy intake from snack | 14.0 (0.4) | 17.2 (0.2) | 19.0 (0.2) | 5.0 (0.5) | 35.7 | < 0.0001 |
Trends in energy intake contribution of NOVA groups for main meals and snacks
From 1998 to 2018, at breakfast, the percentage of energy from minimally processed foods decreased from 86.6 to 68.1% (difference, − 18.6% [SE, 0.6]; P < 0.0001), whereas that from ultra-processed foods increased from 3.3 to 20.3% (difference, + 17.0% [SE, 0.6]; P < 0.0001). At lunch, the percentage of energy from minimally processed foods decreased from 74.7 to 61.6% (difference, − 13.1% [SE, 0.7]; P < 0.0001), whereas that from ultra-processed foods increased from 15.4 to 26.7% (difference, + 11.3% [SE, 0.7]; P < 0.0001). At dinner, the percentage of energy from minimally processed foods decreased from 81.4 to 60.3% (difference, − 21.1% [SE, 0.6]; P < 0.0001), whereas that from ultra-processed foods increased from 8.6 to 26.6% (difference, + 18.0% [SE, 0.6]; P < 0.0001). The percentage of energy from snacking on minimally processed foods decreased from 51.8 to 31.0% (difference, − 20.7% [SE, 1.0]; P < 0.0001), whereas that from ultra-processed foods increased from 31.8 to 62.6% (difference, + 37.0% [SE, 0.9]; P < 0.0001). At all eating occasions, processed culinary ingredients and processed foods contributed the least to total energy intake, and the percentage of energy from those foods did not significantly change (Fig. 1).
Trends in the percentage of energy intake from NOVA groups for main meals and snacks by KNHANES cycle, Korean adults aged 20–69 years.unprocessed or minimally processed foods,processed culinary ingredients,processed foods,ultra-processed foods,National Health and Nutrition Examination Survey,standard error. Note: The estimates were obtained after adjusting for age and sex using the linear regression model.values for trends were estimated with the survey cycles modeled as an orthogonal polynomial MPF PCI PF UPF KNHANES SE P
Trends in different age groups
When compared to their counterparts, adults aged 20–49 years showed greater changes in the percentage of energy from breakfast (difference, − 7.3% for those aged 20–29 years; − 9.5% for those aged 30–39 years; − 9.1% for those aged 40–49 years; − 7.5% for those aged 50–59 years; − 6.0% for those aged 60–69 years; all P < 0.0001) and that from dinner (difference, + 2.8% for those aged 20–29 years; + 4.5% for those aged 30–39 years; + 3.5% for those aged 40–49 years; + 1.3% for those aged 50–59 years; − 0.9% for those aged 60–69 years; all P < 0.05 except for those aged 50–69 years), while adults aged 50–69 years showed a greater upward trend in the percentage of energy intake from snacks (difference, + 3.7% for those aged 20–29 years; + 3.2% for those aged 30–39 years; + 4.8% for those aged 40–49 years; + 7.4% for those aged 50–59 years; + 7.4% for those aged 60–69 years; all P < 0.0001 (Fig. 2).
At all eating occasions except for lunch, adults aged 20–49 years had a greater decline in the percentage of energy from minimally processed foods (breakfast: difference, − 24.2% for those aged 20–29 years; − 22.1% for those aged 30–39 years; − 18.9% for those aged 40–49 years; − 14.7% for those aged 50–59 years; − 11.3% for those aged 60–69 years; all P < 0.0001; dinner: difference, − 22.4% for those aged 20–29 years; − 22.8% for those aged 30–39 years; − 23.0% for those aged 40–49 years; − 17.6% for those aged 50–59 years; − 14.7% for those aged 60–69 years; all P < 0.0001; snacks: difference, − 20.4% for those aged 20–29 years; − 22.4% for those aged 30–39 years; − 23.8% for those aged 40–49 years; − 18.9% for those aged 50–59 years; − 14.7% for those aged 60–69 years; all P < 0.0001) and a greater increase in the percentage of energy from ultra-processed foods than those aged 50–69 years (breakfast: difference, + 23.2% for those aged 20–29 years; + 21.3% for those aged 30–39 years; + 17.1% for those aged 40–49 years; + 12.9% for those aged 50–59 years; + 8.5% for those aged 60–69 years; all P < 0.0001; dinner: difference, + 19.2% for those aged 20–29 years; + 19.7% for those aged 30–39 years; + 19.9% for those aged 40–49 years; + 14.8% for those aged 50–59 years; + 11.8% for those aged 60–69 years; all P < 0.0001; snacks: difference, + 30.1% for those aged 20–29 years; + 34.5% for those aged 30–39 years; + 34.2% for those aged 40–49 years; + 27.9% for those aged 50–59 years; + 18.0% for those aged 60–69 years; all P < 0.0001) (Fig. 3).
Trends in the percentage of energy intake from main meals and snacks by KNHANES cycle, Korean adults aged 20–69 years, stratified by age.National Health and Nutrition Examination Survey,standard error. Note: The estimated mean and SE were obtained after adjusting for sex using the linear regression model.values for trends were estimated with the survey cycles modeled as an orthogonal polynomial.values for interaction were determined by including the interaction term of survey cycle and age in the linear regression model KNHANES SE P P
Trends in the percentage of energy intake from minimally processed and ultra-processed foods for main meals and snacks by KNHANES cycle, Korean adults aged 20–69 years, stratified by age.unprocessed or minimally processed foods,ultra-processed foods,National Health and Nutrition Examination Survey,standard error. Note: The estimates were obtained after adjusting for sex using the linear regression model.values for trends were estimated with the survey cycles modeled as an orthogonal polynomial.values for interaction were determined by including the interaction term of survey cycle and age in the linear regression model MPF UPF KNHANES SE P P
Discussion
From 1998 to 2018, the percentage of energy from breakfast decreased, but that from dinner and snacks increased. This may indicate that a decreased percentage of energy from breakfast was compensated for by an increased percentage of energy from snacks. At all eating occasions, we observed a decline in the percentage of energy from minimally processed foods and an increase in that from ultra-processed foods, indicating that ultra-processed foods have become major sources of total energy intake. When stratified by age, younger adults (< 50 years old) had a greater decline in the percentage of energy from breakfast and an increase in that from dinner, whereas older adults (≥ 50 years old) had a greater increase in the percentage of energy from snacks than their counterparts; younger adults (< 50 years old) had a steeper decrease in the percentage of energy from minimally processed foods and an increase in that from ultra-processed foods for breakfast, dinner, and snacks compared to older adults (≥ 50 years old).
This study using KNHANES data is the first attempt to examine temporal trends in the energy intake contribution of all eating occasions among Korean adults. Globally, eating patterns have shifted toward a more westernized style, which could be harmful to human health [30]. In our study, the percentage of energy from breakfast significantly decreased, while that from lunch almost remained unchanged and that from dinner slightly increased from 1998 to 2018. This finding is in accordance with a previous study that reported increases in the percentage of skipping breakfast but not noticeable changes in the percentages of skipping lunch and dinner between 2011 and 2019 (skipping breakfast: 21.4% in 2011, 31.3% in 2019; skipping lunch: 7.6% in 2011, 8.0% in 2019; and skipping dinner: 4.8% in 2011, 5.5% in 2019) [22].
Our results showed that a reduced proportion of breakfast appeared to be compensated for by increasing snack consumption. From 1998 to 2018, the percentage of energy from breakfast decreased by 32.8% and that from snacks increased by 35.8%, while total energy intake did not change substantially. Furthermore, breakfast contributed more energy than snacks did in 1998 (25.0% from breakfast vs. 14.0% from snacks); however, it reversed in 2016–2018 (16.7% from breakfast vs. 19.0% from snacks). Despite the varied definition of snacking by study (e.g., snacking frequency or energy contribution from snacking), there is strong evidence that snacking contributes to positive energy balance [31, 32]. Yet, it is not confirmed if snacking per se adversely contributes to health outcomes [31, 32] or nutritional quality [33], but even snacking can be helpful in the older population [33]. More importantly, healthy snacks like whole grains, fruits, and vegetables are beneficial to health; however, popular snacks are usually energy dense and nutrient poor (e.g., salty snacks, sweetened beverages) [31]. Thus, the types of foods for snacks need to be given more attention.
Globally and in Korea, more highly processed foods have become popular [23, 34]. Although food processing per se is not necessarily harmful, ultra-processed foods are generally characterized by poor nutritional profiles and corresponding detrimental health outcomes [34]. From 2011 to 2018, the consumption of fruits and vegetables (i.e., minimally processed) decreased, while that of beverages (i.e., ultra-processed) increased in Korea [22]. Furthermore, coke was newly ranked in the ten most frequently consumed foods in Korea in 2019 [35]. Our study showed that the percentage of energy from ultra-processed foods substantially increased, while that from minimally processed foods decreased from 1998 to 2018 at all eating occasions. Such trends imply that eating patterns regarding food sources among Koreans have worsened during the past 20 years. Thus, applicable recommendations are necessary to improve their eating patterns by highlighting the types of foods that should be consumed (e.g., limiting the consumption of ultra-processed foods).
Age differences were evident in our study, indicating that relatively younger adults had greater changes than older adults. They tended to reduce their energy consumption from minimally processed foods but increase it from ultra-processed foods. These trends may account in part for the greater increase in obesity prevalence among younger adults than older adults between 2009 and 2019 in Korea [36]. Between 2010–2012 and 2016–2018, upward trends in the energy consumption of ultra-processed foods were stronger in younger adults than in older adults (+ 2.5% for those aged 20–49 years; + 1.6% for those aged 50–64 years; + 1.0% for those aged 65 years) [23]. Generally, the majority of people have experienced several life events in their adulthood, including undertaking higher education, getting a job, or becoming a parent. The given life events may require independent living, and obtaining food and preparing meals are their own responsibility in this case [37]. According to a previous study, younger adults are more likely to purchase less expensive, energy-dense, and ready-to-eat foods and less likely to plan their meals, especially under stressful situations (e.g., after a long day at work, examination) [37]. Additionally, they might think that healthy foods are expensive, and they tend to adopt energy-dense diets faster than older adults [38].
Currently, the prevailing recommendations in Korea do not include specific guidelines regarding the consumption of processed foods [39]. Given the unfavorable effect of high consumption of ultra-processed foods on health, it is crucial to continuously monitor consumption patterns and conduct further investigations to establish a causal relationship between the intake of ultra-processed foods and health outcomes. These efforts are necessary to generate compelling evidence that can inform more effective dietary recommendations. Also, our study showed that the increase in ultra-processed food is more evident among younger adults than older adults. To address this issue, several public health strategies could be pursued, particularly targeting the younger adult population. These strategies may include the development of education programs, the implementation of informative package labelling, the modification of the food environment, and the consideration of taxation measures [40].
This study has strengths. To our knowledge, this is the first study to characterize temporal trends in the energy intake contribution and food sources of meals and snacks in Korean adults. Another strength of this study is the use of nationally representative dietary data collected using standardized protocols and providing generalizability to Korean adults.
Several limitations should be noted. First, there are inevitable measurement errors (e.g., underreporting of energy intake) in self-reported dietary assessments, including the 24-h dietary recall interview. Especially in Korean adults, there is a 12% underreporting of total energy intake estimated by 24-h dietary recall when compared with the double-labeled water method [41], and underreporting is likely more prevalent among women living alone and with a lower education level than their counterparts [42]. Furthermore, snacks (especially afternoon snacks) are most likely to be underreported [43]; thus, cautious interpretations are needed. Further studies are required to examine whether underreporting has changed over time and to adjust errors due to underreporting in Korea. Second, the use of only a single 24-h dietary recall may not be sufficient to measure individual usual intakes because of day-to-day variations (a source of within-person random error) [46]. Usual dietary intake is often recommended in some cases (e.g., when examining the proportion of the population at or below a certain level of intake). However, a single 24-h dietary recall is sufficient to estimate the mean intake of a population [44]. Third, the use of respondent descriptions for eating occasions may not be interchangeable with those from other studies since there is no consensus definition of eating occasions [45]. Furthermore, there is a lack of objective methods, such as the examination of biomarkers of eating patterns, to determine whether eating patterns in our study were valid. Fourth, we focused on Korean adults aged 20–69 years, and thus, our findings may not be generalizable to other populations with different races/ethnicities, ages (e.g., adolescents), or locations. Fifth, the NOVA classification system has a few unresolved issues, including that its simplistic definition may ignore the features of food itself [46]. Several types of ultra-processed foods, also known as premium ultra-processed foods (e.g., granola with probiotics, brown bread, or plain yogurt), are not necessarily harmful to health. However, it has been considered the most prominent method of food classification considering the extent and purpose of industrial food processing, which may have been overlooked in previous literature focusing only on nutrient profiles [47]. An updated NOVA classification system may be needed to address this issue. Lastly, to discern the consumption of UPFs, the 5-digit food codes were used from the KNHANES data. It is important to note that using this method may not ensure a precise distinction between Group 3 and Group 4 in the NOVA classification due to the reliance on general food codes. In the KNHANES dataset, processed foods lacking specific brand and product names, as well as food-away-from-home items without listed ingredients, are coded using these general food codes. This approach might lead to an underestimation of individuals’ UPF consumption.
In conclusion, the eating patterns of Korean adults changed from 1998 to 2018, with the greatest decrease in the percentage of energy from breakfast and the greatest increase in the percentage of energy from snacks. On all eating occasions, the contribution of minimally processed foods decreased, while that of ultra-processed foods increased, especially among younger adults. These findings highlight the importance of continuous monitoring of both eating patterns and the types of foods selected for all eating occasions to provide more realistic dietary recommendations that reflect the current state of diets and to offer an opportunity for health improvement. Furthermore, additional studies into the factors contributing to such trends are necessary to better understand the context in which negative changes occurred.
Supplementary Information
Below is the link to the electronic supplementary material.
Author contributions
SJ: methodology, data curation, formal analysis, writing—original draft preparation, writing—reviewing and editing; JYK: conceptualization, methodology; SP: conceptualization, methodology, writing—reviewing and editing. All authors gave final approval and agreed to be accountable for all aspects of ensuring integrity and accuracy.
Funding
This work was supported by grants from the National Food Safety Information Service and the National Research Foundation of Korea (NRF) grant funded by the Korea government (Ministry of Science and ICT) (RS-2023-00274240). Sohyun Park’s effort was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT & Future Planning (2021R1A6A1A03044501, 2021R1I1A3060248). The funders had no role in the design, analysis or writing of this article.
Data availability
The KNHANES data are publically available for download.
Declarations
Conflict of interest
The authors declare that they have no conflicts of interest.
Ethical standards
KNHANES was approved by the Korea Disease Control and Prevention Agency ethics review board. All participants gave written informed consent before participation in KNHANES, and all survey data were anonymized before analysis.
Footnotes
Contributor Information
Jee Young Kim, Email: kimjee@foodinfo.or.kr.
Sohyun Park, Email: sopark@hallym.ac.kr.
References
Associated Data
Supplementary Materials
Data Availability Statement
The KNHANES data are publically available for download.