Association of Dietary Patterns with Weight Status and Metabolic Risk Factors among Children and Adolescents

Unhealthy dietary patterns are associated with obesity in children and adolescents. However, few studies have investigated the relationships between dietary patterns and obesity-related metabolic disorders among Asians. We identified dietary patterns in children and adolescents and examined the associations between these patterns and obesity, insulin resistance, and metabolic syndrome in South Korea. This study is a cross-sectional design. We used baseline data from an intervention study of 435 Korean children and adolescents aged 6–17 years. Insulin resistance was assessed as HOMA-IR ≥ 2.6. Metabolic syndrome was diagnosed by cardiovascular disease risk factor clustering. Dietary intakes were estimated using 3-day food records. Factor analysis was used to obtain dietary patterns, and we examined the associations between dietary patterns and obesity-related markers adjusted for potential covariates. Three dietary patterns were identified as fast food and soda (FFS), white rice and kimchi (WRK), and oil and seasoned vegetable (OSV) patterns. Compared with participants in the lower intake of FFS pattern, those in the top intake were associated with a higher waist circumference (WC) (β = 1.55), insulin level (β = 1.25), and body mass index (BMI) (β = 0.53) and it was positively associated with HOMA-IR ≥ 2.6 (OR = 2.11; 95% CI: 1.227–3.638) (p < 0.05). WRK pattern was associated with lower weight and higher HDL cholesterol, and the OSV pattern was associated with a lower weight, WC, and insulin level (p < 0.05). The FFS pattern showed a positive relation with WC, serum insulin, and BMI, and the other two dietary patterns indicated a preventive effect of those parameters. The FFS pattern was associated with significantly elevated insulin resistance among children and adolescents.


Introduction
Childhood obesity is one of the serious public health challenges worldwide [1]. In Korea, the prevalence of obesity in children and adolescents increased 1.5 times from 11.2% in 2007 to 17.3% in 2017, and severe obesity increased 1.6 times during the same period [2]. Overweight and obese children and adolescents are likely to maintain their status into adulthood and are at higher risks for developing chronic diseases such as type 2 diabetes and cardiovascular disease [3].
The causes of childhood obesity are multifactorial, including genetic, behavioral, and environmental factors [4]. Among these, dietary factors are the key cause of overweight and obesity [5]. Previous studies indicated that intake of fast food is associated with obesity [6],

Study Design and Subjects
This cross-sectional study used the baseline data from a prior study that is the Intervention for Children and Adolescent obesity via Activity and Nutrition (ICAAN) for our analysis. The ICAAN study was designed to examine the effects of multidimensional intervention including circuit training and balanced diet in children and adolescents in Korea [15]. A total of 435 Korean children and adolescents aged 6-12 years (n = 336) and aged 13-17 years (n = 99) were recruited for this study, respectively.
We included those who participated in the ICAAN study who met the criteria for normal weight children and adolescents (body mass index (BMI) 5-85th percentile of age and sex-specific BMI), overweight and obese children and adolescents (BMI 85th-120% of the 95th percentile), and severely obese participants (BMI ≥ 120% of the sex-and agespecific 95th percentile BMI) according to the 2017 Korean National Growth Chart [16,17], and excluded 53 subjects, whose anthropometric, biochemical, and dietary intake data were missing. A total of 435 Korean boys (n = 261) and girls (n = 174) were included in the dietary pattern study. This study was approved by the Institutional Review Board of Hallym University Sacred Heart Hospital (approval No. 2016-I135).

Anthropometric Measurements and Blood Pressure
Body weight (Wt), height (Ht), waist circumference (WC), hip circumference (HC), and systolic and diastolic blood pressure (SBP and DBP) were measured by a trained staff. Ht and Wt were measured to the nearest 0.1 cm and 0.1 kg, respectively (DS-103, Dongsan Jenix Co., Seoul, South Korea). Body mass index (BMI) was calculated based on Wt and Ht (kg/m 2 ). In addition, BMI z-score was calculated by BMI-for-age percentiles together with the Lambda-Mu-Sigma (LMS) method of Cole and Green [18], which provides a way of obtaining normalized growth percentile standards, based on 2017 Korean National Growth Charts for Children and Adolescents [16]. WC was measured with participants standing straight using a flexible tapeline at the midpoint of the lower rib and the iliac crest to the nearest 0.1 cm. HC was measured using a flexible tapeline at the horizontal circumference of the highest point of the buttocks to the nearest 0.1 cm. The waist-hip ratio (WHR) was calculated based on recorded WC and HC measurements. Both SBP and DBP were measured two times in the right arms of participants in a seated position following a rest of at least 5 min by an automatic sphygmomanometer (HEM-907, OMRON Healthcare Co., Kyoto, Japan) and the average value was used for analysis.

Criteria of Metabolic Risk Factors
The insulin resistance is determined using the homeostatic model assessment insulin resistance (HOMA-IR). The HOMA-IR criteria for children and adolescents determined the insulin resistance criterion (HOMA-IR ≥ 2.6) using a range that increases the risk of cardiovascular risk [19] and it was calculated by insulin and fasting blood glucose. The metabolic syndrome is defined by criteria from cardiovascular disease risk factor clustering (CVD-RFC) in children and adolescents [20][21][22]. According to the CVD-RFC, metabolic syndrome in children and adolescents is present if two or more of the following factors must be met. Abdominal obesity was defined as WC (cm) ≥ 90th percentile value of age and sex according to the 2007 Korean National Growth Charts, TG ≥ 110 mg/dL, HDL-C < 40 mg/dL, SBP or DBP (mmHg) ≥ 90th percentile value of age and sex according to the 2007 Korean National Growth Charts, FBG ≥ 100 mg/dL.

Dietary Assessments
Dietary intake was estimated using a 3-day food record [23] which is written by each participant and their caregiver. Then, we confirmed the food record through their parents or caregiver. Participants were asked to record their intake of meals and snacks, including beverages during a nonconsecutive period of three days (two weekdays and one weekend day). They also were asked to record ingredients of meals as well as portion sizes. A well-trained dietitian reviewed and confirmed the written 3-day food record using food models during face-to-face interviews to increase precision in reporting. Researchbased typical food intake data were used to calculate averages and energy, macronutrients, and micronutrients were analyzed using a computer-aided nutritional analysis program (CAN-Pro) web version 5.0 (Korean Nutrition Society, 2016).

Dietary Patterns Analysis
All food items were classified into 25 food groups, based on the generalized 17 food groups in the Korean Nutrient Database of the Korean Nutrition Society. This study separated the original grains group into 6 subgroups to indicate representative food items that comprise typical consumption in Korean children and adolescents. Among the food groups, the meat and fish groups were combined into a single food group because they are the most consumed animal protein sources in South Korea [24]. The list of food groups is presented in Supplemental Table S1. We conducted factor analysis (based on principal component analysis (PCA)) to investigate dietary patterns based on the percentages of total energy intake in the 25 food groups. Factor analysis using SAS proc factor (SAS Institute Inc.) was rotated via an orthogonal transformation (varimax rotation) to obtain a simplistic structure and distinct interpretation. To determine the number of dietary patterns, we considered eigenvalues greater than 1.4 and scree plots. Also, to describe the dietary patterns, food groups were retained in a factor if they had a correlation of 0.25 with that factor. The dietary pattern was named based on prominent food items. We divided Factor scores into three tertiles on the basis of their contribution to each dietary pattern, and an increase from T1 to T3 was assumed.

Statistical Analysis
All analyses were performed using SAS software (SAS 9.4; SAS Institute Inc., Cary, NC, USA). The characteristics of participants were compared by Student t-test and Chi-square test. Each dietary pattern was subdivided into tertiles. The tertiles were compared using a generalized linear model adjusted for age, sex, total energy intake, house income and parental education level, followed by Tukey's post hoc tests. We assessed the associations between dietary patterns and obesity-related markers (Wt, WC, BMI z-scores, insulin, TG, LDL-C, and HDL-C) using linear regression. Subsequently, a multivariable logistic regression model was used to estimate the odds of obesity, insulin resistance, and metabolic syndrome for each dietary pattern. This study also adjusted for potential covariates including age, sex, and energy intakes. The variables were described as odds ratio (OR) and 95% confidence intervals (95% CI). Following regression analysis, variable median scores by tertiles of dietary patterns were used to examine trends. All statistical significance was defined at p < 0.05.
Factor analysis revealed three dietary patterns, and they are presented in Table 2. The three patterns were named based on food items. The fast food and soda (FFS) pattern was characterized by high intakes of fast food, carbonated beverages, and instant ramen, and low consumption of soups, vegetables and mushrooms, fruit, and milk and dairy products. The white rice and kimchi (WRK) pattern was characterized by high intakes of white rice, Kimchi, processed food, and eggs, and low intakes of snack and cereals, flour and rice cakes, noodles, and sugars. This represents simplified Korean food. In addition, the oil and seasoned vegetable (OSV) pattern represents a high consumption of seasoned vegetable or vegetable stir-fry, characterized by high intakes of vegetable oil, sugars, seasonings, potatoes, nuts, and vegetable and mushrooms. These patterns accounted for 1.99% (FFS), 1.99% (WRK) and, 1.69% (OSV) of total variance explained in food intakes, respectively (data not shown).  The difference of intakes, anthropometric and biochemical markers according to tertile of dietary patterns is summarized in Table 3. After adjusting for age, sex and energy intake, participants with higher FFS pattern had greater energy intake. Energy and fat intakes gradually increased, whereas carbohydrate intake decreased with higher tertile of FFS pattern (p < 0.000). In contrast, higher tertile of the WRK pattern showed less energy intake and lower fat intake, whereas carbohydrate intake increased with higher tertile of the WRK pattern (p < 0.000). Similarly, the intake of carbohydrates was increased, and fat was decreased in the top tertile of the OSV pattern (p < 0.000). For anthropometric characteristics, Wt, WHR, and, BMI z-score gradually increased with higher tertile of FFS pattern (p < 0.000). On the other hand, Wt gradually decreased, but DBP increased in higher tertiles of WRK pattern (p < 0.000). Additionally, Wt, DBP, WHR, and BMI z-score gradually decreased with higher tertile of the OSV pattern (p < 0.000). In terms of comparing biochemical characteristics across tertiles of the dietary patterns, the higher tertiles of the FFS pattern showed increased levels of insulin, HOMA-IR, and triglyceride, whereas HDL-C gradually declined (p < 0.000). Conversely, insulin level decreased in higher tertiles of the WRK pattern (p < 0.000). Subjects following the OSV pattern showed decreased levels of insulin, HOMA-IR and triglyceride in higher tertiles (p < 0.000). Association between tertiles of discrete dietary pattern and obesity and metabolic risk markers are shown in Table 4. After adjusting for age, sex, and energy intake, the FFS pattern had positive association with WC, SBP, insulin, weight, BMI z-score and HOMA-IR (β = 1.55, 1.31, 1.25, 1.00, 0.16, and 0.34, respectively; p < 0.05), but the pattern was negatively associated with HDL-C (β = −0.03, p < 0.05). The WRK pattern had association with WC, insulin, and HDL-C (β = 0.44, −0.16 and 0.58, respectively; p < 0.05). Similarly, the OSV pattern had reverse association with Wt, SBP, WC, BMI z score, insulin and HOMA-IR (β = −1.57, −0.76, −0.86, −0.10, −1.27 and −0.30, respectively; p < 0.001). However, the OSV pattern was positively associated with HDL-C (β = 0.17, respectively; p < 0.05). The association between obesity, insulin resistance, and metabolic syndrome across tertile of dietary patterns is shown in Table 5. After adjustment, the FFS pattern had twofold higher odds of insulin resistance (OR = 2.11; 95% CI: 1.23-3.64) compared with those in the T1. Participants with the highest intake of FFS pattern had a 1.7 times higher odds of BMI z-score ≥ 2 (OR = 1.67; 95% CI: 1.01-2.76), and nearly 1.8 times more likely to have prevalent metabolic syndrome (OR = 1.79; 95% CI: 1.11-2.87) compared with those in the T1. In the model I (crude), however, the association was not significant after adjustment of age, sex, energy intake, household income, and paternal education level (model II).  )   T1  T2  T3  T1  T2  T3  T1  T2  T3 BMI z score ≥ 2 Abbreviations: OR, odds ratio; CI, confidence interval; T, tertile; BMI, body mass index; WC, waist circumference; SBP, systolic blood pressure; DBP, diastolic blood pressure. (1) Model I was not adjusted results of logistic regression analysis between obesity, insulin resistance, and metabolic syndrome and tertiles of dietary pattern. (2) Model II was the results of logistic regression analysis adjusted for age, sex, and energy intake, household income, and parental education level between obesity, insulin resistance, and metabolic syndrome and tertiles of dietary pattern. (3) Metabolic syndrome is defined according to cardiovascular disease risk factor clustering (CVD-RFC) for children and adolescents. Defined as the presence of over 2 of following risk factors: Obesity (waist circumference) ≥ 90th percentile of adult cut-off, Triglycerides (mg/dL) ≥ 110, HDL cholesterol (mg/dL) < 40, BP (mmHg) SBP or DBP ≥ 90th percentile of adult cut-off, fasting plasma glucose (mg/dL) ≥ 100.

Discussion
This study identified three major dietary patterns among Korean children and adolescents; FFS, WRK, and OSV patterns. After adjusting for covariates, the FFS pattern was positively associated with Wt, SBP, WC, insulin, and HOMA-IR, and caused a significantly higher prevalence of insulin resistance. The WRK pattern was associated with lower Wt and insulin and associated with higher WC and HDL-C. The OSV pattern was associated with lower Wt, WC, insulin, and HOMA-IR, and associated with higher HDL-C.
To date, the majority of published studies on dietary patterns have reported that dietary patterns of children and adolescents in Asia had been increasingly Westernized [25,26]. Korea is not an exception. The secular trends of dietary pattern revealed that the traditional Korean patterns (higher consumptions of white rice, kimchi, vegetables, and fish) have declined, while a Western pattern has increased in children and adolescents aged 10-19 years [27]. According to related previous studies in the United States and European countries, dietary patterns described as high fat intake and high calories increase the risk of obesity compared to dietary patterns of whole grains, low-fat dairy products, vegetables, and fruits [10,28]. However, in a Chinese study, the modern dietary pattern (characterized by high intake of milk, fast foods, eggs, meat, poultry, and cake) has a higher relationship with obesity than traditional Chinese dietary patterns (represented by high consumption of vegetables, rice, pork, and legumes) [25]. This is similar to the results in Asian countries such as Japan [26]. According to the results using Korean national data, Korean traditional meal patterns, Western patterns, and healthy patterns are derived [27]. However, in our study, the proportion of obese children is high, which stem from different patterns from that of the previous Korean dietary patterns study [27].
Consistent with previous research [13,29], our study found a similar dietary pattern designated as FFS pattern which is composed of fast food, carbonated beverages, and instant ramen. According to a previous study about dietary pattern for adolescents, it was found that high consumption of fast food and red meat has association with increasing overweight in Lebanese adolescents aged between 13 and 19 years [13]. Additionally, previous studies have confirmed that high consumption of pizza, burgers, sugar drinks is associated with increasing high BP [14] and increasing the risk of insulin resistance [30]. It was reported to be associated with not only a high intake of energy, sodium, SFA, and sugar, but also a low intake of fiber and PUFA [14,30]. In the present study, high consumption of the FFS pattern did not increase the intake of sodium and sugar (data are not shown), but the results concerning energy, fiber, and PUFA intake were similar to the previous [27,31]. In addition, participants with higher fast food and soda consumption had a higher prevalence of insulin resistance and higher prevalence of the metabolic syndrome, but it was not significant after adjustment that is due to a lack of statistical power. The earlier study confirmed that the incidence of metabolic syndrome is rising commensurately with the rising intake of fast food and soda in certain populations [6]. Because the high-carbohydrate foods contribute to insulin resistance, it prevents glucose from being used as an energy source, increases the synthesis of TG and affects the concentration of lipid and lipoprotein, which causes metabolic disorders to occur simultaneously [32]. Although intake of the FFS pattern in this study does not increase carbohydrate intake, the risk of metabolic syndrome could be increased as it is related to increased fat intake and associated with increased risk of insulin resistance. Also, one other study said that even though fast food is known to be unhealthy, unhealthy dietary patterns including fast food have a stronger correlation with metabolic disorders than fast food consumption on its own [33]. Therefore, we should carefully consider whether eating the FFS pattern might contribute to metabolic syndrome.
A WRK dietary pattern composed of white rice, kimchi, egg, and processed food in Korean children and adolescents might be influenced by modernization [26]. However, a previous study reported that subjects whose weight is close to normal weight are likely to have a Korean style dietary pattern [34], and another study in Korean adults identified that the traditional dietary pattern has a negative association with central obesity and association with decreased risk of metabolic syndrome [35]. Consistent with those results, this study found that the WRK pattern has a negative association with Wt, BMI z-score and insulin level in our subjects. However, it was not shown to be associated with a lower prevalence of metabolic syndrome or obesity. This may be because, while traditional dietary patterns had low-fat content and a high portion of vegetables (established contributors to good health) [35], the WRK pattern in this study did not include the vegetable and instead contained processed food. Especially, adolescents are experiencing a dramatic increase in intake of Westernized food including processed food among adolescents aged 12-14 years in Seoul, Korea [36]. Therefore, further study is necessary to monitor children and adolescents' dietary patterns in the long term to examine periodically their dietary patterns in Korea.
The OSV pattern which is characterized by the vegetable oil, sugars, seasonings, potatoes, nuts and vegetable and mushrooms pattern had a negative association with Wt, WC, BMI z-score and insulin level. Vegetable intake was inversely associated with obesity in children and adolescents in China [31], vegetable and vegetable oils were associated with lower insulin resistance in children and adults [37,38]. Also, the nut consumption had benefits for adolescents that were inversely associated with BMI z-score and linked to reduced glucose and lipid risk for cardiovascular disorders [39]. The benefits from the OSV pattern originate from eating a variety of vegetables and nuts [36,39]. It is because the unsaturated fatty acids which are rich in vegetable oil are known to be inversely associated with WC, BP, and insulin resistance. In addition, dietary fiber which is rich in nuts and vegetables is well known to be related to help the improvement of BP, insulin resistance, and obesity in adult women and obese children and adolescents [37,38,40]. Consistent with other studies, the oil and seasoned vegetable pattern had a high intake of fiber and unsaturated fatty acids [41] (data not shown). The OSV pattern, which includes such nutritional factors, is believed to reduce the risk of cardiometabolic disease factors such as lower Wt, WC, BMI z-score and insulin level and increasing HDL-C. However, according to a previous study, high consumption of vegetable oil may increase energy density [42]. Therefore, it is necessary to consider excessive consumption of vegetable oil.
Additionally, the OSV pattern in this study also included sugars and seasonings such as salt and soy sauce. Accordingly, it is difficult to conclude definitively that this particular OSV pattern is beneficial for childhood because consumption of a salt-rich dietary pattern should increase the risk of high blood pressure [14] and higher sugar intake increases fasting glucose and insulin resistance [43]. Therefore, eating with high consumption of vegetable oil (such as nuts, sesame oil, and perilla oil, etc.) and vegetables are believed to be beneficial to the cardiometabolic disease factors.
The present study has several strengths. This is the first study to examine the association of dietary patterns and cardiometabolic disease factors in children and adolescents, including children and adolescents with severe obesity in Korea. Then, we used a 3-day food record and confirmed it by the participants' caregivers and a well-trained clinical dietitian. Moreover, we used the SES variables for analysis (e.g., Monthly family income, Parental education).
Despite these strengths, several limitations may have affected this study. Since this study was a cross-sectional design, it is not possible to retrospectively evaluate whether dietary patterns are causal with insulin resistance, metabolic syndrome, and other indicators. Second, there was a possible under-reporting of dietary intake that may occur in obese and severely obese participants. Third, even though physical activity is one of the major factors in obesity, we were unable to use physical activity as a variable due to lots of missing data of physical activity on the participants. At least, this study should be interpreted assuming that there are no significant differences in the level of physical activity of the subjects.

Conclusions
We found three dietary patterns of the FFS, WRK, and OSV patterns in Korean children and adolescents. Among these dietary patterns, the high intake of the FFS pattern was associated with an increased risk of insulin resistance. These findings will be of some help to counseling and education in preventing obesity in childhood and adolescence. However, further studies are needed to investigate more objectively the causal relationship between dietary patterns and cardiometabolic risk using prospective data.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/nu13041153/s1, Table S1: Food grouping for Korean children and adolescents in subject.  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.