Hostname: page-component-586b7cd67f-2plfb Total loading time: 0 Render date: 2024-11-30T14:58:04.391Z Has data issue: false hasContentIssue false

Eating frequency is inversely associated with BMI, waist circumference and the proportion of body fat in Korean adults when diet quality is high, but not when it is low: analysis of the Fourth Korea National Health and Nutrition Examination Survey (KNHANES IV)

Published online by Cambridge University Press:  12 April 2018

Sunmi Kim
Affiliation:
Department of Family Medicine, Kangwon National University Hospital, Kangwon National University School of Medicine, 156, Baengnyeong-ro, Chuncheon-si, Gangwon-do 24289, Republic of Korea
Jeong Hee Yang
Affiliation:
Department of Family Medicine, Kangwon National University Hospital, Kangwon National University School of Medicine, 156, Baengnyeong-ro, Chuncheon-si, Gangwon-do 24289, Republic of Korea
Gyeong-Hun Park*
Affiliation:
Department of Dermatology, Dongtan Sacred Heart Hospital, Hallym University College of Medicine, 7, Keunjaebong-gil, Hwaseong-si, Gyeonggi-do 18450, Republic of Korea
*
*Corresponding author: G.-H. Park, fax +82 31 8086 2638, email [email protected]
Rights & Permissions [Opens in a new window]

Abstract

The role of eating frequency (EF) in obesity development has been debated, and few studies have investigated Asian populations. Diet quality might affect the association between EF and obesity. Therefore, we investigated the association between EF and obesity indicators in a representative sample of Korean adults with consideration to diet quality. This cross-sectional study used data of 6951 participants aged 19–93 years (male 49·8 %, female 50·2 %) from the Fourth Korean National Health and Nutrition Examination Survey. EF was assessed using a questionnaire, and diet quality was defined as mean adequacy ratio (MAR). To explore the association between EF and obesity indicators, we used multiple linear regression analyses with and without interaction terms between diet quality and EF. EF was inversely associated with each obesity indicator, including body fat percentage (BF%), BMI and waist circumference (WC), showing a significant linear trend (P<0·001 for BF%, WC and BMI). In addition, the association between EF and each obesity indicator was significantly altered according to diet quality (P value of the interaction term EF×diet quality=0·008 in the regression model for BF%, <0·001 for BMI and 0·043 for WC). In the stratified analyses according to diet quality, EF had a significant inverse association with BF%, WC and BMI in the high diet quality groups, but not in the low diet quality groups. This study suggests that EF is inversely associated with the obesity indicators when diet quality is high, but not when it is low in Korean adults.

Type
Full Papers
Copyright
Copyright © The Authors 2018 

Obesity has become a global epidemic health problem because it is a highly prevalent and major risk factor for CVD and chronic diseases including cancer( Reference Guh, Zhang and Bansback 1 ). The recent global estimates of the WHO revealed that about 13 % (11 % of men and 15 % of women) of the world’s adult population were obese in 2016, and the worldwide prevalence of obesity increased nearly by 30% between 1980 and 2013( 2 , Reference Ng, Fleming and Robinson 3 ). The prevalence of obesity has also been increasing in Asian countries, including Korea( Reference Yoon, Lee and Kim 4 ). The obesity prevalence in Korean adults increased from 26·0 to 36·3 % in men and from 26·5 to 27·6 % in women between 1998 and 2009( Reference Kang, Shim and Lee 5 ).

Diet is a major lifestyle factor contributing to the development of obesity, and it is modifiable if properly managed( Reference Stone and Kushner 6 ). Among various dietary factors, eating frequency (EF) has been suggested as a modifiable aspect of dietary behaviour that may influence the risk of obesity( Reference Holmback, Ericson and Gullberg 7 , Reference House, Shearrer and Miller 8 ). EF is usually defined as the sum of the number of meals and snacks consumed per day( Reference Holmback, Ericson and Gullberg 7 , Reference Bachman, Phelan and Wing 9 ). There have been many previous studies that have investigated the relationship between EF and obesity, but the results have been inconsistent thus far. A previous prospective study reported that decreased EF is predictive of greater 10-year increases in BMI and waist circumference (WC)( Reference Ritchie 10 ). A cross-sectional study also suggested that higher EF is associated with lower WC and reduced cardiometabolic risk factors, including fasting glucose, total cholesterol, LDL cholesterol and TAG, and that these associations are mediated by WC( Reference Smith, Blizzard and McNaughton 11 ). Another study revealed that increased EF is related to lower BMI and WC in overweight Hispanic youth( Reference House, Shearrer and Miller 8 ). In contrast, others reported the positive association between EF and obesity( Reference Murakami and Livingstone 12 ), or they could not detect any association( Reference Kant, Schatzkin and Graubard 13 , Reference Mills, Perry and Reicks 14 ). In particular, the body fat percentage (BF%) showed no association with EF( Reference Jennings, Cassidy and van Sluijs 15 ), or even a positive correlation with EF( Reference Yannakoulia, Melistas and Solomou 16 ). However, the factors that lead to such discrepancies between studies have not yet been clarified.

Previous studies have shown that a higher intake of fruits, vegetables, nuts and whole grains may play a protective role against obesity( Reference Fogelholm, Anderssen and Gunnarsdottir 17 ). In contrast, obesity results from a higher intake of sweets, refined bread and high-energy density food( Reference Fogelholm, Anderssen and Gunnarsdottir 17 ), which can induce an excessive energy intake (EI)( Reference Rolls, Bell and Castellanos 18 ). Korea has experienced a nutrition transition over recent years( Reference Lee and Cho 19 ). Consumption of traditional Korean diet, which is a low-fat and high-vegetable diet, has decreased, whereas consumption of bread, meat and seafood has increased, as Korea has become more westernised( Reference Lee and Cho 19 ). These changes have affected the increase of obesity prevalence( Reference Jung, Park and Choi 20 , Reference Popkin 21 ). Given that the diet quality may also play a role in the development of obesity( Reference Boggs, Rosenberg and Rodriguez-Bernal 22 , Reference Wolongevicz, Zhu and Pencina 23 ), it can be speculated that the relationship between EF and obesity might be modified by diet quality, but this issue has not been studied thus far.

Furthermore, few studies have covered the relationship between EF and obesity in Asian populations( Reference Kang, Ju and Park 24 , Reference Kim, Goh and Lee 25 ), although the typical Asian lifestyle, including diet and eating habits, differs from the typical Western lifestyle. Accordingly, the effect of EF on obesity may differ between Asian and Western populations. Therefore, in this study, we aimed to investigate the association between EF and obesity indicators including BMI, WC and BF% in a representative sample of the Korean adult population, and to determine whether the association between EF and the obesity indicators changes with diet quality.

Methods

Study population

This cross-sectional study was based on data from the 2nd and 3rd years of the Fourth Korea National Health and Nutrition Examination Survey (KNHANES IV), which was conducted by the Korean Centers for Disease Control and Prevention from 2007 to 2009. KNHANES IV is a nationwide representative study, and its target population is non-institutionalised civilians in the Republic of Korea. The sample frame was determined based on the 2005 population and housing census, and the representative households were selected using a stratified multistage clustered probability sampling design. However, family members younger than 1 year in the selected households were excluded from the survey. The survey was composed of four parts: the Health Interview Survey, the Health Behaviour Survey, the Health Examination Survey and the Nutrition Survey. All participants signed an informed consent form and the protocol was approved by the Institutional Review Board of the Korean Centers for Disease Control. The details regarding survey design and methods are specified elsewhere( 26 Reference Lee and Park 30 ).

In the 2008 and 2009 KNHANES IV, 9744 and 10 533 individuals participated, respectively. Of the 20 277 participants, we sequentially excluded 5206 subjects younger than 19 years old; 4480 subjects without measurements for BF%, WC or BMI; 1293 subjects without nutritional data including meal frequency, snack frequency and nutrient intake; 261 subjects without data for socio-economic or lifestyle variables such as household income, education level, smoking status, frequency of alcohol consumption, physical activity, frequency of resistance exercise, stress level and depressed mood; and 2086 subjects who answered that their meal frequency was zero or that they did not eat a meal or a snack as per usual on the survey day. The data of the remaining 6951 participants (aged 19–93 years) were analysed in this study (Fig. 1).

Fig. 1 Study population. BF%, body fat percentage; WC, waist circumference.

Measurement of variables

Measurement of obesity indicators

As indices of obesity, we used BF% measured by dual-energy X-ray absorptiometry (DXA), as well as WC and BMI. Trained medical staff measured height and weight by 0·1-cm units and 0·1-kg units, respectively, according to the standardised procedures at mobile examination centres. BMI was calculated as weight divided by height squared (kg/m2). WC was measured according to the WHO guideline to the nearest 0·1 cm in a horizontal plane at the level of the midpoint between the iliac crest and the costal margin at the end of a normal expiration( 31 ). Body composition status (i.e. BF%) was measured by DXA (Discovery-WTM; Hologic).

Eating frequency and nutritional measurements

EF was defined as the sum of the number of meals and snacks eaten per day as previously described( Reference Kim, Park and Yang 32 , Reference Murakami and Livingstone 33 ). Meal frequency was assessed using the following question: ‘Did you eat breakfast/lunch/dinner yesterday?’ Snack frequency was estimated using the following question: ‘How many times do you eat snacks a day?’ Although these questions have not been formally validated as measures of snack and meal frequency, similar questions have been used in other studies( Reference Kim, Park and Yang 32 , Reference Abdel-Megeid, Abdelkarem and El-Fetouh 34 , Reference Vik, Overby and Lien 35 ). Meal frequency was classified as one, two or three meals per day; snack frequency was classified as none, one, two or three snacks per day; and EF was categorised as less than three, three, four and five or more per day, as previously described( Reference Kim, Park and Yang 32 ). Categories of EF were determined by considering the distribution of each measure in the study population, ensuring adequate number in each group.

Daily energy and nutrient intake, including total EI (kJ/d (kcal/d)), fat intake (g/d), carbohydrate intake (g/d) and protein intake (g/d), were assessed on the basis of the results of the nutrition survey of KNHANES IV( 36 ). The nutrition survey was conducted in person by trained dietary interviewers, who visited participants’ homes and asked the respondents to remember in detail all the food and drinks they consumed during a period of time in the recent past (usually in the previous 24 hours). The 24-h recall method is a cost-effective and applicable dietary assessment for characterising the average population intake( Reference Karvetti and Knuts 37 ). Plastic food models, a set of measuring guides (including bowls, plates, earthen pots, jars, measuring spoons, glasses, mugs and coffee cups), shape charts (circle, ellipse, wedge, triangle and rectangle) and rulers were used to help the respondent report the volume and dimensions of dishes consumed, which was converted to weight using the food portion/weight database( 38 ). Next, the amount of each food ingredient included in a dish was estimated using the recipe database by Korea Health Industry Development Institute( Reference Kim 39 ). Finally, the energy and nutrient intake was calculated using the seventh edition of the food composition table from the Korean National Academy of Agricultural Science( 40 ). The results of these calculations are provided on the website of the Korea Centers for Disease Control and Prevention( 41 ).

Estimated average requirement (EAR) of daily energy was obtained from Dietary Reference Intakes for Koreans 2010 (online Supplementary Table S1)( 42 ). The nutrient adequacy ratio (NAR) was calculated for each of nine nutrients (protein, Ca, P, Fe, vitamin A, thiamine, riboflavin, niacin and vitamin C) using the following formula: NAR=subject’s daily intake of a nutrient/recommended nutrition intake according to sex and age of that nutrient. The nine NAR values were then averaged to yield a mean adequacy ratio (MAR)( Reference Madden, Goodman and Guthrie 43 ). The MAR provides an index of the overall diet quality. A high MAR implies a high-quality diet( Reference Kant 44 ).

Measurement of demographic, socio-economic and lifestyle-related variables

Data on the age group (categorised as 19–29, 30–39, 40–49, 50–59, 60–69 and 70 or more years), sex (male/female), smoking status (categorised as never, past (had smoked ≥100 cigarettes during their lifetime but not smoking currently) and current smoker (had smoked ≥100 cigarettes and still smoking)), alcohol drinking frequency (less than once per month, once per month, two to four times per month, two to three times per week and four times or more per week), education level (elementary school/middle school/high school/university), resistance exercise frequency (none, once per week, twice per week, three times per week, four times per week and five times or more per week) and physical activity assessed by the International Physical Activity Questionnaire scores (metabolic equivalent of task-min per week)( Reference Craig, Marshall and Sjostrom 45 ) were acquired by the Health Interview Survey. The household income was estimated using the following question: ‘What is the approximate total household income for the past 1 year, including wages, real-estate income, pensions, interest, government subsidies, allowances for relatives or children and so on?’ Thereafter, the study population was divided into quartiles (low/mid-low/mid-high/high) on the basis of the total household income. The depressed mood (yes/no) was determined using the following question: ‘Have you ever felt sad or desperate for 2 consecutive weeks during the past 1 year so that your daily life is hindered?’ The stress level (rare/a little/much/too much) was assessed using the following question: ‘How stressed out do you feel in your daily life?’

Statistical analysis

All statistical analyses were conducted using statistical software R version 3.3.1 (The R Foundation for Statistical Computing). As the KNHANES IV adopted a stratified multistage clustered probability sampling design, the survey package for R was used to account for cluster effects and sampling weights( Reference Lumley 46 , Reference Lumley 47 ). The study data included a total of 307 clusters, and three to 45 subjects were included in each cluster. All of the results are presented as weighted values. The general characteristics of the study population are presented as means with their standard errors (SE) for continuous variables and percentages (%) with their standard errors for categorical variables. The nutritional characteristics are also presented as mean values with their standard errors. We explored the association between EF and each obesity indicator (BF%, BMI and WC) using multiple linear regression analyses for complex survey design( Reference Lumley and Scott 48 ). To prevent multicollinearity between variables, we confirmed that all variance inflation factors are smaller than 10. Because the distributions of BF%, BMI and WC were not symmetric, the Box–Cox power transformation was used to make them more normally distributed. The parameters for the transformation of BF%, BMI and WC were 1·124397, −0·04895731 and 0·4656249, respectively. The design effects for the transformed variables from BF%, BMI and WC were 2·6549, 1·7001 and 2·7373, respectively. The transformed variables were used as dependent variables in the multiple regression analyses, and the predicted marginal means and 95 % CI for the dependent variables were estimated in each EF group. From those, the adjusted means and 95 % CI for BF%, BMI and WC were calculated as inverses of the Box–Cox transformation. We also tested for a linear trend of the dependent variables across EF groups after adjusting for potential confounders. The confounding variables were chosen on the basis of the results of previous studies, and included age group( Reference Molarius 49 ), sex( Reference Garawi, Devries and Thorogood 50 ), smoking status( Reference Dare, Mackay and Pell 51 ), alcohol drinking frequency( Reference Molarius 49 ), resistance exercise frequency( Reference Lee, Bacha and Hannon 52 ), physical activity( Reference Molarius 49 ), MAR (quartile)( Reference Yoon and Jang 53 ), EI:EAR ratio( Reference Chamieh, Moore and Summerbell 54 ), depressed mood( Reference Simon, Von Korff and Saunders 55 ), household income( Reference Drewnowski 56 ), education level( Reference Molarius 49 ) and stress level( Reference De Vriendt, Moreno and De Henauw 57 ). The same methods were also used to assess whether snack frequency and meal frequency were significantly associated with each obesity indicator. To determine whether the association between obesity indicators and EF, snack frequency and meal frequency changes according to diet quality, we analysed further multiple linear regression models that included all variables listed above and additional interaction terms between diet quality index (MAR quartiles) and EF, snack frequency or meal frequency. Then, the multiple linear regression analyses were repeated for the subgroups stratified by MAR quartile. Statistical significance was defined by a two-tailed P<0·05.

Results

General and nutritional characteristics of study participants according to eating frequency groups

EF ranged from 1 to 6, and it was classified as less than three (EF<3: 9·60 %), three (EF=3: 33·99 %), four (EF=4: 36·55 %) and five or more per day (EF≥5: 19·87 %). Mean values of obesity indicators and demographic, health behavioural and socio-economic characteristics according to EF groups are presented in Table 1. Table 2 summarises mean daily intakes of energy, macronutrients and micronutrients, and mean MAR values according to EF groups.

Table 1 General characteristics of the study population (Mean values and percentages with their standard errors)

EF, eating frequency; MET, metabolic equivalent of task.

* Means and standard errors using sampling weight for complex sample.

Percentages and standard errors using sampling weight for complex sample.

Table 2 Nutritional characteristics classified according to eating frequency (Mean values with their standard errors)

EF, eating frequency; EAR, estimated average requirement; NAR, nutrient adequacy ratio; RE, retinol equivalents; NE, niacin equivalents.

* EAR was obtained from Dietary Reference Intakes for Koreans 2010.

Mean adequacy ratio: average of NAR for nine nutrients (protein, Ca, P, Fe, vitamin A, thiamine, riboflavin, niacin, vitamin C). NAR=the subject’s daily intake of a nutrient/Korean recommended nutrition intake of that nutrient. All NAR values are truncated at 1·0.

Association between eating frequency and obesity indicators (body fat percentage, BMI and waist circumference)

In the multiple linear regression analyses for survey design with adjustment for covariates, the adjusted means of BF%, BMI and WC were all decreased as EF increased from the low-EF group (EF<3) to the high-EF group (EF≥5), showing a significant linear trend (P<0·001 for BF%, BMI and WC) (Table 3). The highest-EF group (EF≥5) showed statistically significant differences in all obesity indicators (BF%, BMI and WC) compared with all other EF groups (Table 4). As snack frequency increased from zero to three per day, the adjusted means of BF%, BMI and WC were also significantly altered (P=0·001 for BF%, 0·002 for BMI and 0·001 for WC). The group having snack twice a day showed significantly lower BF%, BMI and WC compared with the group having snack once a day or the group having no snack. Each obesity indicator was also significantly changed according to meal frequency (P=0·002 for BF%, <0·001 for BMI and 0·036 for WC). The group having meals three times a day showed significantly lower BF%, BMI and WC compared with the group having meals twice a day.

Table 3 Body fat percentage, BMI and waist circumference according to eating frequency (Adjusted means and 95 % confidence intervals)

* Adjusted for age group, sex, smoking status, alcohol drinking frequency, physical activity, resistance exercise frequency, household income, education level, stress level, mean adequacy ratio (quartile), energy intake:estimated average requirement ratio and depressed mood.

Adjusted for same variables as in * plus meal frequency.

Adjusted for same variables as in * plus snack frequency.

Table 4 P values for pairwise comparisons for body fat percentage, BMI and waist circumference according to eating frequency

* Adjusted for age group, sex, smoking status, alcohol drinking frequency, physical activity, resistance exercise frequency, household income, education level, stress level, mean adequacy ratio (quartile), energy intake:estimated average requirement ratio and depressed mood.

Adjusted for same variables as in * plus meal frequency.

Adjusted for same variables as in * plus snack frequency.

Change in the association between eating frequency and obesity indicators according to diet quality

To assess the diet-quality-associated alteration of the relation between obesity indicators and EF, snack frequency and meal frequency, we analysed multiple linear regression models that included the same variables as above and additional interaction terms between diet quality index (MAR quartiles) and EF, snack frequency or meal frequency. The results showed that the association between EF and each of the three obesity indicators was significantly altered according to diet quality (P value of the interaction term EF×MAR quartile=0·008 in the regression model for BF%, <0·001 for BMI and 0·043 for WC). In addition, the associations between snack frequency and both BMI and WC were significantly changed according to diet quality, but that between snack frequency and BF% was not (P value of the interaction term snack frequency×MAR quartile=0·036 in the regression model for BMI, 0·004 for WC and 0·105 for BF%). The interaction term meal frequency×MAR quartile was statistically significant only in the regression models for WC (P=0·013 in the regression model for WC, 0·100 for BMI and 0·080 for BF%). Thus, further subgroup analyses were performed to examine the association between EF, snack frequency and meal frequency and obesity indicators for which the interaction terms were significant in each group stratified by MAR quartile.

In the stratified multiple linear regression analyses according to diet quality group, EF showed significant inverse associations with every obesity indicator (BF%, BMI and WC) after adjusting for potential confounders in two higher diet quality groups (the third and fourth quartile of MAR), but not in two lower diet quality groups (the first and second quartile of MAR). In the highest and the second-highest diet quality groups, the adjusted means of BF%, BMI and WC were all decreased as EF increased from the low-EF group (EF<3) to the high-EF group (EF≥5), showing a significant linear trend. However, in the lowest and the second-lowest diet quality groups, there was no significant linear trend between EF and obesity indicators. The snack frequency was significantly associated with both BMI and WC in the highest diet quality group, with only WC in the second-highest diet quality group and with neither in the lower diet quality groups (the first and second MAR quartile). There was a significant association between meal frequency and WC only in the highest diet quality group, but not in the lower three diet quality groups (Table 5). Table 6 summarises the results of pairwise comparisons for BF%, BMI and WC according to EF in the subgroup analyses stratified by diet quality index.

Table 5 Body fat percentage, BMI and waist circumference according to eating frequency stratified by diet quality index (Adjusted means and 95 % confidence intervals)

MAR, mean adequacy ratio.

* Adjusted for age group, sex, smoking status, alcohol drinking frequency, physical activity, resistance exercise frequency, household income, education level, stress level, energy intake:estimated average requirement ratio and depressed mood.

Adjusted for same variables as in * plus meal frequency.

Adjusted for same variables as in * plus snack frequency.

Table 6 P values for pairwise comparisons for body fat percentage (BF%), BMI and waist circumference (WC) according to eating frequency stratified by diet quality index

MAR, mean adequacy ratio.

* Adjusted for age group, sex, smoking status, alcohol drinking frequency, physical activity, resistance exercise frequency, household income, education level, stress level, energy intake:estimated average requirement ratio and depressed mood.

Adjusted for same variables as in * plus meal frequency.

Adjusted for same variables as in * plus snack frequency.

Discussion

In this study, we found that EF, snack frequency and meal frequency are inversely associated with obesity indicators, namely BF%, BMI and WC, in Korean adults, but some of those relations were altered according to diet quality. When stratified by diet quality, these inverse associations between EF and the obesity indicators were significant in the higher two diet quality groups, but not in the lower two diet quality groups.

The pathway between EF and obesity has not yet been fully clarified. However, it had previously been indicated that increased EF is associated with reduced insulin concentration, postprandial lipid levels and lipogenesis( Reference Jenkins, Ocana and Jenkins 58 , Reference Jenkins, Wolever and Vuksan 59 ). Increased EF can decrease the postprandial surge of glucose and thus decrease the amount of insulin released in response( Reference Poston, Haddock and Pinkston 60 ). Insulin inhibits lipolysis in adipocytes, primarily through inhibition of lipase enzyme activity, and increases fat deposition( Reference Ma, Bertone and Stanek 61 , Reference Saltiel and Kahn 62 ). Thus, EF may contribute to the development of obesity.

However, despite these theoretical bases, previous clinical studies have shown mixed results. Several studies have revealed the inverse relationship between EF and obesity( Reference Ma, Bertone and Stanek 61 , Reference Ruidavets, Bongard and Bataille 63 ). A study of 499 US adults reported that an EF of four or more was associated with a lower risk of obesity in comparison with an EF of three or fewer( Reference Ma, Bertone and Stanek 61 ). Another study of 330 middle-aged men in France reported a significant inverse relationship between EF and BMI and waist:hip ratio( Reference Ruidavets, Bongard and Bataille 63 ). In contrast, a cross-sectional study of British adults reported a positive association between EF and BMI and between EF and WC( Reference Murakami and Livingstone 12 ). Meanwhile, some clinical trials reported no association between EF and weight loss( Reference Bachman and Raynor 64 , Reference Cameron, Cyr and Doucet 65 ), despite key limitations, namely small sample size and an insufficient follow-up time for making substantive conclusions( Reference Bachman and Raynor 64 , Reference Cameron, Cyr and Doucet 65 ).

The reason for this inconsistency remains unclear, but the controversy with regard to the association between EF and obesity in previous studies might arise from differences in the diet quality. The type of food eaten along with an increased EF may be important in determining the nature of the relationship between EF and obesity risk. A previous study of Australian adults suggested that higher EF is associated with lower WC and reduced fasting glucose and lipid profile in men, but not in women. They hypothesised that the different results between men and women might be owing to the difference in snacking quality. That is, Australian women who ate more often might eat unhealthy snacks( Reference Smith, Blizzard and McNaughton 11 ). In our study, diet quality was also a possible effect modifier on the relationship between EF and obesity indicators. Increased EF with low diet quality might countervail the metabolically positive effect of higher EF, and thus not be helpful in reducing obesity. However, previous studies did not consider diet quality, which might play a role in the discrepancy. The mechanism by which diet quality alters the association between EF and obesity has not been sufficiently clarified. However, previous studies suggested that poor diet quality is accompanied by greater insulin resistance and lower adiponectin levels, which is an adipokine that decreases insulin resistance and inflammation( Reference Johnson-Down, Labonte and Martin 66 Reference Mojiminiyi, Abdella and Al Arouj 68 ). This might counterbalance the advantage of high EF on the development of obesity.

The nutrition transition process in Korea has resulted in dietary changes including an increase in the animal food consumption and a decrease in total cereal intake( Reference Kim, Moon and Popkin 69 ). In this study, we could not find significant differences in EF according to income and education levels. However, the proportions of women were higher in higher-EF groups. A study on Americans also showed no association between diet quality and socio-economic status( Reference Hiza, Casavale and Guenther 70 ). However, the diet quality was improved with income level in adults, and women showed better diet quality. Better diet quality was observed in children and older adults compared with younger and middle-aged adults.

This study has some limitations. First, this is a cross-sectional study, which limited our ability to reveal the causal relationship between EF and obesity. Second, there may be potential recall bias because lifestyle factors, including diet, were based on data retrospectively collected by self-reported questionnaires. Third, dietary variables were estimated by a single 24-h dietary recall instead of three 24-h dietary recalls. This might not reflect the usual diet at the individual level, as day-to-day variation was not considered. Therefore, we only included study subjects who answered that they ate as per usual on the survey day to avoid this bias as much as possible. Finally, the participants of this study were of a single ethnic origin, and thus generalisation of the study results must be undertaken with caution. For all its limitations, the strength of this study is that it is the first to identify the potential effect modification of diet quality on the association between EF and obesity indicators using a representative sample of the Korean population.

In conclusion, we revealed that EF is inversely associated with obesity indicators including BMI, WC and BF%, and the association between EF and these obesity indicators is altered according to diet quality in Korean adults. EF is inversely associated with BMI, WC and BF% when diet quality is high, but not when it is low. Further prospective studies are needed to verify the causal relationship between EF and obesity.

Acknowledgements

This study was supported by the Clinical Research Center of Kangwon National University Hospital (grant of 2010 Research Support Project for Newly Appointed Clinical Professors).

G.-H. P. and S. K. conceived the study and had primary responsibility for final content, and G.-H. P. and J. H. Y. helped design the study. S. K. analysed data, performed statistical analyses and drafted the manuscript, and G.-H. P. helped with the revision of the manuscript. All authors read and approved the final manuscript.

The authors declare that there are no conflicts of interest.

Supplementary material

For supplementary material/s referred to in this article, please visit https://doi.org/10.1017/S0007114518000557

References

1. Guh, DP, Zhang, W, Bansback, N, et al. (2009) The incidence of co-morbidities related to obesity and overweight: a systematic review and meta-analysis. BMC Public Health 9, 88.Google Scholar
2. World Health Organization (2017) Prevalence of obesity among adults, BMI≥30, crude estimates by WHO region. http://apps.who.int/gho/data/view.main.2480A?lang=en (accessed June 2017).Google Scholar
3. Ng, M, Fleming, T, Robinson, M, et al. (2014) Global, regional, and national prevalence of overweight and obesity in children and adults during 1980–2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet 384, 766781.Google Scholar
4. Yoon, KH, Lee, JH, Kim, JW, et al. (2006) Epidemic obesity and type 2 diabetes in Asia. Lancet 368, 16811688.Google Scholar
5. Kang, H-T, Shim, J-Y, Lee, H-R, et al. (2014) Trends in prevalence of overweight and obesity in Korean adults, 1998–2009: The Korean National Health and Nutrition Examination Survey. J Epidemiol 24, 109116.Google Scholar
6. Stone, NJ & Kushner, R (2003) Effects of dietary modification to reduce vascular risks and treatment of obesity. Cardiol Clin 21, 415433.Google Scholar
7. Holmback, I, Ericson, U, Gullberg, B, et al. (2010) A high eating frequency is associated with an overall healthy lifestyle in middle-aged men and women and reduced likelihood of general and central obesity in men. Br J Nutr 104, 10651073.Google Scholar
8. House, BT, Shearrer, GE, Miller, SJ, et al. (2015) Increased eating frequency linked to decreased obesity and improved metabolic outcomes. Int J Obes (Lond) 39, 136141.Google Scholar
9. Bachman, JL, Phelan, S, Wing, RR, et al. (2011) Eating frequency is higher in weight loss maintainers and normal-weight individuals than in overweight individuals. J Am Diet Assoc 111, 17301734.Google Scholar
10. Ritchie, LD (2012) Less frequent eating predicts greater BMI and waist circumference in female adolescents. Am J Clin Nutr 95, 290296.Google Scholar
11. Smith, KJ, Blizzard, L, McNaughton, SA, et al. (2012) Daily eating frequency and cardiometabolic risk factors in young Australian adults: cross-sectional analyses. Br J Nutr 108, 10861094.Google Scholar
12. Murakami, K & Livingstone, MB (2014) Eating frequency in relation to body mass index and waist circumference in British adults. Int J Obes (Lond) 38, 12001206.Google Scholar
13. Kant, AK, Schatzkin, A, Graubard, BI, et al. (1995) Frequency of eating occasions and weight change in the NHANES I Epidemiologic Follow-up Study. Int J Obes Relat Metab Disord 19, 468474.Google Scholar
14. Mills, JP, Perry, CD & Reicks, M (2011) Eating frequency is associated with energy intake but not obesity in midlife women. Obesity (Silver Spring) 19, 552559.Google Scholar
15. Jennings, A, Cassidy, A, van Sluijs, EM, et al. (2012) Associations between eating frequency, adiposity, diet, and activity in 9-10 year old healthy-weight and centrally obese children. Obesity (Silver Spring) 20, 14621468.Google Scholar
16. Yannakoulia, M, Melistas, L, Solomou, E, et al. (2007) Association of eating frequency with body fatness in pre- and postmenopausal women. Obesity (Silver Spring) 15, 100106.Google Scholar
17. Fogelholm, M, Anderssen, S, Gunnarsdottir, I, et al. (2012) Dietary macronutrients and food consumption as determinants of long-term weight change in adult populations: a systematic literature review. Food Nutr Res 56, 19103.Google Scholar
18. Rolls, BJ, Bell, EA, Castellanos, VH, et al. (1999) Energy density but not fat content of foods affected energy intake in lean and obese women. Am J Clin Nutr 69, 863871.Google Scholar
19. Lee, KW & Cho, MS (2014) The traditional Korean dietary pattern is associated with decreased risk of metabolic syndrome: findings from the Korean National Health and Nutrition Examination Survey, 1998–2009. J Med Food 17, 4356.Google Scholar
20. Jung, SJ, Park, SH, Choi, EK, et al. (2014) Beneficial effects of Korean traditional diets in hypertensive and type 2 diabetic patients. J Med Food 17, 161171.Google Scholar
21. Popkin, BM (2001) The nutrition transition and obesity in the developing world. J Nutr 131, 871s873s.Google Scholar
22. Boggs, DA, Rosenberg, L, Rodriguez-Bernal, CL, et al. (2013) Long-term diet quality is associated with lower obesity risk in young African American women with normal BMI at baseline. J Nutr 143, 16361641.Google Scholar
23. Wolongevicz, DM, Zhu, L, Pencina, MJ, et al. (2010) Diet quality and obesity in women: the Framingham Nutrition Studies. Br J Nutr 103, 12231229.Google Scholar
24. Kang, HT, Ju, YS, Park, KH, et al. (2006) [Study on the relationship between childhood obesity and various determinants, including socioeconomic factors, in an urban area]. J Prev Med Public Health 39, 371378.Google Scholar
25. Kim, S, Goh, E, Lee, DR, et al. (2011) The association between eating frequency and metabolic syndrome. Korean J Health Promot 11, 917.Google Scholar
26. The Korean Association for Survey Research (2007) Sampling design of the 4th (2007~2009) KNHANES. http://knhanes.cdc.go.kr/ (accessed February 2012).Google Scholar
27. Kim, HJ, Kim, Y, Cho, Y, et al. (2014) Trends in the prevalence of major cardiovascular disease risk factors among Korean adults: results from the Korea National Health and Nutrition Examination Survey, 1998–2012. Int J Cardiol 174, 6472.Google Scholar
28. Kim, Y & Lee, BK (2012) Associations of blood lead, cadmium, and mercury with estimated glomerular filtration rate in the Korean general population: analysis of 2008–2010 Korean National Health and Nutrition Examination Survey data. Environ Res 118, 124129.Google Scholar
29. Kweon, S, Kim, Y, Jang, MJ, et al. (2014) Data resource profile: the Korea National Health and Nutrition Examination Survey (KNHANES). Int J Epidemiol 43, 6977.Google Scholar
30. Lee, KO & Park, JW (2007) A rolling sampling design for the Korea National Health and Nutrition Examination Survey. Surv Res 8, 6789.Google Scholar
31. World Health Organization Western Pacific Region (2000) The Asia-Pacific Perspective: Redefining Obesity and Its Treatment. Sydney: Health Communications Australia.Google Scholar
32. Kim, S, Park, GH, Yang, JH, et al. (2014) Eating frequency is inversely associated with blood pressure and hypertension in Korean adults: analysis of the Third Korean National Health and Nutrition Examination Survey. Eur J Clin Nutr 68, 481489.Google Scholar
33. Murakami, K & Livingstone, MB (2016) Associations between meal and snack frequency and overweight and abdominal obesity in US children and adolescents from National Health and Nutrition Examination Survey (NHANES) 2003–2012. Br J Nutr 115, 18191829.Google Scholar
34. Abdel-Megeid, FY, Abdelkarem, HM & El-Fetouh, AM (2011) Unhealthy nutritional habits in university students are a risk factor for cardiovascular diseases. Saudi Med J 32, 621627.Google Scholar
35. Vik, FN, Overby, NC, Lien, N, et al. (2010) Number of meals eaten in relation to weight status among Norwegian adolescents. Scand J Public Health 38, 1318.Google Scholar
36. Korea Health Industry Development Institute (2010) Quality control and analysis support on nutrition survey of KNHANES IV (2009). http://cdc.go.kr/CDC/info/CdcKrInfo0201.jsp?menuIds=HOME001-MNU1154-MNU0005-MNU1889&fid=28&q_type=&q_value=&cid=1503&pageNum=1 (accessed August 2017).Google Scholar
37. Karvetti, RL & Knuts, LR (1985) Validity of the 24-hour dietary recall. J Am Diet Assoc 85, 14371442.Google Scholar
38. Korea Health Industry Development Institute (2007) Development of Food and Nutrient Database – Food Portion/Weight Database. Seoul: Korea Health Industry Development Institute (in Korean).Google Scholar
39. Kim, YH (2011) Recipe database for evaluation of food and nutrient intakes. Public Health Wkly Rep Korea CDC 4, 786790.Google Scholar
40. Korean National Academy of Agricultural Science (2007) Food Composition Table, 7th ed. Suwon: Rural Development Administration (in Korean).Google Scholar
41. Korea Centers for Disease Control and Prevention (2017) Korea National Health and Nutrition Examination Survey. https://knhanes.cdc.go.kr/knhanes/eng/index.do (accessed August 2017).Google Scholar
42. The Korean Nutrition Society (2010) Dietary Reference Intakes for Koreans srS. Seoul: The Korean Nutrition Society.Google Scholar
43. Madden, JP, Goodman, SJ & Guthrie, HA (1976) Validity of the 24-hr. recall. Analysis of data obtained from elderly subjects. J Am Diet Assoc 68, 143147.Google Scholar
44. Kant, AK (1996) Indexes of overall diet quality: a review. J Am Diet Assoc 96, 785791.Google Scholar
45. Craig, CL, Marshall, AL, Sjostrom, M, et al. (2003) International physical activity questionnaire: 12-country reliability and validity. Med Sci Sports Exerc 35, 13811395.Google Scholar
46. Lumley, T (2017) Package ‘survey’ https://cran.r-project.org/web/packages/survey/survey.pdf (accessed August 2017).Google Scholar
47. Lumley, T (2010) Complex Surveys: A Guide to Analysis Using R. Hoboken, NJ: John Wiley & Sons, Inc.Google Scholar
48. Lumley, T & Scott, A (2017) Fitting regression models to survey data. Stat Sci 32, 265278.Google Scholar
49. Molarius, A (2003) The contribution of lifestyle factors to socioeconomic differences in obesity in men and women – a population-based study in Sweden. Eur J Epidemiol 18, 227234.Google Scholar
50. Garawi, F, Devries, K, Thorogood, N, et al. (2014) Global differences between women and men in the prevalence of obesity: is there an association with gender inequality? Eur J Clin Nutr 68, 11011106.Google Scholar
51. Dare, S, Mackay, DF & Pell, JP (2015) Relationship between smoking and obesity: a cross-sectional study of 499,504 middle-aged adults in the UK general population. PLOS ONE 10, e0123579.Google Scholar
52. Lee, S, Bacha, F, Hannon, T, et al. (2012) Effects of aerobic versus resistance exercise without caloric restriction on abdominal fat, intrahepatic lipid, and insulin sensitivity in obese adolescent boys: a randomized, controlled trial. Diabetes 61, 27872795.Google Scholar
53. Yoon, JS & Jang, H (2011) Diet quality and food patterns of obese adult women from low income classes: based on 2005 KNHANES. Korean J Community Nutr 16, 706715.Google Scholar
54. Chamieh, MC, Moore, HJ, Summerbell, C, et al. (2015) Diet, physical activity and socio-economic disparities of obesity in Lebanese adults: findings from a national study. BMC Public Health 15, 279.Google Scholar
55. Simon, GE, Von Korff, M, Saunders, K, et al. (2006) Association between obesity and psychiatric disorders in the US adult population. Arch Gen Psychiatry 63, 824830.Google Scholar
56. Drewnowski, A (2009) Obesity, diets, and social inequalities. Nutr Rev 67, Suppl. 1, S36S39.Google Scholar
57. De Vriendt, T, Moreno, LA & De Henauw, S (2009) Chronic stress and obesity in adolescents: scientific evidence and methodological issues for epidemiological research. Nutr Metab Cardiovasc Dis 19, 511519.Google Scholar
58. Jenkins, DJ, Ocana, A, Jenkins, AL, et al. (1992) Metabolic advantages of spreading the nutrient load: effects of increased meal frequency in non-insulin-dependent diabetes. Am J Clin Nutr 55, 461467.Google Scholar
59. Jenkins, DJ, Wolever, TM, Vuksan, V, et al. (1989) Nibbling versus gorging: metabolic advantages of increased meal frequency. N Engl J Med 321, 929934.Google Scholar
60. Poston, WS, Haddock, CK, Pinkston, MM, et al. (2005) Weight loss with meal replacement and meal replacement plus snacks: a randomized trial. Int J Obes (Lond) 29, 11071114.Google Scholar
61. Ma, Y, Bertone, ER, Stanek, EJ 3rd, et al. (2003) Association between eating patterns and obesity in a free-living US adult population. Am J Epidemiol 158, 8592.Google Scholar
62. Saltiel, AR & Kahn, CR (2001) Insulin signalling and the regulation of glucose and lipid metabolism. Nature 414, 799806.Google Scholar
63. Ruidavets, JB, Bongard, V, Bataille, V, et al. (2002) Eating frequency and body fatness in middle-aged men. Int J Obes Relat Metab Disord 26, 14761483.Google Scholar
64. Bachman, JL & Raynor, HA (2012) Effects of manipulating eating frequency during a behavioral weight loss intervention: a pilot randomized controlled trial. Obesity (Silver Spring) 20, 985992.Google Scholar
65. Cameron, JD, Cyr, MJ & Doucet, E (2010) Increased meal frequency does not promote greater weight loss in subjects who were prescribed an 8-week equi-energetic energy-restricted diet. Br J Nutr 103, 10981101.Google Scholar
66. Johnson-Down, L, Labonte, ME, Martin, ID, et al. (2015) Quality of diet is associated with insulin resistance in the Cree (Eeyouch) indigenous population of northern Quebec. Nutr Metab Cardiovasc Dis 25, 8592.Google Scholar
67. Monfort-Pires, M, Folchetti, LD, Previdelli, AN, et al. (2014) Healthy Eating Index is associated with certain markers of inflammation and insulin resistance but not with lipid profile in individuals at cardiometabolic risk. Appl Physiol Nutr Metab 39, 497502.Google Scholar
68. Mojiminiyi, OA, Abdella, NA, Al Arouj, M, et al. (2007) Adiponectin, insulin resistance and clinical expression of the metabolic syndrome in patients with Type 2 diabetes. Int J Obes (Lond) 31, 213220.Google Scholar
69. Kim, S, Moon, S & Popkin, BM (2000) The nutrition transition in South Korea. Am J Clin Nutr 71, 4453.Google Scholar
70. Hiza, HA, Casavale, KO, Guenther, PM, et al. (2013) Diet quality of Americans differs by age, sex, race/ethnicity, income, and education level. J Acad Nutr Diet 113, 297306.Google Scholar
Figure 0

Fig. 1 Study population. BF%, body fat percentage; WC, waist circumference.

Figure 1

Table 1 General characteristics of the study population (Mean values and percentages with their standard errors)

Figure 2

Table 2 Nutritional characteristics classified according to eating frequency (Mean values with their standard errors)

Figure 3

Table 3 Body fat percentage, BMI and waist circumference according to eating frequency (Adjusted means and 95 % confidence intervals)

Figure 4

Table 4 P values for pairwise comparisons for body fat percentage, BMI and waist circumference according to eating frequency

Figure 5

Table 5 Body fat percentage, BMI and waist circumference according to eating frequency stratified by diet quality index (Adjusted means and 95 % confidence intervals)

Figure 6

Table 6 P values for pairwise comparisons for body fat percentage (BF%), BMI and waist circumference (WC) according to eating frequency stratified by diet quality index

Supplementary material: File

Kim et al. supplementary material

Kim et al. supplementary material 1

Download Kim et al. supplementary material(File)
File 17.5 KB