Dietary behaviours associated with metabolic syndrome: a latent class analysis

Background Obese individuals have a number of dietary behaviours that might contribute to their weight-related health risks. However, obesity de�ned solely by body mass index may not re�ect the true heterogeneity of the obese population. The aims of this study were to classify the dietary behaviours of obese individuals into subgroups and to explore the relationship between their patterns of dietary behaviour and cardiometabolic risk factors. Methods The study participants were 259 patients who visited an outpatient weight management clinic at a tertiary hospital and underwent a dietary behaviour assessment between January 2014 and February 2019. Dietary behaviours were assessed in three domains with nine categories, including choice of food (frequent eating out, instant/fast/takeaway food), eating behaviour (irregular meals, frequent snacking, including eating at night, emotional eating, and overeating/binge eating), and nutrient intake (high fat/high calorie foods, salty food, and poorly balanced diet). Latent class analysis (LCA) was used to classify the subjects according to these nine categories of dietary behaviour. Associations between latent class and metabolic syndrome were assessed in a logistic regression model. Results Based on their dietary behaviour, the subjects were classi�ed into three LCA-driven classes, including a referent class of healthy eaters (n=118), a class of emotional eaters (n=53), and a class of irregular unhealthy eaters (n=88). Compared with the referent class, emotional eaters had a signi�cantly higher body mass index (beta=3.40, P<0.001) and metabolic syndrome (odds ratio 2.88, 95% con�dence interval 1.16–7.13). Conclusions Our �nding of three LCA-driven obesity phenotypes could be useful for assessment and management of obesity and metabolic syndrome. The association between higher BMI and metabolic syndrome was stronger in emotional eaters than in healthy eaters and irregular unhealthy eaters. Emotional eaters might bene�t from emotional regulation strategies.


Background
The World Health Organization has declared obesity to be a global epidemic with a prevalence that has tripled since 1975 and complications that have led to the death of at least 2.8 million individuals worldwide. 1Moreover, 35% of Korean adults aged ≥ 18 years are reported to be obese. 2esity is de ned as abnormal or excess accumulation of fat that endangers health.Although the body mass index (BMI) is widely used to classify overweight and obesity, 1 there are some caveats when using this value to determine excess body fat, the main one being that BMI does not discriminate between lean body mass and fat mass.Therefore, BMI can overestimate body fat in muscular athletes and underestimate sarcopenia in older adults. 3Moreover, the cardiometabolic risks of obesity and their heterogeneity cannot be assessed by BMI alone. 4besity is a consequence of multiple genetic, socioeconomic, lifestyle, and environmental factors.
However, the main driver of the obesity pandemic is likely to be changes in the global food system and accompanying changes in dietary behaviour. 5,6Several studies have shown a close link between obesity and dietary behaviour, particularly snacking in the absence of hunger, 7 consumption of fast foods, 8 binge eating or food addiction, 9 vulnerability to external food cues, 10 preference for foods high in calories, fat, sugar, and salt, and frequently eating outside the home. 11 is important to recognise the complex factors that shape and in uence dietary behaviour, which can be classi ed as food choice, eating behaviour, and dietary/nutritional intake. 12Food choice includes behaviours and other factors that precede food intake, such as preferences, frequency of purchase, food preparation, and intentions.Eating behaviour is categorised as eating habits, eating occasions, portion size, dieting, symptoms of disordered eating, and 'pickiness'.Dietary/nutritional intake includes the speci cs of what is consumed, such as dietary pattern, meal pattern, food intake, and food components.
Awareness of the different types of dietary behaviour has prompted a recommendation for a tailored approach to the treatment of obesity. 13Conceptualising and classifying subtypes of obese individuals according to their dietary behaviour might allow more personalised and effective behavioural and nutritional treatments.
5][16] It can also be used to divide a population into mutually exclusive subgroups and classify them exhaustively based on the intersection of multiple observed characteristics. 17e aims of this study were to classify the dietary behaviour of overweight or obese individuals into subgroups using LCA and to explore the relationship between each of these subgroups and cardiometabolic risk factors.

Study design and participants
This cross-sectional study was based on a retrospective review of the medical records of 259 patients who visited an outpatient weight management clinic at a tertiary hospital and underwent a dietary behaviour assessment between January 2014 and February 2019.Patients who were younger than 18 years, those for whom height or weight data were missing, and those with no dietary behaviour assessment records were excluded.

Sociodemographic and clinical variables
Our outpatient weight management clinic is attended by obese patients who want to lose weight or are referred by other departments.Patients who attend this clinic initially complete a self-administered questionnaire designed to collect information on sociodemographic and lifestyle factors, a weight history, and eating-related behaviours, and then attend a consultation with a family physician specialised in obesity management.At the next session, a dietitian performs a dietary assessment, provides the patient with detailed education on nutrition, and prescribes a low-or very-low-calorie diet.Next, the patient attends an appointment with a doctor who checks for adherence or barriers to the prescribed diet or the exercise recommendations and, if needed, prescribes appetite-suppressing medication.These sessions are repeated at intervals of 2 weeks for 1 month, 4 weeks for 2 months, and 8 weeks for 4 months until the patient has achieved a weight decrease of at least 10% of the initial body weight.Patients are then encouraged to attend a weight loss maintenance program, where they learn strategies for maintaining the weight lost, and attend regular follow-up sessions for at least 12 months.
In this study, we collected sociodemographic information on sex, age, income, and level of education from the self-completed questionnaires.Lifestyle factors, such as smoking, hazardous drinking, and frequency of exercise, were included.
Anthropometric measurements, including height, weight, systolic and diastolic blood pressure, and body fat were obtained by bioelectrical impedance analysis using the InBody 720 device (BioSpace Inc., Urbandale, IA).
Using the National Cholesterol Education Program-Adult Treatment Panel III criteria for Asian populations, 18 a diagnosis of metabolic syndrome was made when at least three of the following ve conditions were met: High blood pressure (systolic ≥ 130 mmHg or diastolic ≥ 85 mmHg) or on antihypertensive medication A high FBS level (≥ 100 mg/dL) or current use of oral antidiabetic medication A high TG level (≥ 150 mg/dL) or current use of lipid-lowering medication HDL cholesterol (< 40 mg/dL in men, < 50 mg/dL in women) Abdominal obesity (≥ 90 cm in men, ≥ 85 cm in women)

Dietary behaviour assessment
The self-administered dietary behaviour questionnaire included the following items, each of which was rated on a 5-point Likert scale: frequency of meals, skipping meals, frequent snacking or night eating, overeating or binge eating, preference for fatty foods or carbohydrates, consumption of instant food such as ramen, fast foods such as pizza or hamburgers, greasy food such as fried chicken, intake of sugarsweetened beverages, and frequency of intake of carbohydrates, vegetables and fruit, protein, fat and dairy products; Table 1).Based on previous studies, [19][20][21][22] we organised our dietary behaviour questions into three main domains with the following nine categories: food choice (frequently eating out and consumption of instant/fast food), eating behaviour (irregular meals, frequent snacking including at night, emotional eating, and overeating/binge eating), and nutritional intake (high-fat/high-calorie foods, salty food, and poorly balanced food intake).We reclassi ed the food choice and eating behaviour domains as dichotomous variables and recorded 'yes' if items in those categories were responded to as 'frequently' or 'very frequently'.For nutritional balance, we referenced the nutrition quotient calculation for Korean adults. 23Brie y, the Korean nutrition quotient consists of nutrition balance, food diversity, moderation of amount of food intake, and dietary behaviour.Nutritional balance assesses the intake of vegetable/fruit, protein ( sh, eggs, and beans), nuts and milk, and dairy products.If the calculated nutrition balance score was in the lower quartile, it was classi ed as unbalanced food intake.
High fat/high calorie intake included high consumption of fatty foods, such as fried food, meat, ham, or sausage and high consumption of sugar-sweetened beverages or foods containing sugar.Each score was summed to yield a mean score.If the mean high fat/high calorie score was more than 4, it was classi ed as 'yes', as for a dichotomous variable.Other latent variables were classi ed as 'yes' or 'no' according to the corresponding mean scores.

Statistical analysis
Continuous variables are presented as mean and standard deviation and categorical variables as frequency count and percentage.LCA was performed using PROC LCA (version 1.3.2) [24][25][26] and was used to classify the patients into nine subtypes of dietary behaviour.The Akaike information criterion and Bayesian information criterion were used to nd the best representative number of subgroups in these data.The number of classes was selected by a combination of parsimony and interpretability, and the class number can be possibly used as a meaningful label for each class. 27Each class was described in terms of sociodemographic factors, anthropometric measurements, and laboratory results.
We examined the association between LCA-derived classes and cardiometabolic risk factors using the LCA distal outcome Macro program (http://methodology.psu.edu/downloads).Brie y, LCA with a distal outcome was constituted with a 'three-step' (BCH) method 28 whereby the parameters of the LCA model are rst estimated without the distal outcome, the posterior probabilities of class membership are then used to compute a weighting variable, and nally the weighting variable is used to calculate a weighted average of outcomes for each class.
The LCA-driven classes were subsequently entered into a logistic regression model to test for associations between each class (with the relatively healthy class chosen as the referent) and the risk of metabolic syndrome with adjustment for age, sex, healthy behaviours, BMI, and other clinical variables.
All statistical analyses were performed using SAS version 9.3 (SAS Institute Inc., Cary, NC).

Results
LCA identi ed three classes of study subjects.The model t indices are summarised in Table 2. LCA models with 2-9 classes were compared for model t and interpretability.We chose a three-class model that showed the best t measured by a lower Bayesian information criterion with higher entropy.  1 show the conditional probabilities for each dietary behaviour in each class.Class 1 subjects (healthy eaters, n = 118) generally had regular meals and fewer snacks, reported less emotional eating, and did not eat fast/instant/takeaway food but tended to have a poorly balanced food intake.Class 2 subjects (irregular, unhealthy eaters, n = 88) were characterised by frequently eating out, irregular meal intake, a preference for calorie-dense food, and poorly balanced food intake, but no emotional eating.Class 3 (emotional eaters, n = 53) had a very high probability of emotional eating, binge eating, preference for calorie-dense food, irregular meals, and frequent eating at night.
[Insert Fig. 1 here] Table 3 summarises the sociodemographic and clinical factors and the health-related behaviours in each class.Subjects in the emotional eating group were younger, had a higher BMI, and were more likely to be female than those in the other two groups.Furthermore, the prevalence of metabolic syndrome was 49% in the emotional eating group and 32% in both the healthy eating and irregular unhealthy eating groups.
The emotional eaters were more likely to exercise regularly and less likely to smoke or consume alcohol than the other two groups.Table 4 shows the association between latent class and cardiometabolic risk factors.There were signi cant differences in BMI, waist circumference, and body fat percentage between the three classes.BMI, waist circumference, and body fat percentage values were lower in class 1 (the healthy referent group) than in classes 2 and 3, with no differences between classes 2 and class 3.No signi cant differences were found in the cardiometabolic variables between the three classes, except for GGT, which was higher in class 2 than in class 1.The prevalence of metabolic syndrome was higher in class 3 than in class 1. Logistic regression was performed to evaluate the associations between the three LCA-driven classes and metabolic syndrome using healthy eaters as the referent group (Table 5).Signi cant association of metabolic syndrome with emotional eating (LCA-driven class 3; odds ratio 2.88, 95% con dence interval 1.16-7.13),but not with irregular unhealthy eating (class 2) when compared with the referent group (class 1).Logistic regression analysis showed that metabolic syndrome was associated with a higher BMI, an increase in liver enzyme levels (> 120 mg/dL, equivalent to three times the upper limit of normal), and current smoking.

Discussion
In this study, subtyping of dietary behaviour in overweight or obese individuals identi ed three latent classes (healthy eaters, irregular unhealthy eaters, and emotional eaters).Moreover, we identi ed a signi cant association between emotional eating and a higher BMI, waist circumference, and body fat percentage.Importantly, emotional eaters also had a higher prevalence of metabolic syndrome.
Despite their different dietary pro les, subjects in classes 1 and 2 showed no signi cant differences in cardiometabolic risk factors except for GGT.This might re ect similar distributions of age, sex, and underlying disease.Irregular unhealthy eaters were more likely to have irregular meals, dine out, and consume alcohol (which might explain the increased GGT in this group).In contrast, classes 2 and 3 had similar dietary behaviour patterns, except for emotional eating, which might have contributed to their higher rates of binge eating, irregular meals, and frequent snacking, which in turn could lead a higher BMI and a higher likelihood of metabolic syndrome.
Emotional eating can be de ned as eating in response to negative emotions or stress and is one of the many causes for weight gain or regaining weight after dieting. 29It is known that stress and negative emotions can lead to a higher intake of palatable energy-dense foods, such as chocolate, cakes, biscuits, pizzas, hamburgers, French fries, and sausages. 30In a study performed in the Netherlands, emotional eating was found to be a mediator between depression and 5-year weight gain in mothers. 31In that study, depressive symptoms were associated with higher rates of emotional eating, which resulted in an increase in BMI independent of depressive symptoms.Emotional eating is also closely associated with insu cient sleep and poor sleep quality. 29Insu cient sleep can cause more negative emotions and interferes with regulation of emotion through neurobiological, cognitive, and behavioural pathways.
Therefore, identifying the subtype of obesity, such as emotional eating, can facilitate tailored treatment, such as training in emotional regulation skills or improving other lifestyle factors such as sleep.
The cross-sectional nature of this study meant that we could not detect a difference in the baseline cardiometabolic risk factors between subjects with healthy dietary behaviours and those with unhealthy behaviours (classes 1 and 2).However, it was worthwhile to subtype the group with irregular unhealthy dietary behaviour for the purposes of prediction and intervention.Several studies have suggested that an unhealthy dietary pattern can lead to increased weight, BMI, and waist circumference 32,33 and that this group of patients might achieve greater weight loss if they receive more tailored interventions targeting multiple health behaviours rather than strategies that target a single behaviour. 34 our knowledge, this is the rst study in Korea that has used LCA to classify dietary behaviour and explore its association with metabolic syndrome in overweight or obese adults.Dietary behaviours are highly inter-related, often concurrent, and affected by a complex interplay of multiple risk factors, including socioeconomic status and other health-related behaviours.LCA can address the complexity of dietary behaviour and capture meaningful key patterns. 27gistic regression analysis of metabolic syndrome and LCA-derived emotional eating showed signi cant associations but not when irregular unhealthy eaters were compared with relatively healthy eaters, indicating that some components of dietary behaviour have a greater effect than others.
The ndings of this study suggest a practical approach for identifying different phenotypes among individuals who are overweight or obese.Emotional eaters should be prioritised for emotional regulation and encouragement of emotional well-being.A recent review suggested that mindfulness and meditation have the potential to decrease emotional eating. 35cording to the American Heart Association, the timing and frequency of meals is also important in the management of cardiometabolic risk factors. 36Regular meals with avoidance of late-night snacking can attenuate the risk of heart disease and diabetes mellitus.In this regard, identifying individuals with unhealthy dietary behaviours, such as irregular mealtimes or frequent snacking (class 2), and managing them with the focus on eating patterns might help achieve a healthy cardiometabolic pro le as well as effective weight reduction.
This study has several limitations.First, it was not possible to use a validated dietary questionnaire because the study was based on a retrospective review of medical charts.Second, categorising the components of the dietary behaviour questionnaire into nine items could be considered subjective; however, we attempted to offset this problem by referencing a dietary pattern evaluation tool devised for Koreans. 22Third, the questionnaire used in the study was self-administered, which might have introduced a degree of reporting bias.Fourth, the study data were obtained from a single hospital weight management clinic and might not be generalisable to other populations.Dietary behaviours can be in uenced by sex, age, and socioeconomic factors. 37Neither sex-speci c nor age-speci c LCA could be performed in this study because of the relatively small number of patients enrolled after exclusion of those with missing data.
Despite its drawbacks, this study shows that subtyping obesity-related dietary behaviours could be a guide to prioritising the components that should be put in place for tailored cognitive behavioural therapy.
Further rigorous research is needed for these interventions to be effective in weight management.

Conclusion
We have shown that LCA-driven obesity phenotyping can be a useful tool for assessment and management of obesity as well as metabolic syndrome.In this study, a higher BMI with metabolic syndrome was more strongly associated with emotional eating than with healthy or irregular unhealthy eating.Emotional eaters might bene t from strategies targeting emotional regulation.People with irregular unhealthy dietary behaviours, such as irregular meal times or frequent night snacking, could be candidates for cognitive behavioural therapy focusing on healthy eating behaviours, which can also contribute to a favourable cardiometabolic pro le.

Table 1
Are you eating at least one of the following foods: meat, sh, egg, bean, and tofu when you have a meal?Do you prefer to eat multi-grain rice over white rice?[InsertTable1 here]

Table 2
Model-t indices for the latent class analysis model

Table 5
Odds ratio and 95% con dence interval for metabolic syndrome in association with latent classes and covariates