Hostname: page-component-586b7cd67f-dsjbd Total loading time: 0 Render date: 2024-11-22T20:32:35.287Z Has data issue: false hasContentIssue false

Association of dietary patterns with obesity and metabolically healthy obesity phenotype in Chinese population: a cross-sectional analysis of China Multi-Ethnic Cohort Study

Published online by Cambridge University Press:  10 January 2022

Dan Tang
Affiliation:
West China School of Public Health and West China Fourth Hospital, Sichuan University, Chengdu, People’s Republic of China
Xiong Xiao
Affiliation:
West China School of Public Health and West China Fourth Hospital, Sichuan University, Chengdu, People’s Republic of China
Liling Chen
Affiliation:
Chongqing Municipal Centre for Disease Control and Prevention, Chongqing, People’s Republic of China
Yixi Kangzhu
Affiliation:
Tibet Centre for Disease Control and Prevention, Lhasa, People’s Republic of China
Wei Deng
Affiliation:
Qingbaijiang District Centre for Disease Control and Prevention, Chengdu, People’s Republic of China
Basang
Affiliation:
Tibet University, Lhasa, People’s Republic of China
Shujuan Yang
Affiliation:
West China School of Public Health and West China Fourth Hospital, Sichuan University, Chengdu, People’s Republic of China
Lu Long
Affiliation:
West China School of Public Health and West China Fourth Hospital, Sichuan University, Chengdu, People’s Republic of China
Xiaofen Xie
Affiliation:
West China School of Public Health and West China Fourth Hospital, Sichuan University, Chengdu, People’s Republic of China
Jiaojiao Lu
Affiliation:
West China School of Public Health and West China Fourth Hospital, Sichuan University, Chengdu, People’s Republic of China
Qun Meng
Affiliation:
West China School of Public Health and West China Fourth Hospital, Sichuan University, Chengdu, People’s Republic of China General Supervision Bureau, National Health Commission of the People’s Republic of China, Beijing, People’s Republic of China
Jianzhong Yin*
Affiliation:
School of Public Health, Kunming Medical University, Kunming, People’s Republic of China Baoshan College of Traditional Chinese Medicine, Baoshan, People’s Republic of China
Feng Hong*
Affiliation:
School of Public Health, the key Laboratory of Environmental Pollution Monitoring and Disease Control, Ministry of Education, Guizhou Medical University, Guiyang, People’s Republic of China
Xing Zhao*
Affiliation:
West China School of Public Health and West China Fourth Hospital, Sichuan University, Chengdu, People’s Republic of China
*
*Corresponding authors: Xing Zhao, email [email protected] Jianzhong Yin, email [email protected]; Feng Hong, email [email protected].
*Corresponding authors: Xing Zhao, email [email protected] Jianzhong Yin, email [email protected]; Feng Hong, email [email protected].
*Corresponding authors: Xing Zhao, email [email protected] Jianzhong Yin, email [email protected]; Feng Hong, email [email protected].
Rights & Permissions [Opens in a new window]

Abstract

Metabolically healthy obesity (MHO) might be an alternative valuable target in obesity treatment. We aimed to assess whether alternative Mediterranean (aMED) diet and Dietary Approaches to Stop Hypertension (DASH) diet were favourably associated with obesity and MHO phenotype in a Chinese multi-ethnic population. We conducted this cross-sectional analysis using the baseline data of the China Multi-Ethnic Cohort study that enrolled 99 556 participants from seven diverse ethnic groups. Participants with self-reported cardiometabolic diseases were excluded to eliminate possible reverse causality. Marginal structural logistic models were used to estimate the associations, with confounders determined by directed acyclic graph (DAG). Among 65 699 included participants, 11·2 % were with obesity. MHO phenotype was present in 5·7 % of total population and 52·7 % of population with obesity. Compared with the lowest quintile, the highest quintile of DASH diet score had 23 % decreased odds of obesity (OR = 0·77, 95 % CI 0·71, 0·83, P trend < 0·001) and 27 % increased odds of MHO (OR = 1·27, 95 % CI 1·10, 1·48, P trend = 0·001) in population with obesity. However, aMED diet showed no obvious favourable associations. Further adjusting for BMI did not change the associations between diet scores and MHO. Results were robust to various sensitivity analyses. In conclusion, DASH diet rather than aMED diet is associated with reduced risk of obesity and presents BMI-independent metabolic benefits in this large population-based study. Recommendation for adhering to DASH diet may benefit the prevention of obesity and related metabolic disorders in Chinese population.

Type
Research Article
Copyright
© The Author(s), 2022. Published by Cambridge University Press on behalf of The Nutrition Society

The global obesity prevalence has nearly tripled since 1975(1). According to Global Burden of Diseases Study 2019, high BMI is among the top five leading risks, accounting for 6·3 % of the global disability-adjusted life-years(Reference Murray, Aravkin and Zheng2). In recent years, a novel obesity phenotype with favourable metabolic profiles, referred to as metabolically healthy obesity (MHO), has raised global interest. Many studies have demonstrated that people with MHO have substantially lower risk of cardiometabolic diseases than their metabolically unhealthy obesity (MUO) counterparts; rigorous and stable MHO even have comparable risks with metabolically healthy normal-weight phenotype(Reference Mongraw-Chaffin, Foster and Anderson3Reference Gao, Lv and Yu6). Given the unsatisfied effect of weight loss-centred strategies for obesity treatment and substantially improved prognosis of MHO, identification of this phenotype and associated determinants, especially modifiable ones, is important for the targeted management of obesity and related adverse outcomes(Reference Smith, Mittendorfer and Klein7,Reference Iacobini, Pugliese and Fantauzzi8) .

Dietary factors are a leading modifiable contributor to obesity and cardiometabolic disorders. In particular, dietary pattern has gained great interest because it reflects the holistic effect of diet and facilitates clinical and public dietary recommendations(Reference Mozaffarian9,Reference Hu10) . The Mediterranean diet and Dietary Approaches to Stop Hypertension (DASH) diet, as two most well-recognised healthy dietary patterns, have been demonstrated to be related to reduced risk for the metabolic syndrome(Reference Hosseini, Whiting and Vatanparast11,Reference Kastorini, Milionis and Esposito12) . However, existing empirical studies regarding the associations of Mediterranean or DASH diet with metabolic obesity phenotypes are limited and have produced inconsistent results(Reference Konieczna, Yanez and Monino13Reference Farhadnejad, Darand and Teymoori18). Most of these studies were conducted in Western countries(Reference Konieczna, Yanez and Monino13Reference Phillips, Dillon and Harrington16), and the results suggest that associations may be heterogeneous across different populations. Up to date, it is unclear whether these healthy dietary patterns are favourably associated with MHO in Chinese population, where the obesity and cardiometabolic disease burden have raised rapidly, and the dietary habits, obesity and metabolic phenotypes distribution differs substantially from Western populations(Reference Ma, Xi and Yang19,Reference Huang, Huang and Withers20) .

The China Multi-Ethnic Cohort (CMEC) is a large-scale cohort with nearly 0·1 million heterogeneous participants from seven diverse ethnic groups. Exploring the relationship of Mediterranean and DASH diets with obesity and MHO in this population will provide unique evidence and important implications for prevention strategies of obesity and related metabolic disorders in China and may further reveal the age, sex, ethnicity and socio-economic status (SES) disparities of the association.

Therefore, based on baseline data of the CMEC, this study aimed to (1) examine the prevalence of obesity and different metabolic obesity phenotypes in this population; (2) explore whether the Mediterranean and DASH diets are associated with a reduced risk of obesity and (3) further investigate whether those dietary patterns are associated with the favourable MHO phenotype in population with obesity (MHO v. MUO).

Materials and methods

Study participants

CMEC started in 2018 and enrolled 99 556 subjects from seven ethnic groups (Tibetan, Yi, Miao, Bai, Bouyei, Dong and Han) in Southwest China. Every eligible participant was required to complete thorough electronic questionnaires, physical measurements and clinical laboratory tests. The survey and collection process were conducted by trained staff with standardised operation procedures. All participants provided written informed consent, and Sichuan University Medical Ethical Review Board and institutional review boards at all CMEC sites approved the study. More detailed information about the study design, sampling methods, baseline measurements and population characteristics has been published elsewhere(Reference Zhao, Hong and Yin21).

For the current analysis, we excluded participants aged < 30 or > 79 years (n 1003), without available data to calculate dietary pattern scores (n 343) or outcome indicators (n 4382), with implausible low or high energy intake (< 3347 [800 kcal] or > 17 572 [4200 kcal] kJ/d for men and < 2510 [600 kcal] or > 14 644 [3500 kcal] kJ/d for women, n 2284)(Reference Shan, Li and Baden22) and with BMI < 18·5 kg/m2 (underweight, n 3095). In addition, those reported previously physician-diagnosed hypertension, diabetes, hyperlipidaemia, CHD, stroke or cancer were also excluded (n 22 750) to minimise possible reverse causality. Finally, 65 699 subjects were eligible and included in this analysis (online Supplementary Appendix Fig. S1).

Exposure assessment

Dietary intake was assessed using a validated FFQ. The FFQ full text, the collection and processing of diet data and the validation of the FFQ have been reported in detail elsewhere(Reference Xiao, Qin and Lv23). Briefly, information about the frequency (times per day, week, month or year) and quantity (average grams per time) for each food group during the preceding year was collected with the aid of food moulds. Moreover, we asked about the types, frequency and quantity of alcohol, tea and other beverages at the individual level and the consumption of oil and salt at the family level. Daily consumption of each food group (g/d), total energy and certain nutrient intake were calculated according to the China food exchange lists and the 2018 China food composition tables. Reliability and validity of the FFQ were assessed in subsample population who completed a second FFQ survey or a 3-d non-consecutive 24-h dietary recall survey. In terms of reliability, intraclass correlation coefficients for different food groups ranged from 0·15 to 0·67. In terms of validity, de-attenuated Spearman rank correlation coefficients ranged from 0·10 to 0·66.

A modified DASH index was used to score the participants’ adherence to DASH diet(Reference Fung, Chiuve and McCullough24). Compared with the reference DASH index, we dropped the sweetened beverages component and nuts component and additionally replaced low-fat dairy component with total dairy products, because of limited information and very low regular consumption proportion for them in our population. Similarly, we adopted alternative Mediterranean diet (aMED) index to represent participants’ adherence to Mediterranean diet(Reference Shan, Li and Baden22), with further exclusion of nuts due to no available information. The detailed scoring criteria are shown in Table 1.

Table 1. Scoring criteria for the DASH diet and aMED diet

DASH, Dietary Approaches to Stop Hypertension; aMED, alternative Mediterranean; Q, quintile.

* Alcohol consumption was categorised into five predefined groups: (10,30], (0,10] or (30,40], 0 or (40,45], (45,50] and > 50 g/d for men; (5,15], (0,5] or (15,25], 0 or (25,30], (30,35] and > 35 g/d for women, and then we assigned five points to the first group, four points to the second group, three points to the third group, two points to the fourth group and one point to the fifth group.

Outcome definition

Height to the nearest 0·1 cm and weight to the nearest 0·1 kg were measured. Waist circumference (WC) was measured to the nearest 0·1 cm at the midpoint between the lowest rib margin and the iliac crest. BMI was calculated as weight in kilograms divided by height in meters squared (kg/m2). Systolic and diastolic blood pressures were measured three times after at least 5 min of rest, and the mean values of these measurements were used for analysis. In addition, overnight fasting blood samples were collected, and blood glucose and lipid profiles were tested using an AU5800 automated chemistry analyzer (Beckman Coulter Commercial Enterprise).

Obesity was defined as BMI ≥ 28 kg/m2 according to Chinese criteria. Metabolic health was defined as the absence of the metabolic syndrome. Thus, based on a harmonised definition for the metabolic syndrome(Reference Alberti, Eckel and Grundy25), the presence of two or fewer of the following five criteria was considered metabolically healthy: (1) WC ≥ 90 cm for males and ≥ 85 cm for females; (2) TAG ≥ 1·7 mmol/l; (3) HDL-cholesterol < 1·0 mmol/l for males and < 1·3 mmol/l for females; (4) systolic blood pressure ≥ 130 mmHg or diastolic blood pressure ≥ 85 mmHg and (5) fasting blood glucose ≥ 5·6 mmol/l. MHO individuals were therefore those with both metabolically healthy status and obesity.

Covariate measurements

Information about socio-demographic, lifestyle factors, health status, medical history and family history was obtained from the baseline questionnaire. To avoid incomplete adjustment or overadjustment, we firstly determined a pool of potential confounders through a systematic literature review on this study topic and constructed a directed acyclic graph (DAG) for this study according to the procedure of ‘Evidence Synthesis for Constructing Directed Acyclic Graphs’ method(Reference Ferguson, McCann and Katikireddi26). The minimally sufficient set of confounders based on the constructed DAG was determined (online Supplementary Appendix Text S2), and the subsequent association analysis would adjust for age, sex, ethnic group (Tibetan, Yi, Miao, Bai, Bouyei, Dong or Han), urbanicity (urban or rural), education (no schooling/primary school, middle/high school or college/university), household income (< ¥20 000, ¥20 000–99 999 or ≥ ¥100 000), profession (farmers or non-farmers), marital status (married/cohabiting or other), smoking (current smoker or not), physical activity (metabolic equivalent task, MET h/d), total energy intake (kcal/d), regular intake of soft drinks, dietary supplements (including fish oil/cod liver oil, vitamin D, other vitamins, Ca tablets and other dietary supplements), spicy food and peppery food (yes or no); insomnia symptoms, anxiety symptoms and depression symptoms (presence or absence); menopause status for women (premenopausal, perimenopausal or postmenopausal, men were coded as a distinct category) and family history of cardiometabolic diseases (presence or absence).

Statistical analysis

Age- and sex-adjusted characteristics of the study participants were described as the mean values and standard deviations or percentages across the presence or absence of obesity and across different metabolic phenotypes (MHO and MUO) in population with obesity. Continuous and categorical variables were compared using linear regression analyses and χ 2 tests, respectively. Marginal structural logistic models using the inverse probability of exposure weighting were employed to assess the associations of quintiles of aMED and DASH diet scores with obesity and MHO separately(Reference Ukoumunne, Williamson and Forbes27). We constructed the exposure model using quintiles of diet scores as dependent variable and confounders (described in the ‘Covariate measurements’ section) as independent variables, and we then compared the balances of these confounders across different exposure groups among different candidate weighting methods to determine the optimal weighting method(Reference Greifer28). In our final analysis, we used the entropy balancing weighting method due to its preferable performance in balancing confounders(Reference Hainmueller29) (see detailed balancing results in online Supplementary Appendix Text S3). The linear trend of these associations was tested using the median scores of each quintile as continuous variables in models. In addition, we further explored whether the associations of diet scores with MHO phenotype were dependent or independent of BMI by additionally adjusting for BMI when computing the weights.

To facilitate the interpretation and explore the dominant contributors to the observed associations of overall dietary patterns, we further conducted component analysis. First, we included all component scores simultaneously in multivariable logistic regression (adjusting for all identified confounders described in the ‘Covariate measurements’ section) for each dietary pattern to evaluate the associations of individual components with outcomes. Subsequently, we assessed the contribution of each of the seven or eight components of DASH diet and aMED index to the risk for obesity and MHO, respectively, by dropping one component at a time from the total score(Reference Trichopoulou, Bamia and Trichopoulos30) (see online Supplementary Appendix Text S4 for more details).

To evaluate disparities in the associations of dietary patterns with obesity and MHO, we performed analysis additionally stratified by sex, age, urbanicity, ethnic group and SES. The heterogeneity among different strata was assessed using the I 2 statistic and Q test (α = 0·1, and I 2 > 50 % was considered significantly heterogeneous). We also performed several sensitivity analyses to examine the robustness of our results: including individuals with self-reported hypertension, diabetes, hyperlipidaemia, CHD, stroke or cancer; using the strict definition of metabolic health (the absence of all abovementioned criteria except for WC); replacing the WC criteria with the waist:hip ratio (≥ 0·90 for men and ≥ 0·85 for women) criteria to define metabolic health and additionally adjusting drinking status when estimating the association of DASH diet (without alcohol component) with obesity and MHO. All statistical analyses were performed with the R Project for Statistical Computing version 4.0.2.

Results

Characteristics of participants

Among the 65 699 participating subjects, the mean age was 49·4 (sd10·9) years, and 38·1 % were male. Individuals with obesity accounted for 11·2 % of the total population, and MHO phenotype accounted for 5·7 % of the total population and 52·7 % of population with obesity. The age- and sex-adjusted characteristics of participants with different obesity status and metabolic obesity phenotypes are presented in Table 2. Individuals with obesity were more likely to be male, Tibetans and non-farmers and to have a lower SES (education and household income), unhealthier lifestyles (physical activity, energy intake and soft drink consumption), lower aMED and DASH diet scores and poor physical and biochemical indicators (BMI, WC, blood pressure, blood glucose and lipid profiles) than individuals without obesity. MHO individuals were more likely to be younger, female, Tibetans, rural residents, farmers, non-current smokers or drinkers and to have lower SES, less reported family history of cardiometabolic diseases, lower aMED and higher DASH diet scores, and better physical and biochemical indicators than MUO.

Table 2. Age- and sex-adjusted characteristics of the study participants

(Mean values and standard deviations for continuous variables; absolute and relative frequencies for categorical variables)

MHO, metabolically healthy obesity; MET, metabolic equivalent task; MUO, metabolically unhealthy obesity; aMED, alternative Mediterranean; DASH, Dietary Approaches to Stop Hypertension; WC, waist circumference; SBP, systolic blood pressure; DBP, diastolic blood pressure; FBG, fasting blood glucose.

* The reported means and percentages are standardised to the age and sex distribution of the corresponding study population except for N (not adjusted), age (only adjusted sex), sex (only adjusted age) and menopausal status in women (only adjusted age).

Linear regression analyses and χ 2 tests were used to compare different groups for continuous variables and categorical variables, respectively.

Associations of dietary patterns with obesity and metabolically healthy obesity

Associations of adherence to DASH and aMED diets with obesity and MHO phenotype are shown in Fig. 1. We observed a linearly negative association between quintiles of DASH diet scores and obesity as well as a linearly positive association between DASH index and MHO phenotype in population with obesity. Compared with participants in the lowest quintile, those with the highest quintile of DASH diet scores had a 23 % decrease in the odds of obesity (OR = 0·77, 95 % CI 0·71, 0·83, P trend < 0·001). Among individuals with obesity, those in the fifth v. the first quintile of DASH diet scores had a 27 % increase in the odds of MHO phenotype (OR = 1·27, 95 % CI 1·10, 1·48, P trend = 0·001). However, no linear association of aMED diet scores with obesity or MHO phenotype was observed in this study, only the third and/or fourth quintile showed borderline benefit. Table 3 presents the results of the associations of aMED and DASH diet scores with MHO phenotype after additionally adjusting for BMI. Overall, the estimated OR (95 % CI) of MHO across different quintiles of these two diet scores were almost identical to those in the main analysis.

Fig. 1. Association of aMED and DASH diet scores with obesity and MHO phenotype. All models adjusted for age, sex, ethnic group, urbanicity, education, household income, profession, marital status, smoking, physical activity, total energy intake, regular intake of soft drinks, dietary supplements, spicy food and peppery food, insomnia symptoms, anxiety symptoms, depression symptoms, menopause status and family history of cardiometabolic diseases using logistic regression with inverse probability of exposure weighting (IPEW). Q2–Q5 represent the second to fifth quintiles of diet scores. The filled red dots represent estimated OR and the vertical red lines represent 95 % CI. aMED, alternative Mediterranean; DASH, Dietary Approaches to Stop Hypertension; MHO, metabolically healthy obesity.

Table 3. OR of MHO associated with quintiles of aMED and DASH diet scores after adjusting for BMI*

(Odds ratios and 95 % confidence intervals)

MHO, metabolically healthy obesity; aMED, alternative Mediterranean; DASH, Dietary Approaches to Stop Hypertension; Q, quintiles.

* BMI was further included in the exposure models (rather than the outcome models) to compute the weights.

Regarding component analysis, the discrepant components between the two dietary patterns explained the differences in their overall associations with obesity and MHO (online Supplementary Appendix Tables S1 and S2). Among the three exclusive components of aMED diet, alcohol showed no associations with obesity or MHO; fish were not associated with MHO but positively associated with obesity and MUFA:SFA was detrimental for both obesity and MHO. In contrast, for the two exclusive components of DASH diet, dairy products were beneficial for both obesity and MHO; Na score presented no association with obesity and a positive association with MHO. Table 4 reports the relative contribution of each component to the observed associations of DASH diet with obesity and MHO. In terms of obesity, dairy products, legumes, and red and processed meat components contributed the most, accounting for 63·1, 50·1 and 25·5 % of the overall negative association, respectively. In terms of MHO, Na, dairy products and vegetable components were the dominant contributors, which accounted for 36·3, 32·9 and 32·9 % of the total positive association, respectively.

Table 4. OR of obesity and MHO associated with overall DASH diet score and subtracted DASH diet scores for each component

(Odds ratios and 95 % confidence intervals)

MHO, metabolically healthy obesity; DASH, Dietary Approaches to Stop Hypertension.

* We assumed a linear relationship between dietary scores and outcomes, with OR representing the risk change per 25 % score range increment.

The models of seven DASH minus component scores were further adjusted for the corresponding subtracted component score in addition to confounders in the main analysis, and the estimated coefficients of these logistic regressions were multiplied by 25/29 to correct for different score ranges.

Reduction in apparent effect (%) = (ORDASH overall – ORDASH minus component)/(ORDASH overall – 1) × 100 %.

§ The reduction in apparent effect can be negative when the effect of the corresponding subtracted component score was unexpectedly detrimental in this study. Similarly, the sum of those positive percentages of the reduction in apparent effect can exceed 100 %.

Stratification and sensitivity analyses

Figure 2 reports the results of stratification analysis on the associations of DASH diet scores with obesity and MHO (results of aMED diet are not shown because of its almost null association in the main analysis). In terms of obesity, the negative association was stronger in younger (I 2 = 74·9 %; P = 0·046) and higher educated (I² = 70·4 %; P = 0·066) participants. In terms of MHO, the positive association was greater in urban residents (I² = 87·3 %; P = 0·005) and increased with the rising education levels (I² = 87·3 %; P = 0·005).

Fig. 2. Stratified analysis for associations of DASH diet scores with obesity and MHO according to predefined demographic and socio-economic factors by comparing the highest with the lowest quintiles. All models adjusted for age, sex, ethnic group, urbanicity, education, household income, profession, marital status, smoking, physical activity, total energy intake, regular intake of soft drinks, dietary supplements, spicy food and peppery food, insomnia symptoms, anxiety symptoms, depression symptoms, menopause status and family history of cardiometabolic diseases using logistic regression with inverse probability of exposure weighting (IPEW), with exclusion of the corresponding stratification factor as appropriate. I 2 statistic and Q test were used to assess heterogeneity among different strata of each stratification factor. The filled red dots represent estimated OR and the horizontal red lines represent 95 % CI. DASH, Dietary Approaches to Stop Hypertension; MHO, metabolically healthy obesity.

In the sensitivity analysis, the association of DASH diet and MHO was moderately attenuated when including individuals with self-reported cardiometabolic diseases or cancer (online Supplementary Appendix Fig. S4), indicating the existence of reverse causality. When we used the strict definition of metabolic health or replaced the WC criteria with waist:hip ratio criteria, the estimated results were very similar (online Supplementary Appendix Fig. S5 and S6). Moreover, further adjusting for drinking status in models of DASH diet resulted in almost identical results as in the main analysis (online Supplementary Appendix Fig. S7).

Discussion

Summary of main results

Our findings show that in this multi-ethnic Chinese population, the highest adherence to DASH diet was associated with 23 % lower odds of obesity and 27 % higher odds of MHO in population with obesity, with no obvious benefits found for aMED diet. The favourable effect of DASH diet on metabolic health for population with obesity was independent of BMI. The benefits of DASH diet on obesity or metabolic health were greater in urban residents and younger and higher educated participants.

The prevalence, characteristics and mechanism of metabolically healthy obesity

In recent decades, interest has focused on MHO phenotype, including the definition, prevalence, characteristics, mechanism, prognosis and medical implications of this new concept. In terms of prevalence, existing studies have reported results with very large variability, ranging from 6 to 75 %(Reference Rey-Lopez, de Rezende and Pastor-Valero31). Differences in the definition of MHO used, study design and sample size of the research, age, sex, ethnicity and other characteristics of the included population are likely responsible for the huge disparity(Reference Smith, Mittendorfer and Klein7). In terms of characteristics, MHO seems to be more common in females(Reference Rey-Lopez, de Rezende and Pastor-Valero31), younger individuals(Reference Rey-Lopez, de Rezende and Pastor-Valero31), Asian populations(Reference Rey-Lopez, de Rezende and Pastor-Valero31), more physically active individuals(Reference Ortega, Cadenas-Sanchez and Migueles32) and subjects with better nutritional status(Reference Iacobini, Pugliese and Fantauzzi8), which were almost consistent with our findings. In mechanism, MHO may attribute to better body fat distribution (lower ectopic fat and higher leg fat deposition), preserved insulin sensitivity, normal adipose tissue function, lower levels of inflammatory and greater cardiorespiratory fitness compared with MUO(Reference Bluher33). In our study, 52·7 % of subjects with obesity were ascertained to have MHO, which was relatively high among existing reports. The large proportion of female participants and exclusion of self-reported cardiometabolic diseases may largely explain the disparity. In addition, given the multi-ethnic nature of our study participants, genetic factors which are related to biological mechanisms of MHO may differ across diverse ethnic populations and thus contribute to the observed difference(Reference Schleinitz, Bottcher and Bluher34,Reference Loos and Kilpelainen35) .

Regarding obesity and metabolically healthy obesity, Dietary Approaches to Stop Hypertension diet is a superior dietary recommendation

In this study, we observed significant and robust beneficial effects of DASH diet, which was consistent with previous evidence(Reference Farhadnejad, Darand and Teymoori18,Reference Soltani, Shirani and Chitsazi36) , but no obvious favourable findings for aMED diet. As shown in the component analysis, unexpected detrimental or null associations of the three exclusive components of aMED diet with outcomes were responsible for the overall null association. MUFA:SFA ratio and fish components represent the recommendation for consumption of healthy olive oil and marine fish, which are rich in MUFA(Reference Willett, Sacks and Trichopoulou37,Reference Trichopoulou, Costacou and Bamia38) , but this recommendation may be difficult to follow in non-Mediterranean regions and populations such as ours. The lack of such a superior source of MUFA and the high-temperature and deep cooking style in these populations may eliminate or even reverse the protective effect. In addition, moderate alcohol consumption is recommended in aMED diets, but the benefits of the recommendation were controversial or conditional. A rigorous Mendelian randomisation study based on the largest cohort in China demonstrated that alcohol intake uniformly increases cardiovascular risk without a protective effect of moderate consumption(Reference Millwood, Walters and Mei39). Frequent consumption of beer or strong spirits rather than grape wine may partially explain the inconsistent results.

In contrast, the results of two exclusive components of DASH diet were as expected. Dairy products showed consistent and substantial benefits on both obesity and metabolic health, which was in line with previous studies, especially in populations with generally low consumption of dairy products(Reference Zong, Sun and Yu40,Reference Dehghan, Mente and Rangarajan41) . The Na component score showed no association with obesity but a positive association with MHO because Na mainly influences blood pressure, which is an important component of metabolic health. Moreover, as a holistic dietary pattern, DASH diet is characterised by a low fat, cholesterol and Na content, but high fibre, K, Mg, Ca and antioxidants content, all of which are demonstrated to be beneficial to components of the metabolic syndrome through different mechanisms including improving β-cells function, insulin sensitivity and gut microbiota and reducing adiposity accumulation, systemic inflammation and oxidative stress(Reference Akhlaghi42). Therefore, regarding improvement on obesity per se or metabolic health for population with obesity, DASH diet may represent a more relevant and valuable dietary recommendation in Chinese population than aMED diet. In addition, it is worth noting that the dominant contributors to the benefits of DASH diet were not identical for different outcomes, which implies that under the overall guidance of DASH diet, recommendations could be flexibly focus more on certain components according to different levels of obesity prevention and control goals in practice.

Metabolic health is achievable without weight loss

The concept of MHO is attractive, given the limited success of current weight loss-centred and ‘one size fits all’ approaches to prevent and control obesity. An increasing number of studies point out that management of obesity does not necessarily consider weight loss to be the most important goal, and maintaining favourable cardiometabolic health and ultimately improving adverse health outcomes may be an easier and wiser strategy(Reference Stefan, Haring and Schulze4,Reference Magkos5) . Our results presented BMI-independent metabolic benefits of DASH diet for population with obesity, which implied that cardiometabolic health may be improved by adopting healthy dietary patterns even without weight loss. There are many previous studies confirming the same viewpoint, including the OmniHeart trial(Reference Appel, Sacks and Carey43), PREDIMED trial(Reference Estruch, Martínez-González and Corella44,Reference Estruch, Ros and Salas-Salvado45) , trials conducted by the DASH Collaborative Research Group(Reference Appel, Moore and Obarzanek46) and so on(Reference Bendall, Mayr and Opie47Reference Shirani, Salehi-Abargouei and Azadbakht49), which consistently demonstrate that following several healthy diets, albeit without changes in weight or BMI, can substantially improve BP, lipid profiles, WC, insulin resistance and even cardiovascular events. Of course, we cannot deny that weight loss is an ideal and essential goal in obesity treatment because obesity has many adverse health effects other than cardiometabolic risk, such as respiratory and sleep disorders, musculoskeletal system diseases and certain cancers(Reference Iacobini, Pugliese and Fantauzzi8); furthermore, MHO may not be an absolutely benign and stable status (many MHO individuals may transit to MUO years later)(Reference Magkos5,Reference Smith, Mittendorfer and Klein7) . Fortunately, healthy diets and many other lifestyle recommendations (such as maintaining physical activity) also have beneficial effects on weight loss(Reference Jensen, Ryan and Apovian50,Reference Look, Pi-Sunyer and Blackburn51) ; targeting the more achievable goal of metabolic health can encourage and facilitate adherence to such healthy behaviours and ultimately improve the long-term and overall health status.

The beneficial effect of diet is stronger in populations with higher socio-economic status and in more developed regions

Another interesting finding in the present study was that the benefits of DASH diet on obesity or metabolic health were significantly stronger in urban residents and individuals with higher education levels. In fact, after carefully observing the results, we found a consistent, although non-significant, tendency of strengthened associations with increasing individual SES levels and regional development levels. The disparity in the availability and affordability of high-quality food is one potential explanation for the observed heterogeneous associations. That is, given the same overall dietary pattern (food type and quantity) guidance, there are substantial variability and flexibility in the choice and quality of specific food items, which may in turn modify the magnitude and even the direction of the health effects of diets. Therefore, to further strengthen and broaden the beneficial effect of healthy dietary patterns, improving the diet quality as well as the availability and affordability of such high-quality food for the whole population through multilevel and multicomponent cooperation and efforts is crucial and warranted.

The generalisability of the study

In terms of internal generalisability, the sampling strategy of the CMEC was carefully designed, which was a multistage stratified cluster sampling method with full consideration of representation for geography feature, ethnic structure, urbanicity status and demographic characteristics (such as sex ratio and age ratio)(Reference Zhao, Hong and Yin21). However, as a labour exporting area, the demographic characteristics of residents in southwest China might be very different from what are expected (e.g. much more man go out to work; thus, the female:male ratio is high). In addition, the CMEC is a volunteer-based study, although the sampling and recruiting process are carefully designed, the response rates are various among population with different characteristics (e.g. women are more willing to participant in the survey, especially in rural and less-developed areas). Therefore, some characteristics of our participants seem to be unrepresentative for the general population (such as the sex ratio); however, we have carefully adjusted such confounding variables in the primary statistical analysis to obtain a valid average treatment effect estimation and conducted several stratification analyses to explore the potential heterogeneity which showed overall homogeneous results.

In terms of external generalisability, the dietary patterns and other characteristics of the CMEC population might be different from that of other countries and regions, and we calculated the DASH and MED diet scores according to population-specific quintiles rather than specific amounts recommended in original DASH and MED diets. Therefore, the estimation results in the present study might not be directly comparable with those from other studies and extrapolated to other populations. However, the findings and conclusions of this study may be generalisable, which are shared by many other studies as we discussed above.

Strengths and limitations

To our knowledge, the present study is the first large-scale study to investigate the associations of well-known healthy dietary patterns with obesity and metabolic obesity phenotypes in Chinese multi-ethnic populations. The strengths of our study include a heterogeneous and unique population, rigorous and standardised data collection methods ensuring high-quality data, large sample size and number of cases and careful adjustment for potential confounders according to an evidence-based DAG and causal inference methods. Nevertheless, there are several limitations worthy of consideration. First, the cross-sectional nature of this study limited the determination of a causal relationship, although we adopted the DAG and inverse probability of exposure weighting approach and tried to minimise reverse causality by excluding participants with any self-reported cardiometabolic disease. Second, the dietary data were collected using an FFQ with only thirteen food groups, which was less accurate than an FFQ with specific food items or other methods, such as 24-h dietary recall. However, such a method is more feasible in large-scale epidemiology studies (especially in this less-developed ethnic minority region where many participants are illiterate, and speaking different languages) and may not lead to serious misclassification in assessing adherence to prior dietary patterns, whose components are also food types and almost covered by the FFQ used. Of course, such a method may result in underreporting and underestimating energy intake. Third, there exists inherent memory error in self-reporting of food and beverage consumption, physical activity and diseases history, etc., and insufficient measurements for some outcome indicators which may introduce bias to some extent. Fourth, although we adjusted for confounders based on a DAG, we cannot exclude the possibility of residual confounding due to measurement errors and unmeasured variables. Finally, given the transient nature of MHO, the determination of metabolic status merely on the basis of one time point measurement is not an optimal strategy and may result in misclassification.

Conclusions

In conclusion, higher adherence to DASH diet rather than aMED diet was favourably associated with obesity and MHO in this study, suggesting that DASH diet may be a superior dietary recommendation regarding the prevention of obesity and related metabolic disorders in Chinese population. The observed BMI-independent metabolic benefits of such a healthy dietary pattern imply that metabolic health may be improved even without weight loss, which may provide encouragement and positive reinforcement for individuals with obesity towards long-term adherence and thorough health benefits. Prospective and interventional studies are needed to confirm the findings and to further explore the effect of healthy dietary patterns on the transition and maintenance of MHO phenotype.

Acknowledgements

We thank all the participants and staff of the CMEC study for their brilliant contributions. We are grateful to Prof. Xiaosong Li, the former principal investigator of the CMEC study, for his leadership and fundamental contribution to the establishment of the CMEC. Prof. Li passed away in 2019.

This work was primarily supported by the National Key Research and Development Program of China (grant no. 2017YFC0907305, 2017YFC0907300). This study was also supported by the National Natural Science Foundation of China (grant no. 81903415, 81860597, 81773548, 81973151), the Key R&D Project of Sichuan Province Science and Technology Support Program (grant no. 2020YFS0215), CNS-ZD Tizhi and Health Fund (grant no. CNS-ZD2020-149) and the Science and Technology Support Plan of Science and Technology Department of Guizhou Province ((2018) no.5403). The sponsors had no role in the design, execution, interpretation or writing of the study.

D. T., X. X. and X. Z. designed the study. X. Z. was the principal investigator and J. Y. and F. H. the co-principal investigators of the CMEC study. D. T. and X. X. wrote the analysis plan and draft the manuscript. All other authors contributed to the acquisition and analysis of data, and interpretation of results. All authors made important contributions to editing and critically revising the manuscript for important intellectual content. All authors have approved the final version to be published. X. Z. is the guarantor of this work, has full access to all of the data and takes responsibility for the integrity of the data.

The authors declare no conflict of interest.

Supplementary material

For supplementary material referred to in this article, please visit https://doi.org/10.1017/S0007114521005158.

Footnotes

These authors contributed equally to this work.

References

WHO (2020) Obesity and Overweight: Key Facts. http://www.who.int/mediacentre/factsheets/fs311/en/ (accessed November 2020).Google Scholar
Murray, CJL, Aravkin, AY, Zheng, P, et al. (2020) Global burden of 87 risk factors in 204 countries and territories, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet 396, 12231249.CrossRefGoogle Scholar
Mongraw-Chaffin, M, Foster, MC, Anderson, CAM, et al. (2018) Metabolically healthy obesity, transition to metabolic syndrome, and cardiovascular risk. J Am Coll Cardiol 71, 18571865.CrossRefGoogle ScholarPubMed
Stefan, N, Haring, HU & Schulze, MB (2018) Metabolically healthy obesity: the low-hanging fruit in obesity treatment? Lancet Diabetes Endocrinol 6, 249258.CrossRefGoogle ScholarPubMed
Magkos, F (2019) Metabolically healthy obesity: what’s in a name? Am J Clin Nutr 110, 533539.CrossRefGoogle Scholar
Gao, M, Lv, J, Yu, C, et al. (2020) Metabolically healthy obesity, transition to unhealthy metabolic status, and vascular disease in Chinese adults: a cohort study. PLoS Med 17, e1003351.CrossRefGoogle ScholarPubMed
Smith, GI, Mittendorfer, B & Klein, S (2019) Metabolically healthy obesity: facts and fantasies. J Clin Invest 129, 39783989.CrossRefGoogle ScholarPubMed
Iacobini, C, Pugliese, G, Fantauzzi, CB, et al. (2019) Metabolically healthy v. metabolically unhealthy obesity. Metab Clin Exp 92, 5160.CrossRefGoogle Scholar
Mozaffarian, D (2016) Dietary and policy priorities for cardiovascular disease, diabetes, and obesity: a comprehensive review. Circulation 133, 187225.CrossRefGoogle ScholarPubMed
Hu, FB (2002) Dietary pattern analysis: a new direction in nutritional epidemiology. Curr Opin lipidol 13, 39.CrossRefGoogle ScholarPubMed
Hosseini, Z, Whiting, SJ & Vatanparast, H (2016) Current evidence on the association of the metabolic syndrome and dietary patterns in a global perspective. Nutr Res Rev 29, 152162.CrossRefGoogle Scholar
Kastorini, CM, Milionis, HJ, Esposito, K, et al. (2011) The effect of Mediterranean diet on metabolic syndrome and its components: a meta-analysis of 50 studies and 534 906 individuals. J Am Coll Cardiol 57, 12991313.CrossRefGoogle ScholarPubMed
Konieczna, J, Yanez, A, Monino, M, et al. (2020) Longitudinal changes in Mediterranean diet and transition between different obesity phenotypes. Clin Nutr 39, 966975.CrossRefGoogle ScholarPubMed
Park, YM, Steck, SE, Fung, TT, et al. (2017) Mediterranean diet, Dietary Approaches to Stop Hypertension (DASH) style diet, and metabolic health in U.S. adults. Clin Nutr 36, 13011309.CrossRefGoogle ScholarPubMed
Arenaza, L, Huybrechts, I, Ortega, FB, et al. (2019) Adherence to the Mediterranean diet in metabolically healthy and unhealthy overweight and obese European adolescents: the HELENA study. Eur J Nutr 58, 26152623.CrossRefGoogle Scholar
Phillips, CM, Dillon, C, Harrington, JM, et al. (2013) Defining metabolically healthy obesity: role of dietary and lifestyle factors. PLOS ONE 8, e76188.CrossRefGoogle ScholarPubMed
Rahimi, H, Yuzbashian, E, Zareie, R, et al. (2020) Dietary approaches to stop hypertension (DASH) score and obesity phenotypes in children and adolescents. Nutr J 19, 19.CrossRefGoogle ScholarPubMed
Farhadnejad, H, Darand, M, Teymoori, F, et al. (2019) The association of Dietary Approach to Stop Hypertension (DASH) diet with metabolic healthy and metabolic unhealthy obesity phenotypes. Sci Rep 9, 17.CrossRefGoogle ScholarPubMed
Ma, S, Xi, B, Yang, L, et al. (2021) Trends in the prevalence of overweight, obesity, and abdominal obesity among Chinese adults between 1993 and 2015. Int J Obes 45, 427437.CrossRefGoogle ScholarPubMed
Huang, J, Huang, JLW, Withers, M, et al. (2018) Prevalence of metabolic syndrome in Chinese women and men: a systematic review and meta-analysis of data from 734 511 individuals. Lancet 392, S14.CrossRefGoogle Scholar
Zhao, X, Hong, F, Yin, J, et al. (2020) Cohort profile: the China Multi-Ethnic cohort (CMEC) study. Int J Epidemiol 50, 721721l.CrossRefGoogle Scholar
Shan, Z, Li, Y, Baden, MY, et al. (2020) Association between healthy eating patterns and risk of cardiovascular disease. JAMA Intern Med 180, 10901100.CrossRefGoogle ScholarPubMed
Xiao, X, Qin, Z, Lv, X, et al. (2021) Dietary patterns and cardiometabolic risks in diverse less-developed ethnic minority regions: results from the China Multi-Ethnic Cohort (CMEC) Study. Lancet Regional Health – Western Pacific 15, 100252.CrossRefGoogle ScholarPubMed
Fung, TT, Chiuve, SE, McCullough, ML, et al. (2008) Adherence to a DASH-style diet and risk of coronary heart disease and stroke in women. Arch Intern Med 168, 713720.CrossRefGoogle ScholarPubMed
Alberti, KG, Eckel, RH, Grundy, SM, et al. (2009) Harmonizing the metabolic syndrome: a joint interim statement of the International Diabetes Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; and International Association for the Study of Obesity. Circulation 120, 16401645.CrossRefGoogle Scholar
Ferguson, KD, McCann, M, Katikireddi, SV, et al. (2020) Evidence synthesis for constructing directed acyclic graphs (ESC-DAGs): a novel and systematic method for building directed acyclic graphs. Int J Epidemiol 49, 322329.CrossRefGoogle ScholarPubMed
Ukoumunne, OC, Williamson, E, Forbes, AB, et al. (2010) Confounder-adjusted estimates of the risk difference using propensity score-based weighting. Stat Med 29, 31263136.CrossRefGoogle ScholarPubMed
Greifer, N (2020) WeightIt: Weighting for Covariate Balance in Observational Studies. R Package Version 0.10.2.Google Scholar
Hainmueller, J (2012) Entropy balancing for causal effects: a multivariate reweighting method to produce balanced samples in observational studies. Polit Anal 20, 2546.CrossRefGoogle Scholar
Trichopoulou, A, Bamia, C & Trichopoulos, D (2009) Anatomy of health effects of Mediterranean diet: Greek EPIC prospective cohort study. BMJ 338, b2337.CrossRefGoogle ScholarPubMed
Rey-Lopez, JP, de Rezende, LF, Pastor-Valero, M, et al. (2014) The prevalence of metabolically healthy obesity: a systematic review and critical evaluation of the definitions used. Obes Rev 15, 781790.CrossRefGoogle ScholarPubMed
Ortega, FB, Cadenas-Sanchez, C, Migueles, JH, et al. (2018) Role of physical activity and fitness in the characterization and prognosis of the metabolically healthy obesity phenotype: a systematic review and meta-analysis. Prog Cardiovasc Dis 61, 190205.CrossRefGoogle ScholarPubMed
Bluher, M (2020) Metabolically healthy obesity. Endocr Rev 41, 405420.CrossRefGoogle ScholarPubMed
Schleinitz, D, Bottcher, Y, Bluher, M, et al. (2014) The genetics of fat distribution. Diabetologia 57, 12761286.CrossRefGoogle ScholarPubMed
Loos, RJF & Kilpelainen, TO (2018) Genes that make you fat, but keep you healthy. J Intern Med 284, 450463.CrossRefGoogle Scholar
Soltani, S, Shirani, F, Chitsazi, MJ, et al. (2016) The effect of dietary approaches to stop hypertension (DASH) diet on weight and body composition in adults: a systematic review and meta-analysis of randomized controlled clinical trials. Obes Rev 17, 442454.CrossRefGoogle ScholarPubMed
Willett, WC, Sacks, F, Trichopoulou, A, et al. (1995) Mediterranean diet pyramid: a cultural model for healthy eating. Am J Clin Nutr 61, 1402S1406S.CrossRefGoogle ScholarPubMed
Trichopoulou, A, Costacou, T, Bamia, C, et al. (2003) Adherence to a Mediterranean diet and survival in a Greek population. N Engl J Med 348, 25992608.CrossRefGoogle Scholar
Millwood, IY, Walters, RG, Mei, XW, et al. (2019) Conventional and genetic evidence on alcohol and vascular disease aetiology: a prospective study of 500 000 men and women in China. Lancet 393, 18311842.CrossRefGoogle ScholarPubMed
Zong, G, Sun, Q, Yu, D, et al. (2014) Dairy consumption, type 2 diabetes, and changes in cardiometabolic traits: a prospective cohort study of middle-aged and older Chinese in Beijing and Shanghai. Diabetes Care 37, 5663.CrossRefGoogle ScholarPubMed
Dehghan, M, Mente, A, Rangarajan, S, et al. (2018) Association of dairy intake with cardiovascular disease and mortality in 21 countries from five continents (PURE): a prospective cohort study. Lancet 392, 22882297.CrossRefGoogle ScholarPubMed
Akhlaghi, M (2020) Dietary Approaches to Stop Hypertension (DASH): potential mechanisms of action against risk factors of the metabolic syndrome. Nutr Res Rev 33, 118.CrossRefGoogle ScholarPubMed
Appel, LJ, Sacks, FM, Carey, VJ, et al. (2005) Effects of protein, monounsaturated fat, and carbohydrate intake on blood pressure and serum lipids: results of the OmniHeart randomized trial. JAMA 294, 24552464.CrossRefGoogle ScholarPubMed
Estruch, R, Martínez-González, MA, Corella, D, et al. (2019) Effect of a high-fat Mediterranean diet on bodyweight and waist circumference: a prespecified secondary outcomes analysis of the PREDIMED randomised controlled trial. Lancet Diabetes Endocrinol 7, e6e17.CrossRefGoogle ScholarPubMed
Estruch, R, Ros, E, Salas-Salvado, J, et al. (2018) Primary prevention of cardiovascular disease with a Mediterranean diet supplemented with extra-virgin olive oil or nuts. N Engl J Med 378, e34.CrossRefGoogle ScholarPubMed
Appel, LJ, Moore, TJ, Obarzanek, E, et al. (1997) A clinical trial of the effects of dietary patterns on blood pressure. DASH Collaborative Research Group. N Engl J Med 336, 11171124.CrossRefGoogle ScholarPubMed
Bendall, CL, Mayr, HL, Opie, RS, et al. (2018) Central obesity and the Mediterranean diet: a systematic review of intervention trials. Crit Rev Food Sci Nutr 58, 30703084.CrossRefGoogle ScholarPubMed
Lopes, HF, Martin, KL, Nashar, K, et al. (2003) DASH diet lowers blood pressure and lipid-induced oxidative stress in obesity. Hypertension 41, 422430.CrossRefGoogle ScholarPubMed
Shirani, F, Salehi-Abargouei, A & Azadbakht, L (2013) Effects of Dietary Approaches to Stop Hypertension (DASH) diet on some risk for developing type 2 diabetes: a systematic review and meta-analysis on controlled clinical trials. Nutrition 29, 939947.CrossRefGoogle ScholarPubMed
Jensen, MD, Ryan, DH, Apovian, CM, et al. (2014) 2013 AHA/ACC/TOS guideline for the management of overweight and obesity in adults. Circulation 129, S102S138.CrossRefGoogle ScholarPubMed
Look, ARG, Pi-Sunyer, X, Blackburn, G, et al. (2007) Reduction in weight and cardiovascular disease risk factors in individuals with type 2 diabetes: one-year results of the look AHEAD trial. Diabetes Care 30, 13741383.Google Scholar
Figure 0

Table 1. Scoring criteria for the DASH diet and aMED diet

Figure 1

Table 2. Age- and sex-adjusted characteristics of the study participants(Mean values and standard deviations for continuous variables; absolute and relative frequencies for categorical variables)

Figure 2

Fig. 1. Association of aMED and DASH diet scores with obesity and MHO phenotype. All models adjusted for age, sex, ethnic group, urbanicity, education, household income, profession, marital status, smoking, physical activity, total energy intake, regular intake of soft drinks, dietary supplements, spicy food and peppery food, insomnia symptoms, anxiety symptoms, depression symptoms, menopause status and family history of cardiometabolic diseases using logistic regression with inverse probability of exposure weighting (IPEW). Q2–Q5 represent the second to fifth quintiles of diet scores. The filled red dots represent estimated OR and the vertical red lines represent 95 % CI. aMED, alternative Mediterranean; DASH, Dietary Approaches to Stop Hypertension; MHO, metabolically healthy obesity.

Figure 3

Table 3. OR of MHO associated with quintiles of aMED and DASH diet scores after adjusting for BMI*(Odds ratios and 95 % confidence intervals)

Figure 4

Table 4. OR of obesity and MHO associated with overall DASH diet score and subtracted DASH diet scores for each component(Odds ratios and 95 % confidence intervals)

Figure 5

Fig. 2. Stratified analysis for associations of DASH diet scores with obesity and MHO according to predefined demographic and socio-economic factors by comparing the highest with the lowest quintiles. All models adjusted for age, sex, ethnic group, urbanicity, education, household income, profession, marital status, smoking, physical activity, total energy intake, regular intake of soft drinks, dietary supplements, spicy food and peppery food, insomnia symptoms, anxiety symptoms, depression symptoms, menopause status and family history of cardiometabolic diseases using logistic regression with inverse probability of exposure weighting (IPEW), with exclusion of the corresponding stratification factor as appropriate. I2 statistic and Q test were used to assess heterogeneity among different strata of each stratification factor. The filled red dots represent estimated OR and the horizontal red lines represent 95 % CI. DASH, Dietary Approaches to Stop Hypertension; MHO, metabolically healthy obesity.

Supplementary material: File

Tang et al. supplementary material

Tang et al. supplementary material

Download Tang et al. supplementary material(File)
File 1.2 MB