Prevalence and Associated Factors of Metabolic Syndrome in Chinese Middle-Aged and Elderly Population: A Nationally Cross-Sectional Study

Background: Currently, China has an increasingly aging population. However, the prevalence of metabolic syndrome (MetS) in this high-risk population for metabolic diseases remains unknown. To investigate the age and gender specic prevalence and associated factors of MetS in Chinese middle-aged and elderly population, we did the study on it. Methods: Data were retrieved from China Health and Retirement Longitudinal Study (CHARLS) and subjected to descriptive statistics. Further, univariate logistic regression was used to evaluate the relevant factors, and then multivariate logistic regression was selected to construct the nal model. Results: In present study, a total of 10,834 participants were included. The overall prevalence of MetS was 32.97% as dened by International Diabetes Federation and 29.75% under National Cholesterol Education Program - The Adult Treatment Panel III criteria. With aging, prevalence of MetS descended in males, while ascended in females. In the >70 years old group, the prevalence of MetS was three times higher in females than that in males (50.43% versus 16.03%). Across all age groups and sexes, the prevalence of MetS in urban areas was signicantly higher than in rural areas. In addition, regardless of gender, the prevalence was highest for those living in the north region (28.41% for males & 51.74% for females) and lowest for those living in the southwest region (13.91% for males & 31.58% for females). Finally, two specic correlates, afternoon nap and blood urea nitrogen (BUN), were identied in present study (P < 0.05). Conclusion: The prevalence of MetS varied in different age groups, sexes, living areas and regions and was associated with afternoon nap and BUN.

Among them, prevalence in women and the aged were signi cantly higher than men and the young. Similar conclusions are also seen in Bangladesh [13], Peru [14], Portugal [15] and et al [16]. In 2015, a meta-analysis enrolling 35 studies reported a pooled prevalence of 24.5% in the overall population and this number signi cantly increased in females and the age [17]. These literatures disclosed observational results in the overall population, however, against the background of aging in China, no speci c study targeting Chinese middle-aged and elderly population is designed to investigate the prevalence and associated factors, which limits further understanding of this high-risk group.
To date, diverse associated factors are investigated comprising age [18], living areas [9], educational levels [19], physical activities [20] etc. Although the majority of previous studies [21][22][23] indicated that risk factors and its negative effects in different ages and gender are different, a few studies showed the opposite conclusion [16], needing further exploring. What's more, risk factors identi ed previously are explored in the overall population and few studies are performed in the aged. It was revealed that a total of 111 million people aged 65 years above resided in China (8.2% of the country residents) [24]. Performing speci c study targeting the aged seems imminent, which is what we do in this survey.
In the present study, data from China Health and Retirement Longitudinal Study (CHARLS), was downloaded and subjected to analyses. This program is specially designed to evaluate the health status in Chinese middle-aged and elderly population and owns a good national representation, providing us a chance to scan the age and gender speci c prevalence and associated factors of MetS in the aged in China.

Study sample and data cleansing
To scan the prevalence of MetS in Chinese middle-aged and elderly population, dataset from CHARLS Follow-up Questionnaire (2015) was downloaded and subjected to further data cleansing. In CHARLS, probabilities proportional to size (PPS) was selected to sample across the whole China, which nally covered 28 provinces and 150 counties. Further detailed description regarding this representative program could be accessed in their o cial website (http://charls.pku.edu.cn/) or publications [25]. In this national crosssectional survey, a total of 21,095 participants aged 40 years old above were enrolled. In virtue of the missing values of covariates and unquali ed blood samples, data cleansing was performed (see Fig. 1). Finally, 10,834 participants whose blood were collected under fasting status were remained and subjected to further analyses.

De nition of MetS and data collection
Currently, there were diverse de nitions to construct MetS such as NCEP-ATP III, IDF etc. Of note, the de nition of IDF considers the ethnic characteristics of the Chinese and has lower cutoff values than other de nitions, which bene ts early diagnosis and intervention. Hence, we mainly adopted IDF's de nition to scan the prevalence of MetS. Besides, for comparison with the results of other studies employing different de nitions, we also used the NCEP-ATP III de nition to scan the morbidity of MetS. The de nition of IDF requires abdominal obesity and two or more of the following: hyperglycemia, low HDL cholesterol, high blood triglycerides and hypertension. As for NCEP-ATP III, presence of any three or more of the components stated above is su cient to diagnose MetS. Detailed de nitions of IDF and NCEP-ATP III could be viewed in previous literatures [9].
To construct MetS, ve components, namely waistline, HDL, triglycerides, blood glucose and blood pressure, are requisite. All the measurement of these ve components were carried out by well-trained researchers from Chinese Center for Disease Control and Prevention (CCDC). Tapes were placed over the clothing around the waist at the level of the navel to measure waistlines. Participants were asked to be relaxed and seated until three blood pressure measurements were completed. Final blood pressures were recorded using the average of three measurements. For assessing blood biomarkers, the night before the blood collection, participants were asked to fast until the next morning. Venous blood of participants was collected and centrifuged into plasma, which was immediately stored at − 20 •C for transporting to CCDC in Beijing within two weeks. Blood biomarkers including HDL, blood glucose, LDL etc. were determined using enzymatic colorimetric tests.

Covariates
In present study, covariates comprising individual characteristics, blood biomarkers and medical histories were collected. Individual behaviors included age, gender, marital status, depression, sleeping time, afternoon nap, geographic region, smoking, alcohol consumption and body mass index (BMI). Age was divided into four groups: 40-50 years old, 50-60 years old, 60-70 years old and > 70 years old as Xiong et al did [26].
Marital status consisted of two groups. One included the married or cohabitated and others incorporating the divorced, widowed, married but not living with spouses were classi ed as another group. Participants with scores ≥ 10 assessed by Center for Epidemiologic studies Depression scale (CESD)-10 questionnaire were seen as depressed patients [27]. Sleeping time was strati ed into 0-6 hours, 6-8 hours and > 8 hours.
Afternoon nap was gathered as yes and no. Their living regions were also recorded as rural region and urban region. Rural region only referred to villages and other regions were uni ed as urban region. Smoking and alcohol consumption were also collected and strati ed according to their status. Smoking was reported as yes, no and quitted. Alcohol consumption was divided as never, drunk but less than once a month and drunk but more than once a month.
Blood biomarkers consist of uric acid, low density lipoprotein, total cholesterol and blood urea nitrogen.
Hyperuricemia was de ned as a blood uric acid concentration higher than 420 µmol/L in men and 357 µmol/L in women. LDL was classi ed into two groups: ≥ 120 mg/dl and < 120 mg/dl. Total cholesterol was classi ed into ≥ 200 mg/dl and < 200 mg/dl. As Zhang's study [28], BMI was split into < 18.5 Kg/m 2 , < 24 Kg/m 2 , < 28 Kg/m 2 and ≥ 28Kg/m 2 . After referring to Arihan's research [29], BUN was divided as 0-20 mg/dl and > 20 mg/dl. Furthermore, medical histories including liver diseases, kidney diseases, digestive diseases, arthritis and rheumatism were recorded. Due to the different living and cultural habits, living localities of the participants were also categorized into six regions as previous study did [30] Data, composed of continuous data and categorical data in present study, were presented as mean ± SD and proportions (%), respectively. Descriptive statistics were performed to investigate the prevalence and distribution of MetS in different subgroups. Further, univariate logistic regression was adopted to evaluate the associated factors with MetS. Covariates with P values < 0.05 were enrolled into nal model assessed by multivariate logistic regression. All the analyses were made by R 3.6.3 and all the gures were drawn by R 3.6.0 and GraphPad Prism 8.0 (GraphPad Software Inc., San Diego, CA, USA). P < 0.05 (two-sided) was considered statistically signi cant.

Baseline characteristics and prevalence of MetS in grouped population
In total, 10,834 participants aged 40 years old above were enrolled in this study. A total of 18 covariates were collected and the distribution of participants in these covariates were shown in Table 1. The overall prevalence of MetS was 32.97% (95% CI = 32.09-33.86) under IDF de nition and 29.75% (95% CI = 28.90-30.62) under NCEP-ATP III de nition. The speci c prevalence among different variables was displayed (see Table 1). Age and gender speci c prevalence of MetS and its components It was noted that with aging, prevalence of MetS under IDF de nition in males descended, while ascended in females (see Fig. 2A). In the > 70 years old group, the prevalence of MetS in females was three times higher than that in males (50.43%, 95% CI = 46.98-53.88 versus 16.03%, 95% CI = 13.76-18.59). In Fig. 2B, with aging, the increasing and decreasing trend for the prevalence of abdominal obesity remained identical in females and males, respectively. As for hyperglycemia, prevalence in both overall population and females increased. However, this increasing trend in males was not observed in the > 70 years old group (see Fig. 2C).
What's more, the prevalence of low HDL in females was nearly the same in all four age groups and thereof in males increased from 40-50 years old to 60-70 years, but decreased in the > 70 years old group (see Fig. 2D). In Fig. 2E, the prevalence of high triglycerides in overall population and females uctuated but this prevalence in males remained its downward trend in all four age groups. In Fig. 2F, the increasing trend of hypertensive patients was obvious in both males, females and overall population.
Age and gender speci c prevalence of MetS in urban and rural areas In Fig. 3A In rural areas, as in urban areas, the female and overall population were showed the same increasing and uctuating trends, respectively (see Fig. 3B). However, in the male population, the prevalence of MetS was revealed a uctuating trend in urban areas contrast to a decreasing trend in rural areas. Among the male population in rural areas the prevalence of MetS was 21.70% in 40-50 age group, 20.97% in 50-60 age group, 18.55% in 60-70 age group and 12.76% in > 70 age group.
Gender speci c prevalence in different regions.

Discussion
Considering the huge amount of aging population in China, investigating the age and gender speci c prevalence and associated factors of MetS in Chinese middle-aged and elderly population is of great signi cance. A nationally representative project targeting the aged, CHARLS, helps realize it.
In the present study, the prevalence of MetS in the participants aged 40 years and older was 32.97% based on IDF de nition. Under NCEP-ATP III de nition, it was 29.75%. According to Li's report, in 2010-2012, this gure in Chinese adults aged 18 years and older was 24.2% under NCEP-ATP III de nition [23]. In He's study, in the overall population, it was 9.5% in 2002 and 18.7% in 2010-2012 [31]. These gures revealed that the aged had higher odds of MetS, which were in lines with other studies [9,32]. Based on IDF criteria, the prevalence of MetS was 45.5% in Tunisia, 37.4% in Iran and 53.9% in Gwalior, a city in India [33,34].
Compared with studies in other countries, prevalence reported in the present study seems lower than these countries, which may be attributed to different races, lifestyles etc. In the aging population, the prevalence was 35% in 40-59 years group and 46.7% in the > 60 years group in the U.S [10]. Prevalence of MetS in aging population were high but still lower than developed countries. In addition, in 2009, this number was 31.5% among people over 35 years old in Jiangsu province, China [35], which indicated that the trend of MetS in aging population did not vary wildly.
It is noteworthy that with aging, the prevalence of MetS descends in males, while ascends in females. This trend is also observed in Korean [36], Indian [34], Spanish [37] etc., which may be partly attributed to hormone secretion. According to Jeenduang's study, prevalence of MetS was 29.37% in postmenopausal women, while was 16.97% in premenopausal women [38]. Hormones hidden behind menopause may lead to this alteration [39]. However, one previous article performed in U.S. presented an opposite observational result that males had higher prevalence of MetS than females [40]. This nding may be biased by their limited samples, which still needs more evidences.
Furthermore, our study also found that the prevalence of MetS in urban areas was signi cantly higher than one in rural areas. Sedentary lifestyles in modern cities may account for this nding [41]. Of note, with aging, this upward trend of MetS prevalence in females was not observed in males, especially for those settling in rural areas who displayed a downward trend. This result was in line with previous studies performed in China [17], India [42], sub-Saharan Africa [43] and Mexico [44] but contrary to Lee's study in the middle-aged Koreans [45]. Study from Javier concluded that men were more susceptible to the urbanization-associated worsening of cardiometabolic health [44], which may partly clarify this disparity. Notwithstanding, the speci c reasons remain hazy. Besides hormones and lifestyles stated above, a number of other potential factors may contribute to this difference including household income, educational levels and annual cost of healthcare etc [46,47].
This disparity of MetS prevalence was also detected in different regions. Regardless of gender, participants had the highest prevalence in the north region and the lowest in the southwest region. According to previous reports [48], it may be linked to dietary nutrient intake. It is well-known that residents in north region mainly eat food made of our, such as noodles, which is rich in carbohydrates. It was well known that intake of carbohydrate and sodium was closely associated with the increased risk of MetS [45]. Moreover, a diet rich in carbohydrates was thought to be a principal reason of aggravation glucose intolerance [49] and dyslipidemia [50]. Therefore, improving the diet structure seems sensible for those participants.  [53]. The design of Maria's study was limited in the < 30 minutes/day group, which may constrain this conclusion. More importantly, BUN > 20 mg/dl was also determined to be a negative correlate.
In Arora's study, both metabolically healthy obesity (MHO) and metabolically unhealthy obesity (MUO) displayed normal kidney function [54]. However, their target population was children and adolescents. This risk in different population varied, which raised a reminder of renal function in this particular middle-aged and elderly MetS patients.
The present study has several defects. First, although some associated factors were identi ed here, however, this association still awaits further cohort study to demonstrate in this speci c population. Second, this study enrolled the whole aging population instead of one speci c gender into analyses, which limited further understanding of its relevant factors by gender. We plan to address this issue in future studies.

Conclusion
In Chinese middle-aged and elderly population, the overall prevalence of MetS was 32.    Age and gender speci c prevalence of metabolic syndrome in urban and rural areas Gender speci c prevalence of metabolic syndrome in different regions Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.