Obesity, lifestyle behaviors and dyslipidemia among Chinese adults aged 45 and older

This study investigated the importance of obesity and lifestyle behaviors in affecting dyslipidemia among adults aged 45 and older in China. The strength of our study is in using the decision tree model to clearly rank the importance of those key factors affecting dyslipidemia. Data were taken from the China Health and Retirement Longitudinal Study. A total of 9,038 adults were included in the study. Logistic regression was used to examine the associations between obesity, lifestyle behaviors and dyslipidemia. Decision tree was built to select the best scheme on prevention of dyslipidemia. Overweight and obesity are rapidly growing threats in China. Regular physical activity could positively affect dyslipidemia and produce desirable health status. This will be beneficial evidence for educating those who do not or cannot perform regular and substantial physical activities.


Background
Over the past two decades, Chinese people have experienced the living environment and unsatisfactory lifestyle changes, such as reduced physical activity, more stressed and long-term sedentary work (1-4). Accordingly, the prevalence of overweight and obesity are rapidly increasing in China (5,6). Scientific research has confirmed that overweight and obesity is an important risk factor to induce dyslipidemia, and dyslipidemia plays an important role in the formation and development of cardiovascular diseases (CVDs) (7)(8)(9).
Dyslipidemia as a major risk factor for CVDs has aroused Chinese people's big awareness (7,8), but treatment of dyslipidemia has not improved a lot accordingly in China (10, 11). A cross sectional study involving 25,697 Chinese individuals found that 38.5% of those receiving lipid lowering treatment did not achieve the treatment goal for low density lipoprotein, which is similar to the results of other studies from China (12)(13)(14). Previous research found that lifestyle changes including increased physical activity and decreased body mass index (BMI) had profound effects on preventing dyslipidemia (15)(16)(17)(18). These findings, together with ones from observational and clinical studies, suggested that combination therapy in controlling dyslipidemia was very important; no obesity and healthy lifestyle behavior would be focused to help the treatment of dyslipidemia (16,(19)(20)(21)(22).
Physical activity has been recommended in the prevention of obesity and dyslipidemia by health organizations around the world, because it costs less and generates fewer side effects than isolated medicine (20,23). Most studies on prevention of dyslipidemia adopted the traditional statistical methods and the interaction between demographic characteristics, lifestyle behaviors and dyslipidemia has rarely been incorporated into the research (17,24,25). Moreover, few information was available regarding levels of physical activity done by the elderly in China.
The intensity as well as frequency and duration of physical activities was frequently overlooked or unknown by elderly patients with obesity and dyslipidemia in China (26,27). It is needed to compare the existing schemes in preventing obesity and dyslipidemia, select the best one, and make it optimized. Accordingly, this study presents national-level data from 2015 China Health and Retirement Longitudinal Study (CHARLS) to estimate the relationship between obesity, lifestyle behaviors and dyslipidemia in middle-aged and older adults in China. Specifically, we'll be looking into the evidences in three aspects of physical activities (intensity, frequency and duration) on prevention of obesity and dyslipidemia and select the best scheme for dyslipidemia patients.

Data source
We used data from the 2015 national follow-up survey of CHARLS, which was designed to provide a high-quality nationally representative panel data with a wide range of information serving the needs of study on aging-related problems in China. The data covered information of individuals of age ≥45 years, and included assessments of lifestyle behaviors, the social and economic circumstances of community residents. The CHARLS team created separate crosssectional weights for individuals corrected for non-response and sampling frame errors in each step of the CHARLS. Details of the CHARLS have been described in previous studies (28,29).
A total of 9,038 men and women aged 45 or older were involved in the study. Face-to-face interviews at participants' home monitored how their health changed over time and how they adjusted to these changes. The presence of dyslipidemia and physical exercise practice were calculated according to self-reported information through household interviews.

Definitions
Body mass index (BMI) was calculated as weight in kg/height 2 in m 2 , and was categorized as <18.5 kg/m 2 (underweight), 18.5-24 kg/m 2 (normal weight), 24-28 kg/m 2 (overweight) or ≥28 kg/m 2 (obese) (30). The prevalence of dyslipidemia depended on whether the respondent reported having certain health conditions (includes: elevation of low-density lipoprotein, triglycerides, and total cholesterol, or a low high-density lipoprotein level.) in a previous interview and/or confirmed them in the last interview via a structured questionnaire. Vigorous activities were considered as making breath much harder than normal and might include heavy lifting, digging, plowing, aerobics, fast bicycling, and cycling with a heavy load. Moderate physical activities were considered as making breath somewhat harder than normal and might include carrying light loads, bicycling at a regular pace, or mopping the floor. Light physical activities were considered as making breath normal, such as walking from place to place that you might do solely for recreation, exercise, or leisure.

Statistical analysis
Statistical analysis was performed using R software (Version 3.4.1, https://www.r-project.org/) and considered the sample weights and complex survey design of the CHARLS. The significant differences in categorical variables were analyzed using chi-square test. All prevalence rates were weighted with the cross-sectional biomarker weight which was based on the individual one and a logit regression of whether the individual responses in biomarker. Based on the statistically significant associations provided by the chi-square test, logistic regression was used to examine the lifestyle factors related to the development of dyslipidemia. The results of logistic regression models were presented as odds ratios (OR), adjusted odds ratios (AOR) and 95 % confidence interval (CI). All statistical tests were two-tailed and P ≤ 0.05 was considered statistically significant.
Decision tree model was established to analyze the importance levels of physical activity on prevention of dyslipidemia. Rpart (R package: recursive partitioning for classification, regression and survival trees, version 4.1-13.) was used to create the decision tree model. It was built by the following process: first the single variable was found which best split the data into two groups, and then this process was applied to each sub-group, and so on recursively until no improvement could be made or the subgroups reached a minimum size (10 for this data). The CRT growing method was used to maximize within-node homogeneity. CRT split the data into segments that were as homogeneous as possible with respect to the dependent variable (31,32).
Gini method was used to measure impurity and the minimum decrease in impurity required to split nodes. The Gini impurity measure at a node t was defined as: The Gini splitting criterion was the decrease of impurity defined as: where PL and PR were probabilities of sending a case to the left child node tL and to the right child node tR respectively. They were estimated as PL =P(tL)/p(t) and PR =P(tR)/p(t). The minimum decrease in impurity required to split a node was set to 0.0001. Cross validation was used to assess how well the tree structure generalized to a larger population, and the number of sample folds was set to 10. The cross validated risk estimate for the final tree was calculated as the average of the risks for all of the trees. An overall measure of variable importance was by comparing the goodness of split measures for each variable (33).    Fig.1.   Fig.1 Variable importance on prevention of dyslipidemia.   Fig.2.  Fig.2 The structure of the decision tree model.

Discussion
In this nationally representative longitudinal survey among middle-aged and older adults in China, the prevalence of overweight was 33.98%, and the prevalence of obesity was 12.73%.
Obese participants had a higher rate of dyslipidemia (20.45%) than participants with overweight (11.24%), normal weight (7.73%) and underweight (3.05%), which proved previous studies that high BMI was associated with increased dyslipidemia risk (2,34,35). This study made a detailed classification on healthy lifestyle of middle-aged and older adults in China (Fig.2).

Physical activity as one of healthy life behaviors played a key preventive role in dyslipidemia.
This result may explain why some obese individuals do not develop dyslipidemia, while some individuals with normal weight become dyslipidemic.
Many mixed protective factors can prevent the occurrence of dyslipidemia among middle-aged and older adults. Prevention of dyslipidemia in adults with obesity can also benefit from these factors (Fig.2). The results showed that for middle-aged adults with BMI ≥28 kg/m 2 , vigorous physical activity was far more effective than moderate and light activity on prevention of dyslipidemia; for older adults (age ≥55) with BMI ≥28 kg/m 2 , moderate physical activity was the better choice, while it is worth noting that 10 minutes of vigorous physical activities one week can gradually help reduce the rate of dyslipidemia in this group (Fig.2: Node 23); for adults living in main city zone, physical activity frequency seemed to be more important than the intensity to induce beneficial effects on prevention of dyslipidemia, etc. Combined with previous studies that a significant amount of weekly energy expenditure from physical activity could produce changes in blood lipids (20,36), the targeted population would be suggested focusing on days spent on regular physical activities. But due to the individual physical difference, it is better to select personally suitable physical activity and healthy life style.
The other important variables in the tree model, according to the normalized importance, were residence, BP levels, drinking, smoking, gender, and sleep duration. This finding could support the idea that lifestyle changes could affect blood lipid levels and prevention of dyslipidemia (37,38). Although some influential factors might not be comprehensively considered in the model, physical activity with multiple adjustments in life was the best option for older people.
Besides the strong awareness of proper physical activity, more comprehensive strategies toward the prevention, screening, treatment, and control would be needed in China to lower the prevalence of dyslipidemia, which need further research.
There were several limitations in our study. Firstly, dyslipidemia was not diagnosed by specific blood test one by one, some patients with dyslipidemia had not been identified, and therefore the rates of dyslipidemia in this study was lower than the actual value. Secondly, the crosssectional nature of our study design prevented us from determining the direction of the association between lifestyle behaviors and prevalence of dyslipidemia.

Conclusions
However, these limitations did not considerably affect our conclusions that healthy lifestyle behaviors especially physical activities were effective on prevention of obesity and dyslipidemia. The rapidly growing threats of obesity and dyslipidemia to middle-aged and older adults in China will decrease. For overweight or obese elderly, regular and moderate levels of physical activity were the most effective precautionary measures; for middle-aged adults, vigorous physical activity seemed to be more important than moderate and light activity to induce beneficial effects on prevention of dyslipidemia; all in all, regular physical activity could positively affect dyslipidemia and produce desirable health status. This will be beneficial evidence for educating those who do not or cannot perform regular and substantial physical activities.

BMI: Body mass index
CVDs: cardiovascular diseases

Ethics approval and consent to participate
Ethical approval for the study was granted by the Ethical Review Committee of Peking University. All individuals in this study provided written consent at the time of participation, and written informed consent was obtained from all study participants.

Consent for publication
Not Applicable.

Availability of data and material
All data in CHARLS are maintained at the National School of Development of Peking University and will be accessible to researchers around the world at the study website. http://charls.pku.edu.cn/zh-CN/page/data.

Competing interests
The authors declare that they have no competing interests.

Funding
This study was supported by grants from the National Natural Science Foundation of China