The longitudinal association of smoking behaviors with obesity risk among Chinese adults from the China Health and Nutrition Survey 1991–2015 CURRENT STATUS: UNDER REVIEW

Present study aims to longitudinal explore independent association of smoking status and cigarette consumption with general obesity and abdominal among Chinese adults by gender. This study include 75,348 adults (35,907 males and 39,441 females) aged 18-65 years from the longitudinal data of China Health and Nutrition Survey (CHNS, 1991-2015). Multilevel mixed-effects linear and logistic regression models were performed for association analysis.

Given the metabolic effect of smoking, it is expected that the greater the number of cigarettes smoked, the lower the smoker's body weight, while different studies have reported conflicting results. Some studies have reported that current smokers have more abdominal obesity than nonsmokers [10,17], but other studies have found no evidence for this association [18,19], or have even found the opposite [20,21]. Moreover, longitudinal studies on the impact of smoking intensity on both general obesity and abdominal obesity among Chinese adults are rare.
China, the largest tobacco producer, consumer and manufacturer in the world, is also experiencing a steady rise in obesity rates. Our objectives therefore were to evaluate in a representative longitudinal study of Chinese adults how smoking influenced the BMI and WC, how this was related to general obesity and abdominal obesity, and whether these relationships varied by smoking intensity.

Study population and data collection
We used data of the China Health and Nutrition Survey (CHNS) for the present investigation, which was an ongoing open-cohort, international collaborative project designed to examine the effects of the social and economic transformation of Chinese society on the health and nutritional status of its population [22]. Using a multistage, random-cluster process to draw a sample of over 30,000 individuals in 15 provinces and municipal cities (Heilongjiang, Liaoning, Shandong, Jiangsu, Henan, Hubei, Yunnan, Zhejiang, Hunan, Chongqing, Hunan, Guangxi, Guizhou, Beijing, and Shanghai) that vary substantially in geography, economic development, public resources, and health indicators in the survey [23]. Further information on survey procedures is reported in detail elsewhere [24].
Our analysis used nine rounds of survey data between 1991 and 2015. Of all the adults (aged 18-65 years), we excluded participants who were pregnant, lactating or missing key variables. Hence, our final sample consisted of 75,348 person-year observations (35,907 male and 39,441 female) with complete demographic data and information on smoking status, obesity indicators, and 3-day, 24 h dietary recalls in a survey year.

Dietary data collection
Three consecutive 24-h dietary recalls were used to assess individual dietary intake in each wave of the CHNS [24]. The participants were asked to report the kinds and amount of all the food and beverage items (measured in g) consumed at home and away from home [25]. Based on the Chinese Food Composition

Cigarette smoking
The interviewees were asked if they were active smokers at the time of the survey with the question "Do you smoke?," and if they answered "Yes," they were further asked, "How many cigarettes do you smoke per day?". Two variables were constructed to measure smoking behavior. The first was smoking status, which was classified into "current smoker" or "nonsmoker." The second was smoking intensity, based on the "number of cigarettes consumed daily," where current smokers were further divided into heavy (> 25 cigarette/d), moderate (15-24 cigarette/d), and light (1-14 cigarette/d) smokers [15,25].

Obesity indicators
Well-trained health workers measured the height (model 206, SECA) and weight (model 880, SECA) of participants following standardized procedures. We calculated body mass index (BMI) by dividing the weight (in kg) by the square of the height (in m 2 )and grouped it into thin (< 18.5 kg/m 2 ), normal (18.5-23.9 kg/m 2 ), overweight (24-27.9 kg/m 2 ), and obese (> 28 kg/m 2 ), based on the recommended cut-off points of BMI for overweight and obesity in Chinese adults by the Working Group on Obesity in China [26]. We measured waist circumference (WC) from the midpoint between the lower border of the rib cage and the iliac crest to the nearest 0.1 cm, to reflect the distribution of body fat. We defined participants as having abdominal obesity if WC ≥ 85 cm in females and ≥ 90 cm in males in accordance with the guidelines of the National Health Commission of the People's republic of China [27].

Other variables
We grouped participants into two marital statuses (single and married), three education levels (primary/illiterate, middle, and high school/above) and two drinking status (yea and no) representing whether the respondent had a habit of drinking spirits during the past reported as means and standard errors for continuous variables or as proportions of the total for categorical variables. One-way analysis of variance or analysis of covariance was used for continuous variables, whereas Chi-square tests were used for categorical variables. A multilevel mixed-effects linear regression model stratified by gender was constructed to estimate smoking behaviors in relation to BMI and WC, and multilevel mixed-effects logistic regression was performed to assess the risk of general obesity and abdominal obesity by smoking intensity. All of the adjusting covariates including age, education level, individual income, physical activity, drinking status, dietary energy intake, percentage of dietary energy intake from fat, survey year, and we calculated regression coefficients and odds ratios. Table 1 Characteristics of male and female stratified by smoking intensity, CHNS 1991-2015 [1][2]     The results of the multilevel mixed-effects linear regression modeling used to examine the effects of smoking behaviors on BMI and WC are presented in Table 2. In general,

Smoking and lifestyle choices
Using a representative cohort with a more than 20-year follow-up, in summary, it appears that male current smokers reported more physical activity and higher dietary energy intake than nonsmokers, contradictory to their significantly lower body weight outcomes in this research. Although smoking is more commonly associated with reduced activity and other studies found negative associations [29,30], there is evidence that smoking is associated with increased physical activity due to its prevalence in specific occupational groups and activity types [31]. Participants in this survey, aged between 18 and 65, were more likely to be involved in professional work activities. It may be that people with manual occupations are more likely to smoke and drink heavily, in addition to being more physically active due to the nature of their work, than people with non-manual occupations [32]. Moreover, previous studies performed mostly among working-age people have shown smokers to be leaner than nonsmokers [33,34].
Biologically, there is a general perception that smoking decreases body weight outcomes due to decreases in appetite and calorie intake, enhanced metabolism, and reduced fat accumulation [35]. However, a previous study showed that, despite their lower body weights, smokers not only do not eat less than non-smokers but in fact tend to eat slightly more [36]. Results of a meta-analysis from 15 developed countries [37] reported that current smokers reported significantly higher intakes of energy and alcohol than nonsmokers, consistent with our findings for Chinese male smokers. The connection between smoking and dietary intake is extremely complex. On the one hand, previous studies have demonstrated that smoking is associated with a significant reduction in monoamine oxidase, an enzyme that is associated with mood function, affecting dietary intake by inhibiting appetite. On the other hand, particular food choices might also correspond to an unhealthy lifestyle associated with smoking or a lack of nutrition knowledge.

Smoking intensity and obesity indicators
The association between smoking and body weight has become a central issue in the obesity literate, but the accumulating evidence is conflicting. Some cross-sectional studies indicate that BMIs are lower in current smokers than in nonsmokers [38,39], while other studies report a U-shaped relationship between smoking and BMI [40,41]. Our study examined the relationship between smoking behaviors and BMI in Chinese adults, showing that on average current smokers had a lower BMI than nonsmokers. Previous crosssectional studies indicate that heavy smoking could be associated with a greater risk of obesity, especially among males. One study of German adults found that male heavy smokers were more likely to obese than male light smokers [42]. Similarly, Shimokata et al. reported heavy smoking to be associated with higher BMI in US males [43]. After adjustment for age, working status, marriage status, individual income, education level, physical activity, drinking status, dietary energy intake, and percentage of dietary energy intake from fat, we found that both light and moderate smoking were associated with significantly lower BMI regardless of gender. Moreover, a notable finding of the present study is that light and moderate smoking decreased the risk of the onset of general obesity only among male current smokers. This suggests that there are gender differences in the impact of smoking on obesity risk. Furthermore, male heavy smokers are more likely to be obese than either other smokers or nonsmokers. One possible explanation is that there are biological differences between smokers and nonsmokers; another is that heavy smokers also have behaviors favoring weight gain such as physical inactivity, unhealthy diet, and high alcohol consumption [44]. Moreover, our study indicates that although there was a significant increase of nonsmokers in China from 1991 to 2015, it is of concern that the number of heavy smokers among Chinese adults increased significantly regardless of gender. In our own study, we found that male heavy smokers were more likely to drink alcohol and have a high-calorie/high-fat diet, while female heavy smokers were more likely to drink alcohol and be physically inactive.
Previous studies using WC to examine the association between smoking and abdominal obesity yielded inconsistent results. Xu et al. found that male non-smokers had non-significantly larger mean WC than current smokers [45]. In contrast, Mizuno et al. found that smokers had a significantly larger WC than nonsmokers only in obese males [46], while Liu et al. found that nonsmokers had a larger WC than current smokers [47]. In the present study, we found that male nonsmokers had a significantly larger WC than current smokers, contrary to the findings for females. These contradictory findings show that the association still requires attention. One possible explanation is that smoking affects the fat distribution in the abdominal area through a variety of biological mechanisms, as by affecting the regulation of sex hormones [14] [44]. For example, lower levels of androgens in male smokers have been found to increase abdominal obesity [48]. Furthermore, the amount smoked daily by smokers has also been reported to be positively associated with abdominal obesity [12]. We found that light and moderate smoking significantly decreases the likelihood of general obesity and abdominal obesity regardless of gender. Unhealthy behaviors may favor weight gain and partly explain the accumulation of visceral fat.
Moreover, our study indicates that although there was a significant increase of nonsmokers in China from 1991 to 2015, it is of concern that the number of heavy smokers among Chinese adults increased significantly regardless of gender. One possible explanation is that heavy smokers are more strongly addicted to nicotine and therefore experience more withdrawal symptoms, cravings, and other difficulties when attempting to quit. Zhang et al. reported that heavy smoking was also associated with poorer quality of life, such as greater perceived stress, poorer overall subjective quality of life, and lower satisfaction with finances, health, leisure activities, and social relationships [49]. This suggests that more interventions are needed for heavy smokers.

Strengths and limitations
First of strengths in this study was that we simultaneously analyzed the associations of smoking with general obesity (i.e. BMI) and abdominal obesity (i.e. WC), which allowed us to better clarify the relationships of smoking with fat distribution and related metabolic disorders. Second, we examined the association between smoking and risk of obesity by combine use levels of smoking intensity (based on the number of cigarette smoked in a day) with binary indicator of smoking participation (current smoker and nonsmoker).
Third, the longitudinal study design could confirmative evidence to causality. Forth, we controlled several important sociodemographic and behavioral confounders and assess sex differences in estimated associations, which helped to improve the validity of our findings.

Consent for publication
Not applicable.

Availability of data and materials
We used data of the China Health and Nutrition Survey (CHNS), which is publicly available at https://www.cpc.unc.edu/projects/china/data/datasets.

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