Prevalence of obesity and correlations with lifestyle and dietary factors in Chinese men

Abstract

Objective

To estimate the age-adjusted prevalence of general and centralized obesity among Chinese men living in urban Shanghai.

Research Methods and Procedures

A cross-sectional study was conducted in 61,582 Chinese men aged 40 to 75. Body mass index (BMI, kg/m²) was used to measure overweight (23≤BMI<27.5) and obesity (BMI≥27.5) based on the WHO recommended criteria for Asians. Waist-to-hip ratio (WHR) was used to measure moderate (75^th≤WHR<90^th percentile) and severe (WHR≥90^th percentile) centralized obesity.

Results

The average BMI and WHR were 23.7 kg/m² and 0.90, respectively. The prevalence of overweight was 48.6% and obesity, 10.5%. The prevalence of general and centralized obesity was higher in men with high income or who were unemployed, tea drinkers, or non-ginseng users than their counterparts. Men with high education had a higher prevalence of overweight and centralized obesity, but had a lower prevalence of obesity and severe centralized obesity compared to those with lower education. Current smokers or alcohol drinkers had a lower prevalence of general obesity but higher prevalence of centralized obesity than non-smokers or non-alcohol drinkers. Ex-smokers and ex-alcohol drinkers had a higher prevalence of general and centralized obesity compared to non-smokers and non-alcohol drinkers. Prevalence of obesity was associated with high energy intake and low daily physical activity.

Conclusions

The prevalence of obesity in Chinese men in urban Shanghai was lower than that observed in Western countries but higher than that in other Asian countries, and the prevalence of general and centralized obesity differed by demographic, lifestyle, and dietary factors.

INTRODUCTION

Obesity has become one of the most urgent public health problems and poses a major threat to human health worldwide [1, 2]. A report from the Chinese Health and Nutrition Survey 1989-2000 suggested that the prevalence of overweight or obesity (BMI≥25) among Chinese men has increased from 10.1% in 1989 to 32% in 2000 in urban and from 4.7% to 15.5% in rural areas [3]. Understanding the factors that are contributing to the current epidemic of obesity and overweight is the first and most important step towards controlling the increasing trend.

Many factors related to weight gain are unevenly distributed across social classes [4]. While it is true that obesity is a result of energy intake exceeding energy expenditure, many factors influence this balance, including heredity, age, gender, race, education level, economic status, physical activity, eating habits, and psychological factors. The association between socioeconomic status (SES) and obesity depends on the population in which studies are conducted [5]. For example, inverse associations have been found in Western societies [2, 4-6], while positive associations have been found in developing countries [7-10]. Recently, as a result of economic growth and urbanization in China, changes in lifestyle and diet have led to an increase in life expectancy, as well as an increase in the burden of obesity-related chronic diseases [11, 12].

In this study, we used baseline data from the Shanghai Men’s Health Study (SMHS) to estimate the prevalence of general and centralized obesity by socioeconomic or lifestyle factors in Chinese men .

RESEARCH METHODS AND PROCEDURE

Study subjects

The Shanghai Men’s Health Study (SMHS) is a population-based cohort study with a major focus on investigating the effect of dietary, occupational, and lifestyle factors on the risk of cancers and other major chronic diseases. Recruitment for the SMHS started in April 2002 and was completed in June 2006. A total of 83,107 eligible permanent male residents of urban Shanghai, China age 40 to 74 were approached. Of these eligible men, 61,582 participated in the study, with a participation rate of 74.1 %. Non-participation was due to refusals (21.1 %), absence during the study period (3.1 %), and other miscellaneous reasons including poor health or hearing problems (1.7 %). The study protocols were approved by the Institutional Review Boards of all participating institutes and all participants provided written, informed consent.

An in-person interview was conducted by trained interviewers using a structured questionnaire, which collected information on demographic characteristics, disease and surgery history, lifestyle factors (e.g. cigarette smoking, alcohol consumption, tea consumption, ginseng use, etc), dietary intake, total daily physical activity, residential history, and occupational history. The dietary instrument and physical activity assessment have been validated [13, 14]. Participants were measured for their current weight, circumferences of the waist and hips, and sitting and standing heights according to a standard protocol. All anthropometric measurements were measured twice, and a third measurement was administered if the difference between the first two exceeded tolerances (1 cm for height and 1 kg for weight). The average of the two closest measurements was used in current analyses. All interviewers were retired health professionals and completed a rigorous training program and were certified before conducting an interview.

Definitions of variables

General and centralized obesity were measured using body mass index (BMI, kg/m²) and waist-to-hip ratio (WHR), respectively. Overweight and obesity were defined according to the general WHO criteria (25≤BMI<30 for overweight and BMI≥30 for obesity) and the WHO’s recommended criteria for Asians (23≤BMI<27.5 for overweight and BMI≥27.5 for obesity) [2, 15]. We defined centralized obesity according to percentiles of WHR. Moderate centralized obesity was defined as WHR falling in the 75^th to 89^th percentile and severe centralized obesity was defined as WHR≥90^th percentile. Subjects’ current jobs were classified into four categories according to current job titles: professional/technical, clerical/commercial, manual labors (agricultural/manufacturing), and retired. Subjects’ education was classified into four levels: elementary school or under, middle school, high school, and college or above. Per capita monthly income was classified into four levels: <500, 500-999, 1,000-1,999, and ≥ 2,000 yuan. Smoking status was classified as current, ex-, or non-smoker by asking the subject whether he had ever smoked at least one cigarette per day for more than 6 months and whether he still smoked at the time of the baseline interview. Similarly, alcohol consumption status was defined as current, ex-, or non-drinker with regular drinking defined as at least three times a week for more than six months. Total daily physical activity was estimated by using the weighted average of energy expended in all activities, including exercise/sports, walking and bicycling for transportation and daily activities, and house work, during the five years preceding the interview (MET-hour/week/year). Intake of macronutrients, total meat, and total fruits/vegetables, as well as energy intake from alcohol consumption, was estimated using a validated food frequency questionnaire (FFQ) that includes 81 food items/groups [14]. The correlation coefficients between the FFQ and 24-hour dietary recall average ranged between 0.59 to 0.66 for macronutrients, 0.41 to 0.59 for most micronutrients, and 0.41 to 0.66 for major food groups. Information on major chronic diseases was collected by asking study subjects whether they had been diagnosed by a physician with any of the following 20 chronic diseases: diabetes, high blood pressure, coronary heart disease, myocardial infarction, benign tumor, prostatomegaly, gallbladder stone, bladder stone, chronic hepatitis, hepatocirrhosis, emphysema, pulmonary tuberculosis, chronic gastritis, gastric ulcer, peptic ulcer, chronic bronchitis, asthma, allergy, stroke, and colorectal polyp.

Statistical analysis

Subjects with missing information on any demographic or lifestyle characteristic (n=1,192) were excluded from the analyses, and final analyses were based on 60,390 participants. The age-adjusted (categorized by 5 years intervals) prevalence of general and centralized obesity was calculated by demographic and lifestyle characteristics and differences were tested using Cochran-Mantel-Haenszel statistics. The p-values for trend tests in the group including all obese men and within sub-groups of obese men were derived from the Pearson’s Chi-square test and the Armitage trend test, respectively. In addition, we excluded men more than 65 years of age from analyses of general obesity (n=12,372), because BMI is not a good indicator of obesity after age 65 due to differential loss of lean mass. We applied polychotomous logistic regression models in the multivariate analyses. Differences between mean total daily physical activity, energy intake from alcohol and dietary factors between different categories of general and centralized obesity were evaluated with t-test using multiple linear regression models with adjustment for age, education, income, current job, smoking, alcohol consumption, tea consumption, ginseng use, and total daily physical activity. We further adjusted for total energy intake in the analysis of macronutrients and food intake. All statistical tests were based on two-sided probability. Statistical analyses were carried out using SAS version 9.1 (SAS Institute, Cary, NC).

RESULTS

Characteristics of the study population are presented in Table 1. At study recruitment, 60.9 % of subjects were employed (17.2% were professionals, 23.0% clerks, and 20.7% manual labors). About 60% of participants had a high school education or above, and 9.7% of subjects earned more than 2,000 yuan per month individually. The prevalence of current smoking, current alcohol consumption, regular exercise, and tea consumption were 58.6%, 29.3%, 35.6%, and 64.1%, respectively. The average of total dietary energy intake was 1,909 (±485) kcal per day. The average BMI was 23.7 kg/m² and the average WHR was 0.90. The prevalence of overweight (23≤BMI<27.5) was 48.6% and obesity (BMI≥27.5) was 10.5% based on the WHO recommended obesity criteria for Asians. The prevalence of overweight and obesity based on WHO general obesity criteria (BMI ≥25) was 30.4% and 2.6%. The percentage of men having both general (based on the general WHO criteria) and centralized obesity was 3.56%, general obesity alone was 4.26%, and centralized obesity alone was 0.89%. The correlation between BMI and WHR was 0.57.

Table 1.

Characteristics of study participants (n=60,390)

	Mean ± SD / percent
Age (years)	54.9±9.7
Weight (kg)	68.4±9.9
Height (cm)	170±5.8
Body Mass Index (BMI, kg/m2)	23.7±3.1
WHO criteria for Asian
BMI 23-27.4 (Overweight)	48.6
BMI ≥ 27.5 (Obesity)	10.5
WHO general criteria
BMI 25-29.9 (Overweight)	30.4
BMI ≥30 (Obesity)	2.5
Waist-to-Hip ratio (WHR)	0.90±0.06
0.94-0.97 (75th-89th percentiles)	15.1
≥ 0.98 (≥ 90th percentile)	10.0
Waist circumference (cm)	85.1±8.7
Current job (%)
Retired	39.0
Professional	17.0
Clerical	23.2
Manual labor a	20.8
Education (%)
≤ Elementary	6.8
Middle school	33.5
High school	36.1
≥ College	23.8
Income per capita in yuan/month/person (%)	12.6
<500	42.7
500-<1,000	35.0
1,000-<2,000	9.7
≥2000
Lifestyle factors
Current smoker (%)	58.7
Current alcohol drinker (%)	29.4
Regular exerciser (%)	35.5
Current tea drinker (%)	64.2
Total energy intake (kcal/day)	1909±485

^aManufacturing and labor workers, including 0.32% farmers, among employed subjects

The age-adjusted prevalence of general obesity (based on the WHO recommended criteria for Asians) by demographic and lifestyle factors are shown in Table 2. The prevalence of overweight and obesity was higher among older men (P<0.01 for both), men with higher income (P<0.01 for both) or who were employed (P<0.01 for overweight), men who were ex-smokers or ex-drinkers (P<0.01 for both), tea drinkers (P<0.01 for both), and non-ginseng users (P<0.01 for obesity) than younger men, men with lower income or who were retired, current smokers or drinkers, non-tea drinkers, or ginseng users. Men with higher levels of education had a higher prevalence of overweight (P<0.01) but lower prevalence of obesity (P<0.01) than men with lower education. Multivariate analysis showed that the association between prevalence of overweight or obesity and demographic/lifestyle factors followed a similar pattern of age-adjusted prevalence by each of these factors, with the exception of education (no association between education and overweight was noted in multivariate analyses) (Table 2).

Table 2.

Age-adjusted prevalence of general obesity (BMI, kg/m²) by socioeconomic and lifestyle factors among men less than 65 years old

	Age-adjusted prevalence of general obesity (BMI, kg/m2)					Multivariate regression
	N	<23	23-27.4	≥ 27.5	Pall	OR1 (95% CI)	OR2 (95% CI)
Age
≤ 45	11131	46.55	44.89	8.55		Reference	Reference
46-50	14210	44.48	47.14	8.38		1.08 (1.02-1.14)	0.96 (0.87-1.06)
51-55	10400	38.82	50.39	10.79		1.24 (1.17-1.32)	1.29 (1.16-1.43)
56-60	6852	35.07	52.19	12.74		1.39 (1.29-1.49)	1.66 (1.48-1.86)
61-65	5425	36.24	51.47	13.88		1.35 (1.24-1.46)	1.54 (1.36-1.76)
			P1<0.01	P2<0.01	<0.01	P trend <0.01	P trend <0.01
Education
≤ Elementary	1198	42.43	44.71	12.86		Reference	Reference
Middle	16947	42.18	47.01	10.81		0.93 (0.82-1.06)	0.69 (0.57-0.84)
High	19051	42.81	47.81	9.38		0.89 (0.78-1.01)	0.57 (0.47-0.68)
≥ College	10822	36.86	53.64	9.50		0.98 (0.85-1.12)	0.56 (0.46-0.67)
			P1<0.01	P2<0.01	0.09	P trend = 0.30	P trend <0.01
Income a
<500	6880	45.29	44.51	10.21		Reference	Reference
500-<1,000	18908	42.96	47.12	9.92		1.07 (1.01-1.14)	1.05 (0.95-1.17)
1,000 -<2,000	17093	39.35	50.66	9.98		1.19 (1.12-1.27)	1.18 (1.06-1.32)
≥2,000	5137	36.10	53.04	11.86		1.27 (1.17-1.39)	1.41 (1.22-1.62)
			P1<0.01	P2<0.01	<0.01	P trend <0.01	P trend <0.01
Current job
Retired	12567	45.82	43.74	10.44		Reference	Reference
Professional	9529	37.96	52.23	9.82		1.22 (1.14-1.31)	1.12 (1.01-1.25)
Clerical	13527	40.47	49.49	10.04		1.20 (1.14-1.28)	1.08 (0.98-1.19)
Manual labor b	12395	41.58	48.67	9.75		1.18 (1.11-1.25)	1.01 (0.91-1.11)
			P2<0.01	P2=0.89	0.01
Cigarette smoking
No	12438	37.26	53.02	9.72		Reference	Reference
Ex-smoker	3719	32.16	54.31	13.53		1.21 (1.11-1.31)	1.55 (1.40-1.75)
Current smoker	31861	44.17	46.26	9.58		0.74 (0.71-0.78)	0.78 (0.72-0.85)
			P1<0.01	P2<0.01	<0.01	P trend <0.01	P trend <0.01
Alcohol consumption
No	31490	40.85	49.08	10.07		Reference	Reference
Ex-drinker	1848	34.33	51.33	14.33		1.31 (1.18-1.45)	1.62 (1.39-1.88)
Current drinker	14680	43.63	46.97	9.40		0.94 (0.90-0.98)	0.86 (0.80-0.93)
			P1<0.01	P2<0.01	<0.01	P trend <0.01	P trend <0.01
Tea consumption
No	15921	43.30	47.41	9.29		Reference	Reference
Yes	32097	40.56	49.08	10.36		1.18 (1.13-1.23)	1.29 (1.20-1.39)
			P1<0.01	P2<0.01	<0.01	P trend <0.01	P trend <0.01
Ginseng use
No	35893	41.41	48.22	10.37		Reference	Reference
Yes	12125	41.70	49.38	8.92		0.98 (0.94-1.03)	0.83 (0.77-0.90)
			P1=0.38	P2<0.01	=0.01	P trend = 0.66	P trend <0.01
Total	48018	41.46	48.53	10.01

P_all: comparison of all general obesity groups from Pearson’s Chi-square test

P₁: comparison of < 23 kg/m² group vs. 23-27.4 kg/m² group for linear trend, derived from the Armitage trend test

P₂: comparison of < 23 kg/m² group vs. ≥ 27.5 kg/m² group for linear trend, derived from the Armitage trend test

OR1 (<23 kg/m² group vs. 23-27.4 kg/m² group) and OR2 (<23 kg/m² group vs. ≥ 27.5 kg/m² group) adjusted for all variables in table using a polychotomous logistic regression model

^aYuan /month/person

^bManufacturing and labor workers, including 0.32% farmers, among employed subjects

The age-adjusted prevalence of centralized obesity increased with age (Table 3). The prevalence of moderate centralized obesity was higher among men with high education levels (P=0.05) and high individual monthly incomes (P=0.03), but the prevalence of severe centralized obesity was higher among men with low levels of education (P<0.01). The prevalence of severe centralized obesity was higher among retired men (P<0.01) than employed men. Current and ex-smokers or current and ex-drinkers, tea drinkers (P<0.01), and non-users of ginseng (P<0.01) had a higher prevalence of centralized obesity. Multivariate analysis showed similar patterns of association, with the exception of income. High education level was negatively associated with moderate (P _trend = 0.02) and severe (P _trend < 0.01) centralized obesity. High income was associated with high prevalence of centralized obesity (Table 3).

Table 3.

Age-adjusted prevalence of centralized obesity (WHR, percentile) by socioeconomic/lifestyle factors

	Age-adjusted prevalence of centralized obesity (WHR, percentile)					Multivariate regression
	N	<75th	75th -89th	≥ 90th	Pall	OR1 (95% CI)	OR2 (95% CI)
Age
≤45	11131	79.21	13.26	7.53		Reference	Reference
46-50	14210	79.91	14.71	8.37		1.13 (1.05-1.21)	1.11 (1.01-1.22)
51-55	10400	75.58	14.92	9.50		1.15 (1.06-1.25)	1.25 (1.13-1.38)
56-60	6852	72.93	16.10	10.97		1.30 (1.19-1.42)	1.50 (1.34-1.68)
61-65	5452	71.41	16.42	12.17		1.36 (1.23-1.51)	1.68 (1.48-1.90)
>65	12372	70.85	16.05	13.10		1.33 (1.21-1.46)	1.77 (1.58-1.99)
			P1<0.01	P2<0.01	<0.01	P trend <0.01	P trend <0.01
Education
≤ Elementary	4076	73.24	14.48	12.29		Reference	Reference
Middle	20255	74.11	15.14	10.75		0.97 (0.88-1.07)	0.99 (0.88-1.10)
High	21786	75.39	14.82	9.79		0.95 (0.86-1.05)	0.89 (0.80-1.00)
≥ College	14273	76.16	14.70	9.14		0.91 (0.82-1.02)	0.81 (0.72-0.92)
			P1=0.05	P2<0.01	<0.01	P trend = 0.02	P trend <0.01
Income a
<500	7602	73.87	14.97	11.15		Reference	Reference
500-<1,000	25782	75.69	14.78	9.53		0.98 (0.91-1.06)	0.95 (0.87-1.04)
1,000-<2,000	21162	75.50	15.08	9.42		1.05 (0.97-1.14)	1.12 (1.01-1.23)
≥2,000	5844	77.22	13.78	8.99		1.17 (1.05-1.30)	1.19 (1.05-1.36)
			P1=0.03	P2=0.38	0.10	P trend <0.01	P trend <0.01
Current job
Retired	23579	73.87	14.97	11.15		Reference	Reference
Professional	10264	75.69	14.78	9.53		0.98 (0.91-1.06)	0.92 (0.83-1.01)
Clerical	13992	75.50	15.08	9.42		0.99 (0.93-1.07)	0.86 (0.79-0.93)
Manual labor b	12555	77.22	13.78	8.99		0.95 (0.88-1.02)	0.79 (0.72-0.86)
			P1=0.29	P2<0.01	<0.01
Cigarette smoking
No	18266	77.84	13.99	8.17		Reference	Reference
Ex-smoker	6618	71.60	16.80	11.60		1.24 (1.14-1.34)	1.40 (1.27-1.53)
Current smoker	35506	74.40	15.13	10.47		1.07 (1.01-1.14)	1.23 (1.13-1.23)
			P1<0.01	P2<0.01	<0.01	P trend = 0.02	P trend <0.01
Alcohol consumption
No	40007	76.14	13.99	8.17		Reference	Reference
Ex-drinker	2661	66.86	16.80	11.60		1.36 (1.22-1.51)	1.59 (1.42-1.79)
Current drinker	17722	73.38	15.13	10.47		1.09 (1.03-1.15)	1.15 (1.08-1.23)
			P1<0.01	P2<0.01	<0.01	P trend <0.01	P trend <0.01
Tea consumption
No	21644	76.39	14.41	9.20		Reference	Reference
Yes	38746	74.03	15.45	10.53		1.08 (1.03-1.13)	1.12 (1.06-1.19)
			P1<0.01	P2<0.01	<0.01	P trend <0.01	P trend <0.01
Ginseng use
No	43611	74.53	15.21	10.26		Reference	Reference
Yes	16779	76.11	14.63	9.26		0.94 (0.89-0.99)	0.89 (0.84-0.95)
			P1=0.03	P2<0.01	<0.01	P trend =0.02	P trend <0.01
Total	60390	74.92	15.07	10.02

P_all: comparison of all centralized obesity groups from Pearson’s Chi-square test

P₁: comparison of < 75^th percentile group vs. 75-89^th percentile group for linear trend, derived from the Armitage trend test

P₂: comparison of < 75^th percentile group vs. ≥ 90^th percentile group for linear trend, derived from the Armitage trend test

OR1 (< 75^th percentile group vs. 75-89^th percentile group) and OR2 (<75^th percentile group vs. ≥ 90^th percentile group) adjusted for all variables in table using a polychotomous logistic regression model

^aYuan /month/person

^bManufacturing and labor workers, including 0.32% farmers, among employed subjects

High BMI and WHR were inversely associated with total daily activity and positively associated with high energy intake from diet (P _trend <0.01) (Table 4). WHR but not BMI was positively associated with energy intake from alcohol consumption (P<0.01). Men with general obesity tended to have higher intake of macronutrients and fruits/vegetables than non-obese men (P _trend <0.01l). Men with moderate centralized obesity had higher intake of protein, meat and total fruit and vegetables than men with WHR<75^th percentile. On the other hand, men with severe centralized obesity were likely to have lower intake of energy, carbohydrates, and total fruit and vegetables, but higher intake of protein, fat, and meat compared to men with WHR<75^th percentile.

Table 4.

Difference of mean total daily physical activity, energy from alcohol intake, and dietary factors by general and centralized obesity

	General obesity (BMI, kg/m2)			P1	P2	P trend
	<23	23-27.4	≥ 27.5
Total daily activity (MET-hr/week)	6.9±3.7	6.9±3.8	6.8±3.9	0.13	<0.01	<0.01
Energy intake from alcohol consumption (kcal/day) *	239.3±208.0	225.4±194.4	247.7±210.9	0.16	<0.01	0.36
Energy intake from diet (kcal/day)	1892±483	1966±487	2013±518	<0.01	<0.01	<0.01
Protein (mg/day) **	77.8±23.6	81.3±24.2	83.4±25.3	<0.01	<0.01	<0.01
Fat (mg/day) **	34.5±15.6	35.7±16.3	36.3±17.0	<0.01	<0.01	<0.01
Carbohydrates (mg/day) **	317.4±86.5	329.7±86.3	337.8±92.6	0.06	0.24	<0.01
Total meat (g/day) **	80.8±52.7	84.1±55.5	87.4±59.6	0.99	0.14	0.07
Total fruits and vegetables (g/day) **	464.7±250.2	509.5±263.7	538.9±285.0	<0.01	<0.01	<0.01
	Centralized obesity (WHR, percentile)
	<75th	75th – 89.9th	≥90th
Total daily activity (MET-hr/week)	7.4±4.0	7.1±4.0	7.0±4.1	<0.01	<0.01	<0.01
Energy intake from alcohol consumption (kcal/day) *	220.6±195.6	231.9±194.6	258.8±231.6	<0.01	<0.01	<0.01
Energy intake from diet (kcal/day)	1908±482	1920±490	1895±499	<0.01	<0.01	<0.01
Protein (mg/day) **	78.1±23.8	79.2±24.2	78.6±25.0	<0.01	<0.01	<0.01
Fat (mg/day) **	34.5±15.6	34.9±16.4	34.9±17.0	0.84	<0.01	0.59
Carbohydrates (mg/day) **	321.1±85.6	322.0±90.0	316.6±87.6	0.14	<0.01	0.01
Total meat (g/day) **	78.6±52.6	81.2±55.7	81.4±58.4	<0.01	<0.01	<0.01
Total fruits and vegetables (g/day) **	485.1±257.2	491.9±2595	484.2±264.9	0.01	<0.01	<0.01

P₁ derived from t-test between normal (<23 kg/m² or <75^th percentile for general obesity and centralized obesity, respectively) and overweight (23-27.4 kg/m²) or moderate centralized obesity (75^th-89.9^th percentile) after adjustment for age, education, income, current job, smoking, alcohol consumption, tea consumption, ginseng use, and total daily physical activity.

P₂ derived from t-test between normal (<23 kg/m² or <75^th percentile for general obesity and centralized obesity, respectively) and obesity (≥ 27.5 kg/m²) or severe centralized obesity (≥ 90^th percentile) after adjustment for age, education, income, current job, smoking, alcohol consumption, tea consumption, ginseng use, and total daily physical activity.

P_trend: Trend test was conducted with BMI and WHR as a dependent variable using linear regression analysis after adjustment for age, education, income, current job, smoking, alcohol consumption, tea consumption, ginseng use, and total daily physical activity.

^*Among current drinkers only

^**Additionally adjusted for energy intake from diet

DISCUSSION

The prevalence of general overweight or obesity using the WHO general criteria (BMI≥25) is 32.9 % in this population. This is lower than the prevalence in Western countries, which ranges from 48% to 73% [16, 17], but higher than the prevalence in other Asian countries, which ranges from 17% to 26% [18-21]. The prevalence in our study is also higher than reports from studies conducted in other cities in China, which range from 22% to 32% [3, 22, 23]. The prevalence in our study of overweight and obesity using the WHO recommended criteria for Asians was 59.1%, which is also higher than the prevalence reported by the Ministries of Health and Science and Technology and the National Bureau of Statistics of the People’s Republic of China in 2002 [23]. The prevalence of overweight, however, was lower than age-standardized prevalence for Chinese men living in urban areas (39.9%, for BMI≥25) as reported in the International Collaborative Study of Cardiovascular Disease in ASIA [22]. The differences in prevalence from various reports could be due to differing risk factor profiles across the geographic areas and ethnic groups studied, as well as differences in data collection methods (e.g., self-reports vs. directly measured height and weight). The prevalence of high WHR in non-obese men was 0.9% (data not shown in tables) for this population, which is lower than that in men from urban Argentina [24] and from Japan [25].

We found that the age-adjusted prevalence of general and centralized obesity was higher among older men and men with high individual monthly incomes than in younger men or men with low income. The prevalence of general and centralized obesity was also higher among ex-smokers and ex-drinkers, tea drinkers, and non-users of ginseng and was associated with lower total daily physical activity and high consumption of energy from alcohol. High intake of macronutrients, meat and fruit/vegetables was positively associated with general obesity. In Westernized societies obesity is associated with lower SES status [2, 4-6], while in some developing countries obesity is associated with higher SES status [7-10]. One review of the related literature found that the main socioeconomic indicators associated with obesity were income in developing countries and education in developed countries [9]. The positive association between higher income and moderate general and centralized obesity in this study is consistent with results from developing countries [26] and a cross-sectional study recently conducted in other areas of China [27]. However, our findings regarding the prevalence of severe obesity related to high income were not consistent with previous studies. The latter suggests that changes in social and economic levels in Shanghai may have influenced patterns of obesity, making them more similar to patterns observed in Western countries [2, 4-6].

Former smokers had a higher BMI and WHR than current and non-smokers in the present study, consistent with several previous studies [28, 29]. On the other hand, we found that current smokers had a higher prevalence of centralized obesity than non-smokers. Previous studies have also reported that current smokers have a lower BMI [28, 30] but a higher WHR [31, 32] than former or non-smokers. These observations suggested that, although smokers have lower BMI than non-smokers, they have a more metabolically adverse fat distribution profile.

The age-adjusted prevalence of general obesity was higher in ex-drinkers than current drinkers, and ex-drinkers had higher dietary energy intake (mean: 1953 ±506 kcal per day) than current drinkers (mean: 1832 ± 503 kcal per day). However, information on energy intake from alcohol was not available for all ex-drinkers, which prevented evaluation of the independent effect of alcohol consumption on obesity. Among current drinkers, we did not find that energy from alcohol consumption was associated with BMI. Our finding of a positive association between alcohol consumption and centralized obesity is consistent with most previous studies [33-36]. Although energy from alcohol consumption has been shown to result in accumulation of triglycerides in adipose tissue [37], the exact biological mechanisms underlying the association between alcohol intake and centralized obesity remain to be established. In this study population, employed men had a higher prevalence of moderate general obesity, but severe centralized obesity was more common among retired subjects. The role of physically active occupations in preventing obesity has been reported in several previous studies [38], including a report from a Chinese population [39]. In the present study, men with general and centralized obesity reported less total daily physical activity compared to non-obese men. The daily physical activity required for transportation and daily living activities tends to be more important than physical activity coming from leisure-time exercise/sports in developing countries [40], including our study population [41]

Previous studies have suggested that green tea may increase thermogenesis and fat oxidation, but has no effect on body weight maintenance [42-44]. Inconsistent with these studies, we found that tea consumption was associated with a higher prevalence of obesity. However, among men free of chronic disease, tea consumption was not associated with the prevalence of either type of obesity (P=0.212 and 0.112 for general and centralized obesity, respectively, data not shown in tables). It is possible that diagnosis of a chronic disease may have led to a change of lifestyle, including increased tea consumption in this study population; men with chronic disease (70%) consumed more tea than those without chronic disease (62%). On the other hand, we found that ginseng use was significantly related to a lower prevalence of obesity. Several animal studies have suggested that ginseng may reduce obesity and hyperglycemia [45, 46]. However, this association has not been seen in humans. Nachtigal et al. reported a non-significant association between long-term (10-year) use of ginseng supplements and weight gain [47]. Because ginseng is commonly used to boost energy and increase immunity in Chinese populations, we could not exclude possible confounding from indications of ginseng use. Further studies are needed to investigate the true relationship between ginseng use and obesity.

It is well documented that the balance between total energy intake and energy expenditure determine weight gain and loss [48]. As reported in two cross-sectional studies of Chinese men [36, 49], we also found that higher BMI was related to higher energy intake. However, men with centralized obesity tended to have lower intake of total dietary energy and carbohydrates, but higher intake of protein, fat, and meat in this study. Income is a determinant of energy intake, particularly for fat and animal protein intake in developing countries [50]. Although the centralized obesity and diet association persisted after adjustment for income and education, residual confounding from other socioeconomic factors cannot be excluded.

Given the cross-sectional nature of our study, no causal associations of demographic and lifestyle factors with the prevalence of general and centralized obesity can be established. Since we were not able to gather information from non-responders to the study, we could not evaluate any possible selective participation bias. However, our study is population-based, had a high response rate (75%) and implemented high quality control practices. All anthropometrics were taken by trained health professionals. Dietary intake and total daily physical activity was assessed using validated instruments. Because soft drinks were not commonly consumed among middle-aged and elderly men at the time the study was initiated, we did not collect information on this food item, and thus we could not evaluate the effect of soft drink consumption on obesity prevalence in this study.

In conclusion, in this population of Chinese men, 33.1% were overweight or obese according to the WHO general criteria (BMI≥25), while 59.1% were overweight or obese by the WHO overweight/obesity criteria for Asians (BMI≥23). The age-adjusted prevalence of general and centralized obesity varied with education, income, employment, smoking, alcohol and tea consumption, regular exercise, and dietary intake of major nutrients and foods. Our study provides important information for obesity research and for the development of preventive measures to reduce burden of obesity.

Abbreviations

BMI

Body Mass Index

FFQ

food frequency questionnaire

MET

metabolic equivalent

PAQ

physical activity questionnaire

SES

socioeconomic status

SMHS

Shanghai Men’s Health Study

WHO

World Health Organization

WHR

Waist-to-Hip Ratio