In 2000, the Centers for Disease Control and Prevention of Pudong and Baoshan districts developed a health registration system for local residents and subsequently implemented a management system for type 2 diabetic patients who had been diagnosed according to the World Health Organization (WHO) criteria [6
]. Between December 2006 and August 2007, the total number of registered type 2 diabetic patients was approximately ~3,000 in Pudong and ~3,500 in Baoshan. We randomly chose 6 communities (by different economic status, three communities each in Pudong and Baoshan) and contacted 2,401 cases who met our inclusion criteria (88% responded). Controls were identified from the population registers using the same inclusion and exclusion criteria for cases in the same communities from which cases occurred. A total of 3,234 non-diabetic individuals were randomly selected and invited by letter or telephone to participate in the study. 76% of these responded. All eligible type 2 diabetic patients and comparable controls were identified to participate in this Shanghai Diabetes Study (SDS) where dietary, anthropometrical, and biochemical assessments were conducted. Written informed consents were obtained from all participants. This study was approved by both local authorities and the Ethics Committee of Shanghai Institute for Biological Sciences (ER-SIBS-250701), and the UCLA Institutional Review Board (UCLA IRB #06-05-096-11).
Participant enrollment and data collection
Of approximately 6,500 type 2 diabetes cases available in Baoshan and Pudong, 2,113 type 2 diabetes patients were enrolled in the SDS. They met the following inclusion criteria in the SDS: 40 - 79 years old; of Chinese Han ethnicity; had resident registration records, and lived locally >= 5 years. We excluded participants who had physical disability, severe diseases (e.g. cancer and stroke) during the previous 6 months, or who were unable or unwilling to sign the informed consent form. Aside from the fact that both districts have a diabetes surveillance system, the two independent sets of cases in a well-established health-care network were intended to serve as split-replication samples to each other (to confirm any new findings that may emerge from future genetic studies). The same inclusion and exclusion criteria for cases were also applied to the selection of controls in these two districts, ultimately enrolling 2,458 controls. These apparently healthy were then examined to determine if they had ever been diagnosed with diabetes or if their fasting glucose levels met the WHO criteria for type 2 diabetes diagnosis. 308 individuals were identified to have impaired fasting glucose (IFG, fasting plasma glucose levels from 6.1 mmol/L to 6.9 mmol/L).
All enrollees were examined by centrally-trained staff at clinics for collection of participant information, including a questionnaire, anthropometric indices and overnight fasting (>= 8 hours) blood samples.
Lifestyle, dietary, and anthropometric measurements
All participants completed a standardized questionnaire that included questions about their demographic characteristics, history of chronic diseases, family history of diabetes in first-degree relatives, cigarette smoking, alcohol consumption, and physical activity. Cigarette smoking was defined as "at least once daily," alcohol consumption was defined as "regularly drink more than 50 ml each time and at least twice/week". Low, medium and high levels of physical activity were defined as "daily housework, walking, flower planting and light stretch activity," "jogging, swimming and ping-pong" and "hiking, tennis and exercise in gym" respectively. For diabetes patients, diabetic symptoms, onset time, and hypoglycemic medications were also collected.
A food frequency questionnaire (FFQ) was also administered to rank participants according to the distribution of long-term nutrient and food intake in the SDS. The development and validation of a similar FFQ used for Han Chinese in Shanghai have been reported previously[7
]. In general, the FFQ can reasonably categorize usual intake of nutrients and food groups among Chinese adults. Briefly, our FFQ includes 100 food-items in 10 categories (staple foods, meats, aquatic products, beans and related products, eggs, milks, vegetables, fruits, pickled vegetables and others). For each food item, participants were asked how frequently in reference to the previous year ("never", "times per day", "times per week", "times per month", and "times per year") and the quantities they consumed the specific item.
All measurements were conducted using a standardized protocol. After an overnight fast, participants were asked to sit at ease, rest for ≥ 5 minutes, and avoid smoking and drinking alcoholic beverages and coffee prior to the scheduled appointment. Blood pressure measurements 30 seconds apart were taken from the participant's right arm, using a conventional mercury sphygmomanometer with appropriate cuff size. The average value of two consecutive measurements was recorded. Subjects taking antihypertensive medications and those with systolic blood pressure ≥ 140 mm Hg, diastolic blood pressure ≥ 90 mm Hg (WHO/ISH) were defined as hypertensive. Height (in centimeter, cm), weight (in kilogram, kg), waist circumference, and hip circumference were measured after participants took off their shoes, hats, coats, and sweaters. Waist circumference was measured at the midpoint between the inferior costal margin and the superior border of the iliac crest on the midaxillary line and hip circumference was measured at the maximum extension of the buttocks. Body mass index (BMI) was defined as kg/m2. Following the recommendation by the Working Group on Obesity in China (International Life Science Association, 2001), we defined central obesity using either waist circumference for men >= 85 cm, for women >= 80 cm or waist-hip ratio <= 0.75, 0.76~0.85, >= 0.86. BMI was defined as < 24 for normal, 24~28 for overweight and > 28 for obesity.
Laboratory and biochemical measurements
Fasting blood specimens were collected using vacuum negative pressure tubes. Blood glucose, glycosylated hemoglobulin A1C (HbA1c), and lipids were measured using Roche modular P800 autoanalyzer. All measurements were performed at the biochemistry and immunology laboratory of Dongfang Hospital, a teaching affiliate of Tongji University. The interassay coefficients of variation were 1.7% for glucose, 3.2% for HbA1c, 1.8% for triglycerides, 1.7% for total cholesterol, 1.2% for LDL-cholesterol, and 1.3% for HDL-cholesterol. The definition of abnormal lipid profiles was triglycerides <= 1.70 mmol/L or HDL-Cholesterol >= 1.04 mmol/L (Chinese Cardiovascular Disease Association, 1997).
Definition of metabolic syndrome
We defined metabolic syndromes using two different criteria (Chinese Diabetes Society, CDS, 2004) [9
] and the National Cholesterol Education Program's Adult Treatment Panel III (NCEP ATP III, 2002) [10
]. According to the CDS criteria, a participant has metabolic syndrome if he or she has three or more of the following criteria: BMI >= 25 kg/m2; fasting glucose >= 110 mg/dl or 2-h plasma glucose >= 140 mg/dl or diabetes; blood pressure >= 140/90 mmHg or hypertensive; and triglycerides >= 150 mg/dl or HDL cholesterol < 35 mg/dl in men or < 39 mg/dl in women. The metabolic syndrome was also defined by the NCEP ATPIII, when three or more of the following five risk determinants were present: waist circumference (men > 102 cm, women > 88 cm), triglycerides >= 150 mg/dl, HDL-C (men < 40, women < 50 mg/dl), blood pressure (>= 130/>= 85 mmHg), and fasting glucose >= 110 mg/dl.
We first examined differences in age, sex, education, age at diagnosis, duration of diabetes, family history and hypoglycemia medication by two recruiting districts. The crude medians (ranges) and prevalence of demographic, lifestyle, anthropometric, and biochemical characteristics by outcome status (type 2 diabetes, IFG, and controls) were then calculated and compared using either Kruskal-Wallis test (for continuous) and χ2 test (for categorical variables). We also estimated the prevalence of metabolic syndrome according to different criteria and its individual components by outcome status. Odds ratios (ORs) and 95% confidence intervals (CIs) of type 2 diabetes by demographics and lifestyle risk factors were calculated in a multiple logistic regression model. Finally, we calculated ORs and 95% CIs of type 2 diabetes by components of metabolic syndrome in a multiple logistic regression model adjusted for age, gender, education, family history, smoking status, alcohol drinking and leisure physical activities. All p values were two-sided, and all statistical analyses were conducted using SAS (version 9.2; SAS institute, Cary, NC).