This study was part of a large longitudinal epidemiological study on diabetes in rural Bangladesh which has been described previously [12]. The current cross-sectional study was carried out during March to December 2009 in an urbanizing rural community "Chandra" 40 km north of capital Dhaka city. Some structural changes occurred during the last 10 years due to territorial expansion of Dhaka city. The region can now be considered as semi-urban as the dwellers have access to most of the basic urban facilities such as 100% sanitation, 95% safe water supply, 90% households electric supply and brick road communication. Substantial number of new small industries, garments factories, poultry farm and fisheries were established during last 10-year period. Ten villages were randomly selected from five areas with a population of approximately 20,000 with aged ≥20 years. For this study, 3,000 individuals were randomly selected and among them 2,376 (79.2%) participated. The present analysis is based on 2,293 participants (842 male and 1,451 female) for whom all the variables were available. Exclusion criteria included pregnant women, and those with a diagnosed acute physical or mental illness.

Once selection procedure was completed, participants were requested to visit a nearby field center after an overnight fast of 8 to 12 hours. Then 8 mL of venous blood was collected from each participant. Fasting plasma glucose (FPG), lipid profile, serum insulin, and glycosylated hemoglobin (HbA1c) were determined from fasting blood sample. All subjects other than those with known DM (n=40) were then given a 75 g oral glucose solution (75 g oral glucose in 250 mL of water) to drink. Another 3 mL of venous blood was collected after 2 hours to determine 2 hour post OGTT. The samples for plasma glucose test were collected in a tube containing sodium fluoride and potassium oxalate (1:3) and were centrifuge immediately after collection. Separated plasma samples were sent to laboratory of Bangladesh Institute of Research and Rehabilitation for Diabetes, Endocrine and Metabolic Disorders (BIRDEM) in ice jell packed cooling boxes and stored at -70℃ until laboratory assays was done. Plasma glucose was measured by the glucose oxidase method using Dimalesion RxL Max (Japan) on the same day. Quality control on the blood glucose measurement was checked by measuring the 2-hour plasma glucose values using the glucose oxidase methods in every 10th case. The intra-assay coefficient of variation was 1.24% at a mean of 5.86 mmol/L, and the inter-assay coefficient of variation was 2.10% at a mean of 5.23 mmol/L. HbA1c estimated by high performance liquid chromatography. Insulin was measured later on by a two-site chemiluminescent immunoassay system (Diagnostic Products Co., Los Angeles, CA, USA). We calculated HOMA-IR and HOMA-B using the method of Matthews et al. [13]. Fasting serum lipid profile was estimated by standard enzymatic procedures (Dimalesion RxL Max). High density lipoprotein cholesterol (HDL-C) was estimated by the direct assay method and low density lipoprotein cholesterol (LDL-C) was estimated by Friedewald's formula. Cutoff values for serum lipid profiles were as follows: cholesterol ≥5.0 mmol/L, triglycerides (TG) ≥1.7 mmol/L, LDL ≥3.0 mmol/L, and HDL-C <1.03 mmol/L (for male) and <1.29 mmol/L (for female). Dyslipidemia was defined as TG ≥1.7 mmol/L and HDL-C <1.03 mmol/L (for male) and <1.29 mmol/L (for female).

Diabetes was diagnosed if FPG value was ≥7.0 mmol/L and/or 2-hour post plasma glucose (2hPG) was ≥11.1 mmol/L. IFG was defined if FPG was between 6.1 and <7 mmol/L and 2hPG was <7.8 mmol/L. IGT was defined if FPG was <6.1 mmol/L and 2hPG was between ≥7.8 and <11.1 mmol/L. Combine IFG and IGT was defined if FPG values was between 6.1 and <7 mmol/L and 2hPG was between ≥7.8 and <11.1 mmol/L. For screening of IFG, IGT, and T2DM, diagnostic criteria of WHO was used [11].

During the 2-hour waiting period for OGTT, the respondents were interviewed for socio-demographic information's. Information's were collected using a pretested questionnaire and their anthropometric measurements including height, weight, hip and waist girth were recorded. Measurements of height, weight, and waist and hip circumference were taken with light clothes without shoes. For height, the subject stood in erect posture vertically touching the occiput, back, hip and heels on the wall while gazing horizontally in front and keeping the tragus and lateral orbital margin in the same horizontal plane. Waist circumference was measured by placing a plastic tape horizontally midway between 12th rib and iliac crest on the mid-axillary line. Similarly, hip was measured by taking the extreme end posteriorly and the symphysis pubis anteriorly. Cutoff values for general obesity for both sexes were a BMI of ≥25 kg/m², cutoff values for central obesity including waist circumference for male and women were ≥90 and ≥80 cm, respectively, and waist to hip ratio (WHR) for male ≥0.90 and female for ≥0.80 [14,15]. In addition, blood pressure was measured in both sitting and standing position. Prior to the measurement, 10-minute rest was assured and using standard cuffs for adults fitted with sphygmomanometer minimized variation in measurement. Two readings were taken 5 minutes apart, and the mean of the two was taken as the final blood pressure reading of the individual. Hypertension was defined as a systolic blood pressure (SBP) of 140 mm Hg and/or diastolic blood pressure (DBP) of 90 mm Hg or current treatment with antihypertensive medication [16]. Physical activity was graded on the ordinal scale of 1 to 3, corresponding to light, moderate and heavy, according to the activity level of study population. For the purpose of data analysis, these results were transformed into a binary variable-inactive (grade 1) and active (grade 2 and 3). Smoking habit was classified as either current or non/ex-smoker. Montogomery-Asberg Depression Rating Scale (MADRS) [17] was used for the assessment of depression scores. The questionnaire was translated into local language Bangla by the first author. Fieldworkers, who had received one week training in using this questionnaire prior to the survey, filled it out. The score of each item can range from 0 to 6; hence, the sum for the 10 items can range from 0 to 60. In general, MADRS scores are categorized in four groups: 0 to 12 (healthy), 13 to 19 (mild depression), 20 to 34 (moderate depression), and 35 to 60 (severe depression) [17]. Based to previous experiences [18] and limited samples it was decided to use only two groups 0 to 12 (healthy subjects) and ≥13 (depressed subjects).

The protocol was approved both by Norwegian and Ethical Review Committee of Diabetic Association of Bangladesh. Verbal consent was received from all subjects prior to inclusion in the study. The procedure was adopted in order to avoid selection bias. Verbal orientation of the objectives and the procedure were read aloud to the subjects, including their right to refuse and withdraw at any stage of the study or to bar their data from analyses. All information and data collected were deemed confidential. All patients received a hardcopy of their own biochemical results.

All the continuous variables were presented as mean (95% confidence interval) or geometric mean (95% confidence interval) and categorical data as percentage (95% confidence interval). Highly skewed variables (HbA1c, TG, insulin, HOMA-B, and HOMA-IR) were log transformed before analysis. Age specific and age standardized prevalence by direct standardization method were estimated on the basis of 2001 census data before performing statistical tests [19]. Differences between the two groups of means adjusted for age were tested by analysis of covariance (ANCOVA) and pairwise comparisons between the groups were performed and corrected for multiple testing using Bonferroni method. Logistic regression models were used to examine the statistical difference of proportions and estimate the odd ratios adjusted for age and potential confounding factors. Measure of association between two continuous variables and two categorical variables were tested by Pearson's and Spearman's correlation coefficient, respectively. Kappa was used to evaluate the agreement between FPG and 2hPG. Multiple logistic regression analyses were also used to study the association of risk variables with T2DM and impaired glucose regulation. Two-sided P values were given and compared with significance level 0.05. The estimation of standardized prevalence was performed with STATA 11 for Windows (StataCorp, College Station, TX, USA) and the other analyses were carried out with PASW 18 for Windows (SPSS Inc., Chicago, IL, USA).