This cross-sectional study was carried out between March and December 2011. By convenience, we studied the population of middle-aged men who were staff members of the Federal University of Viçosa, located in the Brazilian southeastern city of Viçosa, Minas Gerais state. This population consisted of 1,774 (N) men aged between 40–59 years. The sample size was calculated using the confidence level of 95% and the prevalence for obesity in Brazilian men (i.e. 17.5%), as detected by the Brazilian Health Minister , and 4% of sampling error, which resulted in 291 (n) participants as a minimum sample size required. The Epi Info software, version 6.04, for cross-sectional studies  was used to estimate sample size.
To select the subjects of this study all 1,774 staff members were listed and numbered alphabetically and those numbered multiple of 6 (N/n: 1,774/291) were chosen. In the event of meeting the exclusion criteria the subject was replaced by his predecessor in the list. Eight hundred fifty-six subjects were interviewed and 300 of them were eligible to take part in the present study.
This study excluded subjects who self-declared: body weight alterations of ≥ 3 kg, altered levels of physical activity and eating habits in the three months preceding the study; thyroid diseases, heart failure, cerebrovascular diseases, infectious and/or inflammatory diseases, diseases of the gastrointestinal tract, liver and chronic kidney and/or history of kidney stones, cancer in the previous ten years, eating disorders (anorexia and bulimia) and food allergies. Subjects using diuretics or drugs that could alter food intake and/or metabolism of nutrients were also excluded. Pacemaker and/or prosthetic users were excluded as it could affect the DXA result analyses. Elite athletes were excluded as they could exhibit an inflammatory condition due to exercise training stress.
This study is in accordance with the resolution 196/96 from the Brazilian Ministry of Health regarding research involving human subjects and was approved by the Ethics Committee on Human Research of the Federal University of Viçosa (Of. Ref. n° 069/2010/CEPH). Only participants who signed the consent form in accordance with the Declaration of Helsinki were selected.
Anthropometry and body composition measurements
Anthropometry and body composition measurements were carried out after a 12-hour fast and the subjects were instructed to perform no physical activities of moderate and high intensity and no caffeine and alcohol ingestion in the 48 hours prior to the test.
Body weight and height were determined following the protocol described by Gordon et al. , using a digital scale with stadiometer (2096PP, Toledo, São Bernardo do Campo, SP, Brazil). Body mass index (BMI) was calculated using the equation proposed by Quetelet and the subjects were categorized as: eutrophic (18.5 to 24.9 kg/m2), overweight (25.0 to 29.9 kg/m2) or obese (≥30 kg/m2), according to the criteria set by the World Health Organization .
Waist circumferences were measured on three anatomical points: (a) narrowest waist (WCNR)  (i.e. nearly 1 cm below the last rib); (b) midpoint between the superior border of the iliac crest and inferior margin of the rib (WCMD) [2, 3] (i.e. nearly 3 cm above the umbilical line); and (c) at the umbilical line (WCUL) . Waist circumferences were measured in triplicate using a flexible, no stretching tape (TR4010, Sanny, São Bernardo do Campo, SP, Brazil) and the average value for each anatomical point was considered for data analyses.
Total body scan was performed by dual beam X-ray absorptiometry (DXA) (LUNAR, GE, Encore software version 13:31, Madison, WI, USA) to determine the percentages of total body fat (%BF) and abdominal area fat (%AAF). Abdominal area fat is the body fat detected in the area between the superior border of the iliac crest and the inferior border of the last rib. Overweight and obesity cutoff values were set at 20% and 25% of %BF [26, 27], respectively. Since there is no cutoff points for %AAF reported, the percentage found in the 75th percentile of %AAF in the present sample was used for central obesity.
Blood pressure, blood glucose, insulin and serum lipid profile measurements
Systolic (SBP) and diastolic blood (DBP) pressure were measured using an automatic inflation blood pressure monitor (BP3AA1-1, G-Tech, OnboElectronicCo, Schenzen, China), registered at ANVISA (No. 80275310004), following the VI Brazilian Guidelines on Hypertension .
Blood samples were collected from the antecubital vein and the serum was separated by centrifugation at 2.225 g for 15 min at room temperature (Sigma 2–3, Sigma Laborzentrifuzen, OsterodeamHarz, Germany). Blood glucose was measured using the glucose oxidase method (Cobas Mira Plus, Roche Diagnostics, GmbH, Montclair, NJ, USA), and insulin was measured by electrochemiluminescence (Modular Analytics, E170, Roche Diagnostics, GmbH, Mannheim, Germany).
Serum total cholesterol, high-density lipoprotein (HDL-C) and triglycerides were determined by an enzymatic colorimetric method (Cobas Mira Plus, Roche Diagnostics GmbH, Montclair, NJ, USA). The atherogenic index was calculated as the total cholesterol to HDL-C ratio .
Determination of metabolic syndrome, insulin resistance and cardiometabolic risk factors
The MetS was considered prevalent in subjects who exhibited three or more factors related to waist circumference (WCMD ≥ 90 cm), hyperglycaemia (glucose ≥ 100 mg/dL), dyslipidaemia (HDL-C < 40 mg/dL), hypertriglyceridemia (≥150 mg/dL) and/or high blood pressure (SBP ≥ 130 mmHg or DBP ≥ 85 mmHg), according to the criteria and cutoff points suggested by Alberti et al. .
The homeostasis model assessment (HOMA-IR) was used to estimate IR by using the equation proposed by Matthews et al. . The cutoff value used for the IR diagnosis was 2.7 as suggested by Geloneze et al. .
The following values were set as cardiometabolic risk factors [32, 33]: triglycerides ≥ 150 mg/dl (hypertriglyceridemia); total cholesterol ≥ 200 mg/dl and HDL-C < 40 mg/dl (dyslipidaemia) and glucose ≥ 99 mg/dl (hyperglycaemia). The participants were classified as hypertensive when systolic and diastolic blood pressures were ≥ 140 and ≥ 90 mmHg, respectively,  and it was considered a cardiometabolic risk when atherogenic index was ≥ 5 .
The subjects who participated in this study occupied working positions classified as levels A, B, C, D, and E, or professor. To evaluate how lifestyle and occupation influenced the level of physical activity they were grouped according to their education level and working positions: Group ABC was composed of technical and administrative staff members, classified as A, B and C, with an education level up to high school. Group DEProf was composed of technical and administrative staff members levels D and E and professors, all college-educated.
The participants were asked about their current smoking status (yes/no) and alcohol consumption (types of alcoholic beverages consumed - beer, wine and/or spirits, frequency and weekly quantity in mL). High alcohol consumption was defined as a weekly intake over 21 units .
The full version of the International Physical Activity Questionnaire  was applied and subjects were categorized as sedentary/moderately active or active/very active.
Data normality was assessed by the “Smirnov-Kolmogorov” test. For data exposure, we used descriptive statistics composed by mean values and standard deviation or median and interquartile range for continuous variables and frequency for categorical variables. After logarithmic transformation the WCNR, WCMD and WCUL values were compared by ANOVA one way followed by the post hoc Tukey test. The physical activity levels were compared using Chi-Square. The Student’s t test was used for independent samples, or its nonparametric equivalent, the Mann-Whitney test, to confirm the existence of differences between mean values per group.
The receiver operating characteristic curve (ROC) was used to detect the best circumference cutoff, sensitivity (Sens) and specificity (Spec) in relation to the cutoff points: 21%BF and 25%BF and 34.6%AAF. The areas under the curve and confidence intervals of 95% (95% CI) were also determined. The univariate and multivariate regression analysis according to Poisson was used to estimate the prevalence ratio (95% CI) of subjects with hyperglycaemia, dyslipidaemia, hypertensive and MetS (dependent variables). In these analyses the WC cutoff point (88.8 cm) served as independent variable and the lifestyle factors as covariates.
Data processing and analysis were carried out with the software SPSS version 16.0 (SPSS Inc. Chicago, IL, USA) and STATA 9.1. The value used for all variables and two-tailed analyses was p # 0.05.