A cross-sectional and comparative study (obese vs overweight vs normal-weight) was conducted in Brazzaville, Republic of Congo, in middle schools. It involved adolescents aged 11 to 18 years attending these schools.

The study sample was determined after randomization according to a two-stage cluster survey, stratified by gender and the commune where the school is located (i.e., the sectoral inspection of the middle schools to which it belongs). We proceeded in the same way within each school. A selection of classes was made by drawing 1/3 by level of study. This selection was made by computer. In practice, 52 schools were selected and 1725 students were chosen: 673 (39.0%) boys and 1052 (61.0%) girls. Due to financial difficulties inherent to the costs of haematological analyses, a draw was made at 1/3 within the group of obese subjects and at 1/5 for overweight and normal-weight subjects. At the end of this process and after consent from the student and/or parents (guardians), 151 adolescents (obese, overweight, normal-weight) were selected. The inclusion criteria for the study were: parental consent for subjects aged 11 to 17 years, and individual consent for the student aged 18 years; a BMI above the 97.5^th percentile or IOTF-30 curve for the age group for obese children, a BMI between the 85^th and 97.5^th percentile for the overweight subject, a BMI between the 25^th and 85^th percentile for the normal-weight subject. Exclusion criteria were any chronic and acute illnesses, the chronic use of alcohol, smoking, metabolic (diabetes mellitus, dyslipidemia), infectious and allergic diseases at the time of blood collection, sickle cell disease continuous use of anti-inflammatory drugs taking an antibiotic treatment, recent history of blood loss, blood transfusion in the last one year, immunization in the last 6 months, major surgical procedures in the past 6 months, having any intestinal and hemoparasites. However, due to the cost of the hematological examinations and the lack of funding for the study, only 82 students (41 boys, 51 girls) participated in the study, of which 45 were obese, 17 were overweight and 20 were normal-weight.

At the beginning of the study, measurements of height (T) using a wall-mounted height gauge (model 220 Seca, Hamburg, Germany, accuracy: 0.1 cm default), weight (P) using an electronic scale (HW200, A&D Mercury Pty Ltd, Thebarton, SA), percentage fat (PCTG), waist circumference (TT) using a tape measure, subscapular (SS) and tricipital (TRI) fold thicknesses using a Harpenden calipers (British Indicators Ltd, West Sussex, UK), were performed. Measurement protocols were as described in the Anthropometric Standardization Reference Manual (Lohman et al., 1991) [15] ; children were dressed lightly. Body mass index (BMI) was calculated using the formula: BMI (kg/m²) = Weight (kg)/[height (m)]².

For a more refined assessment of obesity, the BMI Z-score was calculated in obese subjects using the World Health Organization AnthroPlus software (Geneva, Switzerland). Fat percentage was assessed from the subscapular (SS) and tricipital (TRI) skinfolds using the equations of Brook (1971), reported by Forbes (1986) [16] . The waist-to-waist ratio (WHtR) was calculated.

Blood samples were collected from each participant using the EDTA anticoagulated test tubes from 8 to 9 am from fasting adolescents in an appropriate, hygienic, air-conditioned room (temperature ranging from 25˚C to 27˚C). The number of leukocytes, red blood cells, platelets, hemoglobin and hematocrit were determined by fluorescence cytometry using the Sysmex XT 100TM.

Approximately 4 ml of venous blood was collected in a dry tube from each adolescent at the elbow, forearm, or posterior aspect of the hand (depending on the ease of finding a superficial vein). Samples were labeled with unique identification; they were transported in a cooler cooled to 2˚C - 8˚C and analyzed within 5 hours of collection. Before analysis, each sample stored at 2˚C - 8˚C was placed in the room for 30 minutes. This collected blood was placed in a water bath for use at 37˚C or room temperature. Whole blood samples were used for analysis of hematological tests. Centrifugation of this blood yielded the serum to be assayed. For the assays, pre-prepared kits (reagents) including buffer solution, enzymes and standards were used.

Lifestyle-related behaviors were assessed using a questionnaire, which was distributed and completed after interviews with health professionals and the investigator. It included items on tobacco and alcohol consumption, smoking status, area of residence (urban vs rural), physical activity (intense for ≥3 days/week, moderate (≥30 min/day and ≥3 days/week, or inactivity for <30 min/day and <5 days/week), history of pathologies (diabetes, dyslipidemia, etc.), history of family disease.

Associations between some anthropometric parameters (BMI Z-score, waist circumference, WHtR, skin folds) and blood count parameters were investigated.

Adolescence was defined as the age range from 10 to 19 years.

The child was defined as obese if the BMI was above the 97.5^th percentile or IOTF-30 values for the age range (Schwartz and Weiss, 1991) [12] .

The subject was declared overweight if the BMI was between the 85^th and 97.5^th percentile.

The subject was considered normal-weight if BMI was between the 25^th and 85^th percentile. Percentiles in this study were based on the 2001 growth reference values for French children and adolescents (Rolland-Cachera et al., 1991) [17] , because of the lack of Bantu growth references.

Waist circumference was considered elevated according to the French values [18] and the International Diabetes Federation criteria (Zimmel et al., 2007) [19] : 1) for TT ≥ 90^th percentile values in subjects of both sexes aged 6 to 15 years; 2) if TT strictly greater than 94 cm for boys and 80 cm for girls in subjects aged 16 - 18 years.

The statistical package for the Social Sciences (SPSS, version 21.0) was used for statistical analysis. Data were presented as mean ± standard deviation when the distribution of values was normal; otherwise, they were expressed as median. Categorical variables were presented in terms of numbers and percentages. The analysis of differences between the two groups used the non-parametric Kruskal-Wallis test. Differences between the three groups of students (obese, overweight, normal-weight) were analyzed using analysis of variance (ANOVA) for continuous variables and the Chi-2 test for categorical variables. Adjusted means and standard deviations of hematological indices (leukocytes, red blood cells, platelets, Hb, Hct) were determined for each group from 3 analyses of covariance (ANCOVA) models after adjustment for potential confounders. In the first ANCOVA 1 model, statistical analysis was performed based on body build after adjustment for age and sex for all study participants. Adjusted means and standard deviations of hematological indices were expressed after controlling for age and sex. In ANCOVA models 2 and 3, statistical analyses were also performed according to corpulence after adjustment for age in boys for model 2, and for age in girls for model 3. Adjusted means and standard deviations of hematological indices were after controlling for age for boys and girls. To assess the correlations between some anthropometric parameters (BMI Z-score, waist circumference, WHtR) and hematological indices, Pearson correlation analysis was used. Furthermore, because the measurement of plasma lipoprotein concentrations is commonly used to estimate cardiovascular disease risk, we investigated the potential relationships between 2 anthropometric indicators (WHtR, BMI Z-score) and lipoprotein concentrations in our obese sample. For this, Spearman’s simple correlation analysis was used. For the relationship between BMI Z-score (independent variable) and hematological parameters (dependent variables) in the obese, multiple linear regression analysis was performed. The significance level of the tests was set at 5%. All data were processed using PASW Statistics software, version 18.0 for Windows (SPSS Inc., Chicago, IL, USA).

Table 1 shows the clinical of 151 subjects of the study population.

The proportion of adolescents who drank alcohol was highest in the overweight group. The majority of adolescents lived in urban communes. However, one-third of the normal-weight adolescents resided outside the city of Brazzaville.

Data on hematological parameters (leukocytes, red blood cells, hemoglobin,

hematocrit and platelets) for all three groups of adolescents are shown in Table 2.

Hematological parameters were different among three groups, and obese adolescents had higher adjusted or unadjusted mean levels than did normal-weight adolescents.

The distribution of mean values of hematological parameters by gender is shown in Figure 1 in boys and 2 in girls.

Mean WBC increased significantly in boys and girls across normal-weight, overweight and obese groups. In the boys, means of WBC and Platelets were significantly higher in obese adolescents than the normal-weight adolescents, the differences being respectively of: +17.4%, and +12.3%. Differences between obese and normal-weight groups were found no significant for RBC (+1.2%), Hb (+4.1%), and Hct (+2.4%).

Concerning girls group, means of WBC, Platelets and Hct were significantly higher in obese adolescents than the normal-weight adolescents, the differences being respectively of: +17.4%, +10.3%, and +6.4%. Differences between obese

Abbreviations: NAJ, unadjusted; AJ, adjusted; WBC, white blood cells; RBC, red blood cells.

and normal-weight groups were found no significant for RBC (+2.8%), and Hb (+2.2%).

Table 3 reports the comparative means of CBC counts between obese boys and girls.

The unadjusted means for hematologic parameters were no significantly different among the boys and girls, except for hemoglobin.