Abstract

Cell blood counts are components of hematological parameters and indicators of pro-inflammatory states. They are proposed to be associated with metabolic syndrome (MetS). This study aimed to assess the relationship of the white blood cell (WBC) and the red blood cell (RBC) counts with components of MetS in the pediatric age group. The sample consisted of 300 children (152 boys) aged 6–12 years. Hierarchical Bayesian analysis of the bivariate Poisson regression model was used to estimate the effect of various components of MetS according to the cell blood counts. We found that RBC and WBC counts were correlated with the fasting blood glucose, the waist-to-height ratio, serum triglycerides and the blood pressure levels adjusted for age, the body mass index, gender, total cholesterol, low-density lipoprotein cholesterol and the hip circumference. The high-density lipoprotein cholesterol was correlated with the RBC counts based on 95% high posterior density regions for parameters in the Bayesian model. Our findings may serve as confirmatory evidence for the beginning of inflammatory process related to the cardio-metabolic factors from early life.

Introduction

Metabolic syndrome (MetS) is recognized as one of the most serious clustering of risk factors for dyslipidemia and hypertension and is also deliberated as the secondary target for the treatment of coronary heart disease, as well as type 2 diabetes [1, 2]. This syndrome is no more limited to adult populations and to high-income countries, but is also prevalent in the pediatric population of developing countries [3].

Such experience is limited in Asian developing countries. By using the criteria of Adult Treatment Panel-III (ATP-III) adopted for children and adolescents, a study in Turkish children reported a prevalence of 2.2% for MetS [4]. In another study conducted on Korean adolescents, the prevalence of MetS increased significantly from 6.8% in 1998 to 9.2% in 2001 [5]. According to this study, nearly 500 000 out of 5.4 million South Korean adolescents aged 12–19 years were suggested to have MetS.

In a study in China, by using the International Diabetes Federation (IDF) definition, the prevalence of MetS was 0.8% for children aged >10 years. The proportion of children with at least one, two and three components of MetS were 25.0, 5.4 and 0.9%, respectively. Such metabolic abnormalities were also present in children <10 years of age [6].

Results of a study conducted in Iran reported a prevalence of 10.1% in adolescents aged 10–19 years (10.3% in boys and 9.9% in girls), according to modified ATP-III criteria [7].

A nationwide study on Iranian children and adolescents revealed a prevalence of 14.1% for MetS based on ATP-III criteria; this remarkable prevalence may have resulted from the increasing prevalence of overweight and abdominal obesity among Iranian children and adolescents [8, 9].

Recent studies showed that the prevalence of MetS among overweight/obese children is 10–36% based on the modified ATP-III criteria, 16–44% based on the IDF criteria and 23–42% according to the World Health Organization criteria [10].

Despite differences in overall MetS prevalence estimates in various populations, the current literature consistently indicates that the prevalence of MetS increases with age [11, 12]. These studies also have shown that obesity, high blood pressure, elevated triglycerides (TG) and low levels of high-density lipoprotein cholesterol (HDL-C) are cardiovascular risk factors, which are tracked from childhood to adulthood [13, 14].

Although the exact underlying mechanisms of this important syndrome remain to be determined, many cross-sectional and longitudinal studies showed strong correlation between MetS and some factors based on the inflammation, endothelial dysfunction, oxidative stress, hepatic dysfunction and insulin sensitivity [15–19]. Insulin resistance, considered as an essential core of MetS, is suggested to be associated with hematological parameters, especially with white blood cell (WBC) and red blood cell (RBC) counts. However, limited experience exists on systemic evaluation between various components of MetS and WBC or RBC counts in the pediatric age group [20–25].

Furthermore, as WBC and RBC counts are simply performed in almost all laboratories, they can possibly be used to identify the patients at risk of future cardiovascular diseases.

Current guidelines suggest that although screening for MetS is possibly not necessary in the pediatric population as a whole, its screening is necessary in overweight children [26]. Therefore, screening for the components of MetS in overweight children can simplify the monitoring strategies. It also increases awareness, both at the physician level and in the individual member or families, based on combined risks of disorders.

Ethnic differences are reported for features of MetS in the pediatric population [27]. We, therefore, conducted the present study to investigate the association of important constituents’ hematological parameters with components of MetS among a group of Iranian obese children.

Methods

A total of 300 children, aged 6–12 years, were randomly selected from among obese children referred to the Child Growth and Development Research Center, Isfahan University of Medical Sciences, Isfahan, Iran. The obesity was defined as the age- and sex-specific body mass index (BMI) ≥95 percentiles of the Center for Disease Control and Prevention [28] that are confirmed to be acceptable for Iranian children [29].

Participants were referred from schools, health care centers and public-dependent pediatric clinics of Isfahan University of Medical Sciences. The eligibility criteria included the following: age of 6–12 years; individuals with simple obesity; individuals with a medical history of any chronic disease, e.g. hypertension, genetic syndrome, diabetes, a previous malignancy; individuals with a history of chronic medication use or those receiving medication for acute diseases at least 2 weeks before the study; and the WBC count >15 000 cells/mm3 [30] or the hemoglobin >16 mg/dl [2]. This study was approved by the institutional research and ethics review board at the Faculty of Medicine, Isfahan University of Medical Sciences. The informed written consent was obtained from parents and oral assent from children. The health checkup consisted of physical examination, anthropometric measurements, blood chemistry, including low-density lipoprotein cholesterol (LDL-C), HDL-C, TG, total cholesterol, fasting blood sugar (FBG) and cell blood count including WBC and RBC. The weight was recorded in light clothing to the nearest 0.1 kg on a digital weighing scale, and the height was measured without shoes to the nearest 0.1 cm using a Stadiometer (Seca, Tokyo, Japan). The average of two values for each measurement was used in the data analysis. The BMI was calculated from the weight (kg) divided by the height (m2). The waist circumference (WC) was measured to the nearest 0.1 cm over the skin, midway between the tenth rib and the iliac crest. The hip circumference (HiC) was measured at the widest part of the hip, at the level of the greater trochanter, to the nearest 0.5 cm. A blood sample was collected after a 12-h fast using the standard venipuncture technique by expert nurses. The lipid profile was examined by autoanalyzer (Hitachi, Japan) using standard kits (Pars Azmoun, Iran).

Total WBC and RBC were computed by an automated cell counter (Hitachi, Japan). All measurements were made by the same trained team of general physicians and nurses under the supervision of the same pediatrician, by using calibrated instruments and following standard protocols. Blood samples were taken from the ante-cubital vein between 8:00 and 9:30 a.m. The MetS was defined based on the criteria of the National Cholesterol Education Program (ATP-III) [31] customized for the pediatric population: TG >100 mg/dl, HDL-C <50mg/dl, the systolic and/or the diastolic blood pressure >95th percentile, the waist-to-height ratio (WHtR) >0.5 [32] and the FBG >100 mg/dl [33].

Statistical analysis

We first explored the frequency distributions of demographic and clinical characteristics. All data were summarized and displayed as mean ± standard deviation (SD) for the continuous variables.

Independent t-test and Mann–Whitney U test (where assumptions for the t-test could not be met) were used for the comparison of physical and biochemical profiles.

The WBC and the RBC are considered as main outcomes, which we presume to affect each other. So, their relationship needs to be analyzed by a simultaneous equation model.

Fitting Bivariate Poisson regression model is plausible because of considering correlation to investigate the relationship between two important hematological parameters (WBC and RBC) and MetS components. In this model fitting, the WBC and the RBC were assumed as dependent variables, while the MetS components as independent variables, adjusting the effect of other covariates, such as age, BMI, gender, cholesterol, LDL and HiC.

In this article, we present a Hierarchical Bayesian approach to report the Bayes estimates and make inferences of model parameter.

We used Markov chain Monte Carlo (MCMC) algorithms, such as the Gibbs sampling, to find Bayes estimates and their corresponding credible intervals of the Bivariate Poisson regression model. The Gibbs sampling approach iteratively generates samples from the full conditional posteriors for any complex model and, unlike the classical methods, makes statistical inferences without imposing large sample size of data sets. Furthermore, in many clinical studies, some measurements are likely to be missing, and thus usual classical estimation methods need to be extended while the MCMC approach imputes automatically missing values and appropriately analyze the data set. In our empirical study, the data set contained many missing values and the sample size was not large enough to use classical methods. Thus, we essentially suggested fitting the underlying complex model by the use of Gibbs sampling approach in the Bayesian framework. After the algorithm was converged, generated samples were used to make Bayesian inferences of model parameters based on their corresponding sample means, sample SDs and 2.5th and 97.5th percentiles, or 95% high posterior density (HPD) regions, of marginal posteriors.

Data were entered and analyzed using Statistical Package for Social Sciences version 20.0 (SPSS, Inc., Chicago, IL, USA) and OpenBugs 3.2.2, a free software for the Bayesian analysis of complex statistical models using Gibbs sampling approach [34].

To start fitting the Bayesian model, we first adopted a class of conditionally conjugate prior distributions, in the sense that both prior and posteriors belong to the same family but with different components. Then, we provided a code in OpenBugs to implement Gibbs sampling (see Appendix).

Results

Summary statistics of several clinical parameters and anthropometric factors for gender are shown in Table 1. Table includes mean ± SD as descriptive statistics on enrollment variables and significant results of the comparison mean analysis according to different gender. Mean trends for both RBC and RBC counts of categorized features of the MetS are listed in Table 2. The complementary results of Table 2 including different categories of gender variables are presented in supplementary file. In general, the obese children with MetS features had higher RBC or WBC counts. Based on statistical test results, there is no statistical difference between WBC and RBC counts in low and high HDL groups (p = 0.44 and 0.29, respectively) and the same for RBC counts, in different groups of TG (p = 0.89).

Table 1

Demographic and clinical characteristics of study population

Characteristicsa Male (n = 152) Female (n = 148) pb 
Age (years) 8.5 ± 2.7 7.2 ± 1.8 0.001 
BMI (kg/m223.5 ± 0.4 23.4 ± 0.6 NS 
WC (cm) 83.7 ± 7.6 80.2 ± 3.2 0.040 
HiC (cm) 87.4 ± 8.3 87.3 ± 7.9 NS 
SBP (mm Hg) 94.2 ± 11.03 93.1 ± 9.07 NS 
DBP(mm Hg) 58.2 ± 8.3 57.9 ± 9.2 NS 
HDL-C (mg/dl) 106.2 ± 27.5 106.1 ± 26 0.003 
LDL-C (mg/dl) 45.9 ± 10.5 42.1 ± 8.8 NS 
Cholesterol (mg/dl) 177.8 ± 30.2 175.2 ± 27.5 NS 
TGs (mg/dl) 132.5 ± 80.2 133.2 ± 79 NS 
WBC count (×1037260 ± 1902 7272 ± 1901 0.034 
RBC count (×1064.9 ± 0.4 3.8 ± 0.7 0.027 
Characteristicsa Male (n = 152) Female (n = 148) pb 
Age (years) 8.5 ± 2.7 7.2 ± 1.8 0.001 
BMI (kg/m223.5 ± 0.4 23.4 ± 0.6 NS 
WC (cm) 83.7 ± 7.6 80.2 ± 3.2 0.040 
HiC (cm) 87.4 ± 8.3 87.3 ± 7.9 NS 
SBP (mm Hg) 94.2 ± 11.03 93.1 ± 9.07 NS 
DBP(mm Hg) 58.2 ± 8.3 57.9 ± 9.2 NS 
HDL-C (mg/dl) 106.2 ± 27.5 106.1 ± 26 0.003 
LDL-C (mg/dl) 45.9 ± 10.5 42.1 ± 8.8 NS 
Cholesterol (mg/dl) 177.8 ± 30.2 175.2 ± 27.5 NS 
TGs (mg/dl) 132.5 ± 80.2 133.2 ± 79 NS 
WBC count (×1037260 ± 1902 7272 ± 1901 0.034 
RBC count (×1064.9 ± 0.4 3.8 ± 0.7 0.027 

aMean ± SD.

bp-value based on two independent sample t-test or Mann–Whitney test.

FBG, fasting blood glucose; SBP, systolic blood pressure; DBP, diastolic blood pressure.

Table 2

The means of WBC and RBC counts for different components of MetS

Variables Means of WBC (per mm3p Means of RBC (per mm3p 
HBP 
    SBP ≥120 mm Hg and/or DBP ≥80 mm Hg 7368 ± 1980 <0.001 5.22 ± 0.39 <0.001 
    SBP <120 mm Hg and/or DBP <80 mm Hg 6350 ± 887 4.98 ± 0.10 
HFBS 
    FBS ≥100 mg/dl (6.1 mmol/l) 8127 ± 2212 <0.001 4.72 ± 0.36 0.041 
    FBS <100 mg/dl (6.1 mmol/l) 7247 ± 1973 5.04 ± 0.44 
HTG 
    TG ≥100 mg/dl 7581 ± 1982 0.045 5.01 ± 0.33 0.89 
    TG <100 mg/dl 7092 ± 2121 4.91 ± 0.49 
LHDL 
    HDL <50 mg/dl/≥50 mg/dl 7345 ± 1936 0.44 4.98 ± 0.52 0.29 
    HDL ≥50 mg/dl/<50 mg/dl 7269 ± 2185 4.96 ± 0.39 
HWC 
    WHtR ≥0.5 cm 8267 ± 2264 <0.001 5.14 ± 0.27 0.003 
    WHtR <0.5 cm 7296 ± 2016 4.97 ± 0.44 
Variables Means of WBC (per mm3p Means of RBC (per mm3p 
HBP 
    SBP ≥120 mm Hg and/or DBP ≥80 mm Hg 7368 ± 1980 <0.001 5.22 ± 0.39 <0.001 
    SBP <120 mm Hg and/or DBP <80 mm Hg 6350 ± 887 4.98 ± 0.10 
HFBS 
    FBS ≥100 mg/dl (6.1 mmol/l) 8127 ± 2212 <0.001 4.72 ± 0.36 0.041 
    FBS <100 mg/dl (6.1 mmol/l) 7247 ± 1973 5.04 ± 0.44 
HTG 
    TG ≥100 mg/dl 7581 ± 1982 0.045 5.01 ± 0.33 0.89 
    TG <100 mg/dl 7092 ± 2121 4.91 ± 0.49 
LHDL 
    HDL <50 mg/dl/≥50 mg/dl 7345 ± 1936 0.44 4.98 ± 0.52 0.29 
    HDL ≥50 mg/dl/<50 mg/dl 7269 ± 2185 4.96 ± 0.39 
HWC 
    WHtR ≥0.5 cm 8267 ± 2264 <0.001 5.14 ± 0.27 0.003 
    WHtR <0.5 cm 7296 ± 2016 4.97 ± 0.44 

Means of WBC and RBC counts for HBP, high blood pressure; SBP, systolic blood pressure; DBP, diastolic blood pressure; HFBS, high fasting blood sugar; HTG, high triglyceride; LHDL, low HDL cholesterol; HWC, high waist circumstance.

In this study, the influence of several MetS components on the WBC and the RBC simultaneously were investigated using bivariate Bayesian modeling of count data after controlling potential confounding variables.

Posterior estimates of parameters were obtained using the OpenBugs software. For the underlying model, 10 000 iterations were discarded as burn-in sample to eliminate the impact of starting values and then 100 000 iterations with tuning of 10 followed to obtain Bayes estimates (posterior means and SDs) of parameters. Convergence was assessed by visual assessment of the Markov chain for all parameters. Trace plots of samples and Monte Carlo errors was checked (not shown here). As was suggested for the convergence assessment, ratios of the Monte Carlo errors relative to the respective SDs of the estimates were <0.05.

Based on 95% HPD, both RBC and WBC counts were significantly correlated with the TG, FBG and HDL-C, but not with the blood pressure levels. WC was correlated only with the WBC count. According to the sign of regression coefficients, there was strong positive correlation between WBC, RBC and FBG (βWBC = 0.872, SDβ = 0.0561; βRBC = 0.687, SDβ = 0.0123). Furthermore, positive correlation existed between the WBC and the RBC counts with TG (βWBC = 0.0674, SDβ = 0.002; βRBC = 0.0525, SDβ = 0.001), BP (βWBC = 0.0434, SDβ = 0.011; βRBC = 0.068, SDβ = 0.006) and WHtR (βWBC = 0.032, SDβ = 0.001; βRBC = 0.117, SDβ = 0.001), respectively. In addition, RBC counts had significant negative correlation with HDL-C (βRBC = −0.134, SDβ = 0.009).

Discussion

The MetS is ever more important risk factor in cardiovascular disease. It is found to be associated with improved erythropoiesis, and improved some hematological factors [35–38]. The high prevalence of this syndrome in the presence of multiple risk factors among overweight children could lead to a disproportionate increase in cardiovascular disease in adulthood [39].

The related researches proposed that the hyperinsulinemia, one essential element of MetS, probably affects the chronic activation of the immune system and the inflammation [13, 40]. An increasing number of MetS features would be associated with higher degrees of insulin resistance. This finding might show that an increase in the WBC and the RBC counts could be an expression of an insulin-resistant state. According to these findings, we may suggest that an activated cytokine system might contribute to both increased WBC and insulin resistance [41].

Several large population-based samples found that the total WBC counts or the RBC counts were associated with the risk of MetS and its components [42, 43]. For instance, a study revealed that individuals in the highest quartile of WBC or RBC counts received a 3-fold increase in the odds ratio for MetS compared with those in the lowest quartile of WBC or RBC counts [44]. Additionally, other studies reported that WBC counts significantly increased with increasing numbers of MetS components [45]. The current study also showed that the WBC count was positively correlated with the blood pressure, WHtR, TG and FBS, and negatively with the HDL cholesterol. This is similar to the result of some researches that have shown that the WBC count is associated with the TG [46], the blood pressure [21], FBG and WC, as components of the MetS [47]. Some other epidemiological studies noted a relationship between WBC and lipid abnormalities [19, 38]. Illustrating the mechanisms based on the correlation between WBC counts and MetS components, some studies proposed that the insulin resistance probably resulted from a chronic activation of the immune system and the inflammation, especially in obese children [20].

The WBC count is also a documentary marker of inflammatory conditions, such as acute infections and the tissue damage. An increase in the count of various types of inflammatory cells [48, 49] is closely correlated with the incidence and the mortality rate of cardiovascular disease [50], as well as with advanced atherosclerosis [51].

Furthermore, a number of researchers have already noted an association between some components of MetS, such as lipid abnormalities and high blood pressure with RBC counts [22, 23].

Based on the relationship between RBC counts and some components of MetS, there is some evidence suggesting that the possibility of hypothesizing the association between RBC counts and MetS might be due to hyperinsulinemia in individuals with MetS components. These studies have addressed that hyperinsulinemia/insulin resistance is correlated with erythropoiesis in humans [24]. Another reason for linking RBC counts and MetS components might be due to the fact that an increasing RBC mass may decrease insulin resistance with the high body iron storage and the increased accumulation of iron can interfere with insulin-mediated effects [51].

As there are controversies to determine which mechanism is more plausible and strongly contributed according to association between Mets components and blood cell counts, further surveys are needed.

Study limitations

Our study had some limitations. First, the insulin resistance, the key component of Mets, was not measured by laboratory indexes, and thus a direct relationship between insulin sensitivity and RBC or WBC counts was not demonstrated. Second, this study was cross-sectional in design, and thus did not permit the identification of causal relationship between RBC or WBC counts and MetS. Finally, our study population, although with large counts, consisted of participants who regularly underwent physical checkups; it is possible that lifestyle habits, like smoking and psychosocial behaviors, in this population might be different from those in the general population. The main novelty of our research is investigating young children; this may provide information regarding early stages of development of cardio-metabolic risk factors and their relation with the inflammation from early life.

Conclusion

We noted that blood count parameters were significantly and evenly associated with some MetS components among young obese children. These findings are in line with studies conducted among adult population, and serve as confirmatory evidence of the association between surrogate markers of the inflammation with cardio-metabolic risk factors from the early life. The suggested pro-inflammatory state in obese children may lead to higher risk of chronic diseases in later life.

Supplementary Data

Supplementary data are available at Journal of Tropical Pediatrics online.

Acknowledgements

We wish to express our appreciation to all study participants. This study was supported by Isfahan University of Medical Sciences. There are no other relevant financial disclosures.

References

1
Wilson
PW
D'Agostino
RB
Parise
H
, et al.  . 
Metabolic syndrome as a precursor of cardiovascular disease and type 2 diabetes mellitus
Circulation
 , 
2005
, vol. 
112
 (pg. 
3066
-
72
)
2
Expert Panel on Detection, Evaluation, and Treatment of High Bloo Cholesterol in Adults
Executive summary of the third report of the National Cholesterol Education Program (NCEP) expert panel on detection. Evaluation and treatment of high blood cholesterol in adults (Adult Treatment Panel III)
JAMA
 , 
2001
, vol. 
285
 (pg. 
2486
-
97
)
3
Kelishadi
R
Childhood overweight, obesity, and the metabolic syndrome in developing countries
Epidemiol Rev
 , 
2007
, vol. 
29
 (pg. 
62
-
76
)
4
Agirbasli
M
Cakir
S
Ozme
S
, et al.  . 
Metabolic syndrome in Turkish children and adolescents
Metabolism
 , 
2006
, vol. 
55
 (pg. 
1002
-
6
)
5
Kim
HM
Park
J
Kim
HS
, et al.  . 
Prevalence of the metabolic syndrome in Korean adolescents aged 12–19 years from the Korean National Health and Nutrition Examination Survey 1998 and 2001
Diabetes Res Clin Pract
 , 
2007
, vol. 
75
 (pg. 
111
-
14
)
6
Xu
H
Li
Y
Liu
A
, et al.  . 
Prevalence of the metabolic syndrome among children from six cities of China
BMC Public Health
 , 
2012
, vol. 
12
 pg. 
13
 
7
Esmaillzadeh
A
Mirmiran
P
Azadbakht
L
, et al.  . 
High prevalence of the metabolic syndrome in Iranian adolescents
Obesity (Silver Spring)
 , 
2006
, vol. 
14
 (pg. 
377
-
82
)
8
Kelishadi
R
Ardalan
G
Gheiratmand
R
, et al.  . 
Paediatric metabolic syndrome and associated anthropometric indices: the CASPIAN Study
Acta Paediatrica
 , 
2006
, vol. 
95
 (pg. 
1625
-
34
)
9
Barzin
M
Hosseinpanah
F
Fekri
S
, et al.  . 
Predictive value of body mass index and waist circumference for metabolic syndrome in 6-12-year-olds
Acta Paediatrica
 , 
2011
, vol. 
100
 (pg. 
722
-
7
)
10
Friend
AJ
Craig
LCA
Turner
SW
The prevalence of metabolic syndrome in children—a systematic review
Arch Dis Child
 , 
2012
, vol. 
97
 (pg. 
116
-
17
)
11
LinksHajer
GR
van der Graaf
Y
Olijhoek
JK
, et al.  . 
SMART Study Group
Levels of homocysteine are increased inmetabolic syndrome patients but are not associated with an increased cardiovascular risk, in contrast to patients without the metabolic syndrome
Heart
 , 
2007
, vol. 
93
 (pg. 
216
-
20
)
12
Servais
A
Giral
P
Bernard
M
, et al.  . 
Is serum cystatin-C a reliable marker for metabolic syndrome?
Am J Med
 , 
2008
, vol. 
21
 (pg. 
426
-
32
)
13
Karnik
S
Kanekar
A
Childhood obesity: a global public health crisis
Int J Prev Med
 , 
2012
, vol. 
3
 (pg. 
1
-
7
)
14
McGill
HC
Jr
McMahan
CA
Herderick
EE
, et al.  . 
Obesity accelerates the progression of coronary atherosclerosis in young men
Circulation
 , 
2002
, vol. 
105
 (pg. 
2712
-
18
)
15
Lann
D
LeRoith
D
Insulin resistance as the underlying cause for the metabolic syndrome
Med Clin North Am
 , 
2007
, vol. 
91
 (pg. 
1063
-
77
)
16
Onat
A
Uyarel
H
Hergenç
G
, et al.  . 
Serum uric acid is a determinant of metabolic syndrome in a population-based study
Am J Hypertens
 , 
2006
, vol. 
19
 (pg. 
1055
-
62
)
17
Rantala
AO
Lilja
M
Kauma
H
, et al.  . 
Gammaglutamyl transpeptidase and the metabolic syndrome
J Intern Med
 , 
2000
, vol. 
248
 (pg. 
230
-
8
)
18
Marchesini
G
Brizi
M
Bianchi
G
, et al.  . 
Nonalcoholic fatty liver disease: a feature of the metabolic syndrome
Diabetes
 , 
2001
, vol. 
50
 (pg. 
1844
-
50
)
19
Friedman
GD
Teskawa
I
Grimm
RH
, et al.  . 
The leukocyte count: correlates and relationship to coronary risks factors: the CARDIA study
Int J Epidemiol
 , 
1999
, vol. 
19
 (pg. 
889
-
93
)
20
Hansen
LK
Grimm
RH
Neaton
JD
The relationship of white blood cell count to the other cardiovascular risk factors
Int J Epidemiol
 , 
1990
, vol. 
19
 (pg. 
881
-
8
)
21
Nieto
FJ
Szaklo
M
Folsom
AR
, et al.  . 
Leukocyte count in middle-aged adults: the Atherosclerosis Risk in Communities (ARIC) Study
Am J Epidemiol
 , 
1992
, vol. 
136
 (pg. 
527
-
37
)
22
Lowe
G
Rumley
A
Norrie
J
, et al.  . 
Blood rheology, cardiovascular risk factors, and cardiovascular disease: the West of Scotland Coronary Prevention Study
Thromb Haemost
 , 
2000
, vol. 
84
 (pg. 
553
-
8
)
23
Lowik
MR
Odink
J
Kok
FJ
, et al.  . 
Hematocrit and cardiovascular risk factors among elderly men and women (Dutch Nutrition Surveillance System)
Gerontology
 , 
1992
, vol. 
38
 (pg. 
205
-
13
)
24
Shimakawa
T
Bild
DE
Relationship between hemoglobin and cardiovascular risk factors in young adults
J Clin Epidemiol
 , 
1993
, vol. 
46
 (pg. 
1257
-
66
)
25
Barlow
SE
Dietz
WH
Obesity evaluation and treatment: Expert Committee recommendations. The Maternal and Child Health Bureau, Health Resources and Services Administration and the Department of Health and Human Services
Pediatrics
 , 
1998
, vol. 
102
 pg. 
29
 
26
Schwandt
P
Kelishadi
R
Ribeiro
RQ
, et al.  . 
A three-country study on the components of the metabolic syndrome in youths: the BIG Study
Int J Pediatr Obes
 , 
2010
, vol. 
5
 (pg. 
334
-
41
)
27
Kuczmarski
RJ
Ogden
CL
Guo
SS
, et al.  . 
CDC Growth Charts for the United States: methods and development
Vital Health Stat
 , 
2000
, vol. 
2002
 (pg. 
1
-
190
)
28
Schwandt
P
Kelishadi
R
Haas
GM
First reference curves of waist circumference for German children in comparison to international values: the PEP Family Heart Study
World J Pediatric
 , 
2008
, vol. 
4
 (pg. 
259
-
66
)
29
McClatchey
KD
Clinical laboratory medicine
 , 
2002
2nd edn
Philadelphia, PA
Lippincott Williams & Wilkins
(pg. 
797
-
829
)
30
Daniels
SR
Eckel
RH
Hayman
L
, et al.  . 
Progress and challenges in metabolic syndrome in children and adolescents: a scientific statement from the American Heart Association Atherosclerosis, Hypertension, and Obesity in the Young Committee of the Council on Cardiovascular Disease in the Young; Council on Cardiovascular Nursing; and Council on Nutrition, Physical Activity, and Metabolism
Circulation
 , 
2009
, vol. 
119
 (pg. 
628
-
47
)
31
Jasmeet
SM
Sathanur
RS
Pronabesh
D
, et al.  . 
Utility of waist-to-height ratio in assessing the status of central obesity and related cardiometabolic risk profile among normal weight and overweight/obese children: the Bogalusa Heart Study
BMC Pediatr
 , 
2010
, vol. 
10
 pg. 
73
 
32
Cook
S
Auinger
P
Li
C
, et al.  . 
Metabolic syndrome rates in United States adolescents, from the National Health and Nutrition Examination Survey, 1999-2002
J Pediatr
 , 
2008
, vol. 
152
 (pg. 
165
-
70
)
33
Kelishadi
R
Hashemipour
M
Ashtijou
P
, et al.  . 
Association of cell blood counts and cardio-metabolic risk factors among young obese children
Saudi Med J
 , 
2010
, vol. 
31
 (pg. 
406
-
12
)
34
Lunn
DJ
Thomas
A
Best
N
, et al.  . 
WinBUGS a Bayesian modeling framework: concepts, structure, and extensibility
Stat Comput
 , 
2000
, vol. 
10
 (pg. 
325
-
37
)
35
Koeing
W
Ernst
E
Matrai
A
Blood rheology associated with cardiovascular factors and chronic cardiovascular diseases. Results of an epidemiological cross-sectional study
Angiology
 , 
1988
, vol. 
39
 (pg. 
986
-
95
)
36
Mardi
T
Toker
S
Melamed
S
, et al.  . 
Increased erythropoiesis and subclinical inflammation as part of the metabolic syndrome
Diabetes Res Clin Pract
 , 
2005
, vol. 
69
 (pg. 
249
-
255
)
37
Targher
G
Seidell
JC
Tonoli
M
, et al.  . 
The white blood cell count: its relationship to plasma insulin and other cardiovascular risk factors in healthy male individuals
J Intern Med
 , 
1996
, vol. 
239
 (pg. 
435
-
41
)
38
Fernandez-Real
JM
Ricart
W
Insulin resistance and inflammation in an evolutionary perspective: the contribution of cytokine genotype/phenotype to thriftiness
Diabetologia
 , 
1999
, vol. 
42
 (pg. 
1367
-
74
)
39
Pickup
JC
Crook
MA
Is type II diabetes a disease of the innate immune system?
Diabetologia
 , 
1998
, vol. 
41
 (pg. 
1241
-
8
)
40
Stefanini
M
Urbas
JV
Urbas
JE
Gaisbock’s syndrome: its hematologic, biochemical and hormone parameters
Angiology
 , 
1978
, vol. 
29
 (pg. 
520
-
33
)
41
Lao
XQ
Neil Thomas
G
Jiang
C
, et al.  . 
White blood cell count and the metabolic syndrome in older Chinese: the Guangzhou biobank cohort study
Atherosclerosis
 , 
2008
, vol. 
201
 (pg. 
418
-
24
)
42
Kim
DJ
Noh
JH
Lee
BW
, et al.  . 
The associations of total and differential white blood cell counts with obesity, hypertension, dyslipidemia and glucose intolerance in a Korean population
J Korean Med Sci
 , 
2008
, vol. 
23
 (pg. 
193
-
8
)
43
Nakanishi
N
Suzuki
K
Tatara
K
White blood cell count and clustered features of metabolic syndrome in Japanese male office workers
Occup Med
 , 
2002
, vol. 
52
 (pg. 
213
-
18
)
44
Wang
YY
Lin
SY
Liu
PH
, et al.  . 
Association between hematological parameters and metabolic syndrome components in a Chinese population
J Diabetes Complications
 , 
2004
, vol. 
18
 (pg. 
322
-
7
)
45
Grundy
SM
Hypertriglyceridemia, insulin resistance, and the metabolic syndrome
Am J Cardiol
 , 
1999
, vol. 
83
 (pg. 
25
-
9
)
46
Targher
G
Seidell
JC
Tonoli
M
, et al.  . 
The white blood cell count: its relationship to plasma insulin and other cardiovascular risk factors in healthy male individual
J Intern Med
 , 
1996
, vol. 
239
 (pg. 
435
-
41
)
47
Prentice
RL
Szatrowski
TP
Fujikura
T
, et al.  . 
Leukocyte counts and coronary heart disease in a Japanese cohort
Am J Epidemiol
 , 
1982
, vol. 
116
 (pg. 
496
-
509
)
48
Sweetnam
PM
Thomas
HF
Yarnell
JW
, et al.  . 
Total and differential leukocyte counts as predictors of ischemic heart disease: the Caerphilly and Speedwell studies
Am J Epidemiol
 , 
1997
, vol. 
145
 (pg. 
416
-
21
)
49
Kannel
WB
Anderson
K
Wilson
PW
White blood cell count and cardiovascular disease: insights from the Framingham Study
JAMA
 , 
1992
, vol. 
267
 (pg. 
1253
-
6
)
50
Friedman
GD
Klatsky
AL
Siegelaub
AB
The leukocyte count as a predictor of myocardial infarction
N Engl J Med
 , 
1974
, vol. 
290
 (pg. 
1275
-
8
)
51
Barbieri
M
Ragno
E
Benvenuti
E
, et al.  . 
New aspects of insulin resistance syndrome: impact on haematological parameters
Diabetologia
 , 
2001
, vol. 
44
 (pg. 
1232
-
7
)

Appendix

Specification of Bivariate Poisson Regression Model

Let Yij, i = 1,2, … ,N, j = 1,2, be blood cell counts of individual i for the outcome j (RBC and WBC). Assume that the Yij, conditioned on the heterogeneity effects bi1 and bi2, follow Poisson distribution with rate parameter λij and the correlation of Yi1 and Yi2 is nonzero. To operationalize, let  

formula
where xij denotes a k × 1 covariate vector and βj is a k × 1 vector of corresponding parameters. The correlation between two response variables Yi1 and Yi2 is modeled by assuming that unobserved heterogeneity effects bi1 and bi2 follow bivariate normal distribution N2(0, Σ), where Σ is a covariance matrix. The individual probability density function of counts Yi1 and Yi2 is given by  
formula

The marginal log-likelihood function of the above bivariate Poisson-lognormal, is of the form forumla. This cannot be written in closed form and thus, in a Bayesian framework, the MCMC approaches, such as the Gibbs sampling, may be implemented to make statistical inference of model parameters. The Gibbs sampling implementation using the OpenBugs software is given below.

The OpenBugs Code for Fitting Hierarchical Bayesian Analysis of Bivariate Poisson Regression Model

  • m
    odel
    {for(i in 1:N){wbc1[i] ∼
    dpois (theta.w[i])
    theta.w[i] <- exp(lambda. w[i])

  • lambda.w[i] <-
    beta
    [1
    ,
    1
    ]+ beta[
    2
    ,
    1
    ] *sex[i] + beta[
    3
    ,
    1
    ]*waist[i] + beta [
    4
    ,
    1
    ]*hip[i] + beta[
    5
    ,
    1
    ]*lh[i] + beta [
    6
    ,
    1
    ]*ch[i] + beta[
    7
    ,
    1
    ]*th[i] + beta [
    8
    ,
    1
    ]*age[i] + beta[
    9
    ,
    1
    ]*bmi[i] + beta [
    10
    ,
    1
    ]*fbs[i] + alpha[subj[i],1]

  • rbc1[i] ∼ dpois(theta.r[i])

  • theta.r[i] <- exp(lambda.r[i])

  • lambda.r[i] <-
    beta[
    1
    ,
    2
    ]+ beta[
    2
    ,
    2
    ] *sex[i] + beta[
    3
    ,
    2
    ]*waist[i] + beta [
    4
    ,
    2
    ]*hip[i] + beta[
    5
    ,
    2
    ]*lh[i] + beta [
    6
    ,
    2
    ]*ch[i] + beta[
    7
    ,
    2
    ]*th[i] + beta [
    8
    ,
    2
    ]*age[i] + beta[
    9
    ,
    2
    ]*bmi[i] + beta [
    10
    ,
    2
    ]*fbs[i] + alpha[subj[i],2]

  • alpha[i,1:2] ∼ dmnorm(zero[1:2],Tau [1:2,1:2])}

  • for(k1 in 1:10){for(k2 in 1:2) {beta[k1,k2] ∼ dnorm(0,0.01)}}

  • Tau[1:2,1:2] ∼ dwish(R[1:2,1:2],2)

  • Var[1:2,1:2] <- inverse(Tau[1:2,1:2])

Supplementary data

Comments

0 Comments