Factor analysis of metabolic syndrome among the middle-aged Bengalee Hindu men of Calcutta, India

DIABETES/METABOLISM RESEARCH AND REVIEWS R E S E A R C H A R T I C L EDiabetes Metab Res Rev 2005; 21: 58–64.Published online 13 May 2004 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/dmrr.481

Factor analysis of metabolic syndrome among themiddle-aged Bengalee Hindu men of Calcutta, India

Arnab Ghosh*

Department of Anthropology,University of Calcutta, Calcutta, India

*Correspondence to:Dr Arnab Ghosh, Department ofAnthropology, University ofCalcutta, 35, Ballygunge CircularRoad, Calcutta 700019, India.E-mail: arnab [email protected]

Received: 23 May 2003Revised: 10 February 2004Accepted: 17 February 2004

Abstract

Background The prevalence of coronary heart disease (CHD) is knownto be very high in the people of Indian origin. In India, rates are risingand CHD has been predicted to rank first among the causes of death inthe Indian population by 2015. The reasons for the increased susceptibilityof Indians to CHD are yet to be understood completely. However, studieshinted that clustering of risk variables of the metabolic syndrome (MS)could be responsible for the increasing incidence of CHD in the Indians.Therefore, identification of the components of the MS could be one aspectin the way to curb the increasing incidence of CHD among the AsianIndians.

Methods Principal component factor analysis (PCFA) was undertaken toidentify the components or factors of the metabolic syndrome (MS) amongthe middle-aged Bengalee Hindu men of Calcutta, India, and was comparedwith the findings from other studies. The present cross-sectional studyconsisted of 212 Bengalee Hindu men aged 30 years and above. Besidesanthropometric measures, lipid profile, blood pressure, and fasting plasmaglucose (FPG) were collected from each participant. Waist–hip ratio (WHR),trunk–extremity ratio (TER), and central fat skinfold ratio (CFSR) werecomputed accordingly. The lipid profile measures that were included weretotal cholesterol (TC), fasting triglyceride (FTG), high (HDL-C), low (LDL-C),and very low density lipoprotein cholesterol (VLDL-C).

Results Principal components factor analysis revealed four uncorrelatedfactors that cumulatively explained 72.37% of the observed variance of themetabolic syndrome by measured variables. These four factors could beidentified as central obesity (factor 1), centralized subcutaneous fat (factor2), lipid profile blood glucose (factor 3), and blood pressure (factor 4). Thepresent factor analysis confirms the general finding from other factor analysesof the metabolic syndrome on different ethnic groups that have identifiedthree to four factors. Furthermore, the first two factors, that is, central obesityand centralized subcutaneous fat cumulatively explained 47% of the observedvariance of metabolic syndrome in this population.

Conclusion Since more than one factor was identified for the metabolicsyndrome and as no observed variable loaded on all four factors, therefore,more than one physiological mechanism could be accounted for the clusteringof risk variables of the metabolic syndrome among the Bengalee Hindu men.Copyright 2004 John Wiley & Sons, Ltd.

Keywords anthropometry; metabolic syndrome; obesity; central obesity; lipidprofile; glucose; blood pressure; factor analysis; Bengalee Hindu; India

Copyright 2004 John Wiley & Sons, Ltd.

Factor Analysis of Metabolic Syndrome 59

Introduction

The distribution of body fat is as important as that of lipidsand glucose in predicting mortality and morbidity fromcardiovascular diseases, including coronary heart disease(CHD). In epidemiological and population-based studies,the simple, non-invasive technique of anthropometricmeasures such as circumferential measurements andsubcutaneous skinfold thickness are used to assessadiposity, topography of adiposity, and their relationshipwith metabolic risk factors (e.g. blood lipids, glucose etc.)of CHD [1–12]. Central obesity, as measured by waistcircumference (WC), waist–hip ratio (WHR) conicityindex (CI), and so on, is a major contributor to thedevelopment of the ‘Metabolic Syndrome’ (MS), whichhas been defined as the clustering of cardiovasculardiseases (e.g. CHD), risk factors such as dyslipidaemia,hypertension, glucose intolerance, and hyperinsulinaemia[6,13–15].

The prevalence of CHD is known to be very highamong Indians, both in India and abroad. Indians whoare settled in the United States have a fourfold higherprevalence than Caucasian Americans and a sixfold higherhospitalization rate than Chinese Americans [16]. In Indiaalso, rates are rising and CHD has been predicted to rankfirst among the causes of death in the Indian populationby 2015 [17]. Moreover, in Indians, CHD occurs at least adecade or two earlier than that seen in the Europeans[3,16,18]. The reason for the increased susceptibilityof Indians to CHD is yet to be understood completely.However, studies hinted that clustering of risk variables(mechanism of which is still unknown to us) of the MScould be responsible for the increasing incidence of CHDamong the Indians. This includes glucose intolerance,central obesity, hypertriglyceridaemia, dyslipidaemia, andincreased level of very low density lipoprotein cholesterol(VLDL-C) [3,18–21]. A recent report stated that in Asianpopulations, mortality and morbidity is occurring withlower body mass index (BMI) [22]. Thus, they tend toaccumulate intra-abdominal fat (central obesity) withoutdeveloping generalized obesity (e.g. BMI). The MS isassociated with striking tendency to central obesity inthe south Asian (e.g. Indian) men. However, they areno more overweight than the European men and have amore centralized distribution of body fat, with thick trunkskinfold and markedly higher central obesity for a givenlevel of body fat [3].

Various statistical techniques could be utilized toidentify the components of the metabolic syndrome.Principal component factor analysis (PCFA) is one suchapproach that groups quantitatively measured variablesinto clusters known as factors on the basis of thecorrelation between variables [23]. PCFA was used toidentify the domains of risk variables of the MS. Forexample, if there is a single underlying cause of theclustering of risk variables of the MS, then factor analysisshould produce only one major factor or component.Therefore, identification of component(s) of the MS

(considered to be the leading cause for CHD) is ofutmost necessity for the etiology of CHD. So far asIndia is concerned, very few studies have so far beenundertaken to identify the components of the MS in theIndian population. But to the best of my knowledge, nosuch study has been undertaken on Bengalee Hindus.In view of the above consideration, factor analysis wasundertaken among the middle-aged Bengalee Hindu menof Calcutta, India, to identify the components of the MSand compare them with the findings from similar studiesconcerning Indians as well as other ethnic groups in theworld.

Materials and methods

Study population

The sample of the present cross-sectional study comprises212 male railway employees of the Eastern Railway,Government of India. They belonged to the BengaleeHindu population and were aged 30 years and above. All212 subjects were residents of Calcutta and its suburbs.The study was conducted during the period December1999 to January 2001 at the outpatient department(OPD) of B.R.Singh Hospital, Eastern Railway, Calcutta,India. All the participants in the present study werepart of a ‘health examination programme’ (HEP) initiatedduring this period as part of the collaboration betweenthe Department of Anthropology, University of Calcuttaand the Department of Pathology, B.R.Singh Hospital,Eastern Railway, Calcutta. Prior to the commencementof the programme, written information regarding aimsand objectives of the HEP was served to different officialdepartments of Eastern Railway situated in Calcutta andits suburbs through a proper channel. In addition, theprogramme had clearly mentioned the criteria of selectionof an individual to this programme. In response, 230Bengalee Hindu men who were aged 30 years and abovehad contacted the OPD. The actual age of participantswas calculated from the date of birth. At the end ofthe study, the author, however, was able to examinea total of 220 male individuals under this programme.Eight [8] individuals were later excluded because ofmissing data. In the present study, therefore, the totalsample size was 212 middle-aged Bengalee Hindu maleindividuals. All 212 individuals were free from CHD (ECGwas performed on all participants). Persons affected withCHD or who at least had a medical history of CHDwere purposely excluded from the study. Out of the totalsubjects, 32% and 22% were hypertensive (in which almost40% were under medication and the rest were under dietchange/control) and diabetic (in which 39% were undermedication) respectively. Written consent about theirwillingness to participate in the programme was obtainedfrom all participants prior to the actual commencement ofthe study. The recorder (AG) interviewed all participantsincluded in the present study in the OPD of the Hospitalin the forenoon session. All 212 individuals were engaged

Copyright 2004 John Wiley & Sons, Ltd. Diabetes Metab Res Rev 2005; 21: 58–64.

60 A. Ghosh

in non-manual works and generally led sedentary lives.Only 32% (n = 67) individuals were smokers. Almost allthe participants (n = 204) were non-vegetarian. As faras physical exercise was concerned, almost 70% of theparticipants were reluctant to do any kind of physicalexercise (e.g. brisk walking, jogging, swimming etc). Outof the total participants, 75% individuals (n = 158) werefound to have changed their diet. But most interestingly,out of these 158 individuals, only 32% individuals(n = 68) were found to consume low fat, low salt,and low sugar-containing diets. Both anthropometric andmetabolic variables were collected from each participanton the same day.

Anthropometric measurements

Height, weight, circumferences of waist (WC) and hip,as well as skinfold thickness at subscapular, suprailiac,abdomen, chest, midaxillary, biceps, triceps, forearm,anterior thigh, and medial calf were collected using astandard technique [24] by the author himself. Heightand weight were measured to the nearest 0.1 cm and0.5 kg respectively. Waist and hip circumferences weremeasured with an inelastic tape to the nearest 0.2 cm. AHoltain skinfold caliper was used to measure skinfolds tothe nearest 0.2 mm. BMI and WHR were computed usingthe standard equation.

Trunk–extremity ratio or TER was calculated using thefollowing equation:

TER =∑

(subscapular + suprailiac + abdomen + chest

+ midaxillary)/∑

(biceps + triceps + forearm

+ anteriorthigh + medial calf)

Central fat skinfold ratio or CFSR was calculated usingthe following equation [25]:

CFSR = (subscapular + abdomen)/2/(medial calf

+ triceps + anterior thigh)/3

Metabolic variables

A fasting blood sample was collected from each subjectfor the determination of metabolic variables, namely,total cholesterol (TC), fasting triglyceride (FTG), fastingplasma glucose (FPG), high (HDL-C), low (LDL-C), andvery low density lipoprotein cholesterol (VLDL-C). Thedetailed procedure of lipid profile and plasma glucoseanalyses has been mentioned elsewhere [26].

All biochemical analyses were done at the biochemistryunit of the Department of Pathology, B.R. Singh Hospital,Calcutta. All metabolic variables were measured in mg/dLunit and then converted into mmol/L unit using standardconversion formulae.

Blood pressure variable

Left-arm blood pressure was taken from each participantwith the help of an Omron M1 digital electronic bloodpressure/pulse monitor (Omron Corporation, Tokyo,Japan). Two forenoon blood pressure measurements weretaken and averaged for analysis. A third measurementwas only taken when the difference between thetwo measurements was greater than 5 mmHg andsubsequently the mean was calculated. A 5-min relaxationperiod between measurements was maintained for all212 individuals. The working condition of the instrumentwas checked periodically with the help of a mercurysphygmomanometer and a stethoscope (auscultatorprocedure). All blood pressure measurements were takenin a quiet room and at room temperature. Systolicblood pressure (SBP) and diastolic blood pressure (DBP)were defined as the points of the appearance (Phase I)and disappearance (Phase V) of the Korotkoff soundsrespectively.

Statistical analyses

Descriptive statistics such as mean, standard deviation,and coefficient of skewness were calculated for allthe variables. Factor analysis was then undertaken togroup quantitatively measured variables into clustersknown as factors on the basis of the correlationbetween the variables. It was done in three steps:computation of a correlation matrix for all variablesincluded, factor extraction, and orthogonal rotation tomake factors readily interpretable. Since factor analysisallows incorporating age as covariate, age was includedin the factor model as covariate. Factors were extractedby principal component analysis (PCA) in which linearcombinations of the variables were formed, with the firstprincipal component accounting for the largest amountof variance in the sample. Varimax rotation, which is anorthogonal rotation in which the factors are assumed toact independently (maximum likelihood), was used inthe study. The components were all uncorrelated. Factorloading, which was equivalent to Pearson’s correlationcoefficient between each variable and the factor, wasused to interpret each factor. Variables with factorloading of at least 0.3 have generally been considered forinterpretation, although it is suggested that only loadinggreater than or equal to 0.4, which therefore shares atleast 15% of variance with the factor, should be usedin interpretation [27]. Previous studies have also useda factor loading of 0.4 or greater to interpret the finalrotated factor pattern [12,27]. A factor loading of 0.4or greater was used to interpret the final rotated factorpattern in the present study. Extracted factors (extractionby PCA) were such that each explained at least as much(eigenvalue ≥ 1) or nearly as much variance as any oneobserved variable (eigenvalue = 1).

All statistical analyses were performed using SPSS,Version 10 package. A p value of <0.05 was consideredas statistically significant.



Results

The distributions of all variables were checked fornormality. Log [10] transformation was undertaken tonormalize the distribution of positively skewed variables(TC, FTG, FPG, HDL-C, and VLDL-C). On the otherhand, square root log transformation was undertakento normalize the distribution of negatively skewedvariable (WHR). The mean, standard deviation (SD),and skewness of anthropometric, obesity, metabolic,and blood pressure variables are presented in Table 1.Bivariate correlation of the traits examined in the MSis presented in Table 2. The factor-loading pattern offour factors (components) identified is presented inTable 3. Only variables with loading greater than orequal to 0.4 were considered for interpretation ofresults. After varimax rotation, central obesity measure,namely the waist–hip ratio, loaded positively on factor1, whereas centralized subcutaneous fat, as measured bytrunk–extremity ratio, loaded positively on factor 2. Threemetabolic variables, namely TC, FTG, and FPG, groupedpositively on factor 3 and SBP and DBP grouped positivelyon factor 4. The loading of individual risk variablevaried from 0.55 to 0.96. These four factors explained72% of the total variance of the metabolic syndromein which the first two factors, that is, central obesity andcentralized subcutaneous fat, cumulatively explained 47%of the total variance as explained by four factors in thestudy population. Additional analysis using the variables,typically considered part of the MS (i.e. WC, BMI, FTG,HDL-C, FPG, SBP and DBP), was also undertaken. Theinclusion and exclusion of subcutaneous fat measures,namely TER as input variable in PCFA, revealed that(results were not shown) TER had significantly alteredthe outcomes of the two procedures, that is, PCFA withTER (72% explanation) and PCFA without TER (64%explanation).

Discussion

Even with the availability of so many modern techniqueslike magnetic resonance imaging (MRI), computerized

Table 1. Descriptive statistics of anthropometric, obesity, meta-bolic and blood pressure variables in the present studypopulation (n = 212)

Variable Mean SD Skewness

Age (in years) 50.7 10.1 0.08Waist circumference (WC) 87.2 6.6 −0.26(cm)Hip circumference 90.2 4.9 0.45(cm)Body mass index (BMI) 23.8 2.8 0.15(Kg/m2)Waist–hip ratio (WHR)a 0.96 0.04 −1.94Trunk–extremity ratio (TER) 1.69 0.18 0.34Central fat skinfold ratio (CFSR) 1.48 0.18 0.10Metabolic variables (mmol/L)Total cholesterol (TC)a 5.5 0.91 0.67Fasting triglyceride (FTG)a 2.2 1.02 1.40Fasting plasma glucose (FPG)a 6.2 1.07 1.07High density lipoprotein cholesterol(HDL-C)a

1.24 0.12 2.18

Low density lipoprotein cholesterol(LDL-C)

3.2 0.70 0.41

Very low density lipoprotein cholesterol(VLDL-C)a

1.0 0.55 1.40

Blood pressure variables (mmHg)Systolic blood pressure (SBP) 132.6 18.3 0.50Diastolic blood pressure (DBP) 81.9 10.0 0.22

aSignificantly skewed.

axial tomography (CAT), bioelectrical impedance analysis(BIA), and dual energy X-ray absorptiometry (DEXA),anthropometry still is the most universally applicable,inexpensive, and non-invasive method available to assessthe size, proportion, and composition of the humanbody [28]. The use of anthropometry as proxies orindicators of a state, condition, or risk is a well-established and time-honored concept in the domain ofbiomedical research and has been used extensively inepidemiologic and pathophysiologic research involvingobesity, over weight, body fat distribution, and healthoutcomes [29]. The association of central obesity,glucose intolerance, hypertension, dyslipidaemia, andhyperinsulinaemia, known as MS, has been observed ina number of ethnic groups worldwide. Studies across thepopulation demonstrated that MS occupies a pivotal rolein the occurrence of cardiovascular diseases, including

Table 2. Intercorrelation matrix of variables considered in the Metabolic syndrome

Variable Age WHR BMI TER TC FTG FPG HDL SBP DBP

Age – 0.04 −0.08 0.05 −0.04 −0.13 0.18 0.02 0.23 0.01WHR – 0.46 0.26 0.16 0.16 0.24 −0.12 0.05 0.18BMI – 0.26 0.03 0.07 0.09 −0.07 0.02 0.22TER – 0.05 0.16 0.13 −0.08 0.06 0.13TC – 0.32 0.04 0.31 0.09 0.13FTG – 0.12 0.21 0.13 0.04FPG – 0.12 0.17 0.14HDL – 0.09 −0.02SBP – 0.77DBP –

BMI, body mass index; WHR, s-hip ratio; TER, trunk–extremity ratio; TC, total cholesterol; FTG, fasting triglyceride; FPG, fastingplasma glucose; HDL-C, high-density lipoprotein cholesterol; SBP, systolic blood pressure; DBP, diastolic blood pressure.Significant at 5% level when correlation coefficient is >0.14. 1% level when correlation coefficient is >0.17. 0.1% level whencorrelation coefficient is >0.19.


62 A. Ghosh

Table 3. Factor loading by principal component analysis withvarimax rotation of risk variables in the Metabolic syndrome(n = 212)

Variables Factor 1 Factor 2 Factor 3 Factor 4

Age 0.212 0.216 0.302 0.324BMI 0.210 0.140 0.068 0.295Waist–hip ratioa 0.828c 0.192 0.063 0.155Trunk–extremity ratio 0.245 0.734c 0.041 0.061Total cholesterolb 0.140 0.083 0.551c 0.147Fasting triglycerideb 0.055 0.138 0.932c 0.030Fasting plasma glucoseb 0.255 0.117 0.574c 0.268High-density lipoproteinb 0.146 0.104 0.226 0.052CholesterolSystolic blood pressure 0.038 0.055 −0.017 0.969c

Diastolic blood pressure 0.113 0.062 0.048 0.949c

Variance explained (%) 26.0 21.0 14.0 11.0Cumulative variance (%) 26.0 47.0 61.0 72.0

aSquare root log transformed values were used.bLog (10) transformed values were used.cLoading with absolute value ≥0.4.

CHD. Therefore, identification of the components ofphenotypes of the MS and how its phenotypic expressiondiffers across the ethnic groups would be helpful inunderstanding the etiology of CHD. Various statisticaltechniques could be applied to examine the associationbetween phenotypes (e.g. dyslipidaemia, hypertension,glucose intolerance, etc.) of the MS. Principal componentfactor analysis (PCFA) is one such approach to identifythe aforesaid association. As far as India is concerned,

very little work has so far been undertaken to identifythe underlying factors/components of the MS amongthe Indians. But no such work has been undertaken onBengalee Hindus. In view of the above consideration, thepresent work was undertaken among the middle-agedBengalee Hindu men of Calcutta, India.

PCFA had identified four factors with 72% explainedvariance of the MS among the Bengalee Hindus ofCalcutta. This means that four separate physiologicalpathways underlie the clustering of phenotypes of MS.This finding was consistent with the presence of severaldistinct physiological domains, as had been found inother ethnic groups. Most importantly, neither of thevariables (phenotypes) loaded on all four components.These four factors could be identified as central obesity(factor 1), centralized subcutaneous fat (factor 2), lipidprofile blood glucose (factor 3), and blood pressure(factor 4). The lack of overlap of central obesity measure,as evident in the study, was also observed in otherstudies concerning the Asian Indians [27,30,31]. Fourdistinct cluster domains with almost 70% explanationwere also observed in south Indian non-diabetic malesubjects [30]. Three major clusters of cardiovasculardisease risk variables with almost 65% explanation werereported by another study [31] among the south Indianmen (Table 4). The present factor analysis confirmed thegeneral findings from other factor analyses of the MSon different ethnic groups that had identified three to

Table 4. Summary of the published studies on the Metabolic syndrome using principal component factor analysis (PCFA) among themale cohort

SampleFactors

Authors Population size F I F II F III F IV

Meigs et al. (1997) FraminghamOffspring study

1150 FI, PI, BMI WHR,HDL, TG

FPG, PPG,FI, PI SBP, DBP–

(26.5%) (18.8%) (17.1%)Gray et al. (1998) American Indians

(native)975 BMI, FPG FI SBP, DBP TG, HDL-C

–(35%) (24%) (13%)

Edwards et al. (1998) Japanese-Americans 2760 Wt, WC, FI SBP, DBP TG, HDL-C FPG, FI(25%) (20%) (18.7%) (13.4%)

Ramachandran et al.(2000)

South Indiannon-diabeticSubjects

1196 BMI, WHR, TC, TG BMI, SBP, DBP FPG, 2 h PG WHR, familyhistory

(28%) (15.3%) (14.2%) (12%)Snehalatha et al. (2000) South Indian

non-diabeticSubjects

396 2 hPG, 2 h Ins, IR,BMI

BMI, SBP, DBP BMI, WHR, TC, TG–

(30.4%) (20.4%) (13.8%)Hodge et al. (2001) Multi-ethnic

population ofMauritius (70% AsianIndian origin)

1414 WHR, BMI, Leptin,fasting&2hinsulin, TG, HDL-C

SBP, DBP Uric acid Fasting & 2 hglucose andinsulin

–

(31%) (14%) (10%)Young et al. (2002) Canadian Population

(Native)3830 BMI, WC, HC TC, SBP, DBP TG, HDL-C, FPG

–(29.2%) (22.1%) (17.7%)

Present study Bengalee Hindus ofCalcutta, India

212 WHR TER TC, FTG, FPG SBP, DBP

(25.9%) (20.9%) (14.2%) (11.3%)

BMI, body mass index; WHR, waist–hip ratio; TER, trunk–extremity ratio; SBP, systolic blood pressure; DBP, diastolic blood pressure; TC, totalcholesterol; FTG, fasting triglyceride; FPG, fasting plasma glucose; HDL-C, high-density lipoprotein cholesterol; IR, insulin resistance; PG, plasmaglucose; 2 h, two hours; FI, fasting insulin; Wt, weight; WC, waist circumference; HC, hip circumference; PI, post load insulin.



four factors [12,27,32–36]. Furthermore, in the presentstudy, the first two factors, that is, central obesity(factor 1) and centralized subcutaneous fat (factor 2),cumulatively explained 47% of total variance of theMS in the study population. These two types of fatdistribution were significantly and positively associatedwith lipids, lipoproteins, glucose, and blood pressureacross the ethnic groups. Therefore, the above two typesof adiposity can predispose Bengalee Hindus to MS. It isnoteworthy to mention here that when restricted to themore conventional variables (e.g. WC, BMI, FTG, HDL-C,FPG, SBP, and DBP), PCFA did produce three uncorrelatedfactors with 64% explanation. The inclusion of skinfoldsmeasure (i.e. TER), on the other hand, had identified fouruncorrelated factors with 72% explanation. Increasingbody of evidence suggested that thick trunk skinfolds withcentral obesity for a given level of BMI is the characteristicfeature of the Asian Indians [3,22]. Therefore, inclusionof skinfolds in PCFA is absolutely necessary. The lackof overlapping of the WC across the factors was quiteunexpected because central obesity has been consideredas the major risk factor of CHD among the people of southAsian origin (e.g. Indians). Furthermore, no overlappingof variables on more than one factor vindicated that akey role for any one of these variables in the MS is veryunlikely.

In a study among the multiethnic population ofMauritius (including the people of Asian Indian origin),three uncorrelated factors were identified with nooverlapping of central obesity. The three factors werecentral metabolic syndrome, hypertension, and glucosetolerance [27]. This finding was generally similar withthat of the four-factor model revealed in the presentstudy. ‘Framingham Offspring Study’ had also identifiedthree factors and was quite similar to the finding fromBengalee Hindus of Calcutta [34]. The ‘Strong HeartStudy’ conducted on the Native American Indians alsoidentified three factors in non-diabetic subjects (Table 4).These were glucose, obesity, and hyperinsulinaemia-bloodpressure-hyperlipidaemia [35]. Three factors (obesity,blood pressure, and lipid-glucose), which togetheraccount for 64% of the variance, were also identifiedamong the three native Canadian populations [12].

However, it is noteworthy to mention that resultsfrom different factors’ analysis is limited by differencesin the race, sex, and age composition of the studysamples, the number of risk variables included, samplesize, and the cut-off points of loadings set by theinvestigators. At the same time, to the best of myknowledge, no PCFA of metabolic syndrome has beenundertaken incorporating data on apolipoprotein B genepolymorphism, for example, XbaI, EcorI, and so on, andlipoprotein lipase (LPL) gene polymorphism, for example,HindIII polymorphism and so on (both apolipoprotein andLPL modulates the plasma lipid concentration).

Owing to vast ethnic and cultural heterogeneity inthe Indian population, further research is needed onother ethnic groups residing in rural and urban Indiato see whether the trend observed in the present study

is also observed in these groups. At the same time, theauthor could not do the same work on Bengalee womenbecause of the cultural limitation in taking measurementsfrom females by male investigators. Lastly, it shouldbe mentioned that cross-sectional observation like thepresent one can only highlight the differences in riskfactor profiles; they do not shed detailed insight into theetiology of chronic diseases like CHD. Furthermore, theIndian diaspora offers a unique opportunity to study the‘gene-environment’ interaction involved in the etiology ofCHD. Comparison of results of local populations (Indiansin India) with that of migrants settled elsewhere shouldyield valuable information on the ethnic susceptibility ofIndians to CHD.

Acknowledgements

The author is grateful to Dr J.Chattopadhyay, Dr G. Das Gupta,Dr S. Sengupta, and Dr S. Batyabal of the Department ofPathology, B.R.Singh Hospital, Calcutta, for their cooperationduring the data collection. The author is also grateful to thestaff and technicians for their help in analyzing the metabolicvariables. This work is financially supported by the UniversityGrant Commission, Department of Special Assistance (PHASE-II), Government of India.

References

1. Donahue RP, Abbott RD, Bloom E, Reed DM, Yano K. Centralobesity and coronary heart disease in men. Lancet 1987; 2:821–824.

2. Shimokata H, Tobin JD, Muller DC, Elahi E, Coon PJ, Andress R.Studies in the distribution of body fat: effect of age, sex andobesity. J Geron (Medical Science) 1989; 44: M66–M73.

3. McKeigue PM, Shah B, Marmot MG. Relation of central obesityand insulin resistance with high diabetes prevalence andcardiovascular risk in South Asians. Lancet 1991; 337: 382–386.

4. Mueller WH, Wear ML, Hanis CL, et al. Which measures of bodyfat distribution is best for epidemiological research? Am JEpidemiol 1991; 133: 858–869.

5. Seidell JC. Environmental influences on regional fat distribution.Int J Obes 1991; 15: 31–35.

6. Valdez R, Seidell JC, Ahn YI, Weiss KM. A new index ofabdominal adiposity as indicator of risk for cardiovasculardiseases: a cross population study. Int J Obes 1993; 17: 77–82.

7. Han TS, Van Leer EM, Seidell JC, Lean MEJ. Waistcircumference action levels the identification of cardiovascularrisk factors: prevalence study in a random sample. Br Med J1995; 311: 1401–1405.

8. Seidell JC, Bouchard C. Visceral fat in relation to health: is it amajor culprit or simply an innocent bystander? Int J Obes 1997;21: 626–631.

9. Kim KS, Robbins D, Turner M, Adams-Campbell LL. Anthropo-metric determinants of risk factor in an African Americanpopulation. Am J Hum Biol 1998; 10: 249–258.

10. Lean MEJ, Han TS, Seidell JC. Impairment of health and qualityof life in people with large waist circumference. Lancet 1998;351: 853–856.

11. Gustat J, Elkasabany A, Srinivasan S, Berenson GS. Relation ofabdominal height to cardiovascular risk factors in young adults.The Bogalusa heart study. Am J Epidemiol 2000; 151: 885–891.

12. Young TK, Chateau D, Zhang M. Factor analysis of ethnicvariation in the multiple metabolic (insulin resistance)syndromes in three Canadian populations. Am J Hum Biol 2002;14: 649–658.

13. Kopelman PG. Obesity as medical problem. Nature 2000; 404:635–643.


64 A. Ghosh

14. Zimmet P, Alberti KGMM, Shaw J. Global and societalimplications of the diabetes epidemic. Nature 2001; 414:782–787.

15. Ramachandran A, Sathyamurthy I, Snehalatha C, et al. Riskvariables for coronary artery disease in Asian Indians. Am JCardiol 2001; 87: 267–271.

16. Enas EA. Coronary artery disease epidemic in Indians: a case foralarm and call for action. Indian J Med Assoc 2000; 98: 694–702.

17. Gupta R, Gupta VP. Meta-analysis of coronary heart diseaseprevalence in India. Indian Heart J 1996; 48: 241–245.

18. Enas EA, Yusuf S, Mehta JL. Prevalence of coronary arterydisease in Asian Indians. Am J Cardiol 1992; 70: 945–949.

19. Gupta R, Prakesh H, Majumdar S, Gupta VP. Prevalence ofcoronary heart disease and coronary risk factors in an urbanpopulation of Rajasthan. Indian Heart J 1995; 47: 331–338.

20. Rajmohan L, Deepa R, Mohan V. Risk factors for coronary arterydisease in Indians: emerging trends. Indian Heart J 2000; 52:221–225.

21. Misra A, Sharma R, Pandey RM, Khanna N. Adverse profile ofdietary nutrients, anthropometry and lipids in urban slumdwellers of northern India. Eur J Clin Nutr 2001; 55: 727–734.

22. WHO/IASO/IOTF. The Asia-Pacific Perspective: RedefiningObesity and its Treatment. Health Communication: Australia,2000.

23. Stevens J. Applied Multivariate Statistics for the Social Sciences.Lawrence Erlbaum: Mahwah: NJ, 1996.

24. Lohman TG, Roche AF, Martorell R (eds). AnthropometricStandardization References Manual. Human Kinetics: Chicago,1988.

25. Baumgartner RN, Roche AF, Chumlea WC, Siervogel RM,Glueck CJ. Fatness and fat patterning: association with plasmalipids and blood pressure in adults, 18 to 57 years of age. Am JEpidemiol 1987; 126: 614–628.

26. Ghosh A, Bose K, Das Chaudhuri AB. Association of foodpatterns, central obesity measures and metabolic risk factors forcoronary heart disease (CHD) in middle aged Bengalee Hindumen, Calcutta, India. Asia Pac J Clin Nutr 2003; 12: 166–171.

27. Hodge AM, Boykot EJ, Courlen Mde, et al. Leptin and othercomponents of the metabolic syndrome in Mauritius–a factoranalysis. Int J Obes 2001; 25: 126–131.

28. Wildman REC, Medieros DM. Advanced Human Nutrition. CRCPress: London, 2000.

29. Seidell JC, Kahn HS, Willamson DF, Lissner L, Valdez R. Reportfrom Centers for Disease Control and Prevention workshop onuse of adult anthropometry for public health and primary healthcare. Am J Clin Nutr 2001; 73: 123–126.

30. Ramachandran A, Snehalatha C, Satyavani K, Sivasankari S,Vijay V. Cosegregation of obesity with familial aggregation oftype 2 diabetes mellitus. Diabetes Obes Metab 2000; 2: 149–154.

31. Snehalatha C, Sivasankari S, Satyavani K, Vijay V, Ramachan-dran A. Insulin resistance alone does not explain the clusteringof cardiovascular risk factors in southern India. Diabet Med 2000;17: 152–157.

32. Chen W, Srinivasan SR, Elkasabany A, Berenson GS. Cardiovas-cular risk factors clustering factors of insulin resistance syndrome(Syndrome X) in a biracial (Balck White) population of chil-dren, adolescent and young adults. Am J Epidemiol 1999; 150:667–674.

33. Lindblad U, Langer RD, Wingard DL, Thomas RG, Barett-Connor EL. Metabolic syndrome and ischaemic heart diseasein elderly men and women. Am J Epidemiol 2001; 153:481–489.

34. Meigs J, D’Agostino RSr, Wilson P, Cupples L, Nathan D,Singer D. 1997. Risk variables clustering in the insulin resistancesyndrome. The Framingham Offspring Study. Diabetes 1997; 46:1594–1600.

35. Gray R, Fabsitz R, Cowan L, Lee E, Howard B, Sanger P. Riskfactor clustering in the insulin resistance syndrome. The StrongHeart Study. Am J Epidemiol 1998; 148: 869–878.

36. Edwards L, Burnchfild CM, Sharp DS, et al. Factors of theinsulin-resistance syndrome in non-diabetic and diabeticelderly Japanese- American men. Am J Epidemiol 1998; 147:441–447.


Documents

Factor analysis of metabolic syndrome among the middle-aged Bengalee Hindu men of Calcutta, India