Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Factor analysis of the metabolic syndrome: obesity vs insulin resistance as the central abnormality


OBJECTIVES: To evaluate whether there is one central abnormality contributing to the conditions associated with the metabolic syndrome (MES), or whether one abnormality is contributing on multiple levels.

METHODS: We recruited 145 Chinese subjects aged 17–68 y with varying degrees of insulin-sensitivity (IS): 33 healthy, 59 with type 2 diabetes mellitus, 32 essential hypertensives and 21 dyslipidaemics. IS was evaluated by the short insulin sensitivity test using a 0.1 U/kg intravenous bolus dose of insulin. Blood pressure, anthropometric measures and biochemical parameters associated with IS were also measured. Exploratory factor analyses (EFA) were performed in the entire group of 145 subjects and in the 76 with normal glucose tolerance.

RESULTS: EFA in all 145 subjects defined three distinct, independent factors. Factor 1 was interpreted as general and central adiposity, impaired IS and glucose intolerance, Factor 2 was associated with hypertension and general and central obesity, whilst Factor 3 was strongly related to low HDL-cholesterol and high triglyceride concentrations and weakly to waist circumference. In patients with impaired glucose tolerance, only two factors were identified; factor 1 related to reduced IS, impaired glucose tolerance, dyslipidaemia and general and central adiposity, and factor 2 which was related to blood pressure and general and central adiposity.

CONCLUSIONS: These models suggest that the clustering of variables in MES is a result of multiple factors linked by adiposity and not a single aetiology. Furthermore, increases in blood pressure are related to obesity in these Chinese subjects rather than decreased IS per se.


It was Reaven who first proposed that insulin resistance was at the centre of a syndrome characterised by a clustering of metabolic abnormalities associated with increased cardiovascular risk, impaired glucose tolerance, type 2 diabetes mellitus (DM), dyslipidaemia, hypertension and obesity.1 The concept that these abnormalities might all be facets of one syndrome has sparked much controversy over which single factor amongst the many inter-related variables is present in all the conditions of metabolic syndrome (MES) and provides the link that unifies them. By far the most popular candidate is insulin resistance,2,3,4 although tissue concentrations of triglycerides5 and general and central obesity6,7,8,9 have also been proposed.

The complex nature of the MES presents considerable methodological challenges to researchers in the field. The high degree of intercorrelation between the variables that comprise this constellation of abnormalities makes clear definition of their individual contributions difficult, if not impossible. A number of investigators have dealt with their data by considering the syndrome as a whole rather than assessing the effects of individual components. The statistical methods required to examine data in such a way are complex, and the results of such analyses are arguably equivocal. However, they provide powerful tools for hypo-thesis testing and illuminating complex relationships that might otherwise be obscured in a large set of variables that are highly interrelated. While statistical methods such as multiple regression are useful in predicting independent variables, it is not possible to define a central abnormality predisposing towards others (if one exists) by using such methods. Furthermore, multiple regression equations become unreliable when there is a high degree of correlation between the dependent variables (multi-collinearity).

A number of groups have examined the clustering of variables in MES in Caucasian populations by using exploratory factor analysis (EFA).4,10,11,12 However, the results have not been consistent. Whilst all studies identify insulin resistance as a major factor, some results suggest that it is directly linked to dyslipidaemic and hypertensive traits,4 others suggest that obesity provides the connection between the insulin-resistant, dyslipidaemic and hypertensive factors,12 and some indicate that the factors are distinct.10,11 The hypothesis that hyperinsulinaemia and insulin resistance are central to the MES was developed in Caucasian populations.13,14 Furthermore, with the exception of Donahue et al,4 most such studies rely solely on insulin and glucose concentrations and do not include an insulin sensitivity test.

Members of our group have previously reported results of structural equation modelling in a large population of Hong Kong Chinese.15 This analysis suggested that hyperinsulinaemia is not central to the syndrome, as has been reported in Caucasians.15 The purpose of the present study was to measure insulin sensitivity and examine the clustering of metabolic variables associated with MES in a heterogeneous population of Hong Kong Chinese, in an attempt to show which metabolic aberrations predominate in this population. Also, whether the same abnormality may be contributing on more than one level. The Hong Kong Chinese are at high-risk of developing the conditions associated with MES and the prevalence of type 2 diabetes mellitus (type 2 DM) and obesity has been increasing in this population at an alarming rate.16,17 A better understanding of the syndrome and possible ethnic differences in this population is therefore of prime importance.



The study was approved by the Ethics Committee of the Chinese University of Hong Kong. All subjects gave their written, informed consent. A heterogeneous population of 145 Chinese individuals with varying degrees of insulin resistance was recruited from four groups, healthy (n=33), type 2 DM (n=59), essential hypertensive (n=32) and dyslipidaemic (n=21), aged 17–68 y. After baseline investigations were completed, many were found to have one or more of the other abnormalities. Type 2 DM and impaired glucose tolerance (IGT) were diagnosed according to the 1985 World Health Organization Criteria using a standard oral glucose tolerance test (OGTT) with 75 g of glucose monohydrate.18 Hypertension was diagnosed as a resting systolic blood pressure (SBP) ≥140 mmHg and/or a diastolic blood pressure (DBP) of ≥90 mmHg, or a requirement for antihypertensive medications within the preceding 12 months. Dyslipidaemia was classified as fasting plasma concentrations of low-density lipoprotein cholesterol (LDL-cholesterol) >3.4 mmol/l, high-density lipoprotein cholesterol (HDL-cholesterol)<1.0 mmol/l or triglycerides (TG)>2.0 mmol/l, the limits defined as normal by the Department of Chemical Pathology of The Prince of Wales Hospital, Hong Kong. Obesity was defined as a body mass index (BMI) ≥27 kg/m2 in males and ≥25 kg/m2 in females, according to standard criteria.19 The subjects were also classified for the presence or absence of the metabolic syndrome based on the criteria published by Taskinen et al.20 According to these criteria, the subjects were said to have MES if they had at least three of the parameters known to be associated with insulin resistance, namely obesity, hyperinsulinaemia, glucose intolerance or type 2 DM, dyslipidaemia or hyperuricaemia. Whenever possible, the subjects underwent a 6-week washout period during which they were medication free. However, in cases where this was not ethically possible, details of their medications were recorded and they did not take their medications on the morning of study. In total 57 subjects were on medication which could not be stopped, 39 were on antihypertensives, 38 on oral hypoglycaemics, and three on lipid-lowering agents.


In all subjects, insulin sensitivity, anthropometric measures and biochemical parameters known to be associated with MES were measured. Waist circumference (cm) was measured mid-way between the lower rib margin and the iliac crest and adjusted for stature by height (cm). Hip circumference (cm) was measured at the level of the femoral trochanters and the waist-to-hip ratio calculated. Insulin sensitivity was determined by a frequently sampled short insulin tolerance test (SITT), which we have described previously,21 using a 0.1 U/kg intravenous bolus dose of Actrapid (HM) insulin (Novo Nordisk, 2880 Bagsvaerd, Denmark). The slope of the linear part of the glucose concentration vs time curve divided by fasting plasma glucose (FPG) was taken as the index of insulin sensitivity (kla). Supine blood pressure measurements were made with a Critikon Dinamap 8100 automatic sphygmomanometer after the subject had been resting for 10 min. The mean of three readings taken 3 min apart was used for analysis.


Fasting plasma insulin (FPI) was measured by radioimmunoassay (Pharmacia, Sweden) with an intra- and inter-assay coefficient of variation of 6 and 14%, respectively. FPG was assayed with a glucose oxidase technique using a Sidekick Glucose Analyser (YSI Inc., Yellow Springs, Ohio, USA), the inter- and intra-assay coefficients of variation (CVs) were both less than 2%. Plasma total cholesterol and TG were assayed enzymatically (DuPont Medical Products, Newark, DE, USA). HDL-cholesterol was determined after fractional precipitation with dextran sulphate–MgCl2 and LDL-cholesterol was calculated by Friedwald's formula.22 Uric acid was measured by the uricase method (Dimension, Uric Acid Flex, Dade International Inc., Newark, DE, USA) and total bilirubin by the Doumas reference method (Dimension, Total Bilirubin Flex, Dade International Inc., Newark, DE, USA); both assays had CVs of less than 2%.


Exploratory factor analysis (EFA) with Varimax rotation was used to determine whether the clustering of metabolic variables observed in MES is due to a single factor or multiple factors in this group of subjects. Only variables with a factor loading of at least 0.3 (sharing at least 10% of the variance with a factor) were used for interpretation. The number of eigenvalues greater than 1.0 determined the number of factors.

The variables to be included in the analysis were chosen with the help of a univariate correlation matrix and their communality score (the amount of variance shared with the other variables) once included in the model. Factor analysis is only justified if the variables used have a high communality. For these reasons, mean arterial pressure was used in place of SBP and DBP. In addition, waist circumference was chosen because it correlated less highly with BMI than waist-to-height ratio and because it had a higher communality than waist-to-hip ratio. All the parameters included in the models were first adjusted for age and gender through multiple regression analysis. The Kaiser–Meyer–Olkin (KMO) statistic was used as a measure of sampling adequacy and the Bartlett test of sphericity was used as a measure of the necessity to perform a factor analysis.23,24,25 All analyses were performed on an IBM compatible PC using the Statistical Package for the Social Sciences (SPSS) for Windows, version 6.1 (SPSS Inc., Chicago, IL, USA, 1994).


Factor analysis of the full data set

Table 1 gives the population characteristics in the entire group of 145 subjects and in the subjects with normal glucose tolerance. Initially, factor analysis was undertaken with the data from the full group of 145 Chinese subjects, including healthy volunteers and patients with IGT and type 2 DM. The model selected contained eight variables—FPI, FPG, BMI, MAP, HDL-cholesterol, waist circumference, TG and kla. Inclusion of further lipid parameters, heart rate, plasma urate and bilirubin concentrations was abandoned because the variables either failed the inclusion criteria or failed to provide additional information. The eight parameters chosen were associated with three distinct, independent factors which together explained 73% of the variance in the data set. The three factors and the factor loadings for each variable are shown in Table 2. Figure 1 gives a graphic representation of the three factors and their proposed links. The first, most influential factor explaining 43% of the total variance in the data set can be interpreted as general and central adiposity, impaired insulin sensitivity and glucose intolerance. The second factor is closely associated hypertension and general and central obesity. Raised plasma TG and low HDL-cholesterol load very highly on the third factor, and waist circumference is weakly associated. The sampling adequacy of the model was high (KMO=0.74) and the Bartlett test of sphericity was highly significant (452, P<0.00001), indicating good model acceptability. According to this analysis, insulin resistance, dyslipidaemia and hypertension are determined by three distinct, independent factors and are not facets of a single syndrome. However, the association of obesity, either general or central, with all three factors links them together.

Table 1 The population characteristics of all 145 subjects and of the 76 subjects with normal glucose tolerance (percentage or mean±s.d.)
Table 2 The three factors underlying the clustering of metabolic markers in 145 Chinese patients with varying degrees of metabolic derangement
Figure 1

A diagrammatic representation of the factor analysis performed in the entire group of 145 subjects.

Factor analysis of subjects with normal glucose tolerance

As the relationship between glucose and insulin concentrations in patients with DM or IGT is often no longer proportional, the above analysis was repeated excluding all patients with IGT or type 2 DM. The characteristics of this study population are also given in Table 1. In the remaining 76 subjects with normal glucose tolerance, slightly different results were obtained, (see Table 3 and Figure 2). Only two factors were identified, explaining a total of 60% of the total amount of variance. The first was associated with IGT, reduced insulin sensitivity, dyslipidaemia and general and central adiposity, and the second was associated with hypertension and general and central adiposity. Dyslipidaemia was included with the parameters related to insulin resistance, but once again hypertension was associated with an independent process, linked to insulin resistance through obesity. The sampling adequacy of this model was also high (KMO=0.75) and the Bartlett test of sphericity was 215, P<0.00001, indicating good model acceptability. To test whether all variables would load on one factor if blood pressure were not included in the model, MAP was excluded and the analysis was re-run. Only one factor was identified with which all the variables were strongly correlated (see Table 3).

Table 3 The two factors underlying the clustering of metabolic markers in 76 subjects with varying degrees of metabolic derangement, but normal glucose tolerance
Figure 2

A diagrammatic representation of the factor analysis performed in the group of 76 subjects without type 2 diabetes mellitus or normal glucose tolerance.


If the abnormalities related to MES were indeed ‘spokes on a wheel’ with one central abnormality at the hub,26 then the factor analyses conducted on this data set would be expected to identify one major factor underlying the clustering of metabolic variables. Conversely, three factors were identified when the data from all subjects were used. The first factor, which could be interpreted as IGT and insulin sensitivity, and two others associated with dyslipidaemia and hypertension, respectively. Both general and central obesity were strongly associated with the first two factors and central obesity was weakly associated with the third, dyslipidaemic factor. This result implies that at least three processes underlie the clustering of these variables and that obesity is the link that unifies the syndrome. However, when subjects with IGT were excluded from the analysis, only two factors were identified: one which was closely associated with IGT, dyslipidaemia and obesity, and one which could be interpreted as increased blood pressure in association with both central and general obesity. When MAP was excluded from the analysis, only one factor was identified. These models suggest two things: firstly that hypertension is not a component of the insulin resistance syndrome; secondly, that in Chinese subjects with normal glucose tolerance, increases in fasting plasma lipids, insulin and glucose concentrations and impaired insulin sensitivity may be due to a single underlying factor. The variable that loaded most highly on that single factor was central adiposity, as measured by waist circumference, and the second most important variable was increased BMI. Whether single or multiple causative factors are present, these models suggest that the unifying abnormality is obesity. It is of interest that when patients with diabetes or abnormal glucose tolerance were excluded from the analysis, the factor closely associated with dyslipidaemia was not defined and the variables associated with it loaded on the primary factor, related to impaired insulin sensitivity and glucose tolerance and obesity. The explanation for this may lie with the role of non-esterified fatty acid (NEFA) concentrations in inducing insulin resistance and glucose intolerance. NEFA concentrations are known to impair insulin-mediated glucose uptake by competing with glucose at insulin sensitive tissue sites, ie adipose tissue and skeletal muscle.27 This leads to hyperglycaemia and insulin resistance.28,29,30,31 However, in diabetic subjects in whom insulin resistance may already be present or insulin levels are insufficient to induce a glucose flux into skeletal muscle, NEFA concentrations do not affect glucose uptake at these sites.31,32 Nonetheless, elevated concentrations will increase triglyceride-rich VLDL synthesis in the liver and induce dyslipidaemia.33,34 Plasma NEFA concentrations could not be added to the factor analysis because results were not available for all the subjects. However, NEFA concentrations could link dyslipidaemia and insulin resistance in individuals with normal glucose tolerance, a link that might be absent in a group containing a large number of diabetics. Further studies are planned to measure NEFA concentrations in the different sub-groups of MES, in an attempt to resolve these issues. It is worthy of note that the variables related to glycaemic control (kla, FPI and FPG) still load on a single factor with the lipid and obesity variables, even in individuals with normal glucose tolerance. This illustrates that these relationships operate on a continuum that is evident before the ‘cut-off’ values normally associated with overt pathology. The age- and sex-adjusted partial correlation matrix for these variables bears this out.

A number of studies have employed factor analysis to examine the MES in larger, Caucasian study populations. Donahue et al4 performed a euglycaemic hyperinsulinaemic clamp in 50 participants from the Miami Community Health Study. Twenty-seven members of the cohort were non-Hispanic Caucasian and 23 were African-American. Standard cardiovascular and anthropometric variables were also measured. They hypothesised that if insulin resistance (M) unifies the metabolic disturbances of MES, then it should load on all of the factors identified. If hyperinsulinaemia is also important, it should load on one but not all factors. Factor analysis yielded two factors when all variables were included: one factor that was identified by increases in insulin resistance (M), uric acid, SBP and DBP, TG, waist circumference and reduced HDL-cholesterol; and one that was identified with increased M, FPI, FPG, blood pressure and waist. The analyses were stratified separately by sex and ethnicity and showed similar results. The authors conclude that, as insulin resistance loaded on both factors, it could be the unifying variable for MES. However, blood pressure and obesity also loaded on both factors. No measures of model adequacy were given, and the sample size was small. Edwards et al10 examined data from 281 non-diabetic women and included both fasting and post-load glucose and insulin concentrations, weight, waist, TG, HDL-cholesterol and LDL-cholesterol particle diameter in the analysis. Three factors were identified: one closely related to hyperinsulinaemia, hyperglycaemia and general and central adiposity; one related to increases in both fasting and post-load insulin and glucose concentrations; and one associated with greater LDL-cholesterol particle diameter, increased TG and reduced HDL-cholesterol concentrations. The fact that fasting insulin and glucose concentrations loaded on two factors was taken as evidence of their importance.

Wingard et al12 identified four factors in both non-diabetic men (n=606) and women (n=765) from a data set which was similar in all respects to that of Edwards et al, except that LDL particle diameter was not measured. Pre- and post-load insulin values loaded on separate factors and dyslipidaemia and blood pressure were also related to distinct processes. Meigs et al11 examined 2458 non-diabetic subjects from the Framingham Offspring Study, again using the same variables, with pre- and post-load insulin and glucose concentrations. Three factors were identified, with the major factor relating to general and central obesity, increased fasting and 2 h-insulin concentrations, raised TG and low HDL-cholesterol concentrations. The second factor was related to fasting and 2 h-insulin and glucose concentrations and the third to increased SBP, DBP and BMI.

While all these studies agree that the clustering of metabolic variables in MES is a result of multiple factors and not a single aetiology, the factors defined are not uniform. Most of these studies did not include a test of insulin sensitivity or resistance (other than fasting insulin and glucose) and all were conducted in Caucasian or African-American populations, making a direct comparison with this work difficult. As previously mentioned, EFA cannot be used to test hypotheses of cause and effect; this is in the realm of confirmatory factor analysis (also known as structural equation modelling). An extension of factor analysis, structural equation modelling tests hypotheses about independent causal pathways between the variables. One such analysis has been performed by members of our group in 1513 Hong Kong Chinese subjects.15 Subjects with impaired renal function or who were being treated with anti-hypertensive or anti-diabetic drugs were excluded. In spite of the high degree of inter-correlation among the variables, the results of the model suggested that there was a central mechanism behind changes in vascular and metabolic parameters in these subjects. These changes were influenced most strongly by obesity (as measured by BMI and WHR) and ageing. However, a positive family history of diabetes was also a mediating factor, albeit to a lesser extent. The model also suggested that hyperinsulinaemia was not a central mediating factor and only had a causal relationship with blood pressure. However, age and obesity explained much of the variance in blood pressure. Insulin concentrations were in turn determined by obesity, a family history of diabetes, FPG and TG concentrations. Interestingly, an inverse relationship was found between a positive family history of diabetes and blood pressure. Thus, the results of this structural equation model agree with exploratory factor analysis performed in all 145 subjects which also identified general (BMI) and central (waist circumference) obesity as the parameters which linked the hypertension, dyslipidaemia and IGT facets of MES. The fact that FPI was only weakly associated with the factors identified as dyslipidaemia (low HDL-cholesterol, high TG, and general and central adiposity, r=0.13) and hypertension (raised MAP and waist circumference, r=0.17) belies a strong role for hyperinsulinaemia per se in the aetiology of these conditions. This is further supported from work in both monozygous and dizygous Caucasian twin pairs, which found that the association of insulin with other components of the metabolic syndrome was mainly mediated by non-genetically acquired obesity.35

In summary, the results of EFA in this heterogeneous group of Hong Kong Chinese subjects suggest that it is obesity (both central and general) that is the common link between the major facets of MES. Insulin sensitivity was only associated with one of the three factors underlying the syndrome, albeit the most important one. Blood pressure was related to a distinct factor, regardless of whether subjects with IGT were included in the analysis, only linked to the factor associated with insulin resistance and IGT through obesity. While these analyses are purely exploratory in nature, they provide valid hypotheses on which to base future work. Given the immense complexity of the biological systems controlling glucose homeostasis, lipid metabolism and blood pressure, and the plethora of genetic mutations that are known to produce distinct phenotypes in all these conditions, it is clearly simplistic to look for a single unifying mechanism. Add to this protean syndrome the further complication of potential ethnic variations in all of its components, and it becomes clear that there can be no simple solution to such a conundrum. Internationally agreed guidelines on the definition and diagnosis of MES would help to identify more specific populations in which to examine the mechanisms concerned. However, EFA is a useful tool with which to test hypotheses on which abnormalities should be included in this definition.


  1. 1

    Reaven GM . Banting Lecture 1988: Role of insulin resistance in human disease Diabetes 1988 37: 1595–1607.

    CAS  Article  Google Scholar 

  2. 2

    Defronzo RA, Ferrannini E . Insulin resistance: a multifaceted syndrome responsible for NIDDM, obesity, hypertension, dyslipidemia, and atherosclerotic cardiovascular disease Diabetes Care 1991 14: 173–194.

    CAS  Article  Google Scholar 

  3. 3

    Després JP, Marette A . Relation of components of insulin resistance syndrome to coronary disease risk Curr Opin Lipidol 1994 5: 274–289.

    Article  Google Scholar 

  4. 4

    Donahue RP, Bean JA, Donahue RD, Goldberg RB, Prineas RJ . Does insulin resistance unite the separate components of the insulin resistance syndrome? Evidence from the Miami Community Health Study Arterioscler Thromb Vasc Biol 1997 17: 2413–2417.

    CAS  Article  Google Scholar 

  5. 5

    Koyama K, Chen G, Lee Y, Unger R . Tissue triglycerides, insulin resistance, and insulin production: implications for hyperinsulinaemia of obesity Am J Physiol 1997 273: E708–E713.

    CAS  PubMed  Google Scholar 

  6. 6

    Larson B . Obesity, fat distribution and cardiovascular disease Int J Obes 1991 15: 53–57.

    Google Scholar 

  7. 7

    Arauz-Pacheco C, Ramirez LC, Schnurr-Breen L, Raskin P . Relationship between insulin sensitivity and degree of obesity in mild hypertension Am J Med Sci 1992 304: 225–230.

    CAS  Article  Google Scholar 

  8. 8

    Björntorp P . Abdominal obesity and the development of non-insulin-dependent diabetes mellitus Diabetes Metab Rev 1988 4: 615–622.

    Article  Google Scholar 

  9. 9

    Després JP . Obesity and lipid metabolism: relevance of body fat distribution Curr Opin Lipidol 1991 2: 5–15.

    Article  Google Scholar 

  10. 10

    Edwards KL, Austin MA, Newman B, Mayer E, Krauss RM, Selby JV . Multivariate analysis of the insulin resistance syndrome Arterioscler Thromb 1994 14: 1940–1945.

    CAS  Article  Google Scholar 

  11. 11

    Meigs JB, D'Agostino RB Sr, Wilson PW, Cupples LA, Nathan DM, Singer DE . Risk variable clustering in the insulin resistance syndrome Diabetes 1997 46: 1594–1600.

    CAS  Article  Google Scholar 

  12. 12

    Wingard D, Von Muhlen D, Barrett-Connor E, Kritz-Silverstein D . Factor analysis of proposed components of the insulin resistance syndrome Diabetes 1996 45: 137A.

    Google Scholar 

  13. 13

    MacDonald IA . Advances in our understanding of the role of the sympathetic nervous system in obesity Int J Obes Relat Metab Disord 1995 19 (Suppl 7): S2–S7.

    PubMed  Google Scholar 

  14. 14

    Modan M, Halkin H, Fuchs Z, Lusky A, Chetrit A, Segal P, Eshkol A, Almog S, Shefi M . Hyperinsulinemia: a link between glucose intolerance, obesity, hypertension, dyslipoproteinemia, elevated serum uric acid and internal cation imbalance Diabetes Metab 1987 13: 375–380.

    CAS  Google Scholar 

  15. 15

    Chan JCN, Cheung JCK, Lau EMC, Woo J, Swaminathan R, Cockram CS . The metabolic syndrome in Hong Kong Chinese—the inter-relationships amongst its components analysed by structural equation modeling Diabetes Care 1996 19: 953–959.

    CAS  Article  Google Scholar 

  16. 16

    Cockram CS, Woo J, Lau E, Chan JC, Chan AY, Lau J, Swaminathan R, Donnan SP . The prevalence of diabetes mellitus and impaired glucose tolerance among Hong Kong Chinese adults of working age Diabetes Res Clin Pract 1993 21: 67–73.

    CAS  Article  Google Scholar 

  17. 17

    Janus ED, Cockram CS, Lam KSL, Lam TH . The prevalence of diabetes mellitus and impaired glucose tolerance in the Hong Kong population 3rd International Diabetes Federation-Western Pacific Regional Congress Hong Kong 1996 90.

    Google Scholar 

  18. 18

    World Health Organization . Diabetes mellitus Report of a WHO study group. Technical reports series no. 727 WHO: Geneva 1985

    Google Scholar 

  19. 19

    National Diabetes Data Group . Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance Diabetes 1979 28: 1039–1057.

    Article  Google Scholar 

  20. 20

    Taskinen M-R . Strategies for the diagnosis of metabolic syndrome Curr Opin Lipidol 1993 4: 434–443.

    CAS  Article  Google Scholar 

  21. 21

    Young RP, Critchley JAJH, Anderson PJ, Lau MSW, Lee KKC, Chan JCN . The short insulin tolerance test: feasibility using venous sampling Diabetic Med 1996 13: 429–433.

    CAS  Article  Google Scholar 

  22. 22

    Friedewald WT, Levy RI, Fredrickson DS . Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge Clin Chem 1972 18: 499–502.

    CAS  Google Scholar 

  23. 23

    McClendon MJ . Multiple regression and causal analysis F.E. Peacock: Itasca, IL 1994

    Google Scholar 

  24. 24

    Bernstein IH, Garbin CP, Teng GK . Applied multivariate analysis Springer: New York 1988

    Book  Google Scholar 

  25. 25

    Flury B, Riedwyl H . Multivariate statistics. A practical approach Chapman and Hall: London 1988

    Book  Google Scholar 

  26. 26

    Hopkins PN, Hunt SC, Wu LL, Williams GH, Williams RR . Hypertension, dyslipidaemia, and insulin resistance: links in a chain or spokes on a wheel? Curr Opin Lipidol 1996 7: 241–253.

    CAS  Article  Google Scholar 

  27. 27

    Randle PJ, Hales CN, Garland PB, Newsholme EA . The glucose fatty-acid cycle: its role in insulin sensitivity and the metabolic disturbances of diabetes mellitus Lancet 1963 ii: 785–789.

    Article  Google Scholar 

  28. 28

    Chen YDI, Golay A, Swislocki ALM, Reaven GM . Resistance to insulin suppression of plasma free fatty acids concentrations and insulin stimulation of glucose uptake in NIDDM J Clin Endocrinol Metab 1987 64: 17–21.

    CAS  Article  Google Scholar 

  29. 29

    Warnotte C, Gilon P, Nenquin M, Henquin J-C . Mechanisms of the stimulation of insulin release by saturated fatty acids Diabetes 1994 43: 703–711.

    CAS  Article  Google Scholar 

  30. 30

    Bevilacqua S, Bonadonna R, Buzzigoli G, Boni C, Ciociaro D, Maccari F, Giorico MA, Ferrannini E . Acute elevation of free fatty acid levels leads to hepatic insulin resistance in obese subjects Metabolism 1987 36: 502–506.

    CAS  Article  Google Scholar 

  31. 31

    Ferrannini E, Barrett EJ, Bevilacqua S, Defronzo RA . Effect of fatty acids on glucose production and utilization in man J Clin Invest 1983 72: 1737–1747.

    CAS  Article  Google Scholar 

  32. 32

    Saloranta C, Franssila-Kallunki A, Ekstrand A, Taskinen M-R, Groop L . Modulation of hepatic glucose production by non-esterified fatty in acids in type 2 (non-insulin-dependent) diabetes mellitus Diabetologia 1991 34: 409–415.

    CAS  Article  Google Scholar 

  33. 33

    Unger RH . Lipotoxicity in the pathogenesis of obesity-dependent NIDDM Diabetes 1995 44: 863–870.

    CAS  Article  Google Scholar 

  34. 34

    Després JP, Moorjani S, Lupien PJ, Tremblay A, Nadeau A, Bouchard C . Regional distribution of body fat, plasma lipoproteins, and cardiovascular disease Arteriosclerosis 1990 10: 497–511.

    Article  Google Scholar 

  35. 35

    Mayer EJ, Newman B, Austin MA, Zhang D, Quesenberry CP Jr, Edwards K, Selby JV . Genetic and environmental influences on insulin levels and the insulin resistance syndrome: an analysis of women twins Am J Epidemiol 1996 143: 323–332.

    CAS  Article  Google Scholar 

Download references

Author information



Corresponding author

Correspondence to PJ Anderson.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Anderson, P., Critchley, J., Chan, J. et al. Factor analysis of the metabolic syndrome: obesity vs insulin resistance as the central abnormality. Int J Obes 25, 1782–1788 (2001).

Download citation


  • insulin resistance
  • hypertension
  • dyslipidaemia
  • type 2 diabetes mellitus
  • Chinese

Further reading


Quick links