The metabolic syndrome: time to get off the merry-go-round?


Gerald M. Reaven, MD, Division of Cardiovascular Medicine, Stanford University School of Medicine, Falk CVRD, Stanford Medical Center, 300 Pasteur Drive, Stanford, CA 94305, USA.
(fax: 650 725 1599; e-mail:


Abstract.  Reaven GM (Stanford University School of Medicine). The metabolic syndrome: time to get off the merry-go-round? (Review) J Intern Med 2011; 269: 127–136.

The diagnostic category of the metabolic syndrome (MetS) has received considerable attention over the last decade, and prestigious organizations continue to strive for its incorporation into medical practice. This review has three goals: (i) summarize the history of the several attempts to define a diagnostic category designated as the MetS; (ii) question the aetiological role of abdominal obesity in the development of the other components of the MetS; and (iii) evaluate a diagnosis of the MetS as an effective way to identify apparently healthy individuals at increased risk to develop cardiovascular disease (CVD) or type 2 diabetes (2DM). The most important conclusion is that the MetS seems to be less effective in this population than the Framingham Risk Score in predicting CVD, and no better, if not worse, than fasting plasma glucose concentrations in predicting 2DM.


On 26 September 2010, inserting the phrase ‘metabolic syndrome’ in PubMed identified 32 131 publications. Adding the modifier of ‘cardiovascular’, ‘diabetes’, or ‘prevalence’ identified c. 10 000 additional publications in each of these three categories, and there were 1150 citations listed under the category of ‘metabolic syndrome definition’. Although it would be interesting to evaluate the biological information present in this enormous list of publications, it is not the goal of this review. Instead, the focus will be on three aspects of the diagnostic category referred to as the ‘metabolic syndrome’ (MetS); (i) its history; (ii) relationship between obesity and the other components of the MetS; and (iii) the diagnostic and/or pedagogic utility of making a diagnosis of the MetS.

History of the MetS

Insulin-mediated glucose uptake by muscle varies more than 600% in apparently healthy individuals [1], with approximately 50% of the variability in insulin action [2] resulting from differences in degree of adiposity (25%) and physical fitness (25%). The remaining 50% is likely to be of genetic origin, with powerful familial and ethnic influences [3, 4]. Type 2 diabetes (2DM) develops when insulin-resistant individuals cannot secrete the increased amounts of insulin needed to compensate for the insulin resistance [5–7]. However, the majority of insulin-resistant individuals are able to maintain the degree of hyperinsulinemia required to prevent manifest decompensation of glucose homeostasis. Although compensatory hyperinsulinemia prevents the development of frank hyperglycaemia in insulin-resistant persons, insulin-resistant/hyperinsulinaemic individuals are at greatly increased risk of being somewhat glucose intolerant, with a dyslipidaemia characterized by a high plasma triglyceride (TG) and low high-density lipoprotein cholesterol (HDL-C) concentration, and an increase in blood pressure [5, 8]. These changes increase cardiovascular disease (CVD) risk, and because the importance as CVD risk factors of insulin resistance/compensatory hyperinsulinemia and its associated cluster of abnormalities was not widely appreciated at the time, the term Syndrome X was introduced in 1988 to focus attention on these relationships [5].

Ten years later, the World Health Organization (WHO) proposed a set of clinical criteria for making a diagnosis of the MetS [9]. To qualify for this diagnosis, it was necessary to be insulin resistant by euglycemic clamp criteria, or have 2DM, impaired glucose tolerance or impaired fasting glucose (IFG). In addition, any two of the following abnormalities needed to be present: obesity (abdominal or overall), dyslipidaemia [high TG or low HDL-C)] concentration, elevated blood pressure, or microalbuminuria.

Although the diagnostic criteria for the MetS bear a strong similarity to the components of Syndrome X, there are three major differences. The most fundamental difference is that Syndrome X was a pathophysiological construct, attempting to explain why minor degrees of glucose intolerance, a high TG and low HDL-C, and essential hypertension tend to cluster in the same individual. In contrast, The WHO proposal was an effort to make a diagnosis with which to identify persons at undue risk for CVD. Secondly, Syndrome X did not include 2DM as a component because it was an attempt to explain why CVD was increased in insulin-resistant individuals who were not diabetic, but able to maintain normal, or near-normal, glucose tolerance by virtue of compensatory hyperinsulinemia. The third difference was the inclusion of obesity as one of the components of the MetS, considered as being similar to dyslipidaemia, glucose intolerance and hypertension. Microalbuminuria was also viewed by the WHO as being a fundamental manifestation of insulin resistance to the same degree as the other components.

The second entry into the MetS sweepstakes arrived in 2001 with the publication by the Adult Treatment Panel III (ATP III) of the National Cholesterol Education Program of their criteria for diagnosing the Mets [10]. The components used by the ATP III to diagnose the MetS are similar to the cluster of changes that comprise Syndrome X and used by the WHO in their definition of MetS. However, in contrast to the WHO, the ATP III did not identify one essential criterion, but specified five criteria: abdominal obesity, as estimated by an increased waist circumference (WC); a high TG and a low HDL-C concentration, elevated blood pressure, and glucose intolerance (prediabetes or 2DM). An individual meeting any three of the aforementioned criteria merits the ATP III diagnosis of the MetS. The ATP III did not include microalbuminuria as a criterion, and abdominal obesity is the only acceptable index of excess adiposity.

The International Diabetes Federation (IDF) climbed on board the MetS merry-go-round in 2005, modestly proposing a ‘new worldwide definition’ of the MetS [11]. The IDF criteria included the same five components as the ATP III version, but like the WHO, indicated that one abnormality had to be present to diagnose the MetS; namely, abdominal obesity, as assessed by measuring WC. An enlarged WC, and any two of the remaining four components, suffices to diagnose the MetS.

The appearance of the IDF version of the MetS was met with enthusiasm by many investigators, and a substantial number of studies have been published comparing the ability of the ATP III and IDF versions of the MetS to accomplish a variety of tasks. On the other hand, the lack of a ‘gold standard’ definition of the MetS was of concern to other individuals who felt that extensive use of the MetS as a clinical entity was hampered without generally agreed upon diagnostic criteria. Fortunately, for the latter group, the ATP III and the IDF, joined by several other prestigious organizations, have proposed a ‘harmonized’ definition of the MetS [12]. The details of this version of the MetS are listed in Table 1, and several issues should be noted. As before meeting any three of the criteria is sufficient for the diagnosis. However, an enlarged WC is no longer an essential abnormality and has reverted to being just one of five equally important components. Given this decision, it is somewhat paradoxical to have the authors of the ‘harmonized’ definition state in another publication [13] that ‘Evidence now indicates that the metabolic syndrome all begins with central obesity’. It should also be noted that the cut points for an elevated WC are not the same for all population groups, and ref. [12] contains a second table that contains population-specific values. Finally, drug treatment is sufficient to meet the criteria for the other four components.

Table 1. ‘Harmonized’ criteria for diagnosing the metabolic syndrome [12]
MeasureCut points
  1. HDL-C, high-density lipoprotein cholesterol.

Elevated waist circumferencePopulation and country-specific
Elevated triglycerides≥1.7 mmol L−1
Low HDL-CMen (<1.0 mmol L−1); women (<1.3 mmol L−1
Elevated blood pressureSystolic ≥130 and/or diastolic ≥85 mmHg
Elevated fasting glucose≥100 mg dL−1

Despite the enormous number of publications devoted to the MetS and the enthusiastic support of the prestigious scientific organizations involved in the ‘harmonization’ of the MetS, belief that a diagnosis of the MetS is useful is not shared by all. For example, the American Diabetes Association and the European Association for the Study of Diabetes issued a joint report in 2004 [14] that was highly critical of the notion of the MetS as a diagnostic category. Instead, they suggested that ‘providers should avoid labeling patients with the term metabolic syndrome’. They went on to argue that ‘adults with any major CVD risk factor should be evaluated for the presence of other CVD risk factors’, and ‘that all CVD risk factors should be individually and aggressively treated’. In this context, it seems only fitting that a recent report from the WHO complete the circle [15]. In contrast to their initial report [9], and to subsequent efforts by the ATP III, IDF and the authors of the ‘harmonized’ version of the MetS [10–13], the latest WHO report concludes that the ‘metabolic syndrome should not be a clinical diagnosis’, but rather viewed as ‘a pre-morbid condition, and should thus exclude individuals with established diabetes or cardiovascular disease’. Finally, it was argued that the MetS ‘has limited practical utility as a diagnostic or management tool’ and that ‘efforts to redefine it are inappropriate in the light of current knowledge and understanding’.

This short summary of the history of the MetS as a diagnostic category is intended to serve as background for the more critical analysis of the diagnostic and pedagogical utility of the MetS that will be the focus of succeeding sections.

Obesity and the MetS; what’s in a name?

Four of the five components that make up the MetS were identified as a cluster of CVD risk factors, more likely to occur in insulin-resistant/hyperinsulinaemic, nondiabetic individuals, subsumed under the rubric of Syndrome X. Obesity was not included as one of the components of Syndrome X, and this was not an oversight. A relationship between excess adiposity and a decrease in insulin-mediated glucose uptake had been demonstrated many years previously [16]. However, obesity does not equal insulin resistance, and an understanding of the relationship between obesity, insulin resistance, and cardio-metabolic risk is necessary to view the MetS in perspective.

Metabolically healthy obesity

The fact that not all overweight/obese individuals are insulin resistant, and demonstrate the abnormalities associated with this defect in insulin action, has frequently resulted in the division of these individuals into two groups: metabolically healthy obese (MHO) and metabolically abnormal obese. This distinction is based on the implicit assumption that insulin resistance is a fundamental characteristic of obesity, and insulin-sensitive, obese individuals are an anomaly, and deserve a special designation (MHO). Parenthetically, although a significant number of normal-weight individuals are insulin resistant, with the associated metabolic abnormalities, it has not led to a comparable division; e.g. normal weight healthy versus normal weight abnormal. For example, Wildman et al. [17], using conventional body mass index (BMI) criteria, divided c. 5000 participants in The National Health and Nutrition Examination Surveys (NHANES, 1999–2004) into normal weight, overweight and obese categories. They then used six criteria, the four components of the ATP III definition of the MetS, HOMA-IR (90th percentile) and CRP (90th percentile) to define subjects as metabolically normal (≤1 abnormality) or metabolically abnormal (≥2 abnormalities). With this approach they classified 51% of overweight individuals as metabolically healthy, 32% of obese subjects as metabolically healthy and 24% of normal-weight individuals as metabolically abnormal. Using a different approach [18], we found in 455 apparently healthy individuals that 30% of those whose steady-state plasma glucose (SSPG) concentrations observed during the insulin suppression test [19–21] placed them in the most insulin-sensitive tertile were overweight/obese, and only 36% of those whose SSPG concentration placed them in the most insulin-resistant tertile were obese. Indeed, approximately one of six individuals in the most insulin-resistant tertile were of normal weight. To put this into clinical perspective, in a study of 211 apparently healthy individuals [22], BMI ≥30.0 < 34.9 kg m−2, we found a prevalence of impaired glucose tolerance of 1% in the most insulin-sensitive tertile versus 46% in the most insulin resistant third of this obese group. These data show that a substantial portion of overweight/obese persons are insulin sensitive, without any of the metabolic abnormalities associated with insulin resistance, and insulin resistance and associated metabolic abnormalities are not uncommon in normal-weight persons. Thus, it seems reasonable to omit modifiers like ‘metabolically healthy’ obese, etc., and simply classify individuals, normal weight or obese, as a function of their cardio-metabolic risk.

Interaction between excess adiposity and cardio-metabolic risk

The fact that obese persons can be insulin sensitive does not mean that excess adiposity is without an adverse effect on insulin action and associated metabolic abnormalities. As indicated earlier, when adjusted for differences in degree of fitness, differences in adiposity account for approximately 25% of the sixfold variability in insulin action in the population at large [2].

To provide insight into the relationship between insulin resistance and associated metabolic abnormalities, we determined SSPG concentration by the insulin suppression test in 314 apparently healthy individuals [23]. As results of prospective studies demonstrated that the third of an apparently healthy population that was most insulin resistant had a significant increase in adverse clinical outcomes [24, 25], the population was divided in thirds based on the values of SSPG concentration; insulin sensitive, intermediate and insulin resistant. Figure 1 displays the best-fit lines describing the relationships between BMI and plasma TG and HDL-C concentrations, and the glucose and insulin responses to a 75-g oral glucose challenge. Data in the two left panels indicate that the higher the BMI, the greater the increase in plasma TG concentration and decrease in HDL-C concentration. However, it is clear that the insulin-resistant group had significantly higher (TG) or lower (HDL-C) concentrations at any given BMI than did the insulin-sensitive group. Plasma glucose and insulin responses to the oral glucose challenge are displayed in the two right panels, and demonstrate the relative minor increase in magnitude of the glucose response as BMI increases. In contrast, the plasma insulin response increases to a much greater degree, and it is obvious that the magnitude of the insulin response to oral glucose minimizes the adverse effect of increasing adiposity on glucose response. It is also obvious that the insulin-resistant group has higher glucose and insulin responses at every BMI value.

Figure 1.

Best-fit lines describing the relationship in 314 apparently healthy volunteers between BMI and fasting plasma TG and HDL cholesterol concentrations and the total integrated glucose and insulin responses to a 75-g oral glucose as a function of SSPG tertile; the most insulin sensitive (SSPG concentration = 61 ± 2 mg dL−1), intermediate (SSPG concentration = 114 ± 2 mg dL−1), and most insulin-resistant (210 ± 4 mg dL−1) tertiles. Adapted from the J Am Coll Cardiol 2002; 40: 937–43, with permission from the authors and the journal. BMI, body mass index; SSPG, steady-state plasma glucose; HDL-C, high-density lipoprotein cholesterol.

The data in Fig. 1 demonstrate the degree to which components of the MetS vary substantially as a function of degree of insulin resistance at a given level of adiposity. The importance of differentiating between the adverse impact of obesity, per se, versus insulin resistance on the components of the MetS is also seen when adverse outcomes are examined. Thus, Ninomiya, et al. [26] used data from NHANES III to examine the relationship between the individual components of the MetS and myocardial infarction (MI) and stroke. They found that insulin resistance (a fasting glucose ≥110 mg dL−1 or a diagnosis of 2DM), and the other components of the MetS, with the exception of a high WC, were all independent predictors of MI and stroke. In the absence of an independent relationship between WC and disease, the authors suggested that this might ‘reflect an indirect effect of high WC through other components of the syndrome’ and ‘that in a model excluding the other components, high WC was significantly related to MI/stroke’.

Regional fat distribution and the MetS: does it matter where the fats at?

Abdominal obesity, as assessed by measuring WC, has gone from being one of the components, along with BMI, of the MetS as defined in the first WHO report [9], to the only index of adiposity in the ATPIII definition of the MetS [10], to the essential component as defined by the IDF [11], back to being only one of the five components in the newly ‘harmonized’ version [12]. Not withstanding the findings of Ninomiya et al. [26], the arbiters of the MetS believe that abdominal obesity is a crucial ingredient of the MetS, including the statement that ‘the metabolic syndrome all begins with central obesity [13]’. This point of view, which has important clinical and pedagogical implications, can be challenged. Specifically, there is evidence that the link between overall obesity, as assessed by BMI, and the components of the MetS seems to be comparable to that between WC and the MetS and that the ability of BMI to predict type 2DM or CVD is also comparable to that achieved with measurement of WC.

Figure 2 illustrates the relationship between SSPG concentration and BMI (left panel) and WC (right panel) in 330 nondiabetic, apparently healthy individuals [27]. Four obvious conclusions can be drawn from these data: (i) in general, the greater the BMI or WC, the more insulin resistant (higher SSPG concentration) the individual; (ii) SSPG concentrations vary widely at any given BMI or WC; (iii) individuals with elevated values of WC or BMI can be insulin sensitive, and insulin resistance can be present in those with normal values of BMI or WC; and (iv) the relationship between SSPG and either index of obesity is comparable. Our demonstration that BMI and WC are comparable in their relationship to SSPG concentration (the measure of insulin action) should not be too surprising in the light of the evidence from Ford et al. [28], based on the NHANES data base, that the correlation coefficient between BMI and WC was >0.9, regardless of age, sex and ethnicity of the groups evaluated.

Figure 2.

Relationship between BMI (left panel) and waist circumference (right panel) and SSPG concentrations in 330 apparently healthy volunteers. Reprinted from the Am J Clin Nutr 2006; 83: 47–51, with permission of the Journal and the authors. BMI, body mass index.

Although the data in Fig. 2 show that a comparable relationship exists between SSPG concentration and the two indices of obesity, it has been argued that there is an additional adverse impact of WC when adjustments are made for differences in BMI. Some evidence for this view can be seen in the upper panel of Fig. 3, in which SSPG concentrations are compared in those with normal or abnormal values for WC [29]. As there are very few individuals with an abnormal WC in the normal-weight group (BMI < 25.0 kg m−2) or a normal WC in the obese group (BMI ≥ 30.0 kg m−2), the comparison is limited to the overweight group (BMI ≥ 25.0 < 30 kg m−2). When this is performed, there is a somewhat higher SSPG concentration in the group with an abnormal WC. However, the results in the lower panel of Fig. 3 suggest that when matched for WC, BMI also has an independent effect on SSPG concentration. Thus, in those with a normal WC, SSPG concentrations are higher in the overweight when compared to the normal-weight BMI groups, and SSPG concentrations are also higher within those with an abnormal WC in the obese when compared to the overweight BMI groups. It is worth emphasizing that very few individuals within the normal WC group are obese, and there are very few normal-weight individuals in the abnormal WC group. These data again document that the two indices of adiposity are highly correlated.

Figure 3.

Mean (±SEM) SSPG concentration as a function of waist circumference category in 330 apparently healthy volunteers divided on the basis of BMI category (top panel), and SSPG concentration as a function of BMI in the volunteers divided on the basis of their WC category. Reprinted from the Am J Clin Nutr 2006; 83: 47–51, with permission of the Journal and the authors. SSPG, steady-state plasma glucose; BMI, body mass index.

As BMI and WC seem to be associated with a similar degree with insulin sensitivity, it should not be too surprising if the MetS could develop in the absence of abdominal obesity. This prediction is nicely borne out in a population-based study by Lee et al. [30] of 4334 individuals of Chinese, Malaysian or South Asian Indian ancestry. Approximately 25% of this population met the ATP III criteria for MetS, using WC cut points for Asians/South Asians; one-third of whom did not have abdominal obesity as estimated by WC. Values of the MetS components differed by selection in those with the MetS, but there was no difference between those with or without abdominal obesity. Risk of CVD was also greater in both MetS groups, but the increased risk did not vary as a function of whether or not abdominal obesity was present. The authors concluded from these data that ‘having metabolic syndrome with or without central obesity confers IHD risk’.

Perhaps the most compelling evidence questioning the uniqueness of WC as the measure of obesity most closely associated with CVD and 2DM comes from the results of the International Day for the Evaluation of Abdominal Obesity study [31]. BMI and WC were measured in 168 159 patients of primary care physicians in 63 countries, and the presence of either CVD or 2DM was noted. The major result was reported as the odds ratio (OR) of a subject having either CVD or 2DM associated with one standard deviation increase in either WC or BMI. Data were presented individually for each of the 11 worldwide geographical regions taking part in the study and differed modestly as a function of region. The overall association between CVD and either index of obesity was comparable, with ORs (95% CI) in men of 1.36 (1.33–1.39) for WC and 1.32 (1.29–1.35) for BMI, and 1.40 (1.37–1.43) for WC and 1.38 (1.37–1.41) for BMI in women. In the case of 2DM, the overall ORs for the association between adiposity and disease were 1.59 (1.49–1.56) vs. 1.52 (1.49–1.56) for BMI and WC in men, with values in women of 1.83 (1.79–1.87) and 1.67 (1.64–1.71) for WC and BMI, respectively. The authors concluded that ‘BMI and particularly WC were both strongly linked to CVD and especially to diabetes’. Readers can compare the magnitude of the differences in the association of the two clinical syndromes with BMI versus WC and decide for themselves how appropriate the choice of the word ‘particularly’ is in the authors’ conclusion.

How useful is the MetS as a diagnostic and/or pedagogical category?

As expressed in the ‘harmonized’ version of the MetS [12], ‘at a clinical level, individual patients with the metabolic syndrome need to be identified so that their multiple risk factors, including lifestyle risk factors, can be reduced’. The importance of recognizing and addressing multiple risk factors for CVD and 2DM is obvious, but it is less obvious that making a diagnosis of the MetS accomplishes this goal more effectively than currently available diagnostic tools.

Clinical syndromes as MetS criteria

Decompensation of glucose tolerance, i.e. frank 2DM is one of the MetS criteria, and once an individual is identified by that diagnosis the importance of addressing all CVD risk factors is well-recognized. Is there any clinical benefit in differentiating patients with 2DM on the basis of whether or not they also qualify for the MetS: 2DM with the MetS versus 2DM without the MetS? A similar argument can be made concerning the inclusion of a diagnosis of hypertension as one of the MetS criteria. Viewed in this manner, it is not surprising that the joint report from the ADA and EASD [14] concluded that ‘adults with any major CVD risk factor should be evaluated for the presence of other CVD risk factors’ and ‘that all CVD risk factors should be individually and aggressively treated’.

Is the sum greater than its parts?

There is general agreement that the MetS predicts both CVD and 2DM, with the association being somewhat greater in the case of 2DM [32, 33]. However, it is not so clear that the MetS performs any better in this regard than its individual components. This should not be too surprising, given that glucose intolerance, a high TG and/or low HDL-C concentration, and elevated blood pressure are recognized risk factors for CVD and/or 2DM [5]. For example, several studies have shown that fasting glucose concentration was as good, if not better, than the diagnosis of the MetS in predicting 2DM [33–35]. Similarly, Eddy et al. [36], using the NHANES data base, found that ‘high glucose, by itself, was as good as any definition of the MetS in predicting risk of future MI’. It is not possible within the context of this presentation to discuss all of the publications that have addressed this issue, but I think it is fair to conclude that the ability of the MetS to identify individuals at risk for 2DM and/or CVD is no better than its competent parts.

MetS versus the Framingham Risk Score (FRS) to identity risk of CVD and 2DM

Although several global risk factor approaches to identify individuals at increased CVD risk have been introduced, the FRS appears to be the one most commonly used. In 2004, Stern et al. [37] compared the ability of two global risk factor approaches and the MetS to identify individuals at risk of CVD and/or 2DM. As regards CVD, they found that ‘the Framingham Risk Score had significantly higher sensitivity and, at the same sensitivity, a significantly better false-positive rate than the metabolic syndrome’. Furthermore, when the FRS was combined with the MetS, ‘neither the sensitivity nor the false-positive rate is significantly than the Framingham Risk Score alone’. Finally, the ability of the MetS to predict 2DM was inferior to the Diabetes Risk Score as defined by Stern et al. [38]. Results from the Atherosclerosis Risk in Communities study [39] did not discern any benefit of the MetS as a predictor of CVD risk, concluding that ‘comparison of ROC curves indicated that the metabolic syndrome did not materially improve CHD risk prediction beyond that the level predicted by the FRS in men or women’. Wannamethee et al. [40] also compared the relative abilities of the MetS and the FRS to identify individuals at risk for CVD or 2DM and concluded that the FRS performed better than the MetS in identifying CVD risk and indicated that the ‘FRS was a better discriminator of the combined cases (CHD, stroke, or diabetes) than MetS’. However, MetS was a ‘far stronger predictor of DM2 than of CHD’ and ‘a better predictor of DM2 than was FRS’. The difference between the ability to identify individuals at risk for CVD when compared to 2DM is hardly surprising: (i) glucose tolerance is part of the MetS diagnostic criteria; and (ii) there is evidence that fasting glucose concentration is as good, if not better, than the diagnosis of the MetS in predicting 2DM [33–35]. Thus, there is no evidence that a diagnosis of the MetS improves our ability to identify individuals at increased risk of 2DM or CVD.

The MetS as an educational tool to enhance treatment compliance

It is often argued that diagnosing the MetS may have no direct clinical benefit, but is a useful educational tool when interacting with patients, educating them, increasing compliance with lifestyle interventions, etc. However, I am unaware of published evidence that this is the case. It is also not self-evident why if a patient has 2DM, or hypertension, there is any advantage to informing them that they have, or do not have, the MetS. Finally, an obese patent with IFG, a TG concentration of 155 mg dL−1 and a systolic blood pressure of 135 mmHg meets the diagnostic criteria for the MetS, whereas another obese patient with IFG, the same blood pressure, and a TG concentration of 145 mg dL−1 does not. How different would be the treatment approach in these two patients – one with and one without the MetS. It is possible to conduct studies design studies to evaluate the benefit in the physician–patient encounter of making a diagnosis of the MetS, but until such information is available, a degree of scepticism as to this claim seems justifiable.

Why does not the MetS perform better?

Given the support of so many prestigious organizations, it may seem surprising that there appears to be relatively little objective evidence that diagnosing the MetS accomplishes the purpose for which it was intended. There are many possibilities to account for this apparent discrepancy between promise and performance, but at least three are worthy of discussion.


There is general agreement that glucose intolerance, a high TG and low HDL-C concentration, and elevated blood pressure tend to cluster together, although arguments can be made as whether insulin resistance is the reason why this happens [5, 8, 13, 15]. It is also clear that being overweight/obese increases the likelihood that this cluster of abnormalities will occur. However, this does not necessarily mean that employing these associated traits in making a diagnosis of the MetS is particularly useful. Indeed, the relative failure of the MetS to provide a better approach to identifying individuals at risk for CVD or 2DM may be inherent in being based on closely related abnormalities. As discussed earlier, the FRS provides a more useful way to identify individuals at risk for CVD than the MetS. The FRS uses a number of different, not necessarily related, factors known to predict CVD. For example, by including both total cholesterol and HDL-C concentrations, the FRS recognizes individuals whose basic CVD risk is a high low-density lipoprotein cholesterol concentration, as well as those likely to share glucose intolerance and a high TG concentration by virtue of a low HDL-C concentration. The inclusion of smoking captures the direct link between smoking and CVD, as well as the fact that smokers tend to be insulin resistant, glucose intolerant and dyslipidaemic, with high TG and low HDL-C concentrations [41, 42]. The same reason that the MetS is less effective as a CVD predictor than the FRS makes it a better predictor of 2DM. Specifically, glucose intolerance is one of the criteria for diagnosing the MetS, not the FRS, and as indicated earlier, fasting plasma glucose concentration is a better predictor of 2DM than the MetS. Finally, the FRS uses continuous variables, not dichotomous one – an important difference.

Dichotomous versus continuous variables

The definition of many clinical syndromes is based on the use of dichotomous values; for example, 2DM is diagnosed by having a fasting plasma glucose concentration ≥7.0 mmol L−1. However, when we turn to risk assessment, using dichotomous variables results in loss of crucial information concerning the magnitude of risk factors [36]. The relationship between CVD risk factors such as cholesterol and blood pressure is continuous, a fact taken into account by the FRS. This also seems to be the case as regards glucose concentration and is most likely the reason why an elevated fasting plasma glucose concentration is as good, if not better, than the MetS in identifying risk of 2DM [33–35].

Two versus three: does it matter?

Compounding the problems related to the use of dichotomous versus continuous variables is the manner in which the five criteria needed to make a diagnosis of the MetS are used. Belief in the utility of making this diagnosis necessitates acceptance of at least the three following implicit assumptions; (i) any one of the five criteria are equipotent in contributing to the risk of developing CVD or 2DM; (ii) the combination of any three components indicates increased risk of CVD or 2DM when compared to any two abnormalities; and (iii) it does not matter what three components led to a diagnosis of the MetS – any combination of three components has the same clinical effect.

Many publications have addressed these issues in some form or other. For example, results of the PAMELA study [43] demonstrated that ‘an increase in waist circumference, and increase in plasma TGs, or a reduction in HDL cholesterol did not significantly predict the risk of CVD or all cause mortality of which the risk was significantly related to each of the two remaining MetS components, that is office BP and blood glucose elevations’. Perhaps the most complete analysis of these issues was performed by Wilson et al. [35], using data from 3323 middle-aged individuals enrolled in the Framingham Offspring Study. For example, the age-adjusted relative risk (RR) of any one component to CVD ranged. from 1.6 to 2.0, with no great difference amongst them. However, the risk of 2DM was increased from three- to sixfold when compared to the other four criteria when IFG was present. When any two criteria were present, the RR of CVD varied from 1.7 to 2.6, with the highest risk associated with IFG and abdominal obesity. Any combination of IFG with any one of the other four MetS criteria had the highest RR of developing 2DM, varying from 8.2 to 10.7. Turning now to those who met three criteria, i.e. had the MetS, the RR of CVD did not increase over the value with only two abnormalities. The same phenomenon was seen when 2DM was the outcome, with values for RR of those with the MetS (three abnormalities) that were no different than those with only two abnormalities. As before, if IFG was one of the abnormities, the RR increased approximately twofold compared to any combination of two or three abnormalities that did not include IFG.

These results indicate that none of the three implicit assumptions underlying the utility of diagnosing the MetS were met. Thus, the RR of any one component of the MetS varied from 2.4 (elevated blood pressure) to 12.5 (IFG) in its ability to predict incident 2DM. The RR to develop CVD with two abnormalities varied from 1.7 (high TG, low HDL-C concentration) to 2.6 (IFG and abdominal obesity), and from 3.1 (high TG and elevated blood pressure) to 10.7 (IFG and abdominal obesity) to develop 2DM. The most common combination of components leading to a diagnosis of the MetS (10.3%) – a high TG and low HDL-C concentration and elevated blood pressure – had RRs of 1.9 and 3.5 to develop CVD and 2DM, respectively. In contrast, the RRs of a similar number (10.9%) of individuals with IFG and elevated blood pressure, i.e. not qualifying for the MetS, were comparable in terms of risk of CVD (2.0 vs.1.9) and greater for 2DM (9.7 vs. 3.5). Reliance on a diagnosis of the MetS to assign risk in this instance would end up ignoring the group at most risk. Finally, having two abnormalities was as useful as having three abnormalities in the prediction of both CVD and 2DM. As concluded by the authors, ‘finding that clusters of three traits do not substantially increase risk of outcomes over two traits is consistent with the hypothesis that even a modest degree of risk clustering reflects a global underlying insulin-resistant pathophysiology’.


The MetS as a diagnostic category has received considerable attention over the last decade, and several prestigious organizations continue to strive for its unqualified introduction into every day medical practice. In this review, an attempt has been made to document the ‘back-and-forth’ changes in its definition, question the elevation of abdominal obesity as to the only clinical useful index of adiposity and identify some of the many problems with its usage. The fundamental question to be answered is what have we leaned from the many publications addressing the MetS, and does this information provide evidence that diagnosing the MetS is a useful exercise.

As the result of the enormous number of manuscripts that have been published, we now know the prevalence of the MetS in almost an infinite number of different populations, and how this figure varies as a function of the criteria used. We also know that individuals with the MetS are at increased risk for both CVD and 2DM, with the magnitude of the association being greater for predicting 2DM than CVD. Finally, the more individual components of the MetS present in an individual, the greater the likelihood of developing an adverse outcome.

However, it can be argued that none of this information has provided new pathophysiological insight, nor does it support the clinical utility of the MetS as a diagnostic category. Thus, because all the components of the MetS have been associated with 2DM and CVD, it is not surprising that taken together they predict both 2DM and CVD, and the more the abnormal components, the greater the risk. However, the FRS seems to be more effective than the MetS in predicting CVD, not surprising as the FRS contains some of the components of the MetS, used as continuous variables, as well as potent CVD risk factors like smoking and cholesterol. The fact that the MetS performs best in predicting 2DM is even less surprising, given the fact that glucose intolerance is one of the diagnostic criteria of the MetS. Indeed, as discussed earlier, IFG is a better predictor of 2DM than the MetS, and IFG plus any one of the other four components of the MetS is more powerful at predicting 2DM than a diagnosis of the MetS that does not capture glucose intolerance. It also seems that there is nothing magical about having three abnormalities, and having two abnormalities seems to be as predictive of a bad outcome. Would the failure to have three abnormalities modify the clinical approach to a patient with only one, or two abnormalities? Finally, if a patient meets the diagnostic criteria for 2DM or essential hypertension, does knowing whether or not they also meet the criteria for the MetS going to affect the treatment plan? In short, despite the many publications, and the efforts to define, and re-define the MetS diagnostic criteria, it is not clear that it is a diagnostic category worth continuing. Put most simply, if it is less effective than the FRS in predicting CVD, its original intent, and apparently no better, if not worse, than IFG in predicting 2DM, is there any reason why the MetS as a diagnostic category should not be given its well-deserved rest?

Conflict of interest statement

No conflict of interest was declared.