The constellation of adverse cardiovascular disease (CVD) and metabolic risk factors, which include elevated abdominal obesity, blood pressure, glucose, and triglycerides (TG) and lowered high-density lipoprotein-cholesterol (HDL-C), has been termed the metabolic syndrome 1. Based on the National Cholesterol Education Program Adult Treatment Program III (ATP II) criteria 1 and the International Diabetes Foundation criteria 2 approximately 35% and 39%, respectively, of U.S. adults possess the metabolic syndrome 3. In terms of health outcomes, the metabolic syndrome is associated with an increased risk of all-cause mortality, CVD morbidity and mortality, type 2 diabetes, and some cancers 4.

In 1980, Khoury et al. 5 were perhaps the first to show the clustering of CVD risk factors in 6-19 yr old subjects of the Cincinnati Lipid Research Clinic's Princeton Study. In 1987, Smoak et al. 6 also reported the clustering of CVD risk factors in the Bogalusa Heart Study. However, not until the recent medicalization and coining of the term 'metabolic syndrome' had reports indicated the emergence of the metabolic syndrome during childhood and adolescence 7891011. Data from the United States (U.S.) National Health and Nutrition Examination Survey (NHANES) III (1988-1994) showed that the prevalence rate of the metabolic syndrome was 4% in 12-19 yr old adolescents and 30% in obese adolescents 11. The prevalence rate increased to 6% (>2 million) in NHANES 1999-2000 10. Another paper using the same data (i.e., NHANES III) but different criteria showed that the prevalence rate was nearly 10% 12. Among 7-to-9 yr old children residing in a rural Midwestern US state, the prevalence rate was 5% 9. Prevalence rates in children and adolescents have been reported to be 6.5% in northern Mexico 13, 9% in Korea 14, 2% in Turkey 15, and 10% in Quebec, Canada 16.

Since there is no universal definition of the metabolic syndrome in children or adolescence and the prevalence rate is relatively low, several authors 1718192021222324252627 have derived a continuous score representing a composite CVD/metabolic risk factor profile or index (i.e., the metabolic syndrome score) which has been recommended in a joint statement by the American Diabetes Association and the European Association for the Study of Diabetes 28. Previous procedures to derive the score include principal component analysis 2122, Z scores 17182123242526, and centile rankings 1920. However, the validity of these composite scores in children and adolescents has not been examined. A recent paper in adults 29 showed a continuous metabolic syndrome score (cMetS) derived from principal component analysis was higher in adults with the metabolic syndrome and that the score increased progressively with increasing number of adverse risk factors.

Besides validating the cMetS, another related issue is identifying the cMetS that confers increased risk of the metabolic syndrome. In other words, what is the 'optimal' cutoff for interpreting the extent of the cMetS? Recently, receiver operating characteristics (ROC) analysis has been used to determine waist circumference and insulin resistance cutoff values for identifying the metabolic syndrome in youth 3031. Based on the increased use and utility of the cMetS in pediatric epidemiological research, the primary purpose of this study was to examine the construct validity of a cMetS in children. The secondary purpose was to identify a cutoff for an adverse cMetS based on ROC analysis.

Table 1 shows the number and percentage of subjects by the number of adverse risk factors. Overall, about half of the subjects did not have any adverse risk factors and in turn, the other half possessed at least 1 adverse risk factor. About 5% possessed the metabolic syndrome (≥3 risk factors). No children possessed all five adverse risk factors. There were no significant sex differences in the % of subjects possessing a given number of adverse risk factors or in the occurrence of any given risk factor. Elevated BP was the most common risk factor (37%) and abnormal fasting glucose was the least common (1%)(Table 2). Although not included as a component of the MetS, it should be noted that 44% of the sample was either overweight or obese.

Descriptive statistics for the physical characteristics and individual risk factors by the number of adverse risk factors are presented in Table 2. As would be expected based on the definition of the MetS, there was a general trend for those with more adverse risk factors to be bigger and possess more adverse levels of the risk factors.

Figure 1 shows the graded relationship between the cMetS and the number of adverse risk factors. The cMetS is lowest in the group with no adverse risk factors (-1.59 ± 1.76) and highest in those possessing the metabolic syndrome (≥3 risk factors) (7.05 ± 2.73). The mean cMetS for the total sample is 0.00 ± 3.0.

According to the ROC curve analysis (Figure 2), the cutoff level yielding the maximal sensitivity and specificity for predicting the presence of the metabolic syndrome was a cMetS of 3.72. The sensitivity and specificity using this cutoff value were 100% and 93.9%, respectively. The AUC was 0.978 (0.957-0.990, 95% confidence intervals) (p < 0.001). Other cutpoint values and their corresponding sensitivity and specificity are provided in Table 3 for interpretation by the reader.

As previously shown in this sample 9, approximately 5% of 7-9 yr olds possess the metabolic syndrome based on the criteria of Cook et al. 11. Again, it should be emphasized that there is no universal definition of the metabolic syndrome in children and adolescents and that the prevalence is relatively low. Thus, both of these factors limit the ability to conduct epidemiological studies examining various genetic and environmental risk factors for the metabolic syndrome or the clustering of risk factors in youth. Therefore, we 2324252642 and others 1718 have constructed a cMetS for investigations in pediatric epidemiology and medicine. An advantage to a cMetS is that it is statistically more sensitive and less error prone by comparison to the dichotomous approach 1843. Several researchers have utilized the cMetS in pediatric epidemiological research. For example, results from the Corpus Christi Child Heart Study indicated higher cMetS in Mexican-American boys and girls compared to White boys and girls 21. Recent studies also show that habitual physical activity and aerobic fitness are inversely associated with the cMetS 171823242642, and that aerobic fitness attenuates the cMetS among high fat children and adolescents 242642. The cMetS has also been shown to track from childhood/adolescence into young adulthood 19202225. This latter finding holds importance to the predictive utility of the cMetS on adult disease outcomes.

To date, this is the first study to validate the cMetS score in children. The results indicate a clear graded relationship between the cMetS and the number of adverse risk factors with those possessing ≥3 risk factors (i.e., metabolic syndrome) having the highest cMetS. Furthermore, results of the ROC analysis identified a cMetS of 3.72 as the optimal cutpoint which confers an increased risk of MetS in this sample of 7-9 year old children. The AUC for this index was 98%, which indicates a cMetS of 3.72 is highly accurate at detecting the presence of MetS in this sample. The sensitivity and specificity using this cutoff value were 100% and 93.9%, respectively. In other words, none of the children with metabolic syndrome had a cMetS score lower than 3.72 and only 6% of the children without metabolic syndrome had a score higher than 3.72. This cutoff value also approximates the mean value for subjects possessing 2 adverse risk factors (Table 2). However, other cMetS indicated in Table 3 should also be considered since they also possess high values for sensitivity and specificity. It should also be considered that there may be varying levels of increased risk for MetS (e.g., low, moderate, high) based on the cMetS, particularly since other scores showed good sensitivity and specificity. Hence, cMetS can be used as a preventative measure rather than a diagnostic tool. However, it is important to note that there are several drawbacks to identifying cut-point value(s) for the cMetS based on this sample, and thus we urge researchers to use the approach herein to validate and create a cMetS that is specific to their study population. Some of these limitations are discussed below. First, the cMetS is sample-specific. Second, the cut-off points used herein, e.g., the Cook criteria does not have a solid clinical or health-related outcome basis. Third, the cMetS is based upon the assumption that each selected variable is equally important in defining CVD risk, although a child's being defined as at risk does not necessarily mean that the child has high levels of CVD risk factors per se.

A similar study in adults has also validated the cMetS. In a randomly selected sample of 18- to 75-year-old (mean age = 46 yrs) Flemish adults (571 men and 449 women), the cMetS, determined by principal component analysis, was significantly higher in subjects with cMetS as defined by the International Diabetes Federation (men [12.8%]: 2.03 ± 1.00, women [>8.5%]: 2.63 ± 1.28) versus subjects without (men: -0.30 ± 1.21, women: -0.24 ± 1.16). Moreover, cMetS increased progressively with increasing numbers of risk factors in men (0 risk factors: -1.21 ± 0.96; 1 risk factor: -0.26 ± 0.87; 2 risk factors: 0.67 ± 0.84; 3 risk factors: 1.76 ± 0.73; and 4 risk factors: 3.04 ± 0.94) and women (0 risk factors: -0.96 ± 0.79; 1 risk factors: 0.16 ± 0.82; 2 risk factors: 1.21 ± 0.82; 3 risk factors: 2.17 ± 0.81; and 4 risk factors: 4.09 ± 0.99) 29. These findings are similar to those reported here and support the use of the cMetS in epidemiological analyses.

As previously mentioned, one limitation to this study is that the MetS criteria - e.g., the Cook criteria - have no clinical or health-related outcome basis. Recently, age-specific criteria for the metabolic syndrome were developed using the LMS statistical technique in 12-19 yr old subjects participating in the NHANES 44. Unfortunately, we could not use these values since our subjects were 7-9 yr old. The age of the subjects is both a limitation and strength of the study. In terms of limitations, the cMetS needs to be validated in older children and adolescents. Another limitation is that the cMetS is sample specific. If epidemiological studies are to use the cMetS, a standardized method of calculating the score may prove beneficial for comparing studies. An alternative approach to the sample-specific Z score approach highlighted here may be to compare individual risk factor value to the population median or clinical cut-points such as those recently published 44.

In conclusion, this paper shows the convergent validity for the cMetS that is becoming widely used in pediatric epidemiological research. Furthermore, researchers can use the approach herein to validate and create cMetS that are specific to their study populations. It is recommended that age-specific criteria for the metabolic syndrome such as those recently developed for 12-19 yr olds be used as a criterion in future studies and that such criteria be developed in younger children as well.

The authors declare that they have no competing interests.

JCE was responsible for the conception of the research question, analysis and interpretation of data; and writing the manuscript. KRL has made contributions to the analysis and interpretation of data, and assisted in drafting the manuscript and revising it critically for important intellectual content. KDD was responsible for conception of the research question, interpretation of data, and assisted in drafting the manuscript and revising it critically for important intellectual content. BKS was responsible for data collection and provided critical feedback to the manuscript for important intellectual content. JED was responsible for acquisition of funding, coordinating data collection, and provided critical feedback to the manuscript for important intellectual content. All authors read and approved the final manuscript.