Stern MP, Williams K, González-Villalpando C, Hunt KJ, Haffner SM: Does the metabolic syndrome improve identification of individuals at risk of type 2 diabetes and/or cardiovascular disease? Diabetes Care 27:2676–2681, 2004
Meigs JB: Metabolic syndrome: in search of a clinical role (Editorial). Diabetes Care 27:2761–2763, 2004
In the analysis of San Antonio Heart Study (SAHS) data presented and discussed in the above-cited articles, all SAHS subjects were inadvertently classified as having met the abdominal obesity criteria, and the National Cholesterol Education Program (NCEP) metabolic syndrome was therefore diagnosed when two or more of the four remaining NCEP criteria were met. This error did not affect the Mexico City dataset. In the SAHS dataset used to compute diabetes incidence, the corrected prevalence of metabolic syndrome is 19.4% (16.0% in non-Hispanics and 21.1% in Mexican Americans). In the cardiovascular disease (CVD) dataset, the corrected prevalence of metabolic syndrome is 24.1% (16.8% in non-Hispanics and 27.6% in Mexican Americans). The corrected sensitivities, false-positive rates, areas under the receiver operating characteristic curve (aROCs), and odds ratios are now given in the accompanying tables.
As can be seen from the new Tables 1 and 2, our basic conclusions remain unchanged. At similar false-positive rates, the Diabetes Risk Score and the Framingham Score have higher sensitivities and, at similar sensitivities, lower false-positive rates than the metabolic syndrome. Moreover, the sensitivities, false-positive rates, and aROCs do not change substantially when the metabolic syndrome is added to these predicting models. The univariate odds ratios remain significant for the metabolic syndrome and both predicting models. However, the multivariate odds ratios drop precipitously for the metabolic syndrome but drop only minimally for the predicting models.
Comparison of predicting ability of the metabolic syndrome, the Diabetes Risk Score, and the Framingham Risk Score
| . | aROC . | Sensitivity (%) . | False-positive rate (%) . |
|---|---|---|---|
| Prediction of diabetes in the SAHS | |||
| Metabolic syndrome | 52.8 | 15.1 | |
| Diabetes Risk Score | 0.819 | 60.5 (P = 0.059) | Fixed at 15.1 |
| Diabetes Risk Score and metabolic syndrome | 0.826 (P = 0.055) | 62.6 (P = 0.37) | Fixed at 15.1 |
| Diabetes Risk Score | Fixed at 52.8 | 10.9 (P < 0.0001) | |
| Diabetes Risk Score and metabolic syndrome | Fixed at 52.8 | 9.6 (P = 0.030) | |
| Prediction of CVD in the SAHS | |||
| Metabolic Syndrome | 54.8 | 22.1 | |
| Framingham Risk Score | 0.816 | 69.4 (P = 0.0008) | Fixed at 22.1 |
| Framingham Risk Score and metabolic syndrome | 0.816 (P = 0.85) | 71.3 (P = 0.32) | Fixed at 22.1 |
| Framingham Risk Score | Fixed at 54.8 | 11.4 (P < 0.0001) | |
| Framingham Risk Score and metabolic syndrome | Fixed at 54.8 | 11.7 (P = 0.48) |
| . | aROC . | Sensitivity (%) . | False-positive rate (%) . |
|---|---|---|---|
| Prediction of diabetes in the SAHS | |||
| Metabolic syndrome | 52.8 | 15.1 | |
| Diabetes Risk Score | 0.819 | 60.5 (P = 0.059) | Fixed at 15.1 |
| Diabetes Risk Score and metabolic syndrome | 0.826 (P = 0.055) | 62.6 (P = 0.37) | Fixed at 15.1 |
| Diabetes Risk Score | Fixed at 52.8 | 10.9 (P < 0.0001) | |
| Diabetes Risk Score and metabolic syndrome | Fixed at 52.8 | 9.6 (P = 0.030) | |
| Prediction of CVD in the SAHS | |||
| Metabolic Syndrome | 54.8 | 22.1 | |
| Framingham Risk Score | 0.816 | 69.4 (P = 0.0008) | Fixed at 22.1 |
| Framingham Risk Score and metabolic syndrome | 0.816 (P = 0.85) | 71.3 (P = 0.32) | Fixed at 22.1 |
| Framingham Risk Score | Fixed at 54.8 | 11.4 (P < 0.0001) | |
| Framingham Risk Score and metabolic syndrome | Fixed at 54.8 | 11.7 (P = 0.48) |
P for comparison with the row immediately above.
Univariate and multivariate odds ratios for predicting diabetes and CVD using the metabolic syndrome, Diabetes Risk Score, and Framingham Risk Score*
| . | Univariate . | Multivariate† . |
|---|---|---|
| Prediction of diabetes in the SAHS | ||
| Metabolic syndrome | 6.32 (4.61–8.65) | 1.94 (1.34–2.82) |
| Diabetes Risk Score | 6.46 (4.97–8.40) | 5.18 (3.89–6.91) |
| Prediction of CVD in the SAHS | ||
| Metabolic syndrome | 4.28 (3.08–5.94) | 1.50 (1.03–2.18) |
| Framingham Risk Score | 9.41 (6.53–13.6) | 7.87 (5.29–11.7) |
| . | Univariate . | Multivariate† . |
|---|---|---|
| Prediction of diabetes in the SAHS | ||
| Metabolic syndrome | 6.32 (4.61–8.65) | 1.94 (1.34–2.82) |
| Diabetes Risk Score | 6.46 (4.97–8.40) | 5.18 (3.89–6.91) |
| Prediction of CVD in the SAHS | ||
| Metabolic syndrome | 4.28 (3.08–5.94) | 1.50 (1.03–2.18) |
| Framingham Risk Score | 9.41 (6.53–13.6) | 7.87 (5.29–11.7) |
Data are odds ratios (95% CI).
Odds ratios for multiple regression model–derived scores are based on a 2.00-unit increment in the logit of risk;
odds ratios from a multiple logistic model containing both the Diabetes Risk Score (5) and the NCEP ATP-III–defined metabolic syndrome (2) or from a multiple logistic model containing the Framingham Risk Score (6) and the NCEP ATP-III–defined metabolic syndrome (2).
