In this study of predominantly black young adults the IDF waist circumference failed to identify the majority of male subjects with insulin resistance measured by HOMA-IR. The optimal cut point of 82 cm derived from the study was much lower than that proposed by the IDF in men suggesting the need to lower this value to improve the sensitivity to identify those with metabolic abnormalities. The IDF criteria performed much better in the female participants and were very similar to those derived from this study. Unlike other populations the study derived cut-points for waist circumference to identify insulin resistance were similar in young black men and women.
There are limited data on ideal cut points for the identification of insulin resistant persons of African decent. In a cross sectional study of black populations in Cameroon, Nigeria, Jamaica, St. Lucia and Barbados the waist circumference cut points for identifying persons with hypertension were lower than the IDF criteria and sex differences in these cut points were not as large as for other populations . In a subsequent cohort study of middle aged Jamaicans a waist circumference of 88 cm in men and 84.5 cm in women was found to be the optimal cut points for predicting incident diabetes . A cross sectional analysis of US adults using NHANES data demonstrated racial differences in the waist circumference cut points that correspond to a BMI of 25 kg/m2, - the basis of the cut points used to identify those with metabolic abnormalities. In the Black Americans the waist circumferences corresponding to a BMI of 25 kg/m2 were 86.4 cm in men and 83.5 cm in women compared with 91.3 cm and 83.4 cm in White American men and women respectively . While the waist circumference corresponding to a BMI of 25 kg/m2 was about 5-6 cm higher in white men compared to black men, there was no significant racial difference in the waist circumference corresponding to a BMI of 25 kg/m2 in the women. These studies suggest the need for additional data to develop more appropriate cut points for Black populations which may differ by age and sex. Similar to the findings of this study the sex differences in the waist circumference cut points among blacks in all three studies were not pronounced.
The IDF criteria for the diagnosis of the metabolic syndrome in adolescents was an attempt to standardize how the disease was defined in younger populations worldwide. Unlike the criteria for adults no consideration has been given to racial differences in the waist circumference that confer additional risk. Our data suggests the need for lower cut points in young black populations compared to middle-aged European populations in order to identify those with insulin resistance.
The study derived waist circumference cut point in men improved sensitivity for detecting insulin resistant young adults, however the sensitivity still remained low. In both men and women the study derived waist circumference cut point identified only about half of those who were insulin resistant. A lower cut point could be considered but this would reduce the specificity of the measurement and could result in unnecessary expenditure in assessment and treatment of young adults who are not at higher risk of complications. The final decision on the waist circumference cut point used by any health system should ultimately be determined by the available human, infrastructural and economic resources.
The low sensitivity of waist circumference for identifying insulin resistant young black men and women suggests that central obesity may need to be combined with other abnormalities to improve the ability to identify those with this condition. Our findings support the recent recommendation by the IDF that abdominal obesity should no longer be a prerequisite for the diagnosis of metabolic syndrome but one of several abnormalities that constitute the metabolic syndrome .
The waist circumference was better than chance in identifying low HDL-cholesterol in both sexes and elevated triglycerides only in men. Waist circumference was not better than chance in identifying the other components of the metabolic syndrome. This may have been a result of the low prevalence of some of these abnormalities. For example elevated fasting glucose and elevated triglycerides were present in less than 2 percent of the population. The study also highlights the differences in the metabolic abnormalities in Caribbean black populations which are typically characterized by a high prevalence of "low" HDL-cholesterol and a low prevalence of elevated triglycerides [21, 22].
This study had some limitations. In this analysis we utilized a HOMA-IR score derived from a computer program as the method for the measurement of insulin resistance  and not insulin resistance measured by clamp or insulin sensitivity by the minimal model. The model used in the HOMA-IR calculation was derived from non-linear empirical equations producing an algebraic solution and this is more accurate than measurements derived graphically or from simple mathematical approximations.
As there is no standardized definition of insulin resistance we used the distribution of the HOMA-IR scores from the population to define insulin resistance and classified those in the highest HOMA-IR sex specific quartile as insulin resistant. As a consequence the absolute definition of insulin resistance is likely to differ from that used for middle-aged populations and this may influence the definition of waist cut-points. The decision to define insulin resistance this way was based on the approach taken by the EGIR to define hyperinsulinemia as being in the top quartile of fasting insulin values in a non-diabetic population . We tested differing definitions of insulin resistance based on the HOMA-IR. Defining those in the top tertile, quintile or decile as insulin resistant did not affect the findings from the analysis. Additionally the sample consisted of young adults and it may not be appropriate to extrapolate our findings to older individuals.
In this cross sectional analysis of a young adult population the ability of these waist circumference cut points to predict more established outcomes of insulin resistance such as coronary disease could not be assessed. If we are able to re-examine this cohort, data on intermediate outcomes of cardiovascular disease such as carotid intima-media thickness or endothelial reactivity can also be collected. We are not aware of any other studies of black youth from the Caribbean (a region that is currently undergoing the epidemiologic transition) with detailed laboratory measurement of metabolic abnormalities and insulin resistance.