Our results revealed high prevalence rates for the metabolic syndrome using both diagnostic criteria but much higher with IDF criteria. These rates were significantly higher in women than in men, independently of the criteria used. Concordance between both definitions was low-to-average overall, and by gender, with the best agreement observed in women. Eighty percents of participants classified as not having MS by the IDF criteria were also classified within this category by the NCEP-ATP III criteria, while NCEP-ATP III reclassified the 20% remaining as having MS. Among those diagnosed with MS on the basis of the IDF criteria, the NCEP-ATP III criteria confirmed the diagnosis of MS in 76% and reclassified 24% as not having MS. Combining the two sets of criteria marginally improved the yield beyond that provided by the IDF criteria alone in men, but not in the overall population and in women.
Few prevalence studies of MS have been conducted in Africa, including in patients with diabetes . Available studies in the general population suggest an increasing prevalence of MS with time, and collectively, have provided variable prevalence of MS ranging from low to as high as figures reported in developed countries [12, 13]. Studies in people with diabetes, conducted in large part in Nigeria have been based on different definition criteria and have also provided variables results. One of the earliest was based on a sample of 218 type 2 diabetic outpatients receiving care at the Olabisi Onabanjo University Teaching Hospital in Nigeria from 1999 to 2001 . In this study, Alebiosu and co-workers found a prevalence rate of 25.5%, similarly among men and women using the WHO criteria . This prevalence was half the figure reported subsequently by Adediran et al.  in a group of 408 type 2 diabetic individuals at the University Teaching Hospital in Lagos (Nigeria) applying the same definition criteria. The prevalence rate in this study was 51% overall, 44% in men and 56% in women . This was still lower than the 54–59% rate found by Isezuo and Ezunu among diabetic patients at the Usmanu Danfodiyo University Teaching Hospital, Northwestern Nigeria in 2002 [16, 17]. More recent studies in the same country have found much higher prevalence rates: 63.6% in one study based on IDF criteria , 62.5% in a second based on the NCEP-ATP III criteria ; and 25.5% , 60% , 69% (in study that enrolled only women)  and 86%  in four studies from the same institution/investigators applying the Joint Interim Statement (JIS) criteria . Elsewhere in Africa, a 43% prevalence rate was reported among 109 diabetic subjects in a tertiary care diabetes clinic in Zimbabwe ; and rates of 66.8%, 85.5%, 74% among men and 87.1%, 79.7%, 93% among women, according to the NCEP-ATP III, WHO and IDF criteria, respectively, reported in Seychelles . One study compared the prevalent of MS based on IDF criteria in African and White South African diabetics. Rates of MS in this study were lower in Africans than in Whites (46% vs. 74%) .
It follows from the above that prevalence rates found in our study are among the highest so far reported in sub-Saharan Africa. In general however, available reports from Africa have mostly provided non-standardised prevalence rates, which make direct comparison between studies less reliable. Those reports have also been based on patients with variable duration of diabetes, which can further increase the disparities in the prevalence of MS between studies. Indeed, the prevalence of components of MS is likely to increase with increasing duration of diabetes. Our prevalence rates compare with reports from various settings around the world [28–30]. These similarities suggest that the prevalence of the MS in people with type 2 diabetes seems to be independent of ethnic factors as the prevalence rates remain high independent of the definition criteria used.
While some studies in Africa have simultaneously applied different definition criteria for MS in the same population, very few (and perhaps none in people with diabetes) have directly compared the performance of different sets of criteria. One study in the general population in rural South Africa found a good agreement between the JIS criteria on one side and either the IDF (kappa = 0.90) or the NCEP-ATP III (kappa = 0.77) on the other side . This agreement in major ways is expected based on the fact that JIS more or less is just a loose adaptation of the IDF and NCEP-ATP III criteria . Therefore virtually all patients with MS based on either IDF or NCEP-ATP III criteria would also have MS based on JIS criteria. The South African study however did not directly compare the IDF and NCEP-ATP III criteria . In another study in Seychelles, the agreement between the IDF, NCEP-ATP III and WHO criteria was assessed overall and after exclusion of participants with diabetes . The agreement between the IDF and NCEP-ATP III criteria was acceptable-to-good while those for IDF vs. WHO and NCEP-ATP III vs. WHO were only modest-to-acceptable, with in both cases only a modest decrease in the point estimates upon exclusion of participants with diabetes . Both the South African and Seychelles reports were all based on community samples. In our hospital-based sample of treated diabetic individuals with varying duration of the disease, we found a low-to-acceptable agreement between the IDF and NCEP-ATP III criteria. The agreement was particularly poor among men. This is essentially explained by differences in the contribution of abdominal obesity to the definition of MS based on IDF or NCEP-ATP III criteria [4, 9].
Though both definitions identified a high prevalence of the metabolic syndrome, combining the two sets of criteria marginally improved the yield beyond that provided by the IDF criteria alone in men, but not in the overall population and in women. This may likely imply that using both definitions is of no value but instead a need to use a single and adaptable definition to a particular population. The JIS criteria published in 2009 aimed to achieve the above . While the JIS Committee has agreed on the principle of a definition with no obligatory component, it also explicitly recognized two facts: 1) the abdominal obesity (waist circumference) component should be based on population-specific threshold; 2) Most patients with type 2 diabetes mellitus will have the metabolic syndrome by the JIS criteria. There is currently no specific cut-off of waist circumference for African Africans. Only one study has attempted to define such a cut-off based on cross-sectional data in South-Africa . This cut-off has not yet been validated for incident outcomes such as cardiovascular disease, and may therefore not be recommended for use in other settings. The study however, suggests that African-specific cut-off would likely be different from the Caucasian cut-off currently recommended for use in Africa . As predicted, studies that have applied the JIS criteria found very high rate of MS among people with diabetes in Africa . All participants in our sample (data not shown) qualified for MS based on the JIS criteria. This suggests that, until the appropriate cut-offs for waist circumference specific to Africans are well defined; the application of the JIS criteria in people with diabetes in this setting would be of limited contribution.
Our study has some limitations that must be accounted for when interpreting our findings. This was a cross-sectional study conducted in a referral health institution. Therefore, whether our findings reflect those from a broader population with diabetes at the community or primary care level is not known. However, virtually all studies on MS in Africa have been conducted in tertiary care institutions, which make our findings comparable with those from other countries in the region. Our study was based on patient files and registers, and therefore we had no control on the quality of data used in the study. It is possible that our estimates are imprecise due to measurement errors. While imprecision in the levels of risk factors could affect the prevalence of MS, it was less likely to affect the agreement statistics between sets of criteria.