Metabolic syndrome according to different definitions in a rapidly developing country of the African region

Aims We examined, in a country of the African region, i) the prevalence of the metabolic syndrome (MetS) according to three definitions (ATP, WHO and IDF); ii) the distribution of the MetS criteria; iii) the level of agreement between these three definitions and iv) we also examined these issues upon exclusion of people with diabetes. Methods We conducted an examination survey on a sample representative of the general population aged 25–64 years in the Seychelles (Indian Ocean, African region), attended by 1255 participants (participation rate of 80.3%). Results The prevalence of MetS increased markedly with age. According to the ATP, WHO and IDF definitions, the prevalence of MetS was, respectively, 24.0%, 25.0%, 25.1% in men and 32.2%, 24.6%, 35.4% in women. Approximately 80% of participants with diabetes also had MetS and the prevalence of MetS was approximately 7% lower upon exclusion of diabetic individuals. High blood pressure and adiposity were the criteria found most frequently among MetS holders irrespective of the MetS definitions. Among people with MetS based on any of the three definitions, 78% met both ATP and IDF criteria, 67% both WHO and IDF criteria, 54% both WHO and ATP criteria and only 37% met all three definitions. Conclusion We identified a high prevalence of MetS in this population in epidemiological transition. The prevalence of MetS decreased by approximately 32% upon exclusion of persons with diabetes. Because of limited agreement between the MetS definitions, the fairly similar proportions of MetS based on any of the three MetS definitions classified, to a substantial extent, different subjects as having MetS.


Introduction
The metabolic syndrome (MetS) represents a cluster of metabolic risk factors that co-occur to a greater degree than predicted by chance. While it is clear that MetS is associated with cardiovascular disease and diabetes [1], it is still controversial whether MetS adds predictive value that goes beyond that of its single constituents [1,2]. Insulin resistance was initially the main focus of MetS, but adiposity, sedentary lifestyle, dietary and genetic factors have also received much attention when considering the pathogenesis of MetS [3]. High prevalence of MetS has often been documented in developed countries and increasingly so in developing countries [4][5][6][7][8][9][10][11][12][13][14][15][16][17], although data in developing countries and particularly in the Sub-Saharan region remain scarce [18].
Over the past decade, several definitions of MetS have been coined. In this paper, we consider three major definitions: i) the definition of the World Health Organization (WHO), issued in 1998 [19]; ii) the definition of the National Cholesterol Education Program Adult Treatment Panel III (ATP), issued in 2001 and updated in 2004 and 2005 [20], and the definition of the International Diabetes Federation (IDF), introduced in 2005 [21]. These definitions agree that the core criteria of MetS include: i) blood glucose impairment (hyperglycemia and/or insulin resistance), ii) excess abdominal/body fat (increased waist and/or obesity), iii) dyslipidemia (low HDL-cholesterol and/or high triglycerides), and iv) elevated blood pressure. However, criteria and cut-off values differ between these definitions, implying that different definitions may identify different people, as documented in the few studies that have addressed this question [4,7,8,22].
There is some controversy over whether identification of MetS should exclude people with diabetes since diabetes alone is sufficient to define high cardiovascular risk and MetS has been used as a tool to predict diabetes [2,23]. This analysis had four main objectives. First, we evaluated the prevalence of MetS according to three MetS definitions in the Seychelles, a rapidly developing country of the African region. Second, we examined the distribution of the MetS criteria according to the different MetS definitions. Third, we compared the level of agreement between the different MetS definitions and their ability to identify the same subjects. Fourth, we examined how the prevalence of MetS and the other end points differed upon restricting assessment of the metabolic syndrome to the non-diabetic population.

Survey procedures
The Republic of Seychelles is a group of islands in the Indian Ocean (African region) situated approximately 1800 km east of Kenya. A large majority of the population is of African descent. HIV and cardiovascular diseases account for approximately 1% and 38% of total mortality, respectively [24]. High prevalence of several cardiovascular risk factors including obesity, hypertension and diabe-tes has been documented as early as in 1989 [25] and in 2004 [26][27][28].
A population-based survey of cardiovascular risk factors was conducted in 2004 under the auspices of the Ministry of Health of the Republic of Seychelles (Seychelles Heart Study III). Methods and main results have been published previously [26][27][28]. Briefly, the sampling frame consisted of a sex and age stratified random sample of the entire population aged 25-64 years. Eligible participants were selected from computerized data of a national population census in 2002, thereafter updated by the civil status authorities. The survey was approved by the Ministry of Health after technical and ethical reviews. Eligible participants were invited by a letter to attend the survey at study centers on specified dates. They were free to participate and gave written informed consent.
Blood pressure was defined as the average of the last two of three measurements with a mercury sphygmomanometer taken at intervals longer than 2 minutes after the participants had been sitting for at least 30 minutes. Weight was measured with precision electronic scales (Seca, Hamburg) and height was measured with a fixed stadiometer.
Fasting blood was taken between 7:00 and 10:00 am. Serum was obtained within 2 hours of blood collection and immediately frozen to -20°C. Blood lipids were measured with standard methods. Fasting blood glucose (FBG) was analyzed immediately with a point-of-care analyzer (Cholestec LDX, Hayward, USA). If glucose was ≥ 5.6 mmol/l and the participant was not aware of having diabetes, an additional capillary measurement was performed within 10 minutes with a glucometer that adjusts readings to plasma values (Ascencia Elite, Bayer) and the average of the two readings was considered [27]. An oral glucose tolerance test (OGTT) was performed on all people who had FBG between 5.6 and 7.0 mmol/L but had never been diagnosed with diabetes. Fasting serum insulin was measured using commercial RIA kits (LINCO Research Inc, Missouri, USA). Microalbuminuria was measured with a semi-quantitative method using a Clinitek Status analyzer (Bayer, Leverkussen, Germany) [28].

Definitions of the metabolic syndrome
The prevalence of MetS was calculated according to the standard criteria detailed in Table 1. For the WHO definition, we defined insulin resistance as the upper quartile of the homeostasis model assessment of insulin resistance (HOMA-IR), calculated as [fasting serum insulin level (microU/ml) × fasting blood glucose level (mmol/L)]/ 22.5 [29], which has been shown to be a reliable estimate of insulin resistance both among diabetic and non diabetic subjects [30,31]. For the IDF definition, we used ethnic-specific cutoff values for waist circumference, i.e. the same criteria for Europid and Sub-Saharan subjects (≥ 94 cm for men and ≥ 80 cm for women) [21] and, by extension, for people of 'mixed' descent, and ≥ 90 cm for men and ≥ 80 cm for women of the 4% of participants considered as 'Indian' or 'Chinese'. Information on treatment for dyslipidemia was not available at an individual level and we assumed that no one was under treatment, consistent with the very low number of people treated in Seychelles for dyslipidemia.

Analysis
Analysis was performed with Stata 9.0. All analyses were standardized for age using the new WHO standard population [32]. Agreement between the three MetS definitions was determined using the Kappa statistic (κ). We graphed Venn diagrams for agreement between different definitions of MetS using the MATLAB 7.6 software, which allows representing areas proportionally to the corresponding numbers of cases. P values < 0.05 were considered statistically significant.

Characteristics of the population
From the 1563 eligible participants, 1255 participated in the survey (80.3% participation rate) and 1218 had all necessary measurements to evaluate MetS according the three considered definitions and were included in this study. The distribution of selected risk factors and MetS criteria is presented in Table 2. The prevalence of MetS increased markedly with age for both genders. However, the prevalence of MetS using the ATP definition was highest at age 45-54 for men. With regards to gender, the prevalence was significantly greater in women than in men according to the ATP and IDF definitions while the prevalence was similar in both genders using the WHO definition.

Prevalence of the metabolic syndrome
A large proportion of people with diabetes also had MetS: 66.8%, 85.5%, 74% among men and 87.1%, 79.7%, 93% among women, according to the ATP, WHO and IDF definitions, respectively. However, this proportion was smaller among men at age 45-64 years than in the other age categories (Figure 1).
Upon exclusion of individuals with diabetes, the agestandardized prevalence of MetS at age 25-64 years was Diabetes, IFG, IGT or insulin resistance 1 plus two or more of the remaining criteria.
Any three or more criteria. Central obesity plus two or more of the remaining criteria.

Distribution of criteria of the metabolic syndrome
High blood pressure, obesity and impaired glucose regulation were the most prevalent criteria for all MetS definitions ( Figure 2). Hypertriglyceridemia was found more often in men than in women, while the opposite was true for low HDL-cholesterol. This criteria distribution was similar across age categories. Figure 3 shows the level of agreement between the three MetS definitions. 37% of all subjects aged 25-64 had MetS as based on any of the three MetS definitions. Among these MetS-holders, 69% had MetS based on ATP and IDF, 50% based on ATP and WHO, 53% based on In order to examine possible causes of fairly low agreement between the MetS definitions, we compared agreement between the criteria of the three MetS definitions in the entire population (

Discussion
The main findings of the study are as follows: i) the prevalence of MetS was high in this population of East Africa regardless of which MetS definition was used; ii) the prevalence of MetS decreased by approximately one third upon exclusion of persons with diabetes; and iii) agreement between different MetS definitions was limited and consequently, the similar prevalence of MetS according to either MetS definition actually identified, to a substantial extent, different subjects as having MetS.
In our study, the prevalence of MetS at the age of 25-64 years ranged between 25% and 30%. This is much higher than the 8% reported in Cameroon [18], which is, to the best of our knowledge, the only other population based published assessment of the prevalence of MetS in Sub-Saharan Africa. A previous study in the Seychelles [34] found that plasma aldosterone, but not plasma renin activity, was associated with MetS. However this study included participants from families with hypertension and was therefore not intended to assess the prevalence of MetS in the general population. In order to compare the prevalence in Seychelles with that in other countries, we compiled findings of selected population-based studies that had assessed MetS according to at least two MetS definitions, and included participants of 35-64 years ( Table   Proportion of  ). The prevalence of MetS was similar in Seychelles as in many Western countries [9][10][11]13] and in urban India [15]. The prevalence of MetS in Seychelles was lower than in certain countries, e.g. USA [5], Portugal [8], Samoa [13], Turkey [7] and Tunisia [4] but higher than in Mexico [6] and several Asian countries e.g. Korea [14], Japan [13] and China [17]. The high prevalence of MetS in Seychelles is consistent with high prevalence of several MetS criteria, particularly overweight, hypertension, dyslipidemia and diabetes [25][26][27].
The prevalence of MetS in Seychelles did not differ markedly according to the three different MetS definitions, consistent with observations in several populations [5,9,10,14,17], but not in others [4][5][6][7]12,13]. However, the prevalence of MetS in Seychelles was moderately higher according to IDF than ATP or WHO, a finding reported in most studies that had assessed this issue [4,5,7,8,[10][11][12][13]15], except for a few [16,17]. It has frequently been reported that the prevalence of MetS according to the WHO definition is generally greater among men Agreement between three definitions of the metabolic syndrome (all prevalence estimates are expressed as percent of the general population) Figure 3 Agreement between three definitions of the metabolic syndrome (all prevalence estimates are expressed as percent of the general population). The level of agreement between the different definitions of MetS was not optimal in our study. Less than half of the individuals labeled as having MetS according to any of the three considered definitions were classified as MetS-holders according to all three definitions. We found that agreement between the MetS definitions was better for IDF-ATP than for WHO-IDF and WHO-ATP. This finding is consistent with data in several populations on several continents [7,8,10,13,15]. This difference is expected since MetS is based on the same criteria according to IDF and ATP, except for the adiposity criterion. Agreement between the MetS definitions was generally better among women than men in Seychelles, consistent with previous reports [5,8,14]. This may relate to the fact, at least in Seychelles, that the prevalence of the adiposity criterion was virtually identical and close to 100% across the three MetS definitions among women with MetS, but was lower and largely different between MetS definitions in men. The prevalence of MetS restricted to the non-diabetic population (e.g. according to the EGIR definition, which is a definition of MetS that explicitly excludes diabetes [33]) was similar in Seychelles as in several countries in Europe [37][38][39] and higher than in Japan and Korea [13]. In the Seychelles, about one third of the adults who had MetS also had diabetes and, inversely, most adults who had diabetes also had MetS. Hence, the prevalence of MetS, according to either the ATP, WHO and IDF definitions, decreased markedly (a relative decrease of approximately one third) upon exclusion of persons with diabetes. This difference is larger than reported in other studies [6,13]. We also observed that this decrease in the prevalence of MetS upon exclusion of diabetes was larger according to the WHO definition compared to the ATP and IDF definitions, which has also been observed in Samoa [13] and Mexico [6].

Men Women
The issue of whether MetS should be defined upon exclusion of diabetes is relevant to both clinical practice and epidemiology. It has been argued that MetS status does not add incremental information for cardiovascular management of diabetic individuals [23], since diabetes alone defines high cardiovascular risk. For the same reason, it  has been suggested that the definition of MetS should exclude individuals with known cardiovascular disease [23]. The question of whether MetS should exclude diabetes and/or cardiovascular disease is clearly dependent on the expected purpose of MetS, i.e. whether MetS is used to predict cardiovascular diseases, diabetes, insulin resistance, or a combination of these conditions independent of conventional risk factors. While our cross-sectional study emphasizes that different definitions of MetS tend to classify different individuals as having MetS, the most important question remains to determine, based on longitudinal data, whether MetS predicts cardiovascular events above the risk factors that constitute the syndrome.
Our study has some limitations. First, we cannot exclude a healthy volunteer bias related to voluntary participation to the study, which could tend to underestimate the actual prevalence of MetS in the population. Second, we assessed insulin resistance based on HOMA-IR, which is only partially correlated with the gold standard (hyperinsulinemic euglycemic clamp), yet a valid proxy for insulin resistance [30,31]. Third, OGTT was only performed on participants unaware of having diabetes and who had FBG between 5.6 and 7.0 mmol/l, hence a source of slight underestimation of diabetes. On the other hand, strengths of the study include the population-based design of the study, a fairly large sample size, and a broad panel of investigations that allowed us to directly compare the different MetS definitions.
In conclusion, this study contributes to mapping the prevalence of MetS worldwide, particularly with regards to the African region. The study also further contributes to the longstanding debate regarding the significance of MetS. The substantially different prevalence of MetS upon exclusion of individuals with diabetes underlies the need to clarify whether MetS is a tool for predicting cardiovascular disease, diabetes, insulin resistance or any other risk condition. These issues are further prompted by the finding, in our study, that the different considered MetS definitions identified, to a substantial extent, different individuals. Not withstanding a much needed unified definition of MetS, our findings in Seychelles emphasize the growing burden of lifestyle-related non-communicable diseases in countries in epidemiological transition including in the African region, consistent with the ongoing epidemic of obesity worldwide and in the Seychelles in particular.