Journal of Internal Medicine 2004; 256: 37 47

Transcription

1 Journal of Internal Medicine 2004; 256: High incidence of type 2 diabetes and increasing conversion rates from impaired fasting glucose and impaired glucose tolerance to diabetes in Mauritius S. SÖDERBERG 1,2, P. ZIMMET 1, J. TUOMILEHTO 3, M. DE COURTEN 1, G. K. DOWSE 1, P. CHITSON 4, H. STENLUND 2, H. GAREEBOO 4, K. G. M. M. ALBERTI 5 & J. SHAW 1 From the 1 International Diabetes Institute, Melbourne, Victoria, Australia, 2 Department of Public Health and Clinical Medicine, University of Umeå, Umeå, Sweden, 3 Department of Epidemiology and Health Promotion, National Public Health Institute and Department of Public Health, University of Helsinki, Helsinki, Finland, 4 Ministry of Health, Port Louis, Mauritius, and 5 Department of Metabolic Medicine, Imperial College, London, UK Abstract. Söderberg S, Zimmet P, Tuomilehto J, de Courten M, Dowse GK, Chitson P, Stenlund H, Gareeboo H, Alberti KGMM, Shaw J (International Diabetes Institute, Melbourne, Victoria, Australia; University of Umeå, Umeå, Sweden; National Public Health Institute, University of Helsinki, Helsinki, Finland; Ministry of Health, Port Louis, Mauritius; and Imperial College, London, UK) High incidence of type 2 diabetes and increasing conversion rates from impaired fasting glucose and impaired glucose tolerance to diabetes in Mauritius. J Intern Med 2004; 256: Objective. To describe the incidence of different stages of glucose intolerance in a population from Mauritius followed over 11 years. Research design, methods and subjects. Populationbased surveys were undertaken in the multi-ethnic nation of Mauritius in 1987, 1992 and 1998 with 5083, 6616 and 6291 participants, respectively. Questionnaires, anthropometric measurements, and a 2-h 75-g oral glucose tolerance test were included. Three cohorts aged between 25 and 79 years with classifiable glucose tolerance data were identified; 3680 between 1987 and 1992, 4178 between 1992 and 1998, and 2631 between 1987 and Glucose tolerance was classified according to WHO 1999 criteria. Results. The incidence rate of type 2 diabetes was higher between 1992 and 1998 than between 1987 and In men, the incidence was similar between cohorts (24.5 and 25.4 per 1000 personyears) whereas the incidence increased in women (23.3 and 16.4 per 1000 person-years). The incidence of diabetes peaked in the year age group and then plateaud or fell. The incidences of impaired glucose tolerance (IGT) and impaired fasting glucose (IFG) decreased in both men and women. Of normoglycaemic subjects at baseline, more women than men developed IGT and more men than women developed IFG. Of those labelled as IFG in 1987, 38% developed diabetes after 11 years. The corresponding figure for IGT was 46%. Conclusions. In this study, we report changes in incidence rates of glucose intolerance over a 11-year period. In particular, differences between men and women were observed. The increased incidence of IGT in women compared with men, and increased incidence of IFG in men compared with women was consistent with, and explains the sex biases seen in the prevalences of these states. Keywords: cohort studies, diabetes mellitus, impaired fasting glucose, impaired glucose tolerance, incidence, Mauritius. Introduction Over the last half century there has been rapid socioeconomic development in many countries resulting in a move from a traditional to a modern way of life. In virtually all populations, higher fat diets and decreased physical activity have accompanied the benefits of modernization. These changes in diet and physical activity levels, combined with increasing longevity, have formed the basis for dramatic Ó 2004 Blackwell Publishing Ltd 37

2 38 S. SÖDERBERG et al. increases in the prevalence of type 2 diabetes mellitus in both developed and developing countries. However, the prevalence of various degrees of glucose intolerance, i.e. type 2 diabetes mellitus, impaired glucose tolerance (IGT), and impaired fasting glucose (IFG), varies considerably between populations. Certain ethnic groups, such as North American Indians [1], Mexican Americans [2], Australian Aborigines [3], Micronesian and Polynesian Pacific Islanders [4, 5], Asian Indians [5, 6], and Chinese [5, 7] are highly susceptible to developing glucose intolerance when their life-style changes from a traditional to a more modernized pattern. Since World War II, the southern Indian Ocean island of Mauritius has experienced a rising standard of living associated with industrialization, and there has been a marked change in the mortality profile from predominantly infectious to chronic non-communicable diseases [8]. The development of diabetes and other noncommunicable diseases such as cardiovascular disease in Mauritius may predict the future pattern in many parts of Asia and Africa. Concerned by this phenomenon, the government of Mauritius launched a programme in 1987 which aimed at the prevention and control of these diseases. A baseline population-based cross-sectional study of disease and risk factor prevalence was conducted in 1987, and formed the initial phase of the programme. Follow-up studies were conducted in 1992 and 1998, creating a unique opportunity for evaluation of trends in the prevalence and incidence of glucose intolerance over 11 years, and the impact of various risk markers. In this paper we describe incidence figures over 11 years using the recently modified WHO criteria for glucose intolerance [9]. Methods Background and subjects Mauritius is a subtropical island located in the southwestern Indian Ocean with a population of about 1.2 million. An estimated 70% of the population are of Asian Indian origin (54% Hindu and 16% Muslim), 2% are of Chinese origin and 28% are of the general population, which mainly comprises people with mixed African and Malagasy ancestry with some European and Indian admixture (Creoles) [10]. Population based surveys were undertaken in 1987 (age range 25 74), 1992 (25 years and older) and 1998 (20 years and older). Details of the survey methodology have been published previously [11 13]. In 1987, 10 randomly selected (with probability proportional to size) population clusters and a purposely selected area of Chinatown in the capital, Port Louis, were surveyed, and all eligible residents were invited to participate. In 1992 and 1998, the same clusters were resurveyed, as well as an additional three clusters selected to assess if trends in disease and risk factor distribution observed in the study cohort also occurred in independent clusters [13]. In 1998, the Chinatown cluster was not surveyed. In both 1992 and 1998, all previous participants were invited, plus all other current eligible residents in each cluster. The numbers of participants were 5083 (86% response rate), 6616 (89% response rate) and 6291 (87% response rate), respectively, in 1987, 1992 and Altogether, 9688 individuals participated and 58% of participants took part in more than one survey (31% in two surveys and 27% in three surveys). The proportion of participants from each survey, who were seen in more than one survey, was 78% in 1987, 83% in 1992 and 72% in Three cohorts of nonpregnant participants aged between 25 and 79 years with classifiable data from two separate surveys were identified for the study of incidence rates: 3680 between 1987 and 1992 (72% of the participants in 1987), 4178 between 1992 and 1998 (63% of the participants in 1992), and 2631 between 1987 and 1998 (52% of the participants in 1987). Notably, those participating in three surveys were eligible for all three cohorts. Censored cases were participants with classifiable data 1987 or 1992 who did not attend for the follow-up survey: 1132 between 1987 and 1992, 1818 between 1992 and 1998, and 1885 between 1987 and The survey protocol was reviewed and approved by the local ethics committee (Melbourne, Australia). Survey procedures The survey methodology was similar in all three surveys. All eligible adults were asked to attend a survey site between and hours, after an overnight fast. After registration, trained local

3 INCIDENCE OF GLUCOSE INTOLERANCE IN MAURITIUS 39 nurses administered a questionnaire and anthropometric measurements were performed [11 13]. Ethnicity was determined by self-report. All subjects not taking diabetes medication had a 2-h 75-g oral glucose (glucose monohydrate) tolerance test. Fasting and 2-h venous blood samples were centrifuged and separated immediately. Plasma glucose was measured on-site in 1987 and 1992 using YSI glucose analyzers (Yellow Springs Instruments, OH, USA) within 3 h of collection with quality controls measured several months later in Newcastle upon Tyne, UK, at the reference laboratory, which is a member of the Wellcome quality assurance scheme. The quality controls showed a small, but consistent and systematic fall in glucose over time. In 1998, plasma was frozen immediately and glucose was measured within approximately 4 months in Newcastle upon Tyne, UK, using the same technology. Considering the delay in analysis, the 1998 values were adjusted upwards using an equation (adjusted glucose ¼ measured glucose) based on the difference between on-site values and quality controls from the 1987 and 1992 surveys. Glucose tolerance status was determined according to 1999 WHO criteria [9]. Diabetes was diagnosed if subjects reported a history of diabetes and were taking hypoglycaemic medication, or the fasting plasma glucose level was 7.0 mmol L )1 and/or the 2-h value was 11.1 mmol L )1. Subjects that reported a history of diabetes were labelled as known diabetes (KDM) if they were taking hypoglycaemic medication or if they had diabetic fasting and/or 2-h glucose values. All other cases were labelled newly diagnosed diabetes (NDM). Cases of diabetes were almost exclusively type 2 [14]. Subjects with a fasting plasma glucose <7.0 mmol L )1 but a 2-h value from 7.8 mmol L )1 to <11.1 mmol L )1 were defined as having IGT. Subjects with a fasting plasma glucose from 6.1 mmol L )1 to <7.0 mmol L )1 but a 2-h value <7.8 mmol L )1 were categorized as having IFG. Statistical analysis The incidence data were standardized by the direct method to the person-year distribution of the cohort, and for female data, the male distribution was used. Three separate cohorts were defined (those with data between 1987 and 1992, between 1992 and 1998, and between 1987 and 1998) for estimation of incidence rates of diabetes, IGT and IFG. For calculation of incidence rates, subjects were defined as incident diabetes if they come from IGT, IFG or normal glucose tolerance (NGT), incident IGT if they come from IFG or NGT, and incident IFG from NGT. The estimated incidence rate in cases per 1000 person-years was calculated with the assumption of a date of diagnosis in the middle of the follow-up period. Censored cases contributed with person-years for half of the period. Incidence rates are presented with 95% CI and rates were compared by calculating incidence rate ratios (IRR) with 95% CI. This difference was statistically significant if the CI did not include one. A Poisson regression analysis was used to estimate the gender effect between 1987 and 1998 allowing for adjustments. For the cohorts between 1987 and 1992, and between 1992 and 1998, a stratified analysis was chosen because of significant interactions between period, gender and age. Means and proportions are presented with 95% CIs and significant differences between means were based on comparison of 95% CIs, and between two proportions by calculation of the 95% CI of the difference. This difference was labelled statistically significant if the CI did not include zero. Pregnant women were excluded from all analyses. All calculations were performed with SPSS,version 10.0 (Chicago, IL, USA) or Stata, version 8.0 (College Station, TX, USA). Results Baseline characteristics The characteristics of the three cohorts are presented in Table 1. Participants in the 6-year cohort recruited in 1992 were older, more obese, had higher fasting and postload glucose levels, but had lower blood pressure and were less likely to be smokers than participants in the 5-year and 11-year cohorts recruited in Participants in the 11-year cohort were younger and had lower fasting glucose and lower systolic blood pressure compared with the 5-year cohort. The proportion of women did not differ between the cohorts but was significantly higher in the cohorts than in the groups of participants seen only once in 1987 or 1992 (in both 50%, 95% CI 47 53). Those who only attended

4 40 S. SÖDERBERG et al. Table 1 Baseline characteristics for longitudinal cohorts in Mauritius Number Age (years) 42.8 ( ) 45.2 ( ) a 41.2 ( ) ab Female sex (%) 54 (52 55) 56 (54 57) 56 (54 58) BMI (kg m )2 ) 23.8 ( ) 25.2 ( ) a 23.9 ( ) b Waist circumference (cm) 76.5 ( ) 84.9 ( ) a 76.5 ( ) b Hip circumference (cm) 90.1 ( ) 96.7 ( ) a 90.5 ( ) b WHR (ratio) 0.85 ( ) 0.88 ( ) a 0.84 ( ) b Fasting glucose (mmol L )1 ) c 5.6 ( ) 5.9 ( ) a 5.5 ( ) ab Postload glucose (mmol L )1 ) c 6.6 ( ) 6.8 ( ) a 6.5 ( ) b Systolic blood pressure (mmhg) ( ) ( ) a ( ) ab Diastolic blood pressure (mmhg) 77.8 ( ) 75.3 ( ) a 77.1 ( ) b Daily smoker (%) 29 (28 31) 22 (21 23) a 27 (26 29) b Characteristics for non-pregnant participants at baseline (1987 or 1992) within the age range of years, stratified for cohort. Differences between cohorts: a, significantly different from ; b, significantly different from Values are given as numbers (n) or means ( c geometric) and 95% CI. Table 2 Incidence of IFG, IGT and DM in Mauritius between 1987 and 1992, and Incidence of IFG Incidence of IGT Incidence of DM Group Period n IR (95% CI) n IR (95% CI) n IR (95% CI) All / ( ) 358/ ( ) 295/ ( ) / ( ) a 425/ ( ) a 479/ ( ) a / ( ) 309/ ( ) 456/ ( ) Men / ( ) b 143/ ( ) b 161/ ( ) b / ( ) ab 180/ ( ) bc 219/ ( ) c / ( ) 121/ ( ) b 211/ ( ) Women / ( ) 215/ ( ) 134/ ( ) / ( ) a 245/ ( ) a 260/ ( ) a / ( ) 188/ ( ) 245/ ( ) Censored cases are not included. Values are given as numbers (cases/persons at risk) and adjusted incidence rates per 1000 person years (IR) with 95% CI. IFG, impaired fasting glucose; IGT, impaired glucose tolerance; DM, diabetes mellitus. Significant differences: a, versus ; b, men versus women (within period), and c, men versus women (between periods). once were also younger and less obese than participants in the cohorts (data not shown). Adjusted incidence rates of diabetes The incidence rates of different categories of glucose intolerance are given in Table 2. (Presented data does not include censored cases unless stated.) Women had a higher incidence rate of diabetes during the second period than the first (IRR ¼ 1.42, 95% CI ), but there was no change for men (IRR ¼ 1.03, 95% CI ). Furthermore, more men than women developed diabetes during the first 5-year period (IRR ¼ 1.50, 95% CI ), whereas there was no significant difference between genders in this respect in the subsequent 6-year period (IRR ¼ 1.09, 95% CI ), or in the 11-year period (IRR ¼ 0.91, 95% CI ). Figure 1 shows that the incidence of diabetes was highest in the year age groups in all cohorts, except for women between 1987 and 1992, for whom the incidence was highest in the year-old age group. The incidence rates (per 1000 person-years) of physician diagnosed diabetes (KDM) were 2.3 ( ), 4.6 ( ), and 6.1 ( ) in the 5, 6 and 11-year periods respectively, whilst the incidence rates of diabetes diagnosed through survey glucose testing (NDM) were 18.0 ( ), 19.3 ( ) and 13.8 ( ), respectively. The incidence of KDM increased significantly between the 5 and 6-year periods (IRR ¼ 1.99, 95% CI ), whilst the

5 INCIDENCE OF GLUCOSE INTOLERANCE IN MAURITIUS 41 IR/1000 person years IR men IR men IR women IR women IR/1000 person years IR men IR men IR women IR women Age group Age group Fig. 1 Incidence rates for diabetes stratified for period, gender and age-group. The age group years was omitted because of few cases. Fig. 2 Incidence rates for IGT stratified for period, gender and agegroup. The age group years was omitted because of few cases. incidence of NDM remained stable (IRR ¼ 1.07, 95% CI ). A stratified analysis revealed that the incidence of diabetes increased both in Asian Indians and in Creoles between the 5 and 6-year periods (IRR ¼ 1.15, 95% CI and IRR ¼ 1.41, 95% CI , respectively). Both Creole and Chinese subjects had a lower incidence of diabetes than Asian Indians during the initial 5-year period (IRR ¼ 0.86, 95% CI and IRR ¼ 0.85, 95% CI , respectively). The difference between Creole and Indian subjects disappeared in the subsequent 6-year period (IRR ¼ 1.05, 95% CI ). Any comparison with Chinese subjects in the final 6-year period was not meaningful because of small numbers. Including cases lost to follow-up, the incidence rates of diabetes per 1000 person-years were 17.4 ( ), 19.2 ( ) and 14.4 ( ) in the 5, 6 and 11-year periods respectively. Adjusted incidence rates of IGT and IFG More women than men developed IGT (Table 2, Fig. 2), during the 5-year period (IRR ¼ 1.48, 95% CI ) and the 6-year period (IRR ¼ 1.21, 95% CI ). This difference remained after adjusting for the age distribution during the 11-year period (IRR ¼ 1.38, 95% CI ). In contrast, the incidence of IFG (Table 2, Fig. 3) was lower in women than men during both the 5 and 6-year periods (IRR ¼ 0.50, 95% CI and IRR ¼ 0.61, 95% CI , respectively), a difference that IR/1000 person years IR men IR men IR women IR women Age group Fig. 3 Incidence rates for IFG stratified for period, gender and agegroup. The age group years was omitted because of few cases. remained significant after taking the age distribution into account during the 11-year period (IRR ¼ 0.65, 95% CI ). The incidence of both IGT and IFG was lower in the second time period, than the first (IRR ¼ 0.79, 95% CI and IRR ¼ 0.51, 95% CI , respectively). Asian Indians developed more IGT than Creoles during the first 5-year period (IRR ¼ 1.34, 95% CI ) whereas Creole and Chinese subjects developed more IFG than Indians during the same period (IRR ¼ 1.62, 95% CI and IRR ¼ 2.45, 95% CI , respectively). Compared with Indians, the incidence of IFG remained higher in Creole subjects in the subsequent 6-year period (IRR ¼ 1.43, 95% CI ), whereas the incidence of IGT equalized (IRR ¼ 1.09, 95% CI ).

6 42 S. SÖDERBERG et al. Analysis of clusters Participants in the cohort had similar incidence rates for diabetes, IGT and IFG irrespective of whether they were seen for the first time in 1987 or 1992 (data not shown). Subjects recruited from the three new clusters in 1992 had a higher incidence of diabetes in the period than subjects from the 10 original sites (IRR ¼ 1.19, 95% CI ). However, this finding should be interpreted with caution as the difference was explained by a relative accumulation of diabetes cases in the and year age groups. These groups had fewer participants (31) than other age groups (686). Incidence according to baseline glucose tolerance status Table 3 shows that amongst those with NGT at baseline, the incidence of diabetes increased significantly between the two periods in women (IRR ¼ 1.76, 95% CI ), but not in men (IRR ¼ 0.88, 95% CI ). More women then men developed IGT from NGT (IRR ¼ 1.53, 95% CI and IRR ¼ 1.20, 95% CI , respectively), and more men than women developed IFG from NGT (IRR ¼ 2.00, 95% CI and IRR ¼ 1.65, 95% CI , respectively) in each period. For those with IFG followed over 11 years, 38.0% ( ) converted to NGT, 7.0% ( ) remained in the IFG category, 17.0% ( ) progressed to IGT, and 38.0% ( ) to diabetes. During the first period, more men than women progressed to diabetes (IRR ¼ 1.54, 95% CI ) whereas the rate was higher in women during the final period (IRR ¼ 1.23, 95% CI ). For IGT, the 11-year outcomes were: 23.6% ( ) converted to NGT, 4.1% ( ) to IFG, 26.4% ( ) remained in the IGT category, and 45.9% ( ) progressed to diabetes. The incidence of diabetes from IGT was higher in men than women during the 5 and 6-year periods (IRR ¼ 1.27, 95% CI and IRR ¼ 1.48, 95% CI , respectively). The conversion rates were higher during the final period compared with the initial period in both men (IRR ¼ 1.97, 95% CI ) and women (IRR ¼ 1.69, 95% CI ). During both periods, more women than men remained in the IGT category (IRR ¼ 1.22, 95% CI and IRR ¼ 1.30, 95% CI , respectively) or converted to NGT (IRR ¼ 1.16, 95% CI and IRR ¼ 1.37, 95% CI , respectively). Those with a diagnosis of diabetes usually retained this diagnosis. However, this was restricted to those with KDM, as more than 97% retained their diagnosis in all cohorts. Of those with NDM in the initial screening, more subjects re-surveyed with KDM after the last 6-year period than after the first 5-year period [49.2% ( ) versus 37.7% ( )]. Discussion The findings described in this paper confirm and extend our knowledge about the dynamics of the epidemic of type 2 diabetes mellitus seen in many parts of the world. Mauritius is an interesting focus for this kind of study as similar trends may be expected to appear in the future in parts of Africa and Asia with continued modernization, given the similarity of ethnic backgrounds. The incidence of diabetes in Mauritius is high and comparable with incidence rates described for Australian Aboriginals [3], Pacific Islanders [5] and Mexican Americans [15], but higher than in Black Americans and Europids [16 19]. The comparison with Pima Indians is difficult as different incidence rates are reported in the literature [1, 20]. We found an effect of gender on the incidence during the first 5-year period which is in contrast to most other studies in which no effect of gender on incidence rates was apparent [3, 21, 22]. The overall incidence rates for diabetes were stable during the study period. However, our findings suggest that there are dynamic changes, which could be amenable to preventative measures. For instance, the incidence of diabetes was higher in men during the first 5-year period whereas this gender-related difference was not seen in the second 6-year period due to a catch-up in women. This pattern suggests that Mauritian women may have been exposed to factors initiating the development of glucose intolerance more recently than men. However, increasing age and obesity in the 6-year period compared with the previous 5-year period is also a plausible explanation for this finding [23]. Interestingly, the incidence rate of diabetes during the

7 Table 3 Changes in glucose tolerance status in Mauritius between and NGT at follow-up [IR (95% CI)] IFG at follow-up [IR (95% CI)] IGT at follow-up [IR (95% CI)] DM at follow-up [IR (95% CI)] Category Cohort Men Women Men Women Men Women Men Women Men Women NGT ( ) ( ) c 20.4 ( ) ac 10.2 ( ) 24.8 ( ) 37.9 ( ) c ( ) a ( ) ( ) c ( ) ( ) ( ) c IFG ( ) ( ) bc ( ) ac ( ) b ( ) a ( ) ( ) ac ( ) ( ) c ( ) ( ) ( ) c IGT ( ) a ( ) bc ( ) ac ( ) ( ) a ( ) c ( ) ( ) c ( ) ( ) ( ) ( ) c DM ( ) c ( ) ( ) ( ) ( ) a ( ) bc ( ) ac ( ) ( ) ( ) ( ) ( ) 14.6 ( ) c 12.9 ( ) 54.1 ( ) c 60.5 ( ) 60.7 ( ) c ( ) ac ( ) ( ) 5.8 ( ) 10.3 ( ) b 35.1 ( ) 74.7 ( ) bc 47.9 ( ) 81.0 ( ) b ( ) ( ) bc Values are given as numbers (n) and standardized incidence rates (IR) per 1000 person years with 95% CI stratified for sex. NGT, normal glucose tolerance; IGT, impaired glucose tolerance; IFG, impaired fasting glucose; DM, diabetes mellitus. Significant differences: a, versus (men between cohorts but within category); b, versus (women between cohorts but within category); c, men versus women (within cohort and category). INCIDENCE OF GLUCOSE INTOLERANCE IN MAURITIUS 43

8 44 S. SÖDERBERG et al. 11-year follow-up was roughly similar to the incidence rates during the shorter periods, whereas the incidence of both IFG and IGT was markedly lower during the 11-year follow-up compared with the shorter periods. This phenomenon indicates clearly the transient nature of both IFG and IGT. Over time, many individuals who initially develop IFG or IGT will progress to diabetes and some will revert to normal [24, 25]. The incidence of diabetes was higher amongst those with IGT than those with IFG. This is likely to reflect the criteria for classification of glucose tolerance. By definition, people with IGT can be close to either the fasting or 2-h glucose criteria for diabetes. However, those with IFG can only be close to the fasting criterion, as 2-h glucose has to be under 7.8 mmol L )1. The narrow ranges of glucose levels for a diagnosis of IFG and IGT should also be kept in mind as a repeated measurement of blood glucose can easily lead to reclassification because of the imprecision in the analysis [26, 27]. However, even a transient diagnosis of IGT may be highly significant as a predictor for future diabetes [26]. Studies on the incidence of IGT are less common than those on diabetes incidence. In this report, we have chosen to focus on the progression of glucose intolerance, and converters from a higher degree of glucose intolerance are thus not included in the figures given in Table 2. Two studies of American whites followed over long periods have been reported. The Rancho Bernardo study included both men and women aged years and 2.2 and 2.5% developed IGT per year, respectively [28], compared with 1.2% for men and 1.6% for women in Mauritius. The Normative Aging Study enrolled only males, who were followed over 18 years, and the annual incidence of IGT was 1.2% [29]. In Nauru, two cohorts were followed [4], and the incidence rate declined from 1975 to 1987, and was lower (19.4 per 1000 person-years) than the incidence rate during the last 6-year period (27.3 per 1000 person-years) in Mauritius. Our estimates are thus comparable with existing data, but our results suggest that women are more prone to develop IGT than are men, and indeed women have a higher prevalence of IGT than men, both in Mauritius [11] and elsewhere [25]. However, the male to female ratio of IGT prevalence varies across populations and may reflect variations in other risk factors for IGT [30]. Impaired fasting glucose became a recognized entity only recently [31], and to our knowledge this is the first report on incidence for this condition. In contrast to IGT, both prevalence and incidence of IFG have a male predominance, which was consistently seen in all three cohorts. We also describe a significant decrease of IFG development in men during the second period. This could be the result of more subjects developing more severe glucose intolerance (IGT or diabetes) during the last period and to IFG being an unstable and intermediate condition. This was clearly shown over 11 years as only 7% of those initially diagnosed as IFG remained in this category. Interestingly, the conversion rate (per 1000 person-years) from IFG to diabetes ranged from 46.6 to 66.9 in this study, which is consistent with findings from both Taiwan in a group with persistent fasting hyperglycaemia [32] and from Europids [24, 33, 34]. Despite variation in measurements, these data clearly indicate that IFG represents an early step in the development of glucose intolerance. By contrast with IFG, a relatively high proportion (26%) of those with an initial diagnosis of IGT remained in the same category after 11 years. A larger proportion of those with IGT progressed to diabetes (46%) than reverted to NGT (24%). Our report is based on probably the largest populationbased cohort of IGT subjects followed prospectively, and the range of conversion rates from IGT to diabetes ( ) places Mauritius in the upper half of reported figures ranging from 2 to 14.3% [25], together with other Asian Indian populations [35, 36]. Interestingly, a higher proportion of IGT subjects converted to diabetes in the second period compared with the first, which was counterbalanced by a higher proportion reverting to NGT in the first period. These changes coincided with a National Intervention Program between 1987 and 1992 including a change in the composition of ration oil from primarily palm oil to primarily soyabean oil. This had effects on the fat composition of the diet [37], which was paralleled by a significant decrease in the relative proportions of saturated fatty acids in serum between 1987 and 1992 [37], which could have effects on serum lipids and cholesterol levels [13, 38]. Whether these changes were sustainable and generalized to the whole population of Mauritius remains to be established. Furthermore, it is unknown if changes in lipid levels affect the rate of

9 INCIDENCE OF GLUCOSE INTOLERANCE IN MAURITIUS 45 conversion from IGT, but experimental data support the concept that lipid overload impairs b-cell function [39]. Certainly, there is evidence that conversion rates from intermediate forms of glucose intolerance can be modified by simple nonpharmacological life-style interventions, such as diet changes and exercise [40, 41]. Asian Indians were more prone to develop IGT whereas Creoles had higher IFG rates, and to our knowledge, this has not been described in a multiethnic study. The significance of this finding remains to be established, but it could highlight either genetic and/or lifestyle factors related to glucose and insulin metabolism. Furthermore, Chinese subjects had a lower incidence of diabetes, a finding that we have reported earlier [5]. However, the small number of Chinese participants in 1998 hampers the interpretation of this difference. The estimates of incidence rates for Mauritius presented in this paper may differ slightly from earlier reports [5, 42] because of the use of the new criteria for glucose intolerance [9]. We believe that the presented estimates reflect the true incidence rates of glucose intolerance in Mauritius, given the relatively large samples and cluster sampling method [11, 12]. Indeed these estimates may be too low as subjects aged 80+ and women with gestational-related glucose intolerance were excluded. Furthermore, data on subjects who might have been diagnosed with diabetes and died between baseline and follow-up were not included, except in the analysis, which included those lost to follow-up, where they were assumed to have not developed diabetes. However, mortality data for the first 5-year period are available, and we recently reported that subjects with KDM have higher mortality than those with a survey-diagnosed diabetes (NDM) [43]. The decreasing incidence in the older age groups could indicate that those developing diabetes have a high early mortality (or that they were less willing to return for the follow-up survey). In contrast, the increasing proportion of KDM in the last 6-year period may indicate that the health service in Mauritius has become more focused on diabetes management during the period studied. The adjustment to glucose measurements in 1998 was necessary because of differences in methodology. We do not believe that the up-adjusted glucose values falsely elevated the incidence for the 6 and 11-year periods. Certainly, the incidence of IFG, IGT and survey-diagnosed diabetes did not increase in the latter cohorts compared with the initial 5-year period. Furthermore, the higher period incidence of diabetes seen in the and cohorts compared with the cohort was largely because of an increase in KDM incidence, rather than survey-diagnosed NDM. In conclusion, in this large prospective study, we report a fairly constant high incidence rate of glucose intolerance over 11 years. However, the incidence rates in men and women have changed, with a notable catch-up amongst women, as well as the direction and rate of progression from IFG and IGT. Conflict of interest statement The authors do not have any conflicts of interest in connection with this paper. Acknowledgements This study was undertaken with the support and collaboration of the Ministry of Health (Mauritius), the World Health Organization (Geneva, Switzerland), International Diabetes Institute (Melbourne, Australia), the University of Newcastle upon Tyne (UK), and the National Public Health Institute (Helsinki, Finland). This study was partially funded by US National Institutes of Health Grant DK Dr Söderberg is supported by grants from the Swedish Heart and Lung Foundation and the Swedish Society of Medicine, and Dr Shaw is supported by a grant from the Institute for Diabetes Discovery, Branford, Connecticut. References 1 Knowler WC, Bennett PH, Hamman RF, Miller M. Diabetes incidence and prevalence in Pima Indians: a 19-fold greater incidence than in Rochester, Minnesota. Am J Epidemiol 1978; 108: Harris MI, Flegal KM, Cowie CC et al. Prevalence of diabetes, impaired fasting glucose, and impaired glucose tolerance in U.S. adults. The Third National Health and Nutrition Examination Survey, Diabetes Care 1998; 21: Daniel M, Rowley KG, McDermott R, Mylvaganam A, O Dea K. Diabetes incidence in an Australian aboriginal population. An 8-year follow-up study. Diabetes Care 1999; 22: Dowse GK, Zimmet PZ, Finch CF, Collins VR. Decline in incidence of epidemic glucose intolerance in Nauruans:

10 46 S. SÖDERBERG et al. implications for the thrifty genotype. Am J Epidemiol 1991; 133: Dowse GK. Incidence of NIDDM and the natural history of IGT in Pacific and Indian Ocean populations. Diabetes Res Clin Pract 1996; 34 (Suppl.): S Omar MA, Seedat MA, Dyer RB, Motala AA, Knight LT, Becker PJ. South African Indians show a high prevalence of NIDDM and bimodality in plasma glucose distribution patterns. Diabetes Care 1994; 17: Tan CE, Emmanuel SC, Tan BY, Jacob E. Prevalence of diabetes and ethnic differences in cardiovascular risk factors. The 1992 Singapore National Health Survey. Diabetes Care 1999; 22: Brissonnette LG, Fareed DS. Cardiovascular diseases as a cause of death in the island of Mauritius, World Health Stat Q 1985; 38: World Health Organization. Definition, diagnosis and classification of diabetes mellitus and its complications: Report of a WHO Consultation. Part 1. Diagnosis and classification of diabetes mellitus. Geneva: WHO, Central Intelligence Agency (CIA). The World Factbook. Washington DC, Available from cia/publications/factbook/geos/mp.html. Accessed on 1 March Dowse GK, Gareeboo H, Zimmet PZ et al., Mauritius Noncommunicable Disease Study Group. High prevalence of NIDDM and impaired glucose tolerance in Indian, Creole, and Chinese Mauritians. Diabetes 1990; 39: Dowse GK, Zimmet PZ, Gareeboo H et al. Abdominal obesity and physical inactivity as risk factors for NIDDM and impaired glucose tolerance in Indian, Creole, and Chinese Mauritians. Diabetes Care 1991; 14: Dowse GK, Gareeboo H, Alberti KG et al., Mauritius Noncommunicable Disease Study Group. Changes in population cholesterol concentrations and other cardiovascular risk factor levels after five years of the non-communicable disease intervention programme in Mauritius. BMJ 1995; 311: Tuomilehto J, Dabee J, Karvonen M et al. Incidence of IDDM in Mauritian children and adolescents from 1986 to Diabetes Care 1993; 16: Haffner SM, Miettinen H, Stern MP. Are risk factors for conversion to NIDDM similar in high and low risk populations? Diabetologia 1997; 40: Colditz GA, Willett WC, Stampfer MJ et al. Weight as a risk factor for clinical diabetes in women. Am J Epidemiol 1990; 132: Eriksson KF, Lindgärde F. Poor physical fitness, and impaired early insulin response but late hyperinsulinaemia, as predictors of NIDDM in middle-aged Swedish men. Diabetologia 1996; 39: Lipton RB, Liao Y, Cao G, Cooper RS, McGee D. Determinants of incident non-insulin-dependent diabetes mellitus among blacks and whites in a national sample. The NHANES I Epidemiologic Follow-up Study. Am J Epidemiol 1993; 138: Njølstad I, Arnesen E, Lund-Larsen PG. Sex differences in risk factors for clinical diabetes mellitus in a general population: a 12-year follow-up of the Finnmark Study. Am J Epidemiol 1998; 147: Gabir MM, Hanson RL, Dabelea D et al. The 1997 American Diabetes Association and 1999 World Health Organization criteria for hyperglycemia in the diagnosis and prediction of diabetes. Diabetes Care 2000; 23: Edelstein SL, Knowler WC, Bain RP et al. Predictors of progression from impaired glucose tolerance to NIDDM: an analysis of six prospective studies. Diabetes 1997; 46: Haffner SM, Hazuda HP, Mitchell BD, Patterson JK, Stern MP. Increased incidence of type II diabetes mellitus in Mexican Americans. Diabetes Care 1991; 14: Hodge AM, Dowse GK, Gareeboo H, Tuomilehto J, Alberti KG, Zimmet PZ. Incidence, increasing prevalence, and predictors of change in obesity and fat distribution over 5 years in the rapidly developing population of Mauritius. Int J Obes Relat Metab Disord 1996; 20: de Vegt F, Dekker JM, Jager A et al. Relation of impaired fasting and postload glucose with incident type 2 diabetes in a Dutch population: The Hoorn Study. JAMA 2001; 285: Alberti KG. The clinical implications of impaired glucose tolerance. Diabet Med 1996; 13: Alberti KG. Impaired glucose tolerance fact or fiction. Diabet Med 1996; 13: S Swai AB, McLarty DG, Kitange HM et al. Study in Tanzania of impaired glucose tolerance. Methodological myth? Diabetes 1991; 40: McPhillips JB, Barrett-Connor E, Wingard DL. Cardiovascular disease risk factors prior to the diagnosis of impaired glucose tolerance and non-insulin-dependent diabetes mellitus in a community of older adults. Am J Epidemiol 1990; 131: Cassano PA, Rosner B, Vokonas PS, Weiss ST. Obesity and body fat distribution in relation to the incidence of noninsulin-dependent diabetes mellitus. A prospective cohort study of men in the normative aging study. Am J Epidemiol 1992; 136: Harris MI. Impaired glucose tolerance prevalence and conversion to NIDDM. Diabet Med 1996; 13: S Expert committee on the diagnosis and classification of Diabetes Mellitus. Report of the Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Diabetes Care 1997; 20: Chou P, Li CL, Wu GS, Tsai ST. Progression to type 2 diabetes among high-risk groups in Kin-Chen, Kinmen. Exploring the natural history of type 2 diabetes. Diabetes Care 1998; 21: Charles MA, Fontbonne A, Thibult N, Warnet JM, Rosselin GE, Eschwege E. Risk factors for NIDDM in white population. Paris prospective study. Diabetes 1991; 40: Dinneen SF, Maldonado DIII, Leibson CL et al. Effects of changing diagnostic criteria on the risk of developing diabetes. Diabetes Care 1998; 21: Ramachandran A, Snehalatha C, Naik RA, Mohan V, Shobana R, Viswanathan M. Significance of impaired glucose tolerance in an Asian Indian population: a follow-up study. Diabetes Res Clin Pract 1986; 2: Motala AA, Omar MA, Gouws E. High risk of progression to NIDDM in South-African Indians with impaired glucose tolerance. Diabetes 1993; 42: Uusitalo U, Feskens EJ, Tuomilehto J et al. Fall in total cholesterol concentration over five years in association with changes in fatty acid composition of cooking oil in Mauritius: cross sectional survey. BMJ 1996; 313:

11 INCIDENCE OF GLUCOSE INTOLERANCE IN MAURITIUS Mensink RP, Katan MB. Effect of dietary fatty acids on serum lipids and lipoproteins. A meta-analysis of 27 trials. Arterioscler Thromb 1992; 12: Unger RH. Lipotoxicity in the pathogenesis of obesitydependent NIDDM. Genetic and clinical implications. Diabetes 1995; 44: Tuomilehto J, Lindstrom J, Eriksson JG et al. Prevention of type 2 diabetes mellitus by changes in lifestyle among subjects with impaired glucose tolerance. N Engl J Med 2001; 344: Pan XR, Li GW, Hu YH et al. Effects of diet and exercise in preventing NIDDM in people with impaired glucose tolerance. The Da Qing IGT and Diabetes Study. Diabetes Care 1997; 20: Shaw JE, Zimmet PZ, de Courten M et al. Impaired fasting glucose or impaired glucose tolerance. What best predicts future diabetes in Mauritius? Diabetes Care 1999; 22: Shaw JE, Hodge AM, de Courten M, Chitson P, Zimmet PZ. Isolated post-challenge hyperglycaemia confirmed as a risk factor for mortality. Diabetologia 1999; 42: Correspondence: Dr Stefan Söderberg, Department of Medicine, Umeå University Hospital, SE Umeå, Sweden. (fax: ; stefan.soderberg@medicin.umu.se).