N Melchionda 1, G Forlani 1 *, G Marchesini 1, L Baraldi 1 and S Natale 1

(2002) 26, 90 96 ß 2002 Nature Publishing Group All rights reserved 0307 0565/02 $25.00 www.nature.com/ijo PAPER WHO and ADA criteria for the diagnosis of diabetes mellitus in relation to body mass index. Insulin sensitivity and secretion in resulting subcategories of glucose tolerance N Melchionda 1, G Forlani 1 *, G Marchesini 1, L Baraldi 1 and S Natale 1 1 Unit of Metabolic Diseases, Department of Internal Medicine and Gastroenterology, University of Bologna, Policlinico S Orsola, Bologna, Italy OBJECTIVE: To determine the influence of body mass index (BMI) on agreement between the American Diabetes Association (ADA) and the new World Health Organization diagnostic criteria for the diagnosis of diabetes mellitus and to investigate the metabolic profile of the resulting subcategories. DESIGN: Cross-sectional study SUBJECTS: A total of 3018 subjects with no previous history of diabetes and fasting glucose < 7.8 mmol=l, with a wide range of BMIs. MEASUREMENTS: (1) Prevalence of impaired glucose regulation (IGR) and diabetes (DM) according to ADA and WHO diagnostic criteria; (2) basal and post-load insulin sensitivity and secretion, calculated on the basis of data derived from an oral glucose tolerance test (OGTT). RESULTS: The diagnosis according to the two classifications was concordant in 2490 subjects, discordant in 528 (452 were identified as impaired glucose tolerance (IGT) and 76 as DM only by means of OGTT). The disagreement increased with increasing BMI, being as high as 25.3% in subjects with BMI 35 kg=m 2. Subjects with isolated fasting hyperglycaemia were mainly characterised by reduced insulin sensitivity and secretion in the basal state, but normal first-phase insulin secretion and moderately reduced insulin sensitivity after glucose challenge. Subjects with isolated 2 h hyperglycaemia were mainly characterised by normal basal insulin secretion and by a marked insulin resistance associated with a blunted first-phase insulin secretion after the glucose load. CONCLUSIONS: The disagreement between ADA and WHO classifications is particularly relevant in obesity, making OGTT mandatory in these subjects. Different pathogenic mechanisms are involved in isolated fasting or post-load hyperglycaemia, possibly related to a different site of insulin resistance (hepatic vs peripheral), and=or to a different disregulation of insulin secretion (basal vs post-load). A correct identification of the underlying mechanism(s) is the rationale for future studies to detect the effectiveness of different pharmacological or behavioural approaches. (2002) 26, 90 96. DOI: 10.1038=sj=ijo=0801847 Keywords: diabetes; glucose tolerance; insulin sensitivity; insulin secretion Introduction In 1997 the American Diabetes Association (ADA) proposed a new classification of diabetes mellitus (DM) and other categories of glucose regulation, simply based on fasting blood glucose (BG). 1 The threshold for DM was lowered to *Correspondence: G Forlani, Unità di Malattie del Metabolismo, Azienda Ospedale S. Orsola-Malpighi, Via Massarenti 9, I-40138 Bologna, Italy. E-mail: forlani@orsola-malpighi.med.unibo.it Received 18 October 2000; revised 16 February 2001; accepted 6 July 2001 7.0 mmol=l from the value of 7.8 mmol=l, set almost 20 y earlier by the World Health Organization (WHO). 2 A new category (impaired fasting glucose IFG) was introduced, to include patients with a fasting BG in the range between 6.1 and 7.0 mmol=l. 1 Patients with IFG were supposed to be at the same risk of progression as patients with impaired glucose tolerance (IGT) according to WHO criteria (a 2 h BG during an oral glucose tolerance test (OGTT) in the range between 7.8 and 11.1 mmol=l). Epidemiological studies proved that ADA and WHO diagnostic criteria do not completely overlap. A few patients,

identified by WHO criteria as IGT or DM on the basis of OGTT, are not recognised as IFG or DM by ADA criteria, based on the sole fasting BG. The disagreement was reported to be mainly present in groups at high risk of DM, and an increased body mass index (BMI) was a relevant source of disagreement. 3 5 On this basis, in 1998 the WHO 6 accepted the ADA proposal to lower the cut-off of fasting BG for the diagnosis of DM to 7.0 mmol=l, and the new category of IFG, but supported the need of OGTT to diagnose IGT and DM with post-load hyperglycaemia. IFG and IGT were joined in the category of impaired glucose regulation (IGR). This policy obviously reduces the disagreement between the two classifications. At present, patients with IGR may be stratified into three subcategories, according to fasting and post-load glucose concentrations: (a) isolated IFG (fasting BG between 6.1 and 7.0 mmol=l and 2 h BG < 7.8 mmol=l); (b) isolated IGT (fasting BG < 6.1 and 2 h BG between 7.8 and 11.1); (c) combined IFG=IGT. The same applies to the diagnosis of DM. These patients are now stratified into three subcategories: (a) isolated fasting DM (fasting BG 7.0 and 2 h BG < 11.1); (b) isolated 2 h DM (fasting BG < 7.0 and 2 h BG 11.1 mmol=l); (c) combined fasting and post-load DM. Subjects with isolated IGT or isolated post-load DM are only identified by WHO criteria (discordant diagnoses). These subcategories probably identify populations at different risk to progress from IGR to overt DM, 7 with peculiar metabolic features, 8,9 cardiovascular risk profile 9,10 and death risk. 11 In particular, differences in hepatic and peripheral insulin sensitivity and first-phase and total insulin secretion might play a primary role. 12 Fasting and glucose-stimulated insulin and glucose concentrations may be used to compute several indices of insulin sensitivity and secretion, 13 which proved to correlate with parameters derived from quantitative, dynamic methods (namely, the glucose clamp technique). 13 17 They may be relevant for a better definition of the metabolic features of individual patients, identifying the mechanism(s) responsible for IGR and DM and giving clues to pharmacological therapy, also in relation to overweight and obesity. We report the retrospective analysis of data derived from 3018 consecutive OGTTs, performed in our department in subjects with no previous diagnosis of DM, in order to identify: (a) the prevalence of subcategories of IGR and DM in relation to classes of BMI; (b) the differences in insulin release and insulin resistance in the subcategories of IGR and DM. Materials and methods Subjects Between 1992 and 1999, 3046 Caucasian subjects, with no previous history or symptoms of DM and no previous fasting BG levels 7.8 mmol=l, underwent an OGTT in our Department of Metabolic Diseases. Twenty-eight cases (young females with anorexia nervosa, with a BMI below 16 kg=m 2 ) were excluded from the final analysis. The remaining 3018 subjects, mainly subjects with overweight or obesity screened for DM at the time of first visit, were 1314 males and 1704 females, aged 38 s.d. 11 y (range, 18 78 y). In particular, 450 subjects had a BMI < 25.0 kg=m 2 (males, 194; females, 256; aged 37 12 y), 1140 were overweight (BMI, 25.0 29.9; males, 550; females, 589; aged 38 11), 856 were obese (BMI, 30.0 34.9; males, 384; females, 472; aged 38 11), and the remaining 572 were severely obese (BMI ¼ 35; males, 186; females, 386; aged 39 11). The OGTT was performed after a 10 h overnight fast, according to standard methodology. All subjects were given a 200 ml solution containing 75 g of glucose; venous blood samples were obtained at 0, 30, 60, 90 and 120 min for determination of plasma glucose and insulin concentrations. Glucose was measured by glucose-oxidase reaction; insulin was measured by an immuno-enzymometric assay (AIA- PACK IRI, AIA-1200 system, Tosoh Co., Tokyo, Japan). The protocol of this retrospective analysis was approved by the senior staff committee of the Department. Parameters of insulin sensitivity and secretion Data from the OGTTs were used to calculate the following indices of insulin sensitivity: 1. Homeostasis model assessment insulin resistance (HOMA IR): 14 fasting insulin (mu=l) multiplied by fasting glucose (mmol=l), divided by 22.5. 2. Insulin sensitivity index (ISI composite), according to Matsuda et al: 15 10.000=square root (fasting glucose (mg=dl)fasting insulin (mu=l) mean OGTT glucose (mg=dl)mean OGTT insulin (mu=l)). 3. Sensitivity index (SI), according to Cederholm et al: 16 75000 þ (fasting glucose (mg=l) 7 2 h OGTT glucose (mg=l)0.19body weight (kg)=120log (mean OGTT insulin (mu=l)mean OGTT glucose (mmol=l)). HOMA IR, 14 an index of insulin sensitivity derived simply from basal glucose and insulin concentrations, has been shown to correlate with the index of insulin sensitivity derived from the euglycaemic glucose clamp, 17 universally considered the gold standard for quantifying insulin sensitivity. 18 It reflects hepatic insulin sensitivity and basal hepatic glucose production, due to the role of endogenous glucose output in the determination of fasting hyperglycaemia. 19,20 The sensitivity index (SI), proposed by Cederholm, 16 explores the disposal of the glucose load and is mainly related to peripheral insulin sensitivity and muscular glucose uptake, due to the major role of peripheral muscle tissue in the handling of an oral glucose load. 21 The insulin sensitivity index, proposed by Matsuda 15 (ISI composite), explores both liver and muscle insulin sensitivity. The following indices of insulin release were considered: (1) Homeostasis model assessment insulin secretion (HOMA IS) 14 (20fasting insulin (mu=l)=(fasting glucose 91

92 (mmol=l) 7 3.5); it reflects insulin secretion in basal conditions. (2) Insulin response to glucose (IRG): 16 D Ins 0 30 (mu=l)= Dglucose 0 30 (mmol=l); it explores the first phase of post-load insulin secretion. (3) Area under 0 120 min insulin curve=area under 0 120 min glucose curve (AUC insulin (pmol=l)=auc glucose (mmol=l)); it explores total insulin secretion in response to glucose ingestion. HOMA IS 14 is an index of insulin secretion derived from basal glucose and insulin concentrations, exclusively exploring basal insulin secretion. IRG (insulin response to glucose) 16 and the ratio AUC insulin=auc glucose explore the first phase and total insulin secretion in response to glucose ingestion, respectively. In conclusion, each test provides a different piece of information. Statistical analysis All data were implemented on a personal computer and the statistical analysis was carried out by means of StatView 5.0 2 program (SAS Institute Inc., Cary, NC.). Results were expressed as means s.d. for each subject group and each variable. Statistical comparison between group means was carried out by means of ANOVA and Student s t-test for unpaired data. The chi-square test was used for comparison of prevalence among groups. Since six sets of variables were simultaneously tested, the limit of significance was calculated according pffiffiffiffiffiffiffiffiffiffiffiffi to Duncan s multiple range 22 to P 0 ¼ 1 ðn 1Þ 1 P, where P ¼ 0.05 and n ¼ 6. The final critical value of significance was therefore set at 0.01. (82%) had a concordant diagnosis in terms of glucose regulation (Table 1). In all 2264 subjects had normal fasting glucose and normal glucose tolerance, 68 had isolated IFG, 72 had combined IFG and IGT, 86 had DM (24 isolated fasting DM and 62 combined fasting and 2 h DM). In 528 cases (18%) the two criteria led to discordant diagnoses (Table 1). A total of 452 subjects with isolated IGT and 76 2 h DM would have been misclassified on the basis of fasting BG (according to ADA criteria). The total prevalence of IGR and DM obviously increased when fasting BG was combined with 2 h BG. IGR increased four-fold (from 4.6 to 19.6%; chi-square, 317.6; P < 0.0001) and DM doubled (from 2.8 to 5.4%; chi-square, 24.3; P < 0.0001). Distribution of IGR and DM according to BMI The distribution of IGR and DM was not similar in the various groups of BMI (Table 1), with the expected larger prevalence with increasing BMI (chi-square; RC ¼ 96.4; P < 0.0001). Also the disagreement between ADA and WHO classification increased with increasing BMI: discordant diagnoses ranged from 13.1% in normal-weight subjects to 14.8% in overweight, to 18.1% in obese and finally to 25.3% in severely obese subjects (chi-square RC ¼ 36.3; P < 0.0001). This was due to an increasing prevalence both of isolated IGT with increasing BMI (from 11.8% to 12.7, 16.2 and to 20.1 respectively; chi-square 21.1; P < 0.0001) and of 2 h DM (from 1.3% in normal weight, to 2.1, 1.9 and to 5.2%, respectively; chi-square, 22.2; P < 0.0001). On the contrary the prevalence of isolated IFG and isolated fasting DM was not influenced by BMI. Results Prevalence of IGR and DM according to ADA and WHO criteria When analysed according to ADA (fasting BG) and WHO criteria (fasting BG plus 2 h BG), 2490 out of our 3018 cases Insulin sensitivity The indices of insulin sensitivity were altered in all categories of overweight and obesity. Insulin sensitivity progressively declined from normal subjects to the category of overweight, obese and severely obese patients (Table 2). Table 1 Distribution of subjects among the subcategories of impaired glucose regulation and diabetes, in relation to the class of BMI (number of cases (%)) Class of body mass index Normal weight Overweight Obese Severely obese Total Class of glucose tolerance (n ¼ 450) (n ¼ 1140) (n ¼ 856) (n ¼ 572) (n ¼ 3018) Normal 368 (81.8%) 904 (79.3%) 626 (73.1%) 366 (64.0%) 2264 Isolated IFG 10 (2.2%) 27 (2.3%) 24 (2.8%) 7 (1.2%) 68 Isolated IGT 53 (11.8%) 145 (12.7%) 139 (16.2%) 115 (20.1%) 452 Combined IFG IGT 6 (1.3%) 14 (1.2%) 23 (2.7%) 29 (5.1%) 72 Isolated fasting DM 1 (0.2%) 8 (0.7%) 10 (1.2%) 5 (0.8%) 24 Isolated 2 h DM 6 (1.3%) 24 (2.1%) 16 (1.9%) 30 (5.2%) 76 Combined fasting and 2 h DM 6 (1.3%) 18 (1.6%) 18 (2.1%) 20 (3.5%) 62

When related to the categories of glucose regulation, the indices of insulin sensitivity had a different behaviour (Figure 1). HOMA-IR, expressing basal insulin resistance, was significantly higher in isolated IFG (4.78 3.47%), in comparison to isolated IGT (3.92 2.37; P < 0.01), while the SI showed a lower resistance in response to glucose in isolated IFG (43.0 10.1) in comparison to isolated IGT (35.1 8.8; P < 0.001). Similarly, post-load insulin resistance was lower isolated fasting DM compared with isolated 2 h DM (SI, 22.5 5.9 and 32.0 6.9; P < 0.001). The ISI, a measure of overall insulin sensitivity, did not show any significant differences between discordant IGR and between discordant DM. These differences in insulin sensitivity among the various categories of IGR and DM were maintained in all BMI classes. As an example, in obese subjects (obese plus severely obese subjects, n ¼ 1428) HOMA-IR was higher in isolated IFG (n ¼ 31; 5.86 4.48) than in isolated IGT (n ¼ 254; 4.48 2.55; P < 0.003), and the SI was lower in isolated IFG (42.6 10.9) in comparison to isolated IGT (32.6 7.9; P < 0.0001), and in isolated 2 h DM (n ¼ 46; 20.0 4.5) when compared to isolated fasting DM (n ¼ 15, 30.0 7.8, P < 0.003). Insulin secretion Also the indices of insulin secretion increased with increasing BMI (Table 2), and for each class of BMI significant differences were observed in comparison to the lower class. In relation to glucose regulation (Figure 2), HOMA-IS, exploring basal insulin secretion, was significantly reduced in all categories with fasting hyperglycaemia: isolated IFG (112.1 72.3), combined IFG-IGT (111.8 62.5), isolated fasting DM (87.6 45.1), combined fasting and 2 h DM (89.7 63.1). In contrast, HOMA-IS was normal in subjects with isolated post-load hyperglycaemia: isolated IGT (198.7 131.27) and isolated 2 h DM (173.6 136.3). IRG, exploring the first-phase insulin response to glucose, was reduced only in subjects with post-load hyperglycaemia: IGT either isolated or combined with IFG, and 2 h DM (both isolated and associated with fasting hyperglycaemia). This last subgroup was the sole category characterised by a decreased insulin-to-glucose area (25.5 14.5 vs 52.7 33.3 in controls; P < 0.001). The differences in the indices of insulin secretion were maintained in various classes of BMI, although not always in the statistical range because of the low number of cases. As an example in the whole group of obese subjects HOMA-IS was lower in isolated IFG (137 88) than in isolated IGT (219 140), and IRG was higher (33.4 22.1 vs 24.3 15.6). When diabetic patients were compared with subjects with IGR, isolated fasting DM was mainly associated with a further worsening of all indices of insulin sensitivity, a fall in basal insulin secretion and maintenance of insulin response to glucose. The presence of isolated 2 h DM was mainly associated with a worsening of all indices of insulin sensitivity, a blunted insulin response to glucose and maintenance of basal insulin secretion. DM with both fasting and 2 h hyperglycaemia was accompanied by a marked deterioration of all indices of insulin sensitivity and secretion. Discussion Prevalence of IGR and DM according to ADA and WHO criteria and in relation to BMI Several studies proved that the new ADA criteria, based on the sole fasting BG, lead to underestimate the prevalence of disease. 3,5,7,10 ADA proposal was mainly based on data derived from selected populations, that cannot be extrapolated to all groups of subjects. In our series, based on data mainly derived from overweight and obese subjects, we would have missed 76% of patients with IGR and 93 Table 2 Indices of insulin sensitivity and insulin secretion in relation to classes of BMI (for acronyms and units: see Methods) Class of body mass index Normal-weight Overweight Obese Severely obese ANOVA Insulin sensitivity HOMA IR 2.48 1.46 2.99 1.71* 3.70 2.33* { 4.62 3.12* { P < 0.0001 ISI (composite) 5.98 3.43 4.61 2.57* 3.67 2.02* { 3.01 1.99* { P < 0.0001 SI 56.5 17.1 50.0 14.4* 44.9 14.2* { 39.1 13.7* { P < 0.0001 Insulin secretion HOMA IS 161 117 189 168 { 209 182* { 250 358* { P < 0.0001 IRG 15.5 11.6 20.6 16.9* 27.8 28.6* { 31.0 32.2* { P < 0.0001 AUC insulin=auc glucose 34.8 18.1 46.6 26.7* 59.9 36.1* { 68.8 40.6* { P < 0.0001 For acronyms and units, see Methods. *P < 0.001 vs normal weight. { P < 0.001 vs the corresponding lower category of BMI. { P < 0.005 vs normal weight.

94 Figure 1 Indices of insulin sensitivity according to the classes of glucose regulation identified by ADA and WHO criteria. Open bars are subjects with normal glucose regulation; dashed bars are subjects with isolated fasting hyperglycaemia (either IFG or DM); dotted bars are subjects with isolated 2 h hyperglycaemia (either IGT or DM); closed bars are subjects with IGR or DM with combined fasting and post-load hyperglycaemia. Data are presented as means and 95% confidence intervals. Note that HOMA-IR is a measure of insulin resistance. In IGR and DM, all indices are significantly different from values measured in normal subjects. The P- value of differences between discordant categories of IGR or DM are reported, whenever significant. Figure 2 Indices of insulin secretion according to the classes of glucose regulation identified by ADA and WHO criteria. For classes, see legend to Figure 1. The P-value of differences between discordant categories of IGR or DM are reported, whenever significant. *Significantly different from normal values. sole fasting BG is therefore very high and the OGTT is more and more recommended with increasing BMI. misclassified 47% of patients with DM, by not considering OGTT values. Similar data were published by Mannucci et al 23 in a smaller series of obese patients. We showed that the disagreement between ADA and WHO criteria, which is 13% in normal-weight subjects, increases with the degree of overweight and obesity, being as high as 25% in severely obese subjects. The risk of underestimate the prevalence of IGR and DM on the basis of the Insulin resistance and secretion in IGR and DM categories, and in relation to BMI classes The combined use of WHO and ADA diagnostic criteria highlighted the existence of different subgroups with peculiar metabolic characteristics, probably reflecting different genetic abnormalities, inside the categories of impaired glucose regulation and DM. Patients with isolated fasting hyperglycaemia have decreased hepatic insulin sensitivity

and increased basal hepatic glucose output, whereas subjects with isolated post-load hyperglycaemia have decreased peripheral insulin sensitivity. 8,12,15 There is evidence that such metabolic differences may be relevant from a clinical standpoint: patients with post-load hyperglycaemia have worse cardiovascular risk profiles 10,24 and increased death risks. 11 Applying the HOMA method to a small series of subjects undergoing an OGTT, Davies et al 9 showed that patients with fasting hyperglycaemia had a reduced B-cell function. However, the group with IGT was characterised by decreased insulin sensitivity, confirmed by clinical features commonly associated with the insulin resistance syndrome. In a large population-based study in patients with frank DM and their family members (over 5000 cases and over 2000 OGTT curves diagnostic for IGR or DM), 12 subjects with isolated IFG were characterised by basal insulin resistance, measured by HOMA-IR. The first-phase insulin response to glucose was normal in IFG, and blunted in IGT and in DM. These subjects were on average overweight, but no attempt was made to correlate discordant IGR or DM with BMI. In agreement with previous data from the literature, based on the clamp technique, 8 insulin sensitivity, computed in the present study by simple indices, was remarkably reduced in all groups of IGR and DM. The categories of combined fasting and post-load hyperglycaemia had the most pronounced defect. The various indices point out that insulin resistance is more pronounced at hepatic level for isolated IFG and at peripheral level for isolated IGT. The same applies to the subcategories of DM. Differences are also present in basal and first-phase insulin secretion, as previously demonstrated in the Botnia study. 12 The indices of insulin sensitivity and secretion had the expected behaviour among the different classes of BMI, showing an increasing insulin resistance and secretion with increasing BMI. In normal weight, overweight and obese patients the indices maintained the ability to discriminate between peripheral and hepatic sensitivity to insulin, and between defects in basal and first-phase insulin secretion. In conclusion, the present results, based on parameters easily derived from a commonly performed test, confirm that ADA and WHO classifications identify subcategories of patients with different defects in insulin sensitivity and secretion, and underline the need for OGTT, mainly in overweight and obese subjects. Apparently the progression from normal glucose tolerance to IGR and DM may follow two different pathways. In a few patients, the deterioration of glucose tolerance progresses through IFG and eventually to isolated fasting DM. It is characterised by a progressive increase in hepatic insulin resistance (HOMA-IR), and reduced basal insulin secretion (HOMA-IS). The second leads to isolated IGT and isolated 2 h DM, and is characterised by increased peripheral insulin resistance (SI index) and reduced first-phase insulin secretion (IRG index). The two pathways may meet at the level of combined IFG and IGT or combined fasting and 2 h DM, which represent an awful outcome, where both defects are present. The possibility to identify these different routes by easily computed indices able to discriminate the site of insulin resistance and the defect in insulin secretion (mainly HOMA- IR, HOMA-IS, SI, IRG) might be of help in planning specific pharmacological interventions. Follow-up studies are needed to confirm difference in the risk of passing from IGR to DM, 7,25 or in the risk of cardiovascular disease and mortality associated with fasting and post-load hyperglycaemia and hepatic vs peripheral insulin resistance. References 1 The 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: 1183 1197. 2 National Diabetes Data Group. Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance. Diabetes 1979; 28: 1039 1057. 3 DECODE Study Group, on behalf of the European Diabetes Epidemiology Study Group. Will new diagnostic criteria for diabetes mellitus change phenotype of patients with diabetes? Reanalysis of European epidemiological data. Br Med J 1998; 317: 371 375. 4 Shaw JE, de Courten M, Boyko EJ, Zimmett PZ. Impact of new diagnostic criteria for diabetes on different populations. Diabetes Care 1999; 22: 762 766. 5 Lee ET, Howard BV, Go O, Savage PJ, Fabsitz RR, Robbins DC, Welty TK. Prevalence of undiagnosed diabetes in three America Indian populations. Diabetes Care 2000; 23: 181 186. 6 Alberti KGMM, Zimmet PZ for the WHO Consultation. Definition, diagnosis and classification of diabetes mellitus and its complications. Part 1: diagnosis and classification of diabetes mellitus provisional report of a WHO consultation. Diabetes Med 1998; 15: 539 553. 7 Shaw JE, Zimmet PZ, de Courten M, Dowse GK, Chitson P, Gareboo H, Hemraj F, Fareed D, Tuomilehto J, Alberti KGMM. Impaired fasting glucose or impaired glucose tolerance. Diabetes Care 1999; 22: 399 402. 8 Weyer C, Bogardus C, Pratley RE. Metabolic characteristics of individuals with impaired fasting glucose and=or impaired glucose tolerance. Diabetes 1999; 48: 2197 2203. 9 Davies MJ, Raymond NT, Day JL, Hales CN, Burden AC. Impaired glucose tolerance and fasting hyperglycaemia have different characteristics. Diabetic Med 2000; 17: 433 440. 10 Gimeno SGA, Ferreira SRG, Laercio JF, Iunes M, The Japanese Brazilian Diabetes Study Group. Comparison of glucose tolerance categories according to World Health Organization and American Diabetes Association diagnostic criteria in a population-based study in Brazil. Diabetes Care 1998; 21: 1889 1892. 11 The DECODE Study Group: glucose tolerance and mortality: comparison of WHO and American Diabetes Association diagnostic criteria. Lancet 1999; 354: 617 621. 12 Tripathy D, Carlsson M, Almgren P, Isomaa B, Taskinen M-R, Tuomi T, Groop LC. Insulin secretion and insulin sensitivity in relation to glucose tolerance. Lessons from the Botnia study. Diabetes 2000; 49: 975 980. 13 Stumvoli M, Mitakou A, Pimenta W, Jenseen T, Yki-Jarvinen H, Van Haeften T, Renn W, Gerich J. Use of oral glucose tolerance test to assess insulin release and insulin sensitivity. Diabetes Care 2000; 23: 295 301. 14 Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC. Homeostasis model assessment: insulin resistance and b-cell function from plasma fasting glucose and insulin concentrations in man. Diabetologia 1985; 28: 412 419. 95

96 15 Matsuda M, De Fronzo RA. Insulin sensitivity indices obtained from oral glucose tolerance testing. Diabetes Care 1999; 22: 1462 1470. 16 Cederholm J, Wibell L. Insulin release and peripheral sensitivity at the oral glucose tolerance test. Diabetes Res Clin Pract 1990; 10: 167 175. 17 Bonora E, Targher G, Alberriche M, Bonadonna RCM, Saggiani F, Zenere MB, Monauni T, Muggeo M. Homeostasis model assessment closely mirrors the glucose clamp technique in the assessment of insulin sensitivity: studies in subjects with various degrees of glucose tolerance and insulin sensitivity. Diabetes Care 2000; 23: 57 63. 18 De Fronzo R, Tobin JD, Andres R. Glucose clamp technique: a method for quantifying insulin secretion and resistance. Am J Physiol 1979; 237: 214 223. 19 De Fronzo RA. Pathogenesis of type 2 diabetes: metabolic and molecular implications for identifying diabetes genes. Diabetes Rev 1997; 3: 177 269. 20 DeFronzo RA, Ferrannini E, Simonson DC. Fasting hyperglycemia in non-insulin dependent diabetes mellitus: contributions of excessive hepatic glucose production and impaired tissue glucose uptake. Metabolism 1989; 38: 387 395. 21 Katz LD, Glickman MG, Rapoport S, Ferrannini E, DeFronzo RA. Splanchnic and peripheral disposal of oral glucose in man. Diabetes 1983; 32: 675 679. 22 Duncan DB. Multiple range test for correlated and heteroscedastic means. Biometrics 1957; 13: 164 204. 23 Mannucci E, Bardini G, Ognibene A, Rotella CM. Comparison of ADA and WHO screening methods for diabetes mellitus in obese patients. Diabetic Med 1999; 16: 579 585. 24 Levitt NS, Unwin NC, Bradshaw D, Kitange HM, Mbanya J-CN, Mollentze WF, Omar MAK, Motala AA, Joubert G, Masuki G, Machibya H. Application of the new ADA criteria for the diagnosis of diabetes to populations studies in sub-saharan Africa. Diabetic Med 2000; 17: 381 385. 25 Dinneen SF, Maldonado D III, Leibson CL, Klee GG, Li H, Melton LJ III, Rizza RA. Effects of changing diagnostic criteria on the risk of developing diabetes. Diabetes Care 1998; 21: 1408 1413.