OBJECTIVE—We sought to assess insulin secretion/insulin resistance index in predicting the risk for future type 2 diabetes

RESEARCH DESIGN AND METHODS—A total of 1,551 nondiabetic subjects from the San Antonio Heart Study received an oral glucose tolerance test (OGTT) with measurement of plasma glucose and insulin concentrations at 0, 30, 60, and 120 min at baseline and after 7–8 years of follow-up. Insulin secretion/insulin resistance index was calculated as the product of Matsuda index and ΔI0–30/ΔG0–30 or ΔI0–120/ΔG0–120. The discriminatory power of various prediction models for development of type 2 diabetes was tested with the area under the receiver-operating characteristic (ROC) curve.

RESULTS—Insulin secretion/insulin resistance index (0- to 30- and 0- to 120-min time periods) had the greatest areas under the ROC curve (0.85 and 0.86, respectively), which were significantly greater than the 2-h plasma glucose concentration during the OGTT or the San Antonio Diabetes Prediction Model (SADPM) (P < 0.001 and P < 0.0001, respectively). A model based on the combination of the SADPM and a modified version of the insulin secretion/insulin resistance index or 1-h plasma glucose concentration had equal power to predict the risk for future type 2 diabetes compared with the insulin secretion/insulin resistance index.

CONCLUSIONS—The insulin secretion/insulin resistance index is useful as a predictor of future development of type 2 diabetes. A model based on the combination of the SADPM and either a modified version of the insulin secretion/insulin resistance index or 1-h plasma glucose concentration can equally predict future type 2 diabetes.

The worldwide prevalence of type 2 diabetes is increasing at epidemic proportions. In the year 2000, there were 150 million individuals with type 2 diabetes worldwide, and this number is expected to double in the next 25 years (1). This increase in type 2 diabetes prevalence is also associated with increases in both morbidity and mortality (2), despite the decrease in cardiovascular morbidity and mortality in nondiabetic individuals over the same time period (3). For example, in the year 2000 (when compared with 1990), the rate of cardiac events in nondiabetic subjects in New York City declined by 14%, while it increased by 54% in individuals with type 2 diabetes over the same period (2).

Recent clinical trials have demonstrated that, in subjects at high risk for type 2 diabetes, both lifestyle changes and pharmacological intervention can reduce the incidence of type 2 diabetes. A moderate increase in physical activity, accompanied with modest (5–7%) reduction in body weight, reduced the conversion rate of impaired glucose tolerance (IGT) to type 2 diabetes by 58% (4,5). Pharmacological intervention with metformin, troglitazone, and acarbose also have been shown to reduce conversion rate from IGT to diabetes by 31% (3), 75% (6), and 25% (7), respectively. The results of these long-term, prospective studies emphasize the importance of identifying subjects at high risk for type 2 diabetes in order to offer them an intervention program that will prevent/halt their progression to overt diabetes. Demonstration of IGT most commonly has been used to identify subjects at high risk for the development of type 2 diabetes. Indeed, all published studies that have evaluated intervention strategies for preventing type 2 diabetes have recruited subjects with IGT (4–7). However, in prospective epidemiological studies, only 30–40% of subjects with IGT ultimately develop type 2 diabetes (8–10). Furthermore, ∼40% of subjects who develop type 2 diabetes have normal glucose tolerance (NGT) at baseline (8). This limits the use of IGT as the sole means to identify subjects at high risk for type 2 diabetes. Such observations have prompted the search for alternative models that more accurately identify subjects at high risk for future type 2 diabetes. A number of models for identifying high-risk subjects have been proposed (11–13). These models are based on risk factors for type 2 diabetes, including age, ethnicity, obesity, lipid profile, blood pressure, and fasting plasma glucose concentration. Although all of these models have value for identifying subjects at risk for future development of type 2 diabetes, their predictive power is similar to or only slightly greater than the 2-h plasma glucose concentration (11–13).

Type 2 diabetes is characterized by defects in both insulin secretion and insulin action (14). Both β-cell dysfunction and insulin resistance can be demonstrated long before overt diabetes is diagnosed, e.g., in subjects with IGT or in NGT subjects with positive family history of type 2 diabetes (14,15). Insulin secretion and insulin resistance can be quantified with the hyperglycemic and euglycemic insulin clamp techniques, respectively (16). However, these techniques are labor intensive and are difficult to apply in clinical practice or in large epidemiological studies. Surrogate measures of insulin secretion and insulin sensitivity have been developed from oral glucose tolerance tests (OGTTs) (17–20). These indexes correlate reasonably well with insulin secretion and insulin sensitivity measured with the hyperglycemic and insulin clamp techniques. Previous epidemiological studies have demonstrated that OGTT-derived measures of insulin resistance and impaired insulin secretion can predict future development of type 2 diabetes (21,22). However, no previous study has examined the ability of an insulin secretion/insulin resistance index to predict future type 2 diabetes and compared this index with other predictive models of type 2 diabetes.

All subjects were participants of the San Antonio Heart Study (23–25), which is a population-based, epidemiological study of type 2 diabetes and cardiovascular disease. A total of 2,941 Mexican Americans and non-Hispanic whites, aged 25–68 years, were enrolled in phase 2 of the San Antonio Heart Study. We excluded phase 2 participants with overt diabetes at baseline based on World Health Organization criteria (26). We also excluded all phase 1 participants, since in these participants plasma glucose levels were not measured at 30 and 60 min. A total of 2,616 eligible participants, who were free of type 2 diabetes at baseline, completed a 7- to 8-year follow-up examination. Of these 2,616 participants, 1,551 had plasma glucose and insulin measurements at 0, 30, 60, and 120 min during the baseline OGTT and constitute the study population. The study was approved by the institutional review board of University of Texas Health Science Center at San Antonio. All subjects gave written informed consent.

Definition of variables and outcomes

All studies were performed in a mobile clinic following a 12-h overnight fast. A fasting blood specimen was obtained for determination of plasma glucose and serum insulin and lipid concentrations. Following collection of baseline blood sample, subjects ingested 75 g glucose (Orangedex; Custom Laboratories, Baltimore, MD) and blood obtained at 30, 60, and 120 min for determination of plasma glucose and serum insulin concentrations. Plasma glucose and serum lipids were measured with an Abbott Bichromatic Analyzer (South Pasadena, CA). Serum insulin was measured by radioimmunoassay (Diagnostic Products, Los Angeles, CA), which has a relatively high degree of cross-reactivity with proinsulin (∼70%). The diagnosis of diabetes was based upon World Health Organization criteria (26): 2-h plasma glucose ≥200 mg/dl or fasting plasma glucose ≥126 mg/dl. Subjects on insulin or oral antihyperglycemic medications also were considered to have diabetes.

Calculations

Areas under the glucose and insulin curves were calculated by the trapezoid rule. Matsuda index of insulin sensitivity (17) was calculated as previously reported. A variation of Matsuda index was calculated by using plasma glucose and insulin concentrations at 30 min during the OGTT in place of mean plasma glucose and insulin concentrations. We refer to this index as the modified Matsuda index. The insulinogenic index was calculated by dividing the increment in serum insulin at 30 min by the increment in plasma glucose at 30 min of OGTT. The insulin secretion/insulin resistance (disposition) index was calculated as the product of insulin secretion measured with (ΔI0–30/ΔG0–30 or ΔI0–120/ΔG0–120) and insulin sensitivity index (Matsuda or modified Matsuda indexes). A previously described, multivariate model for predicting future type 2 diabetes, called the San Antonio Diabetes Prediction Model (SADPM) (11) (which includes age, sex, ethnicity, BMI, blood pressure, fasting plasma glucose, triglycerides, and HDL), was also examined, as was the predictive value of the 1-h plasma glucose concentration during the OGTT.

Statistical methods

Variables are presented as means ± SD. The significance of the mean differences was tested with Student's t test. Statistical significance was considered at the level of P < 0.05. Assessment of the predictive discrimination of the various models was made using the receiver-operating characteristic (ROC) curve by plotting the sensitivity against the corresponding false-positive rate. The area under the ROC curve was used as a measure of how well a continuous variable predicts the development of type 2 diabetes. To examine whether differences between two areas under ROC curves were statistically different, the algorithm developed by DeLong et al. (27) was used. Statistical analyses were performed with the SAS statistical software system (Cary, NC).

The study included 1,551 subjects: at baseline, 1,257 had normal NGT, 88 had isolated impaired fasting glucose, 154 had isolated IGT, and 52 had both impaired fasting glucose and IGT. A total of 179 subjects progressed to type 2 diabetes after 7–8 years of follow-up; the remaining 1,372 subjects remained free of diabetes at follow-up. Table 1 presents the anthropometric and metabolic characteristics of the two study groups: progressors and nonprogressors. Subjects who progressed to type 2 diabetes were older and had higher BMI, fasting plasma glucose, insulin and triglyceride concentrations, and 2-h plasma glucose and insulin concentrations during OGTT. Indexes of insulin secretion and insulin sensitivity also were lower in subjects who progressed to type 2 diabetes compared with subjects who remained free of diabetes at follow-up.

The area under the ROC curve is presented in Table 2. Similar to previous studies (11), the 2-h plasma glucose concentration had an area under the ROC curve of 0.79 and was comparable to the area under the ROC curve of the SADPM (0.80). Both the SADPM and 2-h plasma glucose concentration had significantly greater area under the ROC curve (0.80 and 0.79, respectively) compared with fasting plasma glucose concentration (ROC = 0.75, P < 0.05).

The insulin secretion/insulin resistance index, calculated either with ΔI0–30/ΔG0–30 or ΔI0–120/ΔG0–120 (as a measure of insulin secretion) and with the Matsuda index (as a measure of insulin sensitivity), had the greatest areas under the ROC curve (0.85 and 0.86, respectively), and both were significantly greater than that of 2-h plasma glucose concentration and the SADPM (P < 0.0001). Combining the SADPM with the insulin secretion/insulin resistance index did not further improve its predictability (Table 2). The area under the ROC curve for the modified insulin secretion/insulin resistance index measured with the modified (30-min) Matsuda index was 0.83, which is only slightly less than that of the unmodified index. However, the combination of the modified insulin secretion/insulin resistance index with the SADPM had comparable area under the ROC curve (0.86) to the unmodified insulin secretion/insulin resistance index.

Plasma glucose concentration at 1 h during the OGTT had a stronger correlation with insulin secretion, insulin resistance, and insulin secretion/insulin resistance index compared with the 2-h plasma glucose concentration (Table 3). Therefore, we examined the ability of the 1-h plasma glucose concentration to predict future type 2 diabetes. The area under the ROC curve for 1-h plasma glucose (0.84) (Table 2) was significantly greater than the area under the ROC curve for the 2-h plasma glucose concentration and the SADPM (both P = 0.01). Addition of the 1-h plasma glucose concentration to the SADPM significantly improved its predictability at a level similar to that of insulin secretion/insulin resistance index.

When a cut point for continuous variables was used as a threshold for predicting future type 2 diabetes, IGT (2-h plasma glucose = 140–199 mg/dl) had the greatest specificity (92%) but the sensitivity was only 51%. The use of plasma glucose concentration = 155 mg/dl at 60 min during the OGTT as a cut point had 75% sensitivity and 79% specificity. Both the 0- to 30-min and 0- to 120-min insulin secretion/insulin resistance indexes had greater sensitivity compared with IGT, although the specificity declined slightly (Table 2).

Clinical trails consistently have demonstrated that lifestyle intervention in subjects with IGT markedly reduces the risk for conversion to type 2 diabetes. However, the identification of subjects with IGT requires that an OGTT be performed, and the time required (2-h) to complete the OGTT has limited its widespread use in clinical practice. Moreover, although the present results demonstrate that the OGTT diagnosis of IGT has high specificity (92%) in predicting which subjects are at high risk for developing type 2 diabetes, it has much lower sensitivity (51%). Thus, if one relies exclusively on the OGTT diagnosis of IGT for identifying high-risk individuals, about half of those who ultimately convert to type 2 diabetes would not have been identified.

Our results demonstrate that the insulin secretion/insulin resistance index is the best predictor of future type 2 diabetes (Table 2). When it is calculated with either ΔI0–30/ΔG0–30 or ΔI0–120/ΔG0–120 (as the measure of insulin secretion) and the Matsuda index (as the measure of insulin resistance), it has the greatest area under the ROC curve (0.85 and 0.86, respectively). Combining the SADPM predictive model (11) with the insulin secretion/insulin resistance index does not further enhance its predictability. Of note, in previous studies indexes of insulin secretion and insulin resistance were found to be significant independent predictors for future type 2 diabetes (21,22). Although they were not compared with the 2-h plasma glucose concentration, all of the indexes that were tested had an area under the ROC curve <0.8 (21). No previous study has assessed the ability of any insulin secretion/insulin resistance index to predict future type 2 diabetes. One limitation in using the insulin secretion/insulin resistance index for predicting future type 2 diabetes is the need to measure the plasma glucose and insulin concentrations every 30 min during the OGTT. This limitation can be overcome by using the modified insulin secretion/insulin resistance index. Although the area under the ROC curve (0.83) for the modified insulin secretion/insulin resistance index calculated with ΔI0–30/ΔG0–30 and a modified version of the Matsuda index (calculated with plasma glucose and insulin concentrations at fasting and 30 min) is slightly lower than that of the unmodified index, combination of the modified insulin secretion/insulin resistance index and SADPM has an area under its ROC curve that is comparable to that of the unmodified index in predicting future diabetes. Use of this model in clinical practice is more convenient than the OGTT diagnosis of IGT because it requires the measurement of plasma glucose and insulin concentrations only at baseline (fasting) and 30 min following the glucose load.

Simply measuring the plasma glucose concentration at 1 h during the OGTT also is a good predictor for future type 2 diabetes. It has a greater area under the ROC curve compared with the 2-h plasma glucose concentration, and, when combined with the SADPM, its area under the ROC curve is comparable with the insulin secretion/insulin resistance index. Furthermore, if one uses a cut point of 155 mg/dl for the 1-h plasma glucose concentration, this has 50% greater sensitivity in predicting future type 2 diabetes compared with the OGTT diagnosis of IGT with only a modest (∼15%) decline in specificity. Because the glucose assay is universally standardized, the combination of 1-h plasma glucose concentration and the SADPM provides an additional method for predicting future type 2 diabetes. This combination is more accurate than 2-h plasma glucose concentration during the OGTT and more convenient than the diagnosis of IGT because the plasma glucose concentration is measured at 1 h instead of after 2 h.

Prediction models can serve one of two purposes: 1) identification of high-risk target populations for clinical and public health intervention and 2) elucidating the pathogenesis of the disease. The models presented in this study reflect primarily the second goal. They demonstrate that a prediction model based on the pathophysiology of the disease performs superiorly to other clinical models in predicting the risk of future type 2 diabetes. With respect to the first goal, one must weigh the greater accuracy of the presented models against the added cost, both in terms of dollars and inconvenience associated with performing the OGTT, which the SADPM does not require. Depending on the circumstances, one or another approach may be preferred.

In summary, the present results demonstrate that the insulin secretion/insulin resistance index derived from the OGTT provides a superior method for predicting future development of type 2 diabetes compared with the diagnosis of IGT based on the 2-h plasma glucose concentration. The combination of the SADPM with the 1-h plasma glucose concentration and a modified version of the insulin sensitivity/insulin resistance index provides alternative predictive models with similar precision and does not require the measurement of plasma glucose concentration 2 h after ingestion of the glucose load.

Table 1—

Clinical, anthropometric, and metabolic characteristics of the study groups

NonprogressorsProgressorsP
n 1,372 179 NS 
Age (years) 43 ± 11 48 ± 10 NS 
Sex (% male) 44 39 NS 
BMI (kg/m2) 27.1 ± 5.0 31.3 ± 5.2 <0.0001 
Positive family history (%) 29 48 NS 
LDL cholesterol (mg/dl) 123 ± 35 124 ± 35 NS 
Triglycerides (mg/dl) 134 ± 90 192 ± 121 <0.0001 
HDL cholesterol (mg/dl) 47 ± 13 42 ± 11 <0.0001 
Total cholesterol (mg/dl) 196 ± 39 200 ± 38 NS 
Fasting glucose (mg/dl) 85 ± 10 95 ± 12 <0.0001 
2-h plasma glucose (mg/dl) 100 ± 29 140 ± 36 <0.0001 
Fasting insulin (μU/ml) 12 ± 13 22 ± 18 <0.0001 
2-h serum insulin (μU/ml) 84 ± 88 156 ± 125 <0.0001 
ΔG(AUC)0–120 (mg · dl−1 · h−1) 65 ± 43 123 ± 44 <0.0001 
ΔI0–120/ΔG0–120 2.9 ± 2.6 2.1 ± 2.2 <0.0001 
ΔI0–30/ΔG0–30 2.4 ± 1.6 1.3 ± 0.9 <0.0001 
Matsuda index 5.1 ± 5.4 2.3 ± 2.0 <0.0001 
Modified Matsuda index 4.8 ± 5.7 2.6 ± 2.1 <0.0001 
ΔI0–120/ΔG0–120 × Matsuda index 20 ± 14 3.6 ± 3.6 <0.0001 
ΔI0–30/ΔG0–30 × Matsuda index 10.1 ± 12.1 2.5 ± 2.7 <0.0001 
ΔI0–30/ΔG0–30 × modified Matsuda index 9.1 ± 10.7 2.8 ± 2.3 <0.0001 
NonprogressorsProgressorsP
n 1,372 179 NS 
Age (years) 43 ± 11 48 ± 10 NS 
Sex (% male) 44 39 NS 
BMI (kg/m2) 27.1 ± 5.0 31.3 ± 5.2 <0.0001 
Positive family history (%) 29 48 NS 
LDL cholesterol (mg/dl) 123 ± 35 124 ± 35 NS 
Triglycerides (mg/dl) 134 ± 90 192 ± 121 <0.0001 
HDL cholesterol (mg/dl) 47 ± 13 42 ± 11 <0.0001 
Total cholesterol (mg/dl) 196 ± 39 200 ± 38 NS 
Fasting glucose (mg/dl) 85 ± 10 95 ± 12 <0.0001 
2-h plasma glucose (mg/dl) 100 ± 29 140 ± 36 <0.0001 
Fasting insulin (μU/ml) 12 ± 13 22 ± 18 <0.0001 
2-h serum insulin (μU/ml) 84 ± 88 156 ± 125 <0.0001 
ΔG(AUC)0–120 (mg · dl−1 · h−1) 65 ± 43 123 ± 44 <0.0001 
ΔI0–120/ΔG0–120 2.9 ± 2.6 2.1 ± 2.2 <0.0001 
ΔI0–30/ΔG0–30 2.4 ± 1.6 1.3 ± 0.9 <0.0001 
Matsuda index 5.1 ± 5.4 2.3 ± 2.0 <0.0001 
Modified Matsuda index 4.8 ± 5.7 2.6 ± 2.1 <0.0001 
ΔI0–120/ΔG0–120 × Matsuda index 20 ± 14 3.6 ± 3.6 <0.0001 
ΔI0–30/ΔG0–30 × Matsuda index 10.1 ± 12.1 2.5 ± 2.7 <0.0001 
ΔI0–30/ΔG0–30 × modified Matsuda index 9.1 ± 10.7 2.8 ± 2.3 <0.0001 

Data are means ± SD unless otherwise indicated. AUC, area under the curve; NS, not significant.

View Large
Table 2—

Area under the ROC curve for various predictive models for future development of type 2 diabetes

ParameterROCP vs. G120ROC combined model*CutoffSensitivitySpecificity
Fasting plasma glucose 0.75 0.05     
Plasma glucose at 30 min 0.77 0.24     
Plasma glucose at 60 min 0.84 0.01 0.86† 155 0.75 0.79 
Plasma glucose at 120 min 0.79   140 0.51 0.92 
SADPM 0.80      
Glucose (AUC) 0–120 min 0.83 0.06 0.87†  0.83 0.71 
ΔI0–120/ΔG0–120 × Matsuda index 0.86 <0.0001 0.86† 4.5 0.82 0.76 
ΔI0–30/ΔG0–30 × Matsuda index 0.85 0.005 0.87† 3.1 0.83 0.77 
ΔI0–30/ΔG0–30 × modified Matsuda index 0.83 0.1 0.86† 3.5 0.82 0.71 
ParameterROCP vs. G120ROC combined model*CutoffSensitivitySpecificity
Fasting plasma glucose 0.75 0.05     
Plasma glucose at 30 min 0.77 0.24     
Plasma glucose at 60 min 0.84 0.01 0.86† 155 0.75 0.79 
Plasma glucose at 120 min 0.79   140 0.51 0.92 
SADPM 0.80      
Glucose (AUC) 0–120 min 0.83 0.06 0.87†  0.83 0.71 
ΔI0–120/ΔG0–120 × Matsuda index 0.86 <0.0001 0.86† 4.5 0.82 0.76 
ΔI0–30/ΔG0–30 × Matsuda index 0.85 0.005 0.87† 3.1 0.83 0.77 
ΔI0–30/ΔG0–30 × modified Matsuda index 0.83 0.1 0.86† 3.5 0.82 0.71 
*

The combined model is comprised of the combination of each parameter with SADPM.

†

P < 0.0001 vs. SADPM; P = NS vs. ROC. AUC, area under the curve.

View Large
Table 3—

Pearson's correlation coefficients between plasma glucose concentration at 60 and 120 min during the OGTT versus indexes of insulin secretion and insulin resistance

ParameterGlucose at 60 minPGlucose at 120 minP
Fasting plasma glucose 0.5 <0.0001 0.35 <0.0001 
Plasma glucose at 120 min 0.6 <0.0001 — — 
Matsuda index −0.44 <0.0001 −0.39 <0.0001 
ΔI0–30/ΔG0–30 −0.36 <0.0001 −0.14 0.07 
ΔGlucose (AUC) 0–120 min 0.91 <0.0001 0.73 <0.0001 
ΔI0–120/ΔG0–120 × Matsuda index −0.44 <0.0001 −0.34 <0.0001 
ΔI0–30/ΔG0–30 × Matsuda index −0.53 <0.0001 −0.36 <0.0001 
ParameterGlucose at 60 minPGlucose at 120 minP
Fasting plasma glucose 0.5 <0.0001 0.35 <0.0001 
Plasma glucose at 120 min 0.6 <0.0001 — — 
Matsuda index −0.44 <0.0001 −0.39 <0.0001 
ΔI0–30/ΔG0–30 −0.36 <0.0001 −0.14 0.07 
ΔGlucose (AUC) 0–120 min 0.91 <0.0001 0.73 <0.0001 
ΔI0–120/ΔG0–120 × Matsuda index −0.44 <0.0001 −0.34 <0.0001 
ΔI0–30/ΔG0–30 × Matsuda index −0.53 <0.0001 −0.36 <0.0001 

AUC, area under the curve.

View Large

This work was supported by a General Clinical Research Center Grant RR01346 fund from the VA Research Service, National Institutes of Health Grant DK-24092, and a VA Merit Award (to R.A.D.).

1.
King H, Aubert RE, Herman WH: Global burden of diabetes, 1995–2025: prevalence, numerical estimates, and projections.
Diabetes Care
 
21
:
1414
–1431,
1998
2.
Fang J, Alderman MH: Impact of the increasing burden of diabetes on acute myocardial infarction in New York City: 1990–2000.
Diabetes
 
55
:
768
–773,
2006
3.
Gillum RF: Trends in acute myocardial infarction and coronary heart disease in the United States.
J Am Coll Cardiol
 
23
:
1273
–1277,
1994
4.
The Diabetes Prevention Program Group: Reduction in the incidence of type 2 diabetes with lyfe style intervention or metformin.
N Engl J Med
 
346
:
393
–403,
2002
5.
Tuomilehto J, Lindstrom J, Eriksson JG, Valle TT, Hamalainen H, Ilanne-Parikka P, Keinanen-Kiukaanniemi S, Laakso M, Louheranta A, Rastas M, Salminen V, Uusitupa M, the Finnish Diabetes Prevention Study Group: Prevention of type 2 diabetes mellitus by changes in lifestyle among subjects with impaired glucose tolerance.
N Engl J Med
 
344
:
1343
–1350,
2001
6.
Knowler WC, Hamman RF, Edelstein SL, Barrett-Connor E, Ehrmann DA, Walker EA, Fowler SE, Nathan DM, Kahn SE, the Diabetes Prevention Program Research Group: Prevention of type 2 diabetes with troglitazone in the Diabetes Prevention Program.
Diabetes
 
54
:
1150
–1156,
2005
7.
Chiasson JL, Josse RG, Gomis R, Hanefeld M, Karasik A, Laakso M, the STOP-NIDDM Trail Research Group: Acarbose for prevention of type 2 diabetes mellitus: the STOP-NIDDM randomised trial.
Lancet
 
359
:
2072
–2077,
2002
8.
Unwin N, Shaw J, Zimmet P, Alberti KGMM: Impaired glucose tolerance and impaired fasting glycemia: the current status on definition and intervention.
Diabet Med
 
19
:
708
–723,
2002
9.
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
 
3
:
173
–178,
1986
10.
Edelstein SL, Knowler WC, Bain RP, Andres R, Barrett-Connor EL, Dowse GK, Haffner SM, Pettitt DJ, Sorkin JD, Muller DC, Collins VR, Hamman RF: Predictors of progression from impaired glucose tolerance to NIDDM: an analysis of six prospective studies.
Diabetes
 
46
:
701
–710,1997
11.
Stern MP, Williams K, Haffner SM: Identification of persons at high risk for type 2 diabetes mellitus: do we need the oral glucose tolerance test?
Ann Intern Med
 
136
:
575
–581,
2002
12.
Schmidt MI, Duncan BB, Bang H, Pankow JS, Ballantyne CM, Golden SH, Folsom AR, Chambless LE, the Atherosclerosis Risk in Communities Investigators: Identifying individuals at high risk for diabetes: the Atherosclerosis Risk in Communities study.
Diabetes Care
 
28
:
2013
–2018,
2005
13.
Lindstrom J, Tuomilehto J: The diabetes risk score: a practical tool to predict type 2 diabetes risk.
Diabetes Care
 
26
:
725
–731,
2003
14.
DeFronzo RA: Lilly Lecture 1987: The triumvirate: β-cell, muscle, liver: a collusion responsible for NIDDM.
Diabetes
 
37
:
667
–687,
1988
15.
Gulli G, Ferrannini E, Stern M, Haffner S, DeFronzo RA: The metabolic profile of NIDDM is fully established in glucose-tolerant offspring of two Mexican-American NIDDM parents.
Diabetes
 
41
:
1575
–1586,
1992
16.
DeFronzo RA, Tobin JD, Andres R: Glucose clamp technique: a method for quantifying insulin secretion and resistance.
Am J Physiol
 
237
:
E214
–E223,
1979
17.
Matsuda M, DeFronzo RA: Insulin sensitivity indices obtained from oral glucose tolerance testing: comparison with the euglycemic glucose clamp.
Diabetes Care
 
22
:
1462
–1470,
1999
18.
Matthews D, Hosker J, Rudenski A, Naylor B, Treacher D, Turner R: Homeostasis model assessment: insulin resistance and beta cell function from fasting plasma glucose and insulin concentrations in man.
Diabetologia
 
28
:
412
–419,
1985
19.
Stumvoll M, Mitrakou A, Pimenta W, Jenssen T, Yki-Jarvinen H, Van Haeften T, Renn W, Gerich J: Use of the oral glucose tolerance test to assess insulin release and insulin sensitivity.
Diabetes Care
 
23
:
295
–301,
2000
20.
Mari A, Pacini G, Murphy E, Ludvik B, Nolan JJ: A model-based method for assessing insulin sensitivity from the oral glucose tolerance test.
Diabetes Care
 
24
:
539
–548,
2001
21.
Hanley AJ, Williams K, Gonzalez C, D'Agostino RB Jr, Wagenknecht LE, Stern MP, Haffner SM, the San Antonio Heart Study, the Mexico City Diabetes Study, the Insulin Resistance Atherosclerosis Study: Prediction of type 2 diabetes using simple measures of insulin resistance: combined results from the San Antonio Heart Study, the Mexico City Diabetes Study, and the Insulin Resistance Atherosclerosis Study.
Diabetes
 
52
:
463
–469,
2003
22.
Hanson RL, Pratley RE, Bogardus C, Narayan KM, Roumain JM, Imperatore G, Fagot-Campagna A, Pettitt DJ, Bennett PH, Knowler WC: Evaluation of simple indices of insulin sensitivity and insulin secretion for use in epidemiologic studies.
Am J Epidemiol
 
151
:
190
–198,
2000
23.
Stern MP, Rosenthal M, Haffner SM, Hazuda HP, Franco LJ: Sex difference in the effects of sociocultural status on diabetes and cardiovascular risk factors in Mexican Americans: the San Antonio Heart Study.
Am J Epidemiol
 
120
:
834
–851,
1984
24.
Stern MP, Patterson JK, Haffner SM, Hazuda HP, Mitchell BD: Lack of awareness and treatment of hyperlipidemia in type II diabetes in a community survey.
JAMA
 
262
:
360
–364,
1989
25.
Burke JP, Williams K, Gaskill SP, Hazuda HP, Haffner SM, Stern MP: Rapid rise in the incidence of type 2 diabetes from 1987 to 1996: results from the San Antonio Heart Study.
Arch Intern Med
 
159
:
1450
–1456,
1999
26.
WHO Study Group:
Diabetes Mellitus
. Geneva, World Health Org.,
1985
(Tech. Rep. Ser., no. 727)
27.
DeLong ER, DeLong DM, Clarke-Pearson DL: Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach.
Biometrics
 
44
:
837
–845,
1988

Published ahead of print at http://care.diabetesjournals.org on 23 March 2007. DOI: 10.2337/dc06-1331.

A table elsewhere in this issue shows conventional and Système International (SI) units and conversion factors for many substances.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

DIABETES CARE
2007