Comparison of Fasting Plasma Glucose and Glycated Hemoglobin for Diagnosing Diabetes in a Taiwanese Population

Abstract

Background:

The purposes of this study were to examine the impact of using glycated hemoglobin (HbA1c) as a diagnostic criterion for (pre)diabetes and to determine the appropriate HbA1c cutoff value in a Taiwanese population.

Methods:

We used a dataset from the Clinical Informatics Research & Development Center of Taichung Veteran General Hospital. Fasting plasma glucose (FPG) and HbA1c test results were obtained from outpatient laboratory reports produced from January 1, 2011, to April 30, 2012. A total of 4920 blood tests were analyzed. For each potential HbA1c cutoff value, the sensitivity, specificity, positive and negative predictive values, and area under the receiver operator characteristic curve (AUC) were calculated at FPG levels of 100 and 126 mg/dL. Two-sample t- and chi-sqaured tests were used to compare differences in characteristics between individuals matching the definitions of diabetes set by the American Diabetes Association (ADA) in 2009 and 2010.

Results:

Among the 4920 study subjects, 580 people had an FPG value <126 mg/dL and an HbA1c ≥6.5%. After the diagnostic criterion of HbA1c percentage was applied, the numbers of patients with normoglycemia and prediabetes decreased, whereas the number of diabetic patients increased. For FPG levels of 100 and 126 mg/dL, the HbA1c cutoff points were 5.95% and 6.35%, respectively.

Conclusions:

More patients were diagnosed with diabetes when HbA1c was added as a diagnostic criterion. HbA1c thresholds of 5.95% and 6.35% were highly specific for prediabetes and diabetes, respectively, in Taiwanese adults.

Introduction

Type 2 diabetes mellitus (T2DM) is a costly chronic condition whose global prevalence continues to increase rapidly, particularly in Asia.¹ In the first decade of 2000, the number of patients with diabetes in Taiwan increased by more than 70%.² Furthermore, over one-fifth of Taiwan's total healthcare costs are derived from diabetes-related expenditures. These high costs could be reduced through early disease intervention. Thus, screening of high-risk individuals is urgently needed to reduce the incidence of diabetic complications.³

Traditionally, fasting plasma glucose (FPG) has been used as the diagnostic indicator for incident diabetes, in accordance with recommendations of the American Diabetes Association (ADA).⁴ However, over the last 20 years, studies have consistently correlated glycated hemoglobin (HbA1c) levels with diabetes complications, including retinopathy and other microvascular diseases.^5,6 These studies provide evidence supporting the use of HbA1c as a diagnostic marker for diabetes.

In 2009, the International Expert Committee recommended that an HbA1c value ≥6.5% be used as a diagnostic criterion of T2DM, with an HbA1c value of 6.1–6.49% being considered indicative of impaired fasting glucose levels. In 2010, the ADA changed its recommendations to allow the use of either the FPG or HbA1c criterion for diagnosing diabetes.⁷ The Taiwanese Diabetes Association adopted the HbA1c criterion for diabetes diagnosis in 2012.

Because racial and geographic differences may impact the reliability of HbA1c assays,⁸ optimal thresholds for detecting diabetes need to be determined for each region and each ethnic group. Accordingly, throughout the world, researchers have performed various community-based studies of adults who have not been diagnosed with diabetes to determine the appropriate HbA1c cutoff value for diabetes diagnosis. For example, in Shanghai, an HbA1c cutoff value of 6.3% had specificity and sensitivity values of 96.1% and 62.8%, respectively, for diabetes diagnosis.⁹ In Korea, these values were 91% and 68%, respectively, when an HbA1c cutoff value of 5.9% was used. An HbA1c cutoff of 5.6% predicted the 6-year risk of developing diabetes with specificity and sensitivity values of 77% and 59%, respectively.¹⁰ In Japan, an HbA1c cutoff value of 5.7% corresponded to the ADA-recommended FPG cutoff of 100 mg/dL.¹¹ Among Asian Indians, the HbA1c cutoff points were somewhat higher (6.1–6.4%), to ensure an appropriate accuracy of the test.¹² In the United States, an HbA1c cutoff value of 6.5% has been identified as optimal for diabetes diagnosis.¹³

The characteristics of patients and prevalence of diabetes have not been thoroughly examined in Taiwan since the HbA1c was added as a diagnostic criterion. Therefore, the objectives of this study were to investigate how the HbA1c criterion performs with regard to identifying prediabetes and diabetes in a Taiwanese population, and to determine the appropriate HbA1c cutoff points for FPG values of 100 and 126 mg/dL.

Methods

Study population

This study was designed to be cross-sectional and used a dataset from the Clinical Informatics Research & Development Center of Taichung Veteran General Hospital. Outpatient laboratory report data, such as date of birth, sex, diagnosis codes, medication lists, visit dates, treatment, and laboratory data, were included in the dataset. Diagnoses were coded according to the International Classification of Disease, 9^th Revision, Clinical Modification (ICD-9-CM). The dataset was scrambled before release to prevent individual patient identification and to protect patient privacy. Approval was obtained from the Ethics Committee of Clinical Research at Taichung Veterans General Hospital. Patient information was de-identified prior to analysis. Informed consent was not obtained. Between January 1, 2011, and April 30, 2012, the FPG and HbA1c values from 20,371 blood tests were analyzed. We identified 4920 patients with HbA1c and FPG test results that did not have any history of being diagnosed of having diabetes or use of antidiabetic agents (diagnosis of ICD-9-CM codes 250.xx recorded in the medical record or the use of antidiabetic agents).

Study design

Patients were excluded from the study if they were under 20 years old or had a history of diagnosed diabetes, chronic kidney disease at stages 3b–5 [defined by an estimated glomerular filtration rate (eGFR) <45 mL/min per 1.73 m²],¹⁴ or anemia [defined by hemoglobin (Hb) <13.0 grams/dL in men and <12.0 grams/dL in women].¹⁵

Biochemical analyses

Methods used for determining the levels of plasma glucose, serum creatinine, and serum lipids, and the activities of serum glutamate oxaloacetate transaminase (GOT), glutamic pyruvic transaminase (GPT), C-reactive protein (CRP), and each lipid parameter were based on an enzymatic analysis (Hitachi 7600-110 automatic analyzer; Hitachi Co., Tokyo, Japan). The HbA1c value was measured in the central laboratory by high-performance liquid chromatography, using the Primus CLC385 (Primus Corporation, Kansas City, MO). Hb concentrations were measured using the Coulter LH-750 hematology analyzer (Beckman Coulter Inc. Miami, FL).

Statistical analysis

The sensitivity, specificity, positive and negative predictive values (PPV and NPV, respectively), and area under the receiver operator characteristic (ROC) curve (AUC) were calculated at all HbA1c cutoff values. Cutoffs for prediabetes and diabetes were determined with the ROC curve. The 2010 diagnostic criteria of the ADA were used as the gold standard to evaluate the 2009 diagnostic criteria of the ADA, and the sensitivity, specificity, PPV, and NPV were determined. Two-sample t- and chi-squared tests were used to compare patient characteristics, using the 2009 and 2010 definitions from the ADA. Two-tailed P values of less than 0.05 were considered to be statistically significant. Data were analyzed using SAS version 9.2 (SAS Institute, Cary, NC).

Results

Plasma glucose and HbA1c cutoff values

The HbA1c cutoff points were 5.95% (AUC 0.885, sensitivity 81.0%, and specificity 84.8%) for an FPG level of 100 mg/dL, and 6.35% (AUC 0.898, sensitivity 87.4%, and specificity 78.0%) for an FPG level of 126 mg/dL.

Number of patients after adding HbA1c as a diagnostic criterion

After HbA1c was added to the diagnostic criteria in 2010, the number of patients with and without diabetes increased and decreased, respectively. A total of 580 patients had FPG values <126 mg/dL and HbA1c ≥6.5% and were not diagnosed with diabetes in 2009, but were diagnosed with diabetes in 2010. Addition of HbA1c as a diagnostic criterion increased the sensitivity of diagnosis of diabetes and not prediabetes (sensitivity 76.8%, specificity 100.0%, PPV 100%, NPV 80.6%, false-positive rate 0%, and false-negative rate 23.15%).

Patient characteristics after adding HbA1c as a diagnostic criterion

Table 1 shows the numbers and characteristics of patients who were diagnosed with diabetes using the 2009 and 2010 criteria of the ADA. After HbA1c was added to the diagnostic criteria in 2010, the numbers of normoglycemic and prediabetic patients decreased, whereas the number of diabetic patients increased. The average [mean±standard deviation (SD)] FPG and HbA1c levels of the three groups decreased significantly. In the prediabetes group, the FPG decreased from 111.8±7.7 mg/dL to 104.7±11.7 mg/dL, and the HbA1c decreased from 6.3%±0.9% to 5.8%±0.3%. In the diabetic group, the FPG decreased from 173.9±57.3 mg/dL to 158.6±58.0 mg/dL, and the HbA1c decreased from 8.0%±1.8% to 7.8%±1.6%. No significant differences were found in creatinine, eGFR, total cholesterol, triglycerides (TGs), high-density lipoprotein cholesterol (HDL-C), GOT, GPT, CRP, or postprandial glucose.

Table 1.

Patient Characteristics As Categorized According to the American Diabetes Association Diabetes Diagnostic Criteria of 2009 and 2010

		Normoglycemia			Prediabetes			Diabetes
Characteristic	Total	ADA 2009	ADA 2010	P value	ADA 2009	ADA 2010	P value	ADA 2009	ADA 2010	P value
n	4920	1634	1168		1361	1247		1925	2505
Age (years)	61.4±14.5	55.8±14.4	52.8±13.4	<0.001	64.4±13.7	63.4±13.7	0.081	64.0±13.8	63.4±13.9	0.401
Sex (male)	2847(57.9)	887 (54.3)	627 (53.7)	0.752	827 (60.8)	747 (59.9)	0.654	1133 (58.9)	1473 (58.8)	0.971
HbA1c (%)	6.7±1.7	5.6±0.8	5.3±0.2	<0.001	6.3±0.9	5.8±0.3	<0.001	8.0±1.8	7.8±1.6	0.007
FPG (mg/dL)	128.0±52.6	87.3±8.3	87.0±7.1	0.259	111.8±7.7	104.7±11.7	<0.001	173.9±57.3	158.6±58.0	<0.001
PPG (mg/dL)	206.6±87.2	147.0±40.5	146.3±46.8	0.980	156.8±56.2	137.3±32.5	0.202	254.5±85.9	248.8±83.7	0.797
Cr (mg/dL)	0.9±0.3	0.9±0.3	0.8±0.2	0.029	1.0±0.3	1.0±0.3	0.563	1.0±0.3	1.0±0.3	0.821
eGFR (mL/min per 1.73 m²)	87.8±28.3	93.8±27.2	96.4±26.8	0.023	83.3±28.1	83.8±26.8	0.701	83.6±28.6	83.6±28.9	0.991
Lipids (mg/dL)
Cholesterol (mg/dL)	189.5±41.1	195.6±38.8	197.8±38.0	0.181	188.4±40.1	191.5±39.0	0.101	182.5±43.7	181.8±43.4	0.680
TGs (mg/dL)	149.5±112.0	131.8±108.0	131.1±117.0	0.895	149.6±95.8	146.6±93.1	0.528	172.3±124.3	165.6±116.3	0.185
HDL-C (mg/dL)	54.2±15.2	57. 0±16.0	58.0±16.2	0.132	53.4±15.0	54.1±15.0	0.418	50.7±13.4	50.9±13.6	0.820
LDL-C (mg/dL)	110.0±36.7	110.9±38.7	113.8±41.2	0.487	111.6±35.9	112.7±34.9	0.711	108.2±36.1	107.5±36.1	0.755
GOT (IU/L)	28.4±26.4	24.7±15.2	24.2±15.0	0.525	33.0±42.4	30.8±38.9	0.426	35.0±27.9	34.8±27.2	0.918
GPT (IU/L)	37.4±51.3	31.7±28.5	30.9±28.5	0.479	39.6±46.7	38.1±45.4	0.524	43.9±74.7	42.8±67.5	0.742
CRP (mg/dL)	0.4±2.2	0.2±0.6	0.2±0.7	0.905	0.3±1.0	0.3±0.9	0.553	2.0±6.6	1.6±5.8	0.606
Hb (grams/dL)	14.3±1.3	14.3±1.3	14.3±1.3	0.980	14.3±1.2	14.3±1.2	0.892	14.4±1.3	14.4±1.2	0.796

Data are the mean±standard deviation (SD) or n (%) for categorical variables.

P values refer to the statistical significance of the differences between groups, calculated using the two-sample t-test.

ADA, American Diabetes Association; HbAuc, glycated hemoglobin; FPG, fasting plasma glucose; PPG, postprandial glucose; Cr, creatinine; eGFR, estimated glomerular filtration rate; TGs, triglycerides; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; GOT, glutamate oxaloacetate transaminase; GPT, glutamic pyruvic transaminase; CRP, C-reactive protein; Hb, hemoglobin.

Discussion

This study demonstrated the utility of HbA1c testing for diagnosing prediabetes and diabetes in a Taiwanese population. The number of patients diagnosed with diabetes increased after the HbA1c value was included as a diagnostic criterion. Optimal HbA1c cutoff values for diagnosis of diabetes and prediabetes were 6.35% and 5.95%, respectively. Patients were diagnosed with diabetes at lower HbA1c and FPG levels after the HbA1c value was included as a diagnostic criterion.

By definition, HbA1c is a form of Hb exhibiting glycation of the β-chain at one or both of the amino-terminal valines.^16,17 Because of the intrinsic relationship between Hg and red blood cells (RBCs), HbA1c may not be an accurate measure of glycemic control in individuals with RBC dysfunctions (e.g., due to genetic variations, renal failure, chronic liver disease, etc.).^17
–19 For example, anemia due to iron deficiency in women has been associated with an elevated HbA1c (≥5.5%).²⁰ To prevent bias in diabetes diagnosis due to RBC dysfunction, we excluded patients with anemia, renal failure, or liver disease from our study.

Racial and geographic differences may impact the reliability of HbA1c assays.⁸ An HbA1c threshold of 6.35% yielded the highest AUC (0.898) for diabetes detection, corresponding to an FPG cutoff of 126 mg/dL. The proficiency of an HbA1c threshold of 5.95% for prediabetes detection was equivalent to that of an FPG threshold of 100 mg/dL. Our findings are similar to those of study in Shanghai, in which the sensitivity of diabetes diagnosis using an HbA1c cutoff of 6.3% was equivalent to that of an FPG cutoff of 126 mg/dL.⁹

Each year, approximately 5–10% of people with impaired glucose tolerance will develop T2DM.²¹ This fact is relevant when considering how to screen for individuals with diabetes and prediabetes. As indicators of diabetes, the HbA1c, FPG, and postload plasma glucose tests [e.g., the 2-h or 75-gram oral glucose tolerance test (OGTT)] measure different aspects of glycemia.

Among our 4920 study subjects, 580 people had an FPG value <126 mg/dL and an HbA1c ≥6.5%. When HbA1c was added to the diagnostic criteria, these 580 individuals were diagnosed with diabetes. Using the 2010 diagnostic criteria of the ADA as the gold standard to evaluate 2009 diagnostic criteria, we obtained sensitivity and specificity values of 76.8% and 100%, respectively. The PPV and NPV were 100% and 80.6%, respectively. Compared with other studies, the value of sensitivity, specificity, PPV, and NPV in our study is high. Therefore, after the HbA1c values were included in the diagnostic criteria, the number of patients with normoglycemia and prediabetes decreased, whereas the number with diabetes increased. A population-based study in the United States revealed an increased prevalence of prediabetes when HbA1c was included in the diagnostic criteria, which was in conflict with our findings. Individuals were diagnosed with prediabetes when they exhibited an HbA1c of 5.7–6.49% or an FPG of 100–125.99 mg/dL.²² However, another report in the United States found no significant difference in diabetes prevalence when HbA1c rather than FPG was used.²³ In that study, approximately 50% of subjects with FPG ≥126 mg/dL had an HbA1c value of 6.00–6.49%.

Few studies have examined the characteristics of patients since the addition of HbA1c to the diagnostic criteria. In our Taiwanese population, the main differences in patient characteristics after HbA1c was added concerned the FPG and HbA1c levels; no significant differences in any other laboratory values were found. The diagnostic thresholds for mean FPG and HbA1c levels decreased in patients with normoglycemia, prediabetes, and diabetes. The mean age of patients with normoglycemia was also younger after HbA1c was included.

Metabolic syndrome increased risk of heart disease, stroke, and diabetes, and was diagnosed by a co-occurrence of three out of five conditions—abdominal obesity, elevated blood pressure, elevated FPG, high serum TGs, and low HDL levels. Examining the data in Table 1, it is interesting that measurements of TGs and HDL did not differ in subgroups diagnosed using fasting glucose alone versus diagnosed by fasting glucose and HbA1c. This implies that the patients with normal fasting glucose who were identified as prediabetic or diabetic by HbA1c alone may not have the metabolic syndrome and may not differ from the subjects with normal FPG plus normal HbA1c values with respect to fasting glucose, TGs, and HDL levels. This calls into question whether the addition of HbA1c identifies insulin-resistant subjects with prediabetes and suggests that these patients may actually be healthy from a cardiometabolic disease perspective. However, blood pressure and waist circumstance were not available in our dataset, and further analysis is needed.

In clinical practice, the HbA1c test is a reliable assay that does not require fasting and is less uncomfortable for the patient than an FPG or 75-gram OGTT test. Because most residents of Taiwan are covered by the National Health Insurance (coverage rate of ∼97% in 2001), the HbA1c assay could be used nearly everywhere in Taiwan. Moreover, in Taiwan, the cost of an OGTT is 1.5 times higher than that of an HbA1c test. For these reasons, using the HbA1c assay as a diagnostic tool would greatly affect the surveillance of dysglycemia in Taiwan. Despite the continuously increasing and high incidence of diabetes in Taiwan (prevalence of 6.5–6.6% in 2004, standardized by age),²⁴ the disease remains underdiagnosed.²⁵ HbA1c testing could improve the diagnosis of dysglycemia in Taiwan, thereby enabling more timely therapeutic intervention. Although use of this test may increase the costs of early detection and treatment, late complication costs could be saved due to early intervention.

This study investigated the change in patient numbers and characteristics after HbA1c was added to the diagnostic criteria for prediabetes and diabetes in Taiwan. We estimated the appropriate HbA1c cutoff points for FPG levels of 100 and 126 mg/dL. The main limitation of this study was inadequate sample size. We used a dataset of hospital outpatients, which may not represent the general population. In addition, a 75-gram OGTT test was not available in this study. Guo et al. had assessed diagnosis using both fasting and 2-h glucose values of 5395 adults without known diabetes from the National Health and Nutrition Examination Survey 2005–2010, and found that HbA1c had low sensitivity and high specificity for identifying diabetes and prediabetes, which varied as a function of age and race.²⁶ We should place our data in the context of related publications when the 2-h glucose is available.

Our findings suggest that using HbA1c as a diagnostic tool would greatly affect the worldwide surveillance and medical expenditures of dysglycemia. However, our study was cross-sectional in nature. Further longitudinal studies are needed to investigate the medical expenses of early diagnosis and treatment compared to the expenses related to diabetic complications.

In conclusion, this study found that the number of diabetic patients increased after HbA1c measurement was added as a diagnostic criterion, which will lead to earlier diagnosis and treatment at a younger age. Moreover, the quality of life of patients would also be improved if diabetic complications were to be prevented by early diagnosis and aggressive treatment. Thresholds of 5.95% and 6.35% were highly specific for detecting undiagnosed prediabetes and diabetes, respectively, in Taiwanese adults.

Footnotes

Acknowledgments

No funding was received for the research reported in the article.

Author Disclosure Statement

No competing financial interests exist.

References

Wild

, Roglic

, Green

, et al. Global prevalence of diabetes: Estimates for the year 2000 and projections for 2030. Diabetes Care, 2004:27:1047–1053.

Jiang

, Chang

, Tai

, et al. Incidence and prevalence rates of diabetes mellitus in Taiwan: Analysis of the 2000–2009 Nationwide Health Insurance database. J Formos Med Assoc, 2012; 111:599–604.

Chang

, Jiang

, Chang

, et al. Accountability, utilization and providers for diabetes management in Taiwan, 2000–2009: An analysis of the National Health Insurance database. J Formos Med Assoc, 2012:111:605–616.

American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care, 2006:29(Suppl 1):S43–S48.

van Leiden

, Dekker

, Moll

, et al. Risk factors for incident retinopathy in a diabetic and nondiabetic population: The Hoorn study. Arch Ophthalmol, 2003:121:245–251.

Diabetes Control and Complications Trial Research Group. The relationship of glycemic exposure (HbA1c) to the risk of development and progression of retinopathy in the diabetes control and complications trial. Diabetes, 1995:44:968–983.

Standards of medical care in diabetes—2010. Diabetes Care, 2010:33(Suppl 1):S11–S61.

Cohen

. A1C: Does one size fit all?. Diabetes Care, 2007; 30:2756–1758.

Bao

, Ma

, Li

, et al. Glycated haemoglobin A1c for diagnosing diabetes in Chinese population: Cross sectional epidemiological survey. BMJ, 2010; 340:c2249.

10.

Choi

, Kim

, Lim

, et al. Hemoglobin A1c as a diagnostic tool for diabetes screening and new-onset diabetes prediction: A 6-year community-based prospective study. Diabetes Care, 2011; 34:944–949.

11.

Kato

, Noda

, Suga

, et al. Haemoglobin A1c cut-off point to identify a high risk group of future diabetes: Results from the Omiya MA Cohort Study. Diabet Med, 2012; 29:905–910.

12.

Mohan

, Vijayachandrika

, Gokulakrishnan

, et al. A1C cut points to define various glucose intolerance groups in Asian Indians. Diabetes Care, 2010; 33:515–519.

13.

Buell

, Kermah

, Davidson

. Utility of A1C for diabetes screening in the 1999 2004 NHANES population. Diabetes Care, 2007; 30:2233–2235.

14.

Bailie

, Uhlig

, Levey

. Clinical practice guidelines in nephrology: Evaluation, classification, and stratification of chronic kidney disease. Pharmacotherapy, 2005; 25:491–502.

15.

McLean

, Cogswell

, Egli

, et al. Worldwide prevalence of anaemia, WHO Vitamin and Mineral Nutrition Information System, 1993–2005. Public Health Nutr, 2009; 12:444–454.

16.

McCarter

, Hempe

, Chalew

. Mean blood glucose and biological variation have greater influence on HbA1c levels than glucose instability: An analysis of data from the Diabetes Control and Complications Trial. Diabetes Care, 2006; 29:352–355.

17.

Bry

, Chen

, Sacks

. Effects of hemoglobin variants and chemically modified derivatives on assays for glycohemoglobin. Clin Chem, 2001; 47:153–163.

18.

Gould

, Davie

, Yudkin

. Investigation of the mechanism underlying the variability of glycated haemoglobin in non-diabetic subjects not related to glycaemia. Clin Chim Acta, 1997; 260:49–64.

19.

Morse

. Mechanisms of hemolysis in liver disease. Ann Clin Lab Sci, 1990; 20:169–174.

20.

Kim

, Bullard

, Herman

, et al. Association between iron deficiency and A1C levels among adults without diabetes in the National Health and Nutrition Examination Survey, 1999–2006. Diabetes Care, 2010; 33:780–785.

21.

Abdul-Ghani

, Tripathy

, DeFronzo

. Contributions of beta-cell dysfunction and insulin resistance to the pathogenesis of impaired glucose tolerance and impaired fasting glucose. Diabetes Care, 2006; 29:1130–1139.

22.

Bullard

, Saydah

, Imperatore

, et al. Secular changes in U.S. prediabetes prevalence defined by hemoglobin A1c and fasting plasma glucose: National Health and Nutrition Examination Surveys, 1999–2010. Diabetes Care, 2013; 36:2286–2293.

23.

Carson

, Reynolds

, Fonseca

, et al. Comparison of A1C and fasting glucose criteria to diagnose diabetes among U.S. adults. Diabetes Care, 2010; 33:95–97.

24.

Chang

, Shau

, Jiang

, et al. Type 2 diabetes prevalence and incidence among adults in Taiwan during 1999–2004: A national health insurance data set study. Diabet Med, 2010,;27:636–643.

25.

Pan

, Yeh

, Hwu

, et al. Undiagnosed diabetes mellitus in Taiwanese subjects with impaired fasting glycemia: Impact of female sex, central obesity, and short stature. Chin J Physiol, 2001; 44:44–51.

26.

Guo

, Moellering

, Garvey

. Use of HbA1c for diagnoses of diabetes and prediabetes: Comparison with diagnoses based on fasting and 2-hr glucose values and effects of gender, race, and age. Metab Syndr Relat Disord, 2014; 12:258–268.