Diabetes mellitus is a global epidemic with India emerging as one of the leading contributors to its burden. Early identification of undiagnosed diabetes and prediabetes is critical to prevent complications and reduce disease progression. Screening traditionally relies on fasting plasma glucose (FPG) or the 2-hour oral glucose tolerance test (OGTT)(1). However, these methods have limitationsFPG requires fasting and is affected by short-term glycaemic variability, while OGTT is time-consuming, inconvenient, and poorly reproducible. To overcome these challenges, glycated haemoglobin (HbA1c) has gained attention as a potential single-step screening and diagnostic tool(2).

HbA1c reflects the mean blood glucose concentration over 8–12 weeks and is unaffected by short-term fluctuations. The American Diabetes Association (ADA,2010) recommends HbA1c ≥6.5% for diabetes diagnosis and 5.7–6.4% for prediabetes(3). However, the applicability of these thresholds to the Indian population remains uncertain. Ethnic, nutritional, and genetic factors influence HbA1c values independent of glucose levels, necessitating population-specific validation(4).

The current study aimed to evaluate the diagnostic and screening performance ofHbA1c compared to FPG and OGTT in identifying newly diagnosed diabetes in an Indian cohort. The study further assessed the optimal HbA1c cut-off offering the best balance between sensitivity and specificity, using ROC curve analysis.

Previous studies conducted in Indian settings, including Mohan et al. (2010) and Kumar et al. (2010), suggested that HbA1c values between 6.0% and6.1% may be optimal for diabetes detection(5,6). Nevertheless, inconsistency exists due to differences in sample characteristics, assay methods, and diagnostic criteria. This study, therefore, aimed to provide region-specific evidence on the diagnostic efficiency of HbA1c as a screening tool for undiagnosed diabetes in an urban Indian population.

AIMS AND OBJECTIVES:

1. To evaluate HbA1c performance compared to fasting and 2-hour post-load glucose levels in identifying undiagnosed diabetes.
2. To determine whether HbA1c can replace or supplement glucose-based tests for community-level screening.

3. To assess the feasibility of adopting HbA1c as a screening test in clinical and preventive setups in India.

Study Design and Setting: This was a hospital-based, cross-sectional observational study conducted in the Department of Medicine at Tata Main Hospital, Jamshedpur, from June 2012 to May 2014. The study aimed to evaluate the performance of glycated haemoglobin (HbA1c) as a screening and diagnostic tool for detecting previously undiagnosed diabetes and prediabetes, compared with fasting plasma glucose (FPG) and oral glucose tolerance test (OGTT). Study Population: A total of 220 adults aged ≥30 years attending health check-up clinics and general outpatient departments were initially screened. After applying inclusion and exclusion criteria, 203 subjects were enrolled for final analysis. All participants were either asymptomatic individuals undergoing routine screening or those with mild nonspecific symptoms such as fatigue, polyuria, or weight loss, without a prior diagnosis of diabetes. Inclusion Criteria: • Adults aged ≥30 years without previously diagnosed diabetes. • Subjects undergoing simultaneous testing for FPG, 2-hour OGTT, andHbA1c on the same day. • Individuals willing to provide informed consent and participate in the screening procedure. Exclusion Criteria: • Known diabetics on any form of glucose-lowering therapy. • Pregnant women or those with gestational diabetes. • Patients with anaemia, hemoglobinopathies, chronic liver or renal disease. • Individuals on long-term steroids, antiretroviral, or antiepileptic drugs affecting glucose metabolism or HbA1c. • Subjects with recent blood transfusion or acute illness. Data Collection and Investigations: All participants underwent comprehensive clinical evaluation including anthropometry (height, weight, BMI) and blood pressure measurement. Venous blood samples were collected after an overnight fast of 8–10 hours for biochemical testing. 1. Fasting Plasma Glucose (FPG): Measured by Glucose Oxidase-Peroxidase(GOD-POD) enzymatic method. 2. Oral Glucose Tolerance Test (OGTT): Conducted using 75 g of anhydrousglucose dissolved in 250 ml of water, followed by plasma glucose measurement at 2 hours post-load. 3. Glycated Haemoglobin (HbA1c): Estimated by the Turbidimetric InhibitionImmunoassay (TINIA) method standardized to IFCC and DCCT/NGSP reference systems. All assays were performed in an NABL-accredited biochemistry laboratory with strict internal and external quality controls. Diagnostic Criteria Diagnosis of diabetes was based on ADA (2010) and WHO (1999) criteria: • Diabetes: FPG ≥126 mg/dl and/or 2-hour OGTT ≥200 mg/dl. • Prediabetes: FPG 100–125 mg/dl or 2-hour OGTT 140–199 mg/dl. • Normal: FPG <100 mg/dl and 2-hour OGTT <140 mg/dl. For HbA1c-based categorization: • Normal:<5.7% • Prediabetes: 5.7–6.4% • Diabetes: ≥6.5% Statistical Analysis: All data were analysedusing SPSS software version 10.0. Continuous variables were expressed as mean ± SD, while categorical variables were presented as frequencies and percentages. • Pearson’s correlation coefficient (r) was calculated to assess the relationship between HbA1c, FPG, and 2-hour OGTT. • Receiver Operating Characteristic (ROC) curves were plotted to evaluate the diagnostic performance of HbA1c and to determine the optimal screening cut-off. • Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy were calculated for various HbA1c thresholds. • The Area Under the Curve (AUC) was used as a measure of test validity. • Agreement between HbA1c and glucose-based diagnosis was analysed using Kappa statistics, with p< 0.05 considered statistically significant.

Baseline Characteristics of the Study Population

A total of 203 adults were included in the final analysis after exclusions. The mean age was 54.9 ± 11.8 years, and the male-to-female ratio was 1.2:1. The average BMI was 25.8 ± 2.1 kg/m², with 62% of participants classified as overweight according to Asian cut-offs.

Table 1: Baseline Demographic and Clinical Profile (n = 203)

Distribution of Study Participants Based on Glycemic Status

Based on ADA and WHO glucose criteria, participants were categorized as follows:

• Normal Glucose Tolerance (NGT): 90 (44.3%)
• Prediabetes: 49 (24.1%)
• Diabetes Mellitus: 64 (31.6%)

When HbA1c ≥6.5% was applied as per ADA criteria, 59 subjects (29.1%) were classified as diabetic, indicating good concordance between HbA1c-based and glucose-based diagnosis.

Table 2: Glycaemic Status Distribution by FPG/OGTT vs HbA1c

Correlation Between HbA1c, FPG, and OGTT

A significant positive correlation was observed between HbA1c and both fasting and postprandial glucose levels:

• HbA1c vs FPG:r = 0.81, p< 0.001
• HbA1c vs 2-hour OGTT:r = 0.78, p< 0.001

This strong correlation confirms HbA1c as a valid proxy for long-term glycaemic exposure.

Table 3: Correlation of HbA1c with Glucose Parameters

Diagnostic Accuracy of HbA1c as a Screening Test

Using glucose-based criteria as the gold standard, diagnostic indices were calculated for different HbA1c thresholds.

At the ADA-recommended cut-off of 6.5%, HbA1c achieved sensitivity of 77.6% and specificity of88.9%.

However, ROC analysis identified an optimal threshold of 6.2% that maximized both sensitivity (83.7%) and specificity (86.1%).

Table 4: Diagnostic Performance of HbA1c at Different Thresholds

Receiver Operating Characteristic (ROC) Curve Analysis: The ROC curve for HbA1c screening performance demonstrated an AUC of 0.90, signifying excellent discriminative power.

The curve showed the steepest rise between 6.0% and 6.3%, indicating that this range provides the best diagnostic yield for detecting previously undiagnosed diabetes.

• Optimal Cut-off:2%
• AUC:90
• Sensitivity:7%
• Specificity:1%
• Diagnostic Accuracy:0%

Graphically, the ROC curve lay well above the reference line, confirming that HbA1c can serve as a standalone screening test in the Indian context.

Agreement Analysis (Kappa Statistics):The agreement between HbA1c (≥6.5%) and FPG/OGTT-based diagnosis was substantial, with a Kappa value of 0.73 (p < 0.001). When the cut-off was adjusted to 6.2%, the agreement improved slightly to κ = 0.78, demonstrating stronger concordance for this threshold.

Table 5: Agreement Between HbA1c and Glucose-based Diagnosis

• HbA1c showed excellent correlation with both fasting and postprandial glucose, supporting its reliability for screening undiagnosed diabetes.
• The optimal HbA1c cut-off for detecting diabetes in this Indian population was 6.2%, providing higher sensitivity than the ADA threshold.
• AUC 0.90 and κ = 0.78 confirm HbA1c as a powerful standalone tool for mass screening, especially in resource-limited or non-fasting settings.
• At lower thresholds (5.8–6.0%), sensitivity improved but specificity dropped, increasing false positivesmaking 6.2% a practical balance point.

The present study assessed the diagnostic performance of glycated haemoglobin (HbA1c) as a screening tool for previously undiagnosed diabetes in an Indian population and compared its accuracy with conventional glucose-based tests such as fasting plasma glucose (FPG) and the 2-hour oral glucose tolerance test (OGTT). The study demonstrated a strong positive correlation between HbA1c and both FPG (r = 0.81) and OGTT (r = 0.78), reaffirming the biochemical reliability of HbA1c as an integrated indicator of chronic glycemia.

ROC curve analysis identified an optimal HbA1c threshold of 6.2%, which provided the best combination of sensitivity (83.7%) and specificity (86.1%), with an area underthe curve (AUC) of 0.90, confirming excellent discriminative ability. In comparison, the ADA-recommended diagnostic threshold of 6.5% offered higher specificity (88.9%) but lower sensitivity (77.6%), implying that a proportion of truly diabetic individuals could remain undiagnosed if this international standard is applied without regional adjustment. These findings highlight the necessity of tailoring HbA1c cut-offs to suit population-specific characteristics, especially in ethnically and physiologically diverse regions like India.

The observed differences in optimal HbA1c thresholds can be attributed to several biological and environmental factors. Indians tend to develop insulin resistance and beta-cell dysfunction at younger ages and lower body mass indices, contributing to early-onset dysglycemia(7). Furthermore, the prevalence of nutritional deficiencies (such as iron deficiency anaemia) and haemoglobin variants may influence HbA1c levels independent of glycemia. Such variations explain why the same HbA1c value may represent a higher mean plasma glucose concentration in Indian populations compared to Western counterparts(8).

Comparable studies support these findings. Mohan et al. (2010), in a South Indian cohort, proposed 6.0% as the most suitable diagnostic threshold, while Kumar et al.(2011) in North India identified 6.1% as optimal(5,6). A multicentric Indian study by Manisha Nair et al. (2011) also recommended a threshold near 5.8% for population screening(9). The current study, with its optimal cut-off of 6.2%, is consistent with these national observations, providing further evidence that the Indian population benefits from slightly lower HbA1c diagnostic limits. Internationally, similar trends have been reported among Asian and African populations, where HbA1c tends to overestimate glycemia relative to Caucasian standards.

From a public health perspective, HbA1c offers substantial advantages as a screening tool. Unlike glucose-based tests, it does not require fasting, can be performed at any time of day, and reflects long-term glycaemic status rather than single-time fluctuations. This makes it particularly valuable in resource-limited and high-volume outpatient settings, where patient compliance with fasting and timed blood collection is challenging. In mass-screening programs, especially for asymptomatic individuals, an HbA1c threshold between 6.0% and 6.2% can effectively identify high-risk individuals for further confirmatory testing.

However, while lowering the diagnostic threshold increases sensitivity and enables early detection, it also slightly increases false positives. The clinical and economic implications of this must be balanced carefully. The positive predictive value (PPV) and negative predictive value (NPV) in this study (82.3% and 87.0% at 6.2%) demonstrate that HbA1c is both cost-effective and reliable when applied in large-scale screening. Importantly, Kappa agreement analysis revealed substantial concordance (κ = 0.73 at ≥6.5%) and strong concordance (κ = 0.78 at ≥6.2%) between HbA1c and glucose-based diagnosis, confirming its suitability as an alternative primary diagnostic criterion.

It is also noteworthy that HbA1c’s diagnostic role extends beyond diabetes to the identification of prediabetes (5.7–6.4%), allowing clinicians to target lifestyle interventions before irreversible beta-cell decline occurs. In India, where nearly one in three adults may harbour undiagnosed dysglycemia, such early detection has far-reaching preventive implications.

Despite these promising results, certain limitations must be acknowledged. HbA1c levels can be influenced by conditions that affect erythrocyte turnover, such as anaemia, chronic kidney disease, and hemoglobinopathiesconditions not uncommon in Indian patients. Additionally, regional laboratory variability and lack of nationwide standardization in HbA1c assays could impact accuracy. Hence, the implementation of HbA1c-based screening must be accompanied by rigorous standardization following NGSP and IFCC protocols to ensure comparability across laboratories.

The ROC curve pattern from this study confirms that the optimal discriminative region lies between 6.0% and 6.3%, beyond which sensitivity declines steeply. This observation underscores that while the ADA’s 6.5% cut-off maintains high specificity, it may not be sensitive enough for early detection in Indian adults. Adopting a threshold around 6.2% could identify a greater number of undiagnosed cases at an earlier, potentially reversible stage, contributing significantly to reducing diabetes-related morbidity.

HbA1c demonstrates excellent validity, convenience, and diagnosticpower as a screening tool for undiagnosed diabetes in India. Establishing a slightly lower, regionally validated diagnostic threshold would enhance the precision of diabetes detection strategies, strengthen community-based screening programs, and support India’s efforts in combating the diabetes epidemic through earlier diagnosis and intervention.

The study establishes that HbA1c is an effective, practical, and reliable screening tool for detecting undiagnosed diabetes in Indian populations. The optimal HbA1c cut-off identified was 6.2%, providing superior diagnostic balance compared to the ADA-recommended 6.5%. With an AUC of 0.90, sensitivity of 83.7%, and specificity of 86.1%, HbA1c offers excellent screening performance and strong agreement with glucose-based tests. These findings emphasize the need for population-specificcalibration of HbA1c thresholds in India to enhance early diabetes detection and prevention, enabling cost-effective, non-fasting, and convenient diagnostic implementation across primary healthcare settings.

1. Pradeepa R, Mohan V. Epidemiology of type 2 diabetes in India. Indian J Ophthalmol. 2021;69(11):2932–8.
2. Sheng L, Yang G, Chai X, Zhou Y, Sun X, Xing Z. Glycemic variability evaluated by HbA1c rather than fasting plasma glucose is associated with adverse cardiovascular events. Front Endocrinol (Lausanne). 2024;15:1323571.
3. Li G, Han L, Wang Y, Zhao Y, Li Y, Fu J, et al. Evaluation of ADA HbA1c criteria in the diagnosis of pre-diabetes and diabetes in a population of Chinese adolescents and young adults at high risk for diabetes: a cross-sectional study. BMJ Open. 2018;8(8):e020665.
4. Pemberton JS, Fang Z, Chalew SA, Uday S. Ethnic disparities in HbA1c and hypoglycemia among youth with type 1 diabetes: beyond access to technology, social deprivation and mean blood glucose. BMJ Open Diabetes Res Care. 2025;13(1):e004369.
5. Kumar PR, Bhansali A, Ravikiran M, Bhansali S, Dutta P, Thakur JS, et al. Utility of glycated hemoglobin in diagnosing type 2 diabetes mellitus: a community-based study. J Clin Endocrinol Metab. 2010;95(6):2832–5.
6. Mohan V, Vijayachandrika V, Gokulakrishnan K, Anjana RM, Ganesan A, Weber MB, et al. A1C cut points to define various glucose intolerance groups in Asian Indians. Diabetes Care. 2010;33(3):515–9.
7. Siddiqui MK, Anjana RM, Dawed AY, Martoeau C, Srinivasan S, Saravanan J, et al. Young-onset diabetes in Asian Indians is associated with lower measured and genetically determined beta cell function. Diabetologia. 2022;65(6):973–83.
8. Christy AL, Manjrekar PA, Babu RP, Hegde A, Rukmini MS. Influence of iron deficiency anemia on hemoglobin A1C levels in diabetic individuals with controlled plasma glucose levels. Iran Biomed J. 2014;18(2):88–93.

9.             Nair M, Prabhakaran D, Narayan KMV, Sinha R, Lakshmy R, Devasenapathy N, et al. HbA1c values for defining diabetes and impaired fasting glucose in Asian Indians. Prim Care Diabetes. 2011;5(2):95–102.