COMPARISON OF DIAGNOSTIC PERFORMANCE AND CLINICAL UTILITY OF DIFFERENT HBA1C CRITERIA AGAINST ORAL GLUCOSE TOLERANCE TEST (OGTT) IN SCREEN DETECTION OF DIABETES IN PENANG, MALAYSIA

Abstract

INTRODUCTION
Glycated haemoglobin (HbA1c) cut-off of 6.5% has been recommended for diagnosing diabetes mellitus. However, HbA1c levels can vary due to functional glucose-independent attributes like ethnicity, which could influence diagnostic performance across different populations. The Malaysian Clinical Practice guideline advocates a lower HbA1c diagnostic cut-off of 6.3% based on one large national study. This study compared the diagnostic performance and clinical utility of HbA1c cut-offs of 6.5% and 6.3%, against the gold standard OGTT in diagnosing diabetes.

METHODOLOGY
A total of 298 subjects without diabetes aged ≥30 years old were purposively sampled across general hospitals, health clinics and community centers in Penang between February 2016 and December 2017. HbA1c cut-offs of 6.3% and 6.5% were validated against the 2-hour OGTT results. Sensitivity, specificity and predictive values were analyzed using SPSS version 22.0. Likelihood ratios (LR) with 95% CIs for diagnostic accuracy were yielded using MedCalc statistical software.

RESULTS
The sample constituted of 175 (58.7%) women and 123 (41.3%) men with mean age (50.9±11.5 years) and mean BMI (26.4±5.50 kg/m2). The prevalence of diabetes was 21.8% using OGTT, 22.8% using an HbA1c cut-off of 6.5% and 30.2% using an HbA1c cut-off of 6.3%. At HbA1c level of 6.5%, the sensitivity was 72.3% and specificity was 91% with positive and negative predictive values of 69% and 92%. Likelihood ratios (LR+ and LR-, 95% CI) were 8.0 (5.2- 12.39) and 0.3 (0.2-0.5), respectively. In contrast, HbA1c cut-off of 6.3% yielded a sensitivity of 84.6%, specificity of 85% with positive and negative predictive values of 61% and 95%. Likelihood ratios (LR+ and LR-, 95% CI) were 5.6 (4.1-7.8) and 0.2 (0.1-0.3), respectively.

CONCLUSION
HbA1c cut off of 6.5% demonstrated better diagnostic performance and clinical utility compared to 6.3%. Larger studies are needed to identify the optimal HbA1c diagnostic criteria for multiethnic Malaysia.