Currently, the diagnostic criteria for diabetes include fasting glucose (FPG), 2-hour oral glucose tolerance test (2 h OGTT) and glycated hemoglobin (HbA1c) test. So, which is the most reliable? FPG is the blood glucose concentration measured after 12 hours of fasting (except water); 2 h OGTT is to detect the change of blood glucose after eating the prescribed amount of glucose; HbA1c does not directly detect the blood glucose level, but can represent the average blood glucose level in the past 2-3 months. In 2009, the American Diabetes Association, the European Association for the Study of Diabetes and the International Diabetes Federation jointly recommended the use of HbA1c laboratory tests for the diagnosis of type 2 diabetes, with HbA1c ≥ 6.5% (48 mmol/mol) being diagnostic of diabetes, although diabetes cannot be excluded below this threshold. Researchers in population-based studies have used different biological indicators to diagnose diabetes in different countries and at different times, but the specificity and accuracy of these tests are unclear. The diversity and inconsistency of diagnostic criteria make it challenging to assess the prevalence of diabetes and the effectiveness of its treatment. In addition, many countries have limited resources to monitor risk factors using only the simple and standardized STEPS method published by the World Health Organization (WHO) in 2005.The main risk factors for STEPS include demographic information, health behaviors, body mass index (BMI), waist circumference, and blood pressure. WHO recommends that countries with adequate resources monitor FPG, optionally with OGTT, however, currently only 57% of the 221 countries worldwide are able to conduct quality diabetes surveys and only 19% have OGTT-based results. A report in The Lancet Diabetes & Endocrinology included 96 studies that investigated the sensitivity and specificity of glucose tests, as well as diagnostic methods based on HbA1c, FPG and 2 hOGTT. The investigators compared the prevalence of diabetes diagnosed using HbA1c (HbA1c ≥ 6.5%, previous history of diabetes or use of insulin and glucose-lowering drugs) with FPG alone and FPG combined with 2 hOGTT (FPG ≥ 7?0 mmol/L, 2 hOGTT ≥ 11.1 mmol/L, previous history of diabetes or use of insulin and glucose-lowering drugs). These subjects used at least two diagnostic criteria in assessing the prevalence of diabetes. The results showed that: 1. The diagnosis of FPG combined with 2 h OGTT was strongly correlated with the prevalence of diabetes diagnosed using FPG only, but the prevalence level was 2-6% higher, while the correlation with HbA1c was slightly lower. The prevalence of HbA1c-based diagnosis was lower than FPG in 42.8% of the surveyed population, higher than FPG in 41.6%, and equal in 15.6% for both diagnostic methods. 3. In previously undiagnosed subjects, the sensitivity (true positive rate) of HbA1c ≥ 6.5% compared to FPG ≥ 7.0 mmol/L was 52.8% and the specificity (true negative rate) was 99.74%, while the sensitivity of HbA1c was 30.5% compared to FPG combined with 2 hOGTT. 4. In the regression analysis, the prevalence of diabetes diagnosed by HbA1c was lower than that diagnosed based on blood glucose. In other words, different biochemical indicators and diagnostic criteria diagnose the undiagnosed diabetic population differently, leading to differences in the assessment of the prevalence of diabetes. Therefore, the use of HbA1c tests alone is not sufficient to diagnose the majority of the undiagnosed population in a healthy population survey, which is likely to be diagnosed with diabetes if glucose tests are used. Therefore, it is advisable to use HbA1c along with FPG to provide additional evidence to assess the relevance of the two tests. According to the results of this study by the NCD Risk Factor Collaborative Group, the prevalence of glucose and HbA1c diagnoses varies with national income, years of study, and BMI, as well as the decreased sensitivity of HbA1c to diagnose diabetes, and HbA1c should be used as a supplement to methods for assessing the prevalence of diabetes based on glucose testing based on existing diagnostic criteria. However, due to the high correlation of several diagnostic methods, it is recommended to add appropriate flexibility and use different assessment strategies under different conditions to facilitate the development of a global epidemiological assessment model of diabetes.