The cutoffs and performance of glycated hemoglobin for diagnosing diabetes and prediabetes in a young and middle-aged population and in an elderly population

- Cutoffs of HbA1c were evaluated in two diverse population-based cohorts.
- HbA1c levels increased with age.
- Age-specific HbA1c cutoffs for detecting diabetes and prediabetes were appropriate.
- The performance of HbA1c as a diagnostic tool was poor.

The aims were to compare the appropriate cutoffs of glycated hemoglobin (HbA1c) in a population of varying ages and to evaluate the performance of HbA1c for diagnosing diabetes and prediabetes. A total of 1064 participants in the young and middle-aged group and 1671 in the elderly group were included and underwent HbA1c testing and an oral glucose tolerance test (OGTT). Sensitivity, specificity, and area under the receiver operating characteristic curve (AUC) were calculated to evaluate the optimal HbA1c cutoffs. Kappa coefficients were used to test for agreement between HbA1c categorization and OGTT-based diagnoses. The optimal HbA1c cutoffs for diagnosing diabetes were 5.7% (39 mmol/mol) in the young and middle-aged group with a sensitivity of 66.7%, specificity of 86.7%, and AUC of 0.821 (95% CI: 0.686, 0.955) and 5.9% (41 mmol/mol) in the elderly group with a sensitivity of 80.4%, specificity of 73.3%, and AUC of 0.831 (0.801, 0.861). The optimal cutoffs for diagnosing prediabetes were 5.6% (38 mmol/mol) and 5.7% (39 mmol/mol) in the young and middle-aged group and in the elderly group, respectively. Agreement between the OGTT-based diagnosis of diabetes or prediabetes and the optimal HbA1c cutoff was low (all kappa coefficients <0.4). The combination of HbA1c and fasting plasma glucose increased diagnostic sensitivities or specificities. In conclusion, age-specific HbA1c cutoffs for diagnosing diabetes or prediabetes were appropriate. Furthermore, the performance of HbA1c for diagnosing diabetes and prediabetes was poor. HbA1c should be used in combination with traditional glucose criteria when detecting and diagnosing diabetes or prediabetes.