Ethnic differences in glycemic markers in patients with type 2 diabetes

Abstract

Objective: Recent studies have reported hemoglobin A1c (HbA1c) differences across ethnic groups that could limit its use in clinical practice. The authors of the A1C-Derived Average Glucose study have advocated to report HbA1c in estimated average glucose (AG) equivalents. The aim of this study was to assess the relationships between HbA1c and the mean of three 7-point self-monitored blood glucose (BG) profiles, and to assess whether estimated AG is an accurate measure of glycemia in different ethnic groups.

Research design and methods: We evaluated 1,879 participants with type 2 diabetes in the DURABLE trial who were 30 to 80 years of age, from 11 countries, and, according to self-reported ethnic origin, were Caucasian, of African descent (black), Asian, or Hispanic. We performed logistic regression of the relationship between the mean self-monitored BG and HbA1c, and estimated AG, according to ethnic background.

Results: Baseline mean (SD) HbA1c was 9.0% (1.3) (75 [SD, 14] mmol/mol), and mean self-monitored BG was 12.1 mmol/L (3.1) (217 [SD, 55] mg/dL). In the clinically relevant HbA1c range of 7.0-9.0% (53-75 mmol/mol), non-Caucasian ethnic groups had 0.2-0.5% (2-6 mmol/mol) higher HbA1c compared with Caucasians for a given BG level. At the mean self-monitored BG levels≤11.6 mmol/L, estimated AG overestimated the actual average BG; at levels>11.6 mmol/L, estimated AG underestimated the actual BG levels.

Conclusions: For a given degree of glycemia, HbA1c levels vary among different ethnic groups. Ethnicity needs to be taken into account when using HbA1c to assess glycemic control or to set glycemic targets. Estimated AG is not a reliable marker for mean glycemia and therefore is of limited clinical value.

Trial registration: ClinicalTrials.gov NCT00279201.

Figures

Figure 1

A–D: The relationship between HbA1c and mean SMBG in the Caucasian, Asian, Hispanic, and African descent ethnic groups. The regression equations are as follows: (A) Caucasian HbA1c (%) = 0.27 × mean SMBG (mmol/L) + 5.66; (B) Asian HbA1c (%) = 0.25 × mean SMBG (mmol/L) + 6.05; (C) Hispanic HbA1c (%) = 0.28 × mean SMBG (mmol/L) + 5.92; and (D) African descent HbA1c (%) = 0.26 × mean SMBG (mmol/L) + 6.17, where mean SMBG is the mean of the SMBG profiles.

Figure 2

A: Relationship between the estimated AG and the mean SMBG values. The observed regression line (solid line) differs significantly from the expected regression line (dotted line). B: Relationship between the mean BG index (difference between estimated BG and measured BG) and the mean SMBG values. At mean SMBG levels of ≤11.6 mmol/L (210 mg/dL), estimated AG overestimates the actual average BG; at BG levels >11.6 mmol/L (210 mg/dL), estimated AG underestimates the true BG levels.