Independent effect of ethnicity on glycemia in South Asians and white Europeans.

OBJECTIVE: HbA(1c) levels are higher in most ethnic groups compared with white Europeans (WEs) independent of glycemic control. This comparison has not been performed between South Asians (SAs) and WEs. We analyzed the independent effect of ethnicity on HbA(1c) and fasting and 2-h plasma glucose (FPG and 2 hrPG, respectively) between these groups. RESEARCH DESIGN AND METHODS: Analysis of the ADDITION-Leicester study, in which 4,688 WEs and 1,352 SAs underwent oral glucose tolerance testing, HbA(1c), and other risk factor measurements.
RESULTS: Significant associations with HbA(1c) included ethnicity, FPG, 2 hrPG, and homeostasis model assessment of β-cell function (P < 0.001); age and sex (P < 0.01); and fasting insulin and potassium (P < 0.05). After adjusting for these and other risk factors, SAs demonstrated higher HbA(1c) (6.22 and 6.02%, mean difference 0.20%, 0.10-0.30, P < 0.001), FPG (5.15 and 5.30 mmol/L, mean difference 0.15 mmol/L, 0.09-0.21, P < 0.001), and 2 hrPG (5.82 and 6.57 mmol/L, mean difference 0.75 mmol/L, 0.59-0.92, P < 0.001) compared with WEs, respectively.
CONCLUSIONS: HbA(1c), FPG, and 2 hrPG levels were higher in SAs independent of factors affecting glycemic control.

Glycated hemoglobin (HbA1c) is now recommended as a diagnostic tool for detecting type 2 diabetes, alongside fasting and 2-h plasma glucose (FPG and 2hrPG, respectively), and remains the standard test for monitoring disease progression (1). Previous studies demonstrate HbA1c values are higher in some black and minority ethnic groups compared with white Caucasians independent of glycemic control or factors that differ between ethnic groups (2–5). These studies suggest HbA1c levels are higher in African Americans by 0.2–0.4%, in Hispanics by 0.1–0.3%, and in Southeast Asians by 0.2–0.3% (2–5). Because this analysis has not been performed in South Asians (people of Indian, Pakistani, and Bangladeshi origin), our aim was to evaluate the independent effect of ethnicity on glycemia among South Asians and white Europeans and to quantify the magnitude of any differences.

RESEARCH DESIGN AND METHODS

The analysis was performed using cross-sectional data from the Anglo-Danish-Dutch Study of Intensive Treatment in People with Screen Detected Diabetes in Primary Care (ADDITION)-Leicester population-based diabetes screening study. An in-depth description of study methods has been published previously (6). In brief, primary care participants aged 40 to 75 years underwent an oral glucose tolerance test (OGTT), classified using World Health Organization 1999 criteria (7), and other measurements, including HbA1c, from 2005 to 2009. HbA1c samples were measured on a Bio-Rad VARIANT II high-performance liquid chromatography instrument (Hemel Hempstead, U.K.), which is standardized to current recommendations for diagnosis of diabetes and has a coefficient of variation <2% (1). This analyzer detected hemoglobinopathies (S and C) and such results were excluded.

Statistical analysis was performed using SPSS version 18.0 (Chicago, IL). Multiple regression analysis was used to determine all significant associations of HbA1c. Insulin resistance and β-cell function were calculated using homeostasis model assessment equations (8). Ethnicity was classified using U.K. national census categories (9). ANCOVA modeling was used to calculate the mean difference of HbA1c between South Asians and white Europeans using stepwise models. Model 1 compared unadjusted HbA1c values. Model 2 adjusted HbA1c levels for age, sex, BMI, waist circumference, systolic and diastolic blood pressure, LDL and HDL cholesterol, triglycerides, creatinine, albumin-to-creatinine ratio, FPG, and 2hrPG. Model 3 included fasting insulin as well. Model 4 was similar to model 2 but excluded FPG and 2hrPG. Adjustments for multiple comparisons were made using Bonferroni corrections. P < 0.05 was considered significant.

RESULTS

There were 6,040 people (4,688 white Europeans and 1,352 South Asians) included in the analysis. The significant associations of HbA1c were ethnicity, FPG, 2hrPG, and homeostasis model assessment of β-cell function (P < 0.001); age and sex (P < 0.01); and insulin and potassium (P < 0.05), producing an adjusted R2 of 0.639.

The mean (SE) crude HbA1c in white Europeans and South Asians was 5.65 (0.01) and 5.81% (0.01), respectively, producing a mean difference of 0.22% (95% CI 0.18–0.25; P < 0.001) (Table 1). After adjustment for risk factors, HbA1c remained higher in South Asians, with a mean difference of 0.19% (0.11–0.27; P < 0.001). Stratification by OGTT result demonstrated similar findings. When FPG was the dependent variable, mean crude values were 5.18 (0.01) and 5.27 mmol/L (0.03) in white Europeans and South Asians, respectively, a mean difference of 0.09 mmol/L (0.03–0.14; P < 0.01). After adjustment, these values were 5.15 (0.01) and 5.30 mmol/L (0.03), a mean difference of 0.15 mmol/L (0.09–0.21; P < 0.001) higher in South Asians. Using 2hrPG as the dependent variable, the mean crude values were 5.89 (0.08) and 6.46 mmol/L (0.07) in white Europeans and South Asians, respectively, producing a mean difference of 0.58 mmol/L (0.43–0.73; P < 0.001). After adjustment, these values were 5.82 (0.04) and 6.57 mmol/L (0.07), a mean difference higher in South Asians of 0.75 mmol/L (0.59–0.92; P < 0.001).

Table 1

A comparison of crude and adjusted differences for HbA1c in white Europeans and South Asians

CONCLUSIONS

In this multiethnic cohort of adults undergoing an OGTT, HbA1c values were 0.2% higher in South Asians than white Europeans, even in analysis stratified by glucose intolerance status. The current study is the first to demonstrate this effect persisted after adjusting for factors that may affect glycemia or that differed between these ethnic groups. The strengths of this study include the large numbers of white Europeans and South Asians who underwent robust measurement of risk factors, allowing detection of any clinically significant differences. The diabetes risk factors included in the multiple regression analysis explained 63.9% of the variation in HbA1c, which is relatively higher than other studies (3). However, there may be other unmeasured factors that influence HbA1c. FPG and 2hrPG levels may not give a robust representation of 24-h glucose profile, a problem recognized in similar studies (3,4). Other examples include dietary intake, genetic influences, and iron deficiency anemia (10,11). Therefore, our finding that sex independently associates with HbA1c should be interpreted with caution. Studies that account for either hematocrit or hemoglobin provide contradictory reports of an independent effect of sex on HbA1c (3,4). Our results showing a higher HbA1c level of 0.2% in South Asians was consistent when separated by males and females (data not shown).

Ethnic variation in HbA1c levels could be attributed predominantly to biological variation in hemoglobin glycation and differential erythrocyte survival. However, African Americans, who also possess higher HbA1c levels than white Caucasians, have more adverse profiles of glycemic markers unaffected by hematological factors, suggesting this does not explain HbA1c differences (2).

Implications for policy makers and clinicians

First, international organizations have recommended using ethnic-specific cut points for South Asians in relation to BMI, waist circumference, and metabolic syndrome, which came as a response to high rates of diabetes within this group (12). However, there is no suggestion of ethnic-specific cut points for diagnosis of diabetes using HbA1c (1). The prevalence of diabetes using HbA1c ≥6.5% is higher in South Asians than white Europeans compared with using an OGTT, with a similar finding for detecting high-risk individuals (13,14). Second, it is reported that a greater proportion of South Asians with established diabetes do not achieve glycemic guideline targets in comparison with white Europeans (15). Because our study demonstrates independently higher HbA1c, FPG, and 2hrPG levels in South Asians, this result may be partially explained by factors related to glycemia. Future research should address the relationship between HbA1c and the onset of diabetes complications, including prevalent retinopathy, between South Asians and white Europeans in well-designed outcome studies to determine if ethnic-specific cut points are required for diabetes diagnosis in South Asians.