Erythrocytosis and Performance of HbA1c in Detecting Diabetes on an Oxygen-Deficient Plateau: A Population Based Study.

CONTEXT: The hemoglobin A1c (HbA1c) test is a standard test for diabetes screening and diagnosis. OBJECTIVE: To evaluate A1c performance for diabetes screening in high-altitude polycythemia compared to a population with a high proportion of people living in an oxygen-deficient environment. DESIGN: A population-based epidemiological survey was conducted. SETTING: The cities Lhasa and Shigatse were selected. Volunteers were recruited through educational advertisements about diabetes. PARTICIPANTS: A total of 1401 Tibetan adults without known diabetes. INTERVENTIONS: Oral glucose tolerance test (OGTT), HbA1c, and complete blood cell count were performed. Hemoglobin A1c was evaluated using high-performance liquid chromatography, and serum glucose level, using the hexokinase method. MAIN OUTCOME MEASURES: World Health Organization criteria were used to define diabetes and prediabetes. Hemoglobin A1c test performance was evaluated using receiver operating characteristic analysis. RESULTS: The participants' mean age was 44.3 +/- 15.0 years; 33.3% of the participants were men and 38.6% lived in urban areas. The prediabetes and diabetes prevalence rates were 7.5% and 3.6%, respectively. The optimal HbA1c cutoff for detecting diabetes was 46 mmol/mol (6.4%), with a sensitivity and specificity of 60.8% and 93.6%, respectively. The cutoff for detecting diabetes was 6.7% (50 mmol/mol) in subjects with high-altitude polycythemia (HAPC). The relationship between red blood cell (RBC) counts and HbA1c was significant (P < 0.001), while there was no correlation between hemoglobin (Hb) and HbA1c (P = 0.085). Multiple linear regression analysis showed that after adjusting for age and fasting serum glucose or 2-hour OGTT (OGTT2h) serum glucose, RBC count and not Hb level was an independent risk factor for HbA1c (beta = 0.140, P < 0.001). CONCLUSIONS: The optimal HbA1c cutoff for detecting diabetes was 46 mmol/mol (6.4%) in Tibet. Red blood cell count was an independent risk factor for elevated HbA1c, and HAPC may affect the predictive ability of HbA1c.