Leonard Reeves, MD, FAAFP
In clinical and community populations, what is the accuracy of hemoglobin A1c (HbA1c) levels for screening and diagnosing type 2 diabetes?
Blinded comparison of HbA1c levels with oral glucose tolerance test (OGTT) results.
2494 patients referred by their medical practitioners for OGTTs in 2003 to 2008 (Melbourne Pathology, clinical group) and 6014 participants from the 2004 to 2005 follow-up of an Australian national population-based study (AusDiab, community group) who had concurrent HbA1c and OGTT measurements.
HbA1c levels determined using Diabetes Control and Complications Trial (DCCT)–aligned caption-exchange chromatography or boronate-affinity chromatography converted to DCCT-aligned HbA1c values.
Plasma glucose measured using hexokinase and classified according to American Diabetes Association criteria for OGTT.
HbA1c levels that rule in and rule out diabetes.
35% of the clinical group and 4.6% of the community group had undiagnosed diabetes. HbA1c levels ≤ 5.5% indicated a very low probability of diabetes; levels ≥ 7.0% indicated an extremely high probability, and levels ≥6.5% were almost as discriminating (Table). 65% of the clinical group and 24% of the community group had intermediate HbA1c levels (5.6% to 6.9%) with progressively increasing probabilities of diabetes.
In clinical and community populations, HbA1c levels ≤ 5.5% ruled out most cases of diabetes and levels ≥ 7.0% ruled in the disease.
HbA1c levels for screening and diagnosing diabetes in clinical (Melbourne Pathology group) and community (AusDiab group) populations*
*Diagnostic terms defined in Glossary.
Leonard Reeves. HbA1c levels ≤ 5.5% ruled out, and levels ≥ 7.0% ruled in, type 2 diabetes in clinical and community populations. Ann Intern Med. 2010;153:JC2–11. doi: 10.7326/0003-4819-153-4-201008170-02011
Download citation file:
Published: Ann Intern Med. 2010;153(4):JC2-11.
Cardiology, Coronary Risk Factors, Diabetes, Endocrine and Metabolism.
Results provided by:
Copyright © 2017 American College of Physicians. All Rights Reserved.
Print ISSN: 0003-4819 | Online ISSN: 1539-3704
Conditions of Use