DIABETICMedicine
DOI: 10.1111/j.1464-5491.2010.03159.x
Short Report: Epidemiology
HbA1c measurement for the diagnosis of diabetes: is it
enough?
G. Cavagnolli*, J. Comerlato†, C. Comerlato†, P. B. Renz†, J. L. Gross*‡ and J. L. Camargo*†
*Postgraduate Program in Endocrinology, Federal University of Rio Grande do Sul (UFRGS), †Clinical Pathology Division and ‡Endocrinology Division, Hospital de
Clı́nicas de Porto Alegre, Porto Alegre, Brazil
Accepted 4 October 2010
Abstract
To analyse the performance of HbA1c in diagnosing Type 2 diabetes based on fasting plasma glucose and ⁄ or 2-h plasma
glucose measurements after a 75-g oral glucose tolerance test.
Aim
Methods This is a study of diagnostic test accuracy in individuals referred to the Clinical Pathology Department for oral
glucose tolerance testing. After fasting overnight, HbA1c, fasting plasma glucose and 2-h plasma glucose were measured. The
receiver operating characteristic curve was used to evaluate the diagnostic performance of HbA1c.
Results Four hundred and ninety-eight subjects (195 male, mean age 56 years) were enrolled and 115 (23.1%) were diagnosed
with diabetes according to glucose-based methods and only 56 (11.2%) individuals were identified by HbA1c ‡ 6.5%
(48 mmol ⁄ mol) (sensitivity 20.9%, specificity 95.3%). There is poor agreement between the newly recommended criterion and
the current glucose-based diagnostic criteria (j = 0.217; P < 0.001), probably because the diagnostic methods identify different
populations of patients. Adding a glucose-based method into an algorithm, as proposed by the UK Department of Health,
improved HbA1c performance.
HbA1c ‡ 6.5% (48 mmol ⁄ mol) showed limited sensitivity to diabetes diagnosis, although with high specificity.
The results suggest that this cut-off point would not be enough to diagnose diabetes. Its use as the sole diabetes diagnostic test
should be interpreted with caution to assure the correct classification of diabetic individuals.
Conclusions
Diabet. Med. 28, 31–35 (2011)
Keywords
diabetes, diagnosis, diagnostic tests, glycated hemoglobin
Introduction
Recently, an International Expert Committee, composed of
members from the European Association for the Study of
Diabetes, the International Diabetes Federation and the
American Diabetes Association reported values of HbA1c
‡ 6.5% (48 mmol ⁄ mol) as the cut-off point for diagnosing
diabetes [1]. The selection of this cut-off point was to emphasize
specificity rather than sensitivity; however, the diagnostic
performance of this specific cut-off point is not mentioned.
Other studies had suggested lower HbA1c values, such as 5.8%
(40 mmol ⁄ mol) and 6.0% (42 mmol ⁄ mol), for this purpose
[2–5]. The United Kingdom Department of Health recommends
the use of algorithms for diabetes screening in high-risk
individuals that include traditional glucose diagnostic criteria
Correspondence to: Joiza Lins Camargo, Serviço de Patologia Clı́nica, Hospital
de Clı́nicas de Porto Alegre, Rua Ramiro Barcellos, 2350; 2 andar, Porto
Alegre, RS, 90035-903, Brazil. E-mail: [email protected]
ª 2010 The Authors.
Diabetic Medicine ª 2010 Diabetes UK
or, alternatively, HbA1c measurements combined or not with
glucose measurements. In non-symptomatic patients, a
confirmed HbA1c ‡ 6.5% (48 mmol ⁄ mol) is enough to
diagnose Type 2 diabetes [6]. However, this guideline also
recommends that patients with HbA1c levels ‡ 6.0%
(42 mmol ⁄ mol) and < 6.5% (48 mmol ⁄ mol) should undergo
an oral glucose tolerance test to establish a diabetes diagnosis. No
further test is required for those patients presenting with HbA1c
levels < 6.0% (42 mmol ⁄ mol), but health lifestyle advice is
given.
Until recently, the lack of HbA1c standardization prevented the
use of HbA1c as part of a potential screening and diagnostic
programme. The introduction of a new reference method
improved HbA1c assay standardization worldwide and its role
in the screening and diagnosis of Type 2 diabetes has been
considered [7–9].
Taking into account that increased prevalence of
retinopathy in different populations is already observed
around HbA1c values of 6.0% (42 mmol ⁄ mol) [1], the risk
31
DIABETICMedicine
for cardiovascular disease is greater at the HbA1c level of
5.6–6.1% (38–43 mmol ⁄ mol) [10] and, bearing in mind that
these individuals would benefit from early intervention, the
aim of this report was to analyse the performance of
the recommended HbA1c ‡ 6.5% (48 mmol ⁄ mol) value for
the diagnosis of Type 2 diabetes [1,6] according to current
diagnostic criteria based on fasting plasma glucose and ⁄ or
2-h plasma glucose after a 75-g oral glucose tolerance test
[11,12].
Research design and methods
This is a study of diagnostic accuracy to evaluate the HbA1c test
for diagnosis of diabetes (see also Supporting Information,
Fig. S1). Conventional diagnostic tests of fasting plamsa glucose
‡ 7.0 mmol ⁄ l and ⁄ or 2-h plasma glucose ‡ 11.1 mmol ⁄ l were
considered to be the reference standard [11–13].
Patients
The study included individuals at high risk of developing or
having diabetes being referred to the Clinical Pathology
Department of the Hospital de Clinicas de Porto Alegre,
between September 2008 and May 2009, to perform an oral
glucose tolerance test. Porto Alegre is situated in Southern Brazil,
where the majority of the population has European ancestry and
approximately 86% of the population is of White ethnicity. All
conditions known to interfere or lead to misinterpretation of
HbA1c results were excluded [7,8]: patients with anaemia [14],
estimated glomerular filtration rate < 60 ml min)1 1.73 m)2
[15] and ⁄ or the presence of variant haemoglobin [16]. Patients
signed an informed consent form and answered a standardized
questionnaire. The protocol was approved by the Ethics
Committee.
Laboratory analysis
After overnight fast, blood samples were drawn to determine
HbA1c, lipid profile, creatinine and glucose levels,and an oral
glucose tolerance test was also performed according to World
Health Organization recommendations [12]. Plasma glucose
was determined by an enzymatic method and creatinine by Jaffé
reaction (Modular P; Roche Diagnostics, Basel, Switzerland).
HbA1c was determined by HPLC method (Tosoh 2.2
Plus HbA1c; Tosoh Corporation, Tokyo, Japan). This is a
National Glycohemoglobin Standardization Program (NGSP)certified method (http://www.ngsp.org/prog/index.html) and
also International Federation of Clinical Chemistry and
Laboratory Medicine (IFCC) aligned [8]. The Clinical
Pathology Department is a participant in and meets the
standards of the HbA1c External Quality Assurance Program.
Diagnosis of diabetes and impaired glucose tolerance were made
according to current diagnosis criteria [11,12]. Impaired fasting
glucose was classified as stated by American Diabetes
Association criteria [11].
32
HbA1c test and diabetes diagnosis • G. Cavagnolli et al.
Statistical analysis
Data are expressed as mean and sd when normally distributed
and as median (range) for non-Gaussian variables. Student’s
t-test, the Mann–Whitney U-test, McNemar test and kappa
coefficient were used as appropriate. The receiver operating
characteristic curve was used to analyse the performance of
the HbA1c test to diagnose diabetes, using fasting plamsa
glucose and ⁄ or 2-h plasma glucose as reference diagnostic
criteria. In addition, the UK Department of Health algorithm
based on HbA1c results was also used to verify if an additional
blood glucose measurement would improve the performance
of HbA1c in diagnosing diabetes. A significance level of 5%
was adopted.
Results
Four hundred and ninety-eight individuals were selected to
participate in this study; 418 (83.8%) of those were of White
ethnicity. The clinical and laboratory characteristics of all
individuals are summarized in Table 1. There was poor
agreement between the diagnosis of diabetes using HbA1c
‡ 6.5% (48 mmol ⁄ mol) and any of the glycaemic tests
(j = 0.217; P < 0.001). According to the traditional criteria,
115 (23.1%) individuals were diagnosed with diabetes: 26 by
fasting plasma glucose, 54 by 2-h plasma glucose and 35 using
both tests. In contrast, although the HbA1c cut-off value of 6.5%
(48 mmol ⁄ mol) has a specificity of 95.3%, this criterion
identified only 56 (11.2%) individuals with diabetes; of those,
29 (25.2%) had the diagnosis established by current criteria: 22
by fasting plasma glucose and 2-h plasma glucose and seven by
either fasting plasma glucose or 2-h plasma glucose. Therefore,
the proportion of individuals with and without diabetes
significantly differs between the glucose- and HbA1c-based
methods (P < 0.001).
There were 86 individuals with a diagnosis established by
current criteria and HbA1c < 6.5% (48 mmol ⁄ mol),
corresponding to 74.8% of all diabetes cases. The sensitivity
and specificity of the HbA1c cut-off value of ‡ 6.5%
(48 mmol ⁄ mol), compared with the diagnosis of diabetes
established by fasting plasma glucose and ⁄ or 2-h plasma
glucose were 20.9 and 95.3%, respectively, resulting in a
negative predictive value of 80.5%.
Based on receiver operating characteristic curve analysis, and
considering fasting plasma glucose alone as the reference
criterion, the cut-off point obtained by the point with the best
equilibrium between sensitivity and specificity (100%-to-100%
diagonal), a value of HbA1c of 6.0% (42 mmol ⁄ mol)
was obtained. The sensitivity and specificity were 74.2 and
72.0%, respectively. The cut-off points of HbA1c of 7.0%
(53 mmol ⁄ mol)and 8.0% (64 mmol ⁄ mol) presented specificities
of 99.5 and 99.8%.
Alternatively, when fasting plasma glucose and ⁄ or 2-h plasma
glucose and 2-h plasma glucose alone were considered as the
reference criteria, an HbA1c level of 5.9% (41 mmol ⁄ mol) was
ª 2010 The Authors.
Diabetic Medicine ª 2010 Diabetes UK
Original article
DIABETICMedicine
Table 1 Characteristics of individuals participating in the study according to HbA1c values
n
Sex (man ⁄ woman)
Age (years)
BMI
Waist circumference (cm)
Without diabetes
Pre-diabetes
Diabetes mellitus
Hypertension
Family history of diabetes
Family history of hypertension
Family history of cardiovascular disease
Fasting plasma glucose (mmol ⁄ l)
2-h plasma glucose (mmol ⁄ l)
HbA1c (%)
Creatinine ((mol ⁄ l)
HbA1c
< 6.0%
(< 42 mmol ⁄ mol)
6–6.4%
(42–46 mmol ⁄ mol)
‡ 6.5%
(‡ 48 mmol ⁄ mol)
331
139 ⁄ 192
53 13
27 5.6
94 17
113 (34.1%)
168 (50.7%)
50 (15.1%)
162 (48.9%)
162 (48.9%)
238 (71.9%)
207 (62.5%)
5.7 0.7
7.4 2.9
5.3 0.4
74.2 17.7
111
35 ⁄ 76
61 11
28 6.0
97 18
8 (7.2%)
67 (60.3%)
36 (32.4%)
79 (71.1%)
61 (54.9%)
75 (67.5%)
72 (64.8%)
6.2 0.8
8.8 4.0
6.1 0.14
81.3 17.7
56
21 ⁄ 35
61 12
30 6.3
103 13
1 (1.78%)
26 (46.4%)
29 (51.7%)
44 (78.5%)
30 (53.5%)
40 (71.4%)
40 (71.4%)
7.6 2.5
12.7 5.8
7.2 1.5
83.0 16.8
n = 498.
Data are mean sd or median (interval).
1
Sensitivity
0.8
0.6
0.4
0.2
0
0
0.2
0.4
0.6
1 Specificity
0.8
1
FIGURE 1 HbA1c receiver operating characteristic curves for diabetes
diagnosis: HbA1c vs. fasting plasma glucose ( ); HbA1c vs. 2-h plasma
glucose ( ); HbA1c vs. fasting plasma glucose and ⁄ or 2-h plasma glucose
( ).
the equilibrium point for sensitivity and specificity (63.5 ⁄ 66.1
and 62.9 ⁄ 64.1%, respectively) (Fig. 1).
In our study, 167 patients had HbA1c ‡ 6.0% (42 mmol ⁄ mol).
Sixty-five individuals were diagnosed with diabetes by the HbA1c
‡ 6.0% (42 mmol ⁄ mol) criterion and 50 were misclassified.
There were 102 false positive results. Only nine individuals of
those had normoglycaemia; however, 93 presented with
hyperglycaemia (41 with impaired fasting glucose and 52 with
impaired glucose tolerance). The sensitivity and specificity of the
HbA1c cut-off value of ‡ 6.0% (42 mmol ⁄ mol) were 51.3 and
78.3%, respectively. The agreement between this cut-off point
ª 2010 The Authors.
Diabetic Medicine ª 2010 Diabetes UK
and current diagnostic criteria was also poor (j = 0.274;
P < 0.001).
Applying the UK Department of Health algorithm to verify if
an additional blood glucose measurement would improve the
performance of HbA1c in diagnosing diabetes, 82 individuals
would be diagnosed with diabetes: in 56 individuals, HbA1c was
‡ 6.5% (48 mmol ⁄ mol) and, in 26, the HbA1c level was
between 6.0 and 6.5% (42–48 mmol ⁄ mol) and they had a
positive oral glucose tolerance test. Fifty individuals with HbA1c
< 6.0% (42 mmol ⁄ mol) and diabetes by the current criteria
remained undiagnosed. There were 168 individuals presenting
with HbA1c < 6.0% (42 mmol ⁄ mol) and an intermediate
hyperglycaemia (50.7%). The sensitivity and specificity for
this strategy were 47.8 and 92.9%, respectively. The agreement
between this approach and current diagnostic criteria improved
(j = 0.416; P < 0.001).
Discussion
This study shows that there is poor agreement between the
newly recommended criterion and the current glucose-based
diagnostic criteria and, consequently, the proportion of patients
diagnosed with diabetes by one or other method is significantly
different.
Our results are in agreement with recent reports carried out
in larger populations, which pointed out the poor diagnostic
performance of HbA1c [17–19] and that it is not superior to
fasting plasma glucose and ⁄ or 2-h plasma glucose [20]. The
cut-off point of HbA1c ‡ 6.5% (48 mmol ⁄ mol) had a
sensitivity of only 25% to diagnose diabetes. However, by
adopting the cut-off of HbA1c ‡ 6.0% (42 mmol ⁄ mol), the
sensitivity increased to 51.3%. Even so, approximately 50%
33
DIABETICMedicine
of patients with diabetes will not be identified. In addition, we
observed an increase in false positive results using the cut-off
point of HbA1c ‡ 6.0% (42 mmol ⁄ mol); 93% of the
individuals misclassified presented some degree of glucose
abnormality.
These findings suggest that the HbA1c method and the glucosebased methods identify different patient populations.
Several studies had suggested lower values than the present
recommended values of HbA1c ‡ 6.5% (48 mmol ⁄ mol) to
diagnose diabetes [3–5]. Besides, the cut-off value of HbA1c
> 6.1% (43 mmol ⁄ mol) presented high sensitivity and
specificity to screen for diabetes in high-risk individuals,
and the combination of HbA1c and fasting plasma glucose
improved the sensitivity for both tests alone [4]. Moreover,
there is a continuous relationship between fasting plasma
glucose and HbA1c and cardiovascular risk. It has been
suggested that individuals with an HbA1c level of 5.6–6.1%
(38–43 mmol ⁄ mol) and a fasting plamsa glucose level of
5.6–6.3 mmol ⁄ l are at greatest risk for cardiovascular disease
and should be targeted for further evaluation [10]. In a
systematic review, the value of HbA1c > 6.1% (43 mmol ⁄
mol) was the recommended optimum cut-off point for the
diagnose of diabetes in most reviewed studies, although it is
also stressed that cut-off points may be population-specific
and may vary by race, age, gender and population prevalence
of diabetes [5].
Alternatively, when the UK Department of Health algorithm
[6] was applied and a glucose measurement was added to the
HbA1c results between 6.0% (42 mmol ⁄ mol) and 6.5%
(48 mmol ⁄ mol) to classify diabetic patients, the agreement
between HbA1c and current diagnostic criteria improved.
According to this algorithm, our data showed that individuals
with HbA1c between 6.0% (42 mmol ⁄ mol) and 6.5%
(48 mmol ⁄ mol) are likely to have an altered oral glucose
tolerance test. A total of 111 individuals presented with HbA1c
levels ‡ 6.0% (42 mmol ⁄ mol) and < 6.5% (48 mmol ⁄ mol) in
our study, only nine had normoglycaemia and 102 presented
some type of hyperglycaemia. The stigma and costs of a false
positive diagnosis would be compensated for by identifying
people with intermediate hyperglycaemia, considering the
important clinical consequences of delaying the diagnosis
[21,22]. However, 50 patients with HbA1c < 6.0%
(42 mmol ⁄ mol) remained undiagnosed for diabetes by using
this algorithm and would benefit from an additional test to
establish their diagnosis. Additionally, there were 168
individuals presenting with HbA1c < 6.0% (42 mmol ⁄ mol)
and an intermediate hyperglycaemia (50.7%).
Therefore, lowering the HbA1c cut-off level and adding a
glucose-based method improved the HbA1c performance in
diagnosing diabetes, suggesting that each method identifies
different patient populations.
In conclusion, HbA1c ‡ 6.5% (48 mmol ⁄ mol) showed high
specificity but limited sensitivity to a diabetes diagnosis. The
results suggest that the cut-off point of ‡ 6.5% would not be
enough to diagnose diabetes. Its use as the sole diabetes
34
HbA1c test and diabetes diagnosis • G. Cavagnolli et al.
diagnostic test should be interpreted with caution to assure the
correct classification of diabetic individuals.
Competing interests
Nothing to declare.
Acknowledgements
We thank the staff of the Pre-Analytical Unit and the Clinical
Biochemistry Unit at the Clinical Pathology Department for
OGTT performance and HbA1c analysis, respectively. This study
was supported by Fundo de Incentivo a Pesquisa (FIPE) to
Hospital de Clı́nicas de Porto Alegre (HCPA). GC and JC
were recipients of scholarships from Coordenação de
Aperfeiçoamento de Pessoal de Ensino Superior (CAPES) and
Fundação de Amparo à Pesquisa do Rio Grande do Sul
(FAPERGS), respectively.
References
1 International Expert Committee. The International Expert Committee report on the role of the HbA1c assay in the diagnosis of
diabetes. Diabetes Care 2009; 32: 1327–1334.
2 Buell C, Kermah D, Davidson MB. Utility of HbA1c for diabetes
screening in the 1999 2004 NHANES population. Diabetes Care
2007; 30: 2233–2235.
3 Tanaka Y, Atsumi Y, Matsuoka K, Mokubo A, Asahina T,
Hosokawa K et al. Usefulness of stable HbA1c for supportive
marker to diagnose diabetes mellitus in Japanese subjects. Diabetes
Res Clin Pract 2001; 53: 41–45.
4 Kim KS, Kim SK, Lee YK, Park SW, Cho YW. Diagnostic value of
glycated haemoglobin (HbA1c) for the early detection of diabetes in
high-risk subjects. Diabet Med 2008; 25: 997–1000.
5 Bennett CM, Guo M, Dharmage SC. HbA1c as a screening tool for
detection of Type 2 diabetes: a systematic review. Diabet Med
2007; 24: 333–343.
6 NHS Health Check. Vascular Risk Assessment and Management
Best Practice Guidance. 3 April 2009; 9: 19–24. Available at http://
www.dh.gov.uk/prod_consum_dh/groups/dh_digitalassets/documents/
ditalasset/dh_098410.pdf Last accessed 30 August 2010.
7 Sacks DB, Bruns DE, Goldstein DE, Maclaren NK, McDonald JM,
Parrott M. Guidelines and recommendations for laboratory analysis
in the diagnosis and management of diabetes mellitus. Clin Chem
2002; 48: 436–472.
8 American Diabetes Association, European Association for the Study
of Diabetes, International Federation of Clinical Chemistry and
Laboratory Medicine and the International Diabetes Federation
Consensus Committee. Consensus statement on the worldwide
standardization of the hemoglobin A1c measurement. Diabetes Care
2007; 30: 2399–2400.
9 Saudek CD, Herman WH, Sacks DB, Bergenstal RM, Edelman D,
Davidson MB. A new look at screening and diagnosing diabetes
mellitus. J Clin Endocrinol Metab 2008; 93: 2447–2453.
10 Jesudason DR, Dunstan K, Leong D, Wittert GA. Macrovascular
risk and diagnostic criteria for type 2 diabetes. Implications for the
use of FPG and HbA1c for cost-effective screening. Diabetes Care
2003; 26: 485–490.
11 The Expert Committee on the Diagnosis and Classification of
Diabetes Mellitus. Follow-up report on the diagnosis of diabetes
mellitus. Diabetes Care 2003; 26: 3160–3167.
ª 2010 The Authors.
Diabetic Medicine ª 2010 Diabetes UK
Original article
12 WHO. Definition and Diagnosis of Diabetes Mellitus and Intermediate Hyperglycemia: Report of a WHO ⁄ IDF Consultation.
Geneva: World Health Organization, 2006.
13 Bossuyt PM, Reitsma JB, Bruns DE, Gatsonis CA, Glasziou PP,
Irwig LM et al. The STARD statement for reporting studies of
diagnostic accuracy: explanation and elaboration. Ann Intern Med
2003; 138: W1–12.
14 Astor BC, Muntner P, Levin A. Association of kidney function with
anemia: the Third National Health and Nutrition Examination
Survey (1988–1994). Arch Intern Med 2002; 162: 1401–1408.
15 Rule AD, Larson TS, Bergstralh EJ, Slevak JM, Jacobsen SJ. Using
serum creatinine to estimate glomerular filtration rate: accuracy in
good health and chronic disease. Ann Intern Med 2004; 141:
929–937.
16 Camargo JL, Gross JL. Conditions associated with very low values
of glycohaemoglobin measured by an HPLC method. J Clin Path
2004; 57: 344–345.
17 van’t Riet E, Alssema M, Rijkelijkhuizen JM, Kostense PJ, Nijpels
G, Dekker JM. Relationship between A1c and glucose levels in the
general Dutch population: the new Hoorn study. Diabetes Care
2010; 33: 61–66.
18 Carson AP, Reynolds K, Fonseca VA, Muntner P. Comparison of
A1c and fasting glucose criteria to diagnose diabetes among US
adults. Diabetes Care 2010; 33: 95–97.
19 Kramer CK, Araneta MR, Barrett-Connor E. A1c and diabetes
diagnosis: The Rancho Bernardo Study. Diabetes Care 2010; 33:
101–103.
ª 2010 The Authors.
Diabetic Medicine ª 2010 Diabetes UK
DIABETICMedicine
20 Kilpatrick E, Bloomgarden Z, Zimmet P. Is haemoglobin A1c a step
forward for diagnosing diabetes? Br Med J 2009; 339: 1288–1290.
21 Bertram MY, Lim SS, Barendregt JJ, Voss T. Assessing the costeffectiveness of drug and lifestyle intervention following opportunistic screening for pre-diabetes in primary care. Diabetologia
2010; 53: 875–881.
22 Selvin E, Steffes MW, Zhu H, Matsushita K, Wagenknecht L,
Pankow J et al. Glycated hemoglobin, diabetes, and cardiovascular
risk in non-diabetic adults. N Engl J Med 2010; 362: 800–811.
Supporting Information
Additional Supporting Information may be found in the online
version of this article:
Figure S1. Study flow diagram.
Please note: Wiley-Blackwell are not responsible for the content
or functionality of any supporting materials supplied by the
authors. Any queries (other than for missing material) should be
directed to the corresponding author for the article.
35