• 
• 

Three Part Question

In [patients with signs and symptoms of diabetes] is [an oral glucose tolerance test or elevated HbA1c] a [better diagnostic test for diabetes mellitus]?

Clinical Scenario

A 5 year old girl presents to the Emergency Department with a one week history of polydipsia, polyphagia, polyuria, and nocturia. She has also had one day of non-localized abdominal pain and low grade fever (T 99ºF). Labs at the urgent care center one day prior show a non-fasting blood glucose of 284; labs at the PCP's office the day of the ED visit include a non-fasting CBG of 190 and HbA1c of 7.3. You wonder if the elevated glycosylated hemoglobin is acceptable for diagnosing diabetes or if she needs further testing with an oral glucose load.

Search Strategy

Medline 1966-07/06 using the OVID interface.
[OGTT or oral glucose tolerance test] AND [HbA1c or hemoglobin A1c or glycosylated hemoglobin or glycated hemoglobin] AND [new onset diabetes or diagnosing diabetes or diagnostic of diabetes or diagnosis of diabetes].

Search Outcome

25 papers found - 10 relevant, 5 unavailable, remaining 5 shown in the table below.

Relevant Paper(s)

Author, date and country	Patient group	Study type (level of evidence)	Outcomes	Key results	Study Weaknesses
Dods RF and Bolmey C, 1979, USA	23 patients with DM, 23-82 yo 35 healthy men and women, 23-80 yo	Diagnostic test study	1. Comparison of 6 scoring methods for OGTT with HbA1c assay results	37-64% of subjects diagnosed with DM by OGTT judged normal by HbA1c	Old study Small sample size HbA1c 9.5 cutoff too high per current standards Too many different methods of evaluating OGTT
				HbA1c correlates with serum glucose in diabetic patients (r=0.57, p<0.001)
			2. Comparison of HbA1c to HDL levels	HbA1c agrees with OGTT only when OGTT evaluated by conservative standards
				Significant inverse relation between HbA1c and HDL (r=-0.42, p<0.001)
Svendsen PA et al, 1981, Sweden	178 people referred for diagnostic OGTT, 10-80+ yo	Diagnostic test study	PG prior to and at 30, 60, 90, 120, 150 min after 50-g oral glucose load	In diabetic OGTT group, 86-94% had elevated HbA1c	No control group Old study
				HbA1c had 0.96 sensitivity and 0.78 specificity for DM
			HbA1c level (4.1-6.4 considered normal range)	Only subjects with substantially reduced glucose tolerance diagnosed by HbA1c
Little RR et al, 1988, USA	381 Pima Indians, 14-89 yo	Diagnostic test study	PG 2 hours after 75-g oral carbohydrate load	Significant correlation between HbA1c and FPG (r=0.91, p<0.0001) and HbA1c and 2-h PG (r=0.88, p<0.0001)	Very specific sample group with high prevalence of NIDDM Normal range HbA1c based on 95% CI for nondiabetic, mostly Caucasian population Old study
				HbA1c shows 85% sensitivity and 91% specificity for identifying DM
			HbA1c level (4.07-6.03 considered normal range)	Normal HbA1c does not exclude DM diagnosis by OGTT
Peters AL et al, 1996, USA	11,276 patients from 18 studies from 1966-1994	Meta-analysis		HbA1c 7.0 is 99.6% sensitive for diabetic cases, 99.9% specific for normal cases	Possible transformed data calculation errors No first-hand experimental evidence Different methods to measure HbA1c
			Evaluation of FPG, OGTT, and HbA1c to define their relationship	89% of patients with HbA1c>7.0 had DM by OGTT
				42% of patients with DM diagnosed by OGTT had HbA1c>7.0
Barr RG et al, 2002, USA	None	Review		Random PG>200 mg/dL and HbA1c>2 SDs above lab mean confirm diagnosis	Review No first-hand experimental evidence for conclusions No p-values or CI given
			Analysis of FPG vs. HbA1c vs. OGTT vs. combo of the above with respect to the 1997 ADA diagnostic criteria for diabetes mellitus	At 2 SDs above mean, HbA1c as a diagnostic test has 66% sensitivity and 98% specificity
				HbA1c reliable and correlates well with FPG and 2-h PG
				OGTT reproducibility only 65.6%

Comment(s)

Although some of these studies indicate that HbA1c has high diagnostic sensitivity and specificity for identifying diabetes mellitus, there is no definitive evidence to indicate it is a better diagnostic test than the OGTT. The sample sizes in the diagnostic test studies were small putting into question their actual power. Moreover, only one of these papers was published after 1997, the year the American Diabetes Association came out with new guidelines for diagnosing diabetes; this paper was also only a review and conclusions were not based on actual experimental evidence. Newer experimental studies are required before deciding whether HbA1c could be used as a diagnostic test for diabetes. The diagnostic test studies were also performed in subjects from wide age ranges. Presenting signs, symptoms, physical findings, and lab results may differ among children, adults, and elderly, so it is necessary to narrow the age range of subjects involved in experiments to come up with significant conclusions for specific age groups. This could additionally help determine the efficacy of using OGTT or HbA1c as diagnostic tests separately in Type 1 and Type 2 diabetes, instead of combining the two under the heading of diabetes mellitus, as these studies did, since these two diseases are different. Several of these papers mention that HbA1c is a better measure of actual plasma glucose levels because it is a physiologic test compared to the OGTT which is based on a non-physiologic stimulus. Glycosylated hemoglobin levels are easier to obtain, and they correlate with the development and progression of complications from diabetes. On the other hand, the OGTT is time-consuming and inconvenient for subjects, does not take into account usual dietary and activity habits of subjects, and is expensive. Taking all of this into account, it does appear there is a need and drive to find a better test, such as measuring the HbA1c, to diagnose diabetes.

Clinical Bottom Line

There is no clear evidence showing HbA1c is better than OGTT as a screening test for diagnosing diabetes mellitus.

References

1. Dods RF and Bolmey C. Glycosylated Hemoglobin Assay and Oral Glucose Tolerance Test Compared for Detection of Diabetes Mellitus. Clinical Chemistry. 1979;25(5):764-68.
2. Svendsen PA et al. HbA1c and the diagnosis of diabetes mellitus. Acta Medica Scandinavica. 1981;210(4):313-16.
3. Little RR et al. Relationship of glycosylated hemoglobin to oral glucose tolerance. Implications for diabetes screening. Diabetes. 1988;37(1):60-4.
4. Peters AL et al. A Clinical Approach for the Diagnosis of Diabetes Mellitus: An Analysis Using Glycosylated Hemoglobin Levels. JAMA. 1996;276(15):1246-52.
5. Barr RG et al. Tests of Glycemia for the Diagnosis of Type 2 Diabetes Mellitus. Ann Intern Med. 2002;137:263-72.