Can a Single Blood Sample Sufficiently Diagnose Diabetes?

Multiple testings for diabetes still recommended for certain groups.

Current guidelines preferentially recommend diabetes be diagnosed with two tests on two separate blood samples, but also recommend that two biochemical tests on a single blood sample may be used.

In general, a patient is diagnosed with diabetes mellitus by a glucose measurement and HbA1c. When a patient is suspected of having hyperglycemia, repeated glucose and HbA1c measurements are expected in an effort to minimize false-positives. According to the American Diabetes Association, diabetes can be diagnosed by two methods. The first being assessment of fasting blood glucose, 2-hour plasma glucose post oral glucose tolerance test, or with HbA1c. Once those results are obtained, confirmatory repeat testing is recommended with a new blood sample to ensure an accurate diagnosis. The second option relies on diagnosing by two separate tests (i.e. HbA1c and plasma glucose) on a single blood sample but only if the results are well above diagnostic thresholds.

Selvin, et al sought to determine if two different tests performed on one blood sample could sufficiently confirm a diabetes diagnosis. Participants for this current study were obtained from the Atherosclerosis Risk in Communities (ARIC), a 25-year-long prospective study performed in four communities across the United States. Patients from the ARIC study were included if they had both glucose and HbA1c measurements. Exclusions occurred if variables or disease data were missing. Important definitions specific to this study can be found in Table 1: Diabetes Definitions.

Table 1: Diabetes Definitions
Prevalent diagnosed diabetes	Self-reported diagnosis by a physician OR use of antidiabetic medication
Confirmed undiagnosed diabetes	Elevated HbA1c ≥ 6.5% AND fasting blood glucose ≥ 126 mg/dL without a diabetes diagnosis
Unconfirmed undiagnosed diabetes	No diagnosis of diabetes AND either elevated HbA1c OR fasting blood glucose

Analysis revealed those with confirmed undiagnosed diabetes (CUD) were at higher risk for diabetes and cardiovascular disease compared to those with unconfirmed undiagnosed diabetes (UUD). Additionally, those with CUD were quite similar to individuals belonging to the diagnosed diabetes group. Individuals belonging to the CUD group were at higher risk for future diagnosis of diabetes (HR 25.00, 95% CI 22.10-28.28). Similarly, those in the UUD group were also at risk for a possible diabetes diagnosis (HR 5.75, 95% CI 5.16-6.40), however a weaker association was found. As would be expected, having CUD was strongly associated with peripheral artery disease (p < 0.001), cardiovascular disease (p = 0.016), and a diabetes diagnosis (p < 0.001). The UUD group showed significance in most of the outcomes as well.

In assessing the prognostic performance of the various definitions (refer to Table 1), the results showed CUD definition had a high specificity to diagnose diabetes at five years (98.1%) and at 15 years (99.6%), but showed a low sensitivity. Looking at the positive predictive value in the CUD group showed better results compared to the combination group (i.e. both CUD plus UUD). At 5, 10, and 15 years the positive predictive value increased (39.7%, 83.7%, and 88.7%, respectively). Negative predictive values were high for either group. The CUD group showed higher incidence of diabetes diagnosis (5 year: 42.0% and 15 years: 97.3%) compared to the UUD group (5 year: 9.9% and 15 years: 71.7%).

In the end, Selvin, et al were able to show that the confirmatory definition has both high specificity and positive predictive value, therefore suggesting that those who fit the criteria are at an increased risk of diabetes diagnosis. While the positive predictive value for the CUD group was lower in the beginning (i.e. potential for increased false-positives), the results by year 15 suggest that those individuals later received a diabetes diagnosis. Lastly, the low sensitivity noted with the CUD group indicates that those fitting the unconfirmed definition, showing only a single elevated test (i.e. HbA1c or glucose), should perform confirmatory repeat testing. Ultimately, the investigators suggest the evidence supports a diabetes diagnosis from a single blood sample.

Practice Pearls:

• Diabetes can be diagnosed by two methods, including: 1) Assessing fasting blood glucose, 2-hour plasma glucose following an oral glucose tolerance test, or with HbA1c followed by confirmatory repeat testing with a new blood sample, or 2) HbA1c or glucose results well above diagnostic thresholds in a single blood sample.
• Selvin, et al showed that those fitting the confirmed and unconfirmed undiagnosed diabetes definition are at an increased risk of future diabetes diagnosis, however those in the confirmed group showed stronger associations.
• The confirmed undiagnosed definition showed high specificity and high positive predictive value suggesting a single blood sample could be used to diagnose diabetes.
• The low sensitivity of the confirmed undiagnosed definition suggests that those with one elevated measurement, either HbA1c or glucose, should undergo repeat testing to confirm the diagnosis.

References:

Citation: Selvin E, Wang D, Matsushita K, Grams ME, Coresh J. Prognostic Implications of Single-Sample Confirmatory Testing for Undiagnosed Diabetes: A Prospective Cohort Study. Ann Intern Med. 19 June 2018. doi: 10.7326/M18-0091.

American Diabetes Association. 2. Classification and Diagnosis of Diabetes: Standards of Medical Care in Diabetes – 2018. Diabetes Care. 41. Supplement 1 (2018): S13-S27.

Kaytie A. Weierstahl, Pharm.D. Candidate, LECOM School of Pharmacy