Abstract P059: Analytic Variability of the Glucose Test: A Comparison among Different Specimens and Anticoagulants
Background: There is a marked variability (both biologic and analytic) involved in glucose testing. Even though biological variability is substantially greater than the analytic, a variability in the measurement of glucose is introduced by the manner in which the blood sample is collected and handled before its arrival to the laboratory. While plasma is recommended as the preferred specimen, some clinicians might be using results obtained on serum specimens. This study aimed to assess systematic difference in the glucose test results using different specimens and anticoagulants.
Methods: Blood samples from 84 men and 101 women 25-64 years old involved in the Italian Health Examination Survey/Osservatorio Epidemiologico Cardiovascolare (HES/OEC) were used. Fasting glucose test was assessed using serum and plasma by the addition of fluoride or EDTA to the blood collection tube. Biochemical tests were assayed in a unique laboratory. Comparison among mean values were assessed using t-test for paired sample.
Results: For men, mean level of fasting glucose was 89 mg/dl (standard deviation: 11 mg/dl) using serum specimens, 87 mg/dl (12 mg/dl) using plasma with EDTA and 93 mg/dl (11 mg/dl) using plasma with fluoride (p<0.0001 plasma fluoride vs serum and plasma EDTA). For women 84 mg/dl (11 mg/dl) using serum specimens, 84 mg/dl (13 mg/dl) using plasma with EDTA and 88 mg/dl (12 mg/dl) using plasma with fluoride (p<0.0001). Difference between plasma EDTA and plasma fluoride does not increase by serum glucose tertiles. Higher Persons correlation was found between serum glucose and plasma fluoride glucose both in men (r=0.9217) and women (r=0.9088). Prevalence of diabetes (fasting glucose >=126 mg/dl or under glucose treatment) was 6.0% in men and 5.9% in women using the three methods.
Conclusions: On the basis of this data, fasting serum and plasma EDTA glucose resulted about 5% lower than plasma fluoride glucose. Prevalence of diabetes in this sample does not seem to be affected by the use of different specimens and anticoagulants, however this is based on only one glucose assessment.
- © 2013 by American Heart Association, Inc.