Correlation between Glycosyiated Hemoglobin in Specimens Containing Potassium EDTA

We recently started estimating glycosylated hemoglobin (HbA1) by using an agarose electrophoresis system (Corning Medical and Scientific, Palo Alto, CA 94306). It is recommended in the manufacturer’s literature that potassium EDTA-anticoagulated samples should be used. We routinely estimate glucose in fluoride/oxalate anticoagulated samples, so we decided to compare results for HbA1 samples anticoagulated in the two ways. Blood specimens from 80 patients who were attending the diabetic outpatient clinic were collected, part of each in a potassium EDTA-containing tube and another part in a fluoride/oxalatecontaining tube. The samples were then analyzed as detailed by Corning. Each pair of samples was electrophoresed on the same gel to avoid between-batch variation. Comparison of the EDTA-containing samples (x) with the fluoride/oxalate-containing samples (y) gave the following linear regression equation y = 0.97x + 0.59; r = 0.98. A pair difference t-test gave the following result: mean difference 0.26, S.E. ofdifference 0.07, t = 3.670 (p < .001). The difference is statistically significant, but too small to affect clinical interpretation of the results. Consequently, we have started estimating both glucose and HbA1 in fluoride! oxalate-treated samples. This has the obvious advantage to the patient of requiring the withdrawal ofonly one 5mL sample.