Precision of a dry-chemistry method of lipid screening

The purpose of this study was to compare total capillary cholesterol values acquired using the Reflotron with a venous sample taken simultaneously, to determine if the Reflotron meets the guidelines of the National Cholesterol Education Program (NCEP) for accuracy.

An announcement and a registration form for a cholesterol screening programme were distributed with employee pay slips at a large southern university. Approximately 15 employees were scheduled for each screening date, with walk-ins allowed at the health assessment site as space allowed. Capillary and venous samples were collected from screening participants (n=285).

Approximately 20 ml of blood was collected from each participant, after fasting for 12 h, using standardized venepuncture techniques in the antecubital vein in the bend of the elbow. In order to overcome technician error, two drops of blood (30 microl) were collected immediately from the previously drawn venous sample by drawing blood into the capillary tube from the opening in the top of the venous tube before centrifuging the venous sample, rather than 'sticking' the finger.

A Kolmogorov-Smirnov (KS) test of normality was calculated for total capillary cholesterol (KS=1.27, P=0.79) and total venous cholesterol (KS=0.99, P=0.28), which revealed insufficient evidence that the distributions were not normal. Participants' total capillary cholesterol values averaged 213.27 mg/dl [standard deviation (SD)=44.66 mg/dl)] when analysed on the Reflotron, and slightly higher (228.86 mg/dl, SD=40.50 mg/dl) for venepuncture. A paired t-test for variance between groups revealed significant differences in total capillary and total venous cholesterol values (t=-41.93, P<0.0001). A mean centered coefficient of variation was performed, revealing a 3.3% error rate, i.e. greater than the 3% allowable by the NCEP III guidelines. The mean percent bias was -7.28% (SD=3.10%) and the absolute mean percent bias was 7.46% (SD=2.64%). The percentage of participants with total cholesterol misclassified was 16.85%. Concomitantly, Spearman correlation coefficients were high (r2=0.94, P=0.01).

Although the Reflotron met most of the NCEP III guidelines for accuracy, the portable analyser provided clinically relevant underestimations of total cholesterol values, especially for the lower and upper values. Consequently, lipid values obtained using the Reflotron may be useful for screening, but the Reflotron should not be used as a diagnostic and management tool.