Evaluation of seven Cholestech L.D.X analyzers for total cholesterol determinations

We assessed the performance of seven Cholestech L.D.X lipid analyzers under tightly controlled laboratory conditions for accuracy and precision in accordance with analytical guidelines of the National Cholesterol Education Program (NCEP). Venous heparinized whole blood (VB) and plasma (VP), venous serum (VS), and capillary fingerstick whole blood (FB) were collected from 18 individuals. Total cholesterol (TC) concentration was measured in VB, VP, and VS on all seven instruments. Three instruments were used for TC measurements of FB. Reference cholesterol values for each individual were generated in the same laboratory with a standardized method. The within-run coefficients of variation (CVs) for all instruments with a Level I pool (1560 mg/L, n = 10) ranged from 1.3% to 1.8% (mean = 1.59%). The between-run CVs with the same pool ranged from 2.2% to 3.4% (mean = 2.84%, n = 10). Correlation coefficients derived from comparison of total cholesterol values generated by the instruments for each specimen type vs the reference cholesterol values were all > 0.97. The average bias for all instruments for each sample type was 1.9% (FB), 4.3% (VB), 6.6% (VP), and 7.0% (VS). Predicted cholesterol concentration for each sample type from regression curves for total cholesterol at the suggested NCEP clinical decision cutoff values of 2000 and 2400 mg/L, respectively, were 2049 and 2431 mg/L for FB, 2081 and 2469 mg/L for VB, 2122 and 2522 mg/L for VP, and 2121 and 2521 mg/L for VS.