Estimating the biases of the Korean national cholesterol proficiency test

It is recommended that clinical laboratories keep the bias of serum total cholesterol analysis at < or =3.0% compared to a reference method. In Korea, national cholesterol proficiency testing has long been available, but there has been little information about the magnitude of analytical bias. The authors calculated the bias of the peer group mean for Korea's national cholesterol proficiency test through an indirect approach that overcomes the potential matrix effect of proficiency test materials. One laboratory was selected among the proficiency test participants to represent Korean laboratories. Total cholesterol levels of six fresh serums spanning a wide range of concentrations were measured by the representative laboratory and three reference laboratories. The relationship between the proficiency test mean and the reference method mean was established by linear regression analysis. The peer group mean of the proficiency test was calculated to have a bias of +2.4 to +2.5% at the medical decision levels. When grouped by instrument and reagent, 29 to 66% of the laboratories showed biases <3.0%. Thus it was determined that the peer group mean of the Korean cholesterol proficiency test has an acceptable level of positive bias. The indirect approach used in this study provides a practical model for estimating cholesterol analytical bias for proficiency testing.