Cholesterol in a Regional Program—Reduction of Interlaboratory Variability by Correcting Systematic Bias

Abstract

Reference methods for cholesterol, on which prediction of coronary heart disease is based, yield values consistently lower than routine methods. Discrepancies among results obtained from different laboratories are produced by systematic bias rather than lack of intralaboratory precision. Each of 22 hospital laboratories measured the cholesterol concentrations in two common serum pools, over a month’s time, as part of a continuing regional quality control program. The mean values for each laboratory were divided by the means of a laboratory using a reference method. The ratios were applied as correction factors to assayed values of unknown serum samples. Interlaboratory differences caused by systematic bias are greatly reduced by the correction, possibly permitting transfer of accurate data among laboratories. The value and limitations of mathematic correction of systematic bias are discussed.