Calculated values of serum LDL-cholesterol (LDL-C) – for better or worse?

Abstract Background and Aims The use of Friedewald’s formula to calculate serum low-density lipoprotein cholesterol (LDL-C) is well-known to have limitations. A modification of it, in 2013, has been proposed to be superior. However, it was not known whether LDL-C values (calculated by the modified formula) meet laboratory performance criteria for their estimation. This study aimed to evaluate this. Methods and Results LDL-C values were calculated for 129,821 lipid profiles, using both Friedewald’s formula and its modified version. Kappa statistics and intra-class correlation coefficient (ICC) were used to determine degree of agreement between directly measured and calculated values for LDL-C. Bias and total percentage error of the values were calculated. LDL-C concentrations calculated by the modified formula showed a greater degree of agreement with directly measured values (kappa = 0.713) than those calculated by Friedewald’s formula (kappa = 0.595). Both the formulae produced values with negative biases (-3.47 for the modified formula and -7.62 for Friedewald’s formula) and total percentage errors above the recommended limit of 12% (15.57 % for the modified formula and 21.77 % for Friedewald’s formula). ICC showed that values calculated by the modified formula showed a greater degree of agreement with directly measured values, across a range of LDL-C values. Conclusion Calculated LDL-C values, using the modified formula, showed better agreement with directly measured values, and less bias and percentage total error than those obtained by use of Friedewald’s formula. However, the percentage total error with use of the modified formula exceeded the recommended limit for LDL-C.