Concordance between LDL-C estimated by various formulas and directly measured LDL-C

Background: Although direct measurement of LDL cholesterol (LDL-C) in blood is possible, there are several formulas for its estimation. The performance and concordance of these formulas have not been evaluated in Colombia.

Objective: To determine the concordance between LDL-C directly measured using the enzymatic technique and existing methods to calculate it.

Methods: Study of diagnostic tests, and concordance. We analyzed complete lipid profile samples, which included direct measurement of LDL-C, from 2014 to 2022 at Hospital Universitario San Ignacio (Bogotá, Colombia). The direct LDL-C measurements were compared with estimations using the DeLong, Sampson, Friedewald, extended Martin/Hopkins, Anandaraja, and Cordova methods. Lin's concordance correlation coefficient (CCC) and Bland-Altman plots were employed, conducting subgroup analyses based on triglycerides (TG), and LDL-C levels. Kappa coefficients assessed agreement in LDL-C risk categories according to dyslipidemia guidelines.

Results: A total of 2144 samples were evaluated. The formulas with the best CCC were DeLong (0.971) and Sampson (0.969), with no relevant differences. The extended Martin/Hopkins formula (0.964) and the Friedewald formula (0.964) also performed well. The Anandaraja (0.921) and Cordova (0.881) equations exhibited inferior performance. For all formulas, a decrease in concordance was observed when triglycerides were ≥400 mg/dL or when LDL-C was <100 mg/dL. Most formulas demonstrated optimal agreement when assessed using risk categories according to dyslipidemia guidelines, except for Anandaraja and Cordova.

Conclusions: The DeLong, Sampson, extended Martin/Hopkins, and Friedewald formulas show the best concordance with directly measured LDL-C, so in most cases the results can be considered interchangeable. However, the Anandaraja and Cordova formulas are not recommended.