The assessment of the performance of three low-density lipoprotein cholesterol equations in hypertriglyceridaemia samples from an academic hospital in Gauteng province

Abstract

Background: Accurate low-density lipoprotein cholesterol (LDL-C) measurement is crucial for cardiovascular risk assessment. Calculated LDL-C values are often preferred over direct enzymatic measurements because of their lower cost and wider accessibility, particularly in resource-constrained laboratory settings. While the Friedewald equation is widely used to calculate LDL-C, its limitations have led to the development of alternative equations such as the Sampson and Martin–Hopkins equations.
Aim: Therefore, this study aimed to assess the performance of three LDL-C equations in hypertriglyceridaemia samples.
Setting: An academic hospital in the Gauteng province.
Methods: A retrospective analysis was conducted on 9337 lipid profile patient results, with 2758 (29.5%) exhibiting hypertriglyceridaemia, defined as triglycerides ≥ 1.7 mmol/L. Low-density lipoprotein cholesterol was determined using a direct enzymatic assay and was calculated using the three equations: Martin–Hopkins, Sampson and Friedewald. Regression analysis between the direct LDL method and the equations was done using Passing-Bablok. Bias and agreement between the different LDL-C measurement methods were evaluated using the Bland-Altman plot, and where agreement between methods was present at clinically relevant concentrations, it was assessed using weighted Cohen’s Kappa statistics. Differences between LDL-C methods were evaluated using the Kruskal–Wallis test.
Results: In hypertriglyceridaemic samples, the Martin–Hopkins equation demonstrated the strongest correlation and the least bias compared to direct LDL measurement. All equations underestimated LDL-C, resulting in patient misclassification. However, Martin–Hopkins had the least misclassified LDL results, particularly in the < 1.8 mmol/L concentration range.
Conclusion: The findings suggest that LDL-C equations exhibit satisfactory agreement with direct methods, with the Martin-Hopkins equation being the best-performing calculated method, while the Friedewald and Sampson equations exhibited considerable underestimation of LDL-C levels.
Contribution: The underestimation of LDL-C equations shows potential of patient misclassification with regards to treatment intervention limits which may lead to erroneous clinical decisions, emphasising the need for careful consideration in clinical implementation.