Directly measured low density lipoprotein cholesterol (DLDLC) has been reported to be more accurate than calculated low density lipoprotein cholesterol (CLDLC) using the Friedewald equation. However, some limitations of DLDLC have been reported. In this study, we evaluated differences between CLDLC and DLDLC measured using HiSens reagents.

Data were collected from 582 persons undergoing routine physical examinations at a general hospital. LDLC measurements were made directly or estimated using the Friedewald formula, and were classified according to the National Cholesterol Education Program's Adult Treatment Panel III guidelines. The relationship between these differences and other clinically relevant factors, such as triglyceride (TG) levels, were examined using multiple logistic regression analysis.

The DLDLC and CLDLC were strongly correlated according to simple linear regression analysis (r=0.917, P<0.001) but the mean difference between measurements was -11.0±15.3 (-62 to 90.5) mg/dL (P<0.001). For more than 10 mg/dL of their absolute differences, the DLDLC was typically lower than the CLDLC. The highest discrepancies in LDLC measurements occurred when LDLC was more than 160 mg/dL and less than 190 mg/dL. Differences in LDLC measurements were prone to striking negative and positive biases dependent on CLDLC and TG concentrations, respectively (all r>0.5).

Unlike other studies, DLDLC was significantly lower than CLDLC and the large differences in LDLC concentrations were not dependent on TG concentration. Our work suggests that verification of DLDLC accuracy is needed and differences in LDLC measurements should be accounted for in making clinical decisions.