Accuracy of Non-Fasting Lipid Profile for the Assessment of Lipoprotein Coronary Risk.

Abstract

To determine the diagnostic accuracy of non-fasting lipid profile in the diagnosis of hyperlipidemia, taking fasting lipid profile as gold standard, in adult population. Cross-sectional validation study. Department of Chemical Pathology and Endocrinology, Armed Forces Institute of Pathology, Rawalpindi, from July to December 2014. One hundred seventy-five adult patients coming for fasting lipid profile were included; their non-fasting samples were taken on the next day. Patients on anti-cholesterol treatment and indoor patients were excluded. Total cholesterol (TC), high density lipoprotein-cholestrol (HDL-C), and triglycerides were measured by direct enzymatic colorimetric method by Modular p-800®. Low density lipoprotein-cholesterol (LDL-C) was calculated by Friedewald's formula, but when triglyceride was greater than 4.5 mmol/l, then LDL-C was measured directly by homogenous enzymatic colorimetric method. Non-HDL-C was calculated by simple equation, i.e. TC-HDL-C. Non-fasting lipid profile had 93% specificity , 51% sensitivity, 94% positive predictive value and 49% negative predictive value; and 65% accuracy with 7.28 positive likelihood ratio and 0.52 negative likelihood ratio. Non-fasting TC and non-HDL-C were significantly higher than fasting TC and non-HDL-C by mean difference of 0.2 mmol/l each with p=0.001 and p=0.004, respectively. Fasting and non fasting HDL-C are comparable to each other with mean difference of 0.01 mmol/l (p=0.745). Receiver operating curve (ROC) of non-fasting non-HDL-C showed 0.804 (95%CI (0.738-0.870), (p=0.000) area under the curve (AUC) indicating that it was a significant test for ruling out hyperlipidemia. Bland-Altmann plot showed a significant difference between non-fasting, non-HDL-C and fasting LDL-C and non-fasting, non-HDL-C -0.087540 with bias -0.00109; therefore, these cannot be alternative to each other. Diagnostic accuracy of non-fasting lipid profile was found significantly higher than fasting lipid profile (p=0.004) for the assessment of lipoprotein coronary risk on the basis of non-HDL-C, which seemed to be significant test for ruling out hyperlipidemia.