ABSTRACT
BACKGROUND: A number of relatively inexpensive compact analyzers are available for use in physician offices and outpatient clinics to measure total cholesterol and, more recently, high-density lipoprotein (HDL) cholesterol and triglycerides. This study was designed to document the analytical performance of two of them, the Abbott Vision and the Kodak Ektachem DT60, for assays of total cholesterol, HDL cholesterol, triglycerides, and calculated low-density lipoprotein (LDL) cholesterol. METHODS: Lipid profiles were measured from venous blood samples of 70 subjects with each test device, and results were compared with those from a laboratory standardized to the Centers for Disease Control. Coefficient of variation (CV) of multiple measurements from three pools of human serum (ie, precision), mean percent difference between device and standard laboratory results (ie, accuracy or bias), and 95% tolerance intervals (total error) were determined. The correct classification of patients into risk categories with device results was compared with the standardized laboratory results. RESULTS: The average CVs for total cholesterol, triglycerides, and HDL cholesterol with the Vision analyzer were 3.6%, 4.4%, and 10.5%, respectively, and with the DT60, 5.0%, 4.1%, and 6.8%, respectively. The average percent biases for the same analytes with the Vision analyzer were 0.2%, 4.0%, and -2.3%, respectively, and with the DT60, -2.1%, 12.1%, and 0.1%, respectively. Total error assessments indicated that total and HDL cholesterol measurements in individual patients met the guidelines of the National Cholesterol Education Program with both devices, but that triglycerides and LDL cholesterol measurements did not. Classification of subjects into risk groups based on total or LDL cholesterol gave clinically satisfactory results with either device. CONCLUSIONS: More precise measurement technology for LDL cholesterol is needed. Physicians and others who rely on compact analyzer results for diagnosis and treatment decisions should consider the degree of inaccuracy and imprecision in these values.
