RESUMEN
BACKGROUND: Analytical performance of 24 Beckman-Coulter AU 5800 methods was verified against recognized quality goals and manufacturer's expected imprecision and bias. METHODS: AU5800 method imprecision, bias, agreement with a comparative method, and linearity were studied using CLSI protocols, commercial control material, patient samples, and linearity test kit solutions. Repeat patient testing and IQC were also used for imprecision. Commutability of control material was tested. Total analytical error (TAE) was estimated for each method and between the tested and the comparative method, the Beckman-Coulter Unicel DxC800. RESULTS: CLSI EP15 total imprecision CV (TCV) < 3.2%. Duplicate patient imprecision CV < 2.8%. IQC imprecision CV < 5.1%, except for low level ALP (CV = 7.4%). Sodium and urate IQC imprecision were higher than manufacturer's specifications. TAE for all methods met accepted quality goals. Correlation between methods was > 0.975, except for Cl (0.971), TP (0.964), and Na (0.948). Average bias versus Unicel DxC800 is high for ALP (17.3%), GGT (37%), LD (20%), TBIL (-23%), and TP (8%) and was confirmed in other laboratories. TAE between methods met allowable total error for 21 analytes. For GGT, between method TAE (23 to 51%) was predictable from expected bias and combined method imprecision. For LD and TP several between method differences were outside boundaries describing expected bias. Linearity was excellent with R2 > 0.997 and deviations met accepted goals. CONCLUSIONS: The Beckman-Coulter AU 5800 demonstrates good linearity, low imprecision, and good correlation with previous methods. Observed between method differences suggest ALP, GGT, LD, TBIL, and TP harmonization should be considered.
