Extent of Equivalence of Results for Urine Albumin among 3 Candidate Mass Spectrometry Reference Measurement Procedures

Abstract Background Urine albumin (UA) is an important biomarker of chronic kidney disease. Current in vitro diagnostic medical devices (IVD-MDs) for measuring UA are not standardized, and median results among IVD-MDs differ by approximately 45%. Since fixed decision values are used to interpret UA, higher-order reference measurement procedures (RMPs) are needed for metrological traceability. Three candidate liquid chromatography-tandem mass spectrometry RMPs have been developed for UA. Methods Eight single-donation human urine samples were measured by 3 candidate RMPs. Results were compared using t-test and variance component analysis. Results The mean results for each urine sample from each RMP laboratory were not statistically different from the overall mean value by t-test. The median total CV including contributions from bias and imprecision among the 3 RMP laboratories was 6.23% using variance component analysis for each sample. The allowable bias to the RMP for an end-user IVD-MD was ≦9.0% or ≦3.0% based on the desirable or optimal total allowable error of 30% or 24%, respectively. A maximum allowable standard uncertainty for an RMP result was determined to be 4.3% or 3.3% for desirable or optimal performance, respectively. The standard uncertainties for all of the RMP laboratories meet the desirable and optimal standard uncertainty specifications. Conclusion The candidate RMPs for UA in these 3 laboratories have suitable agreement of results and uncertainties for use as higher-order RMPs in the metrological traceability of end-user IVD-MDs for measuring UA.