Long-term stability of laboratory tests and practical implications for quality management

Background: Long-term stability of analytical performance is required for adequate patient management. We investigated the use of patient data to document test stability, and the relevance of observed instabilities on a surrogate medical outcome. We used multiyear patient and internal quality control (IQC) data from two laboratories for tests to monitor chronic kidney and thyroid disease. Methods: We plotted moving means of the 50th percentiles of stratified patient data and of the daily IQC means. We evaluated observed instabilities based on goals inferred from the analytes' biological variation and investigated their effect on classification of results against reference intervals. Results: Patient and IQC data generally matched well, except for analytes, for which other than analytical variation sources prevailed. Analytical instabilities were predominantly due to reagent/calibrator lot changes, however, for immunoassays also to within-lot instabilities, urging frequent recalibrations. The relevance of biased results on medical decisions ranged from negligible to very pronounced, indicating the need for assessment of analytical performance in relation to quality goals inferred from biological variation. Conclusions: Patient percentiles offer great potential to assess/monitor the medium-to long-term analytical stability of a test within certain constraints. Differences in analytical quality between assays can significantly affect medical outcome.