Mark A Mackay1, Tony C Badrick2. 1. RCPA Quality Assurance Programs, St Leonards, Sydney, NSW, Australia 2065. 2. RCPA Quality Assurance Programs, St Leonards, Sydney, NSW, Australia 2065. Electronic address: Tony.badrick@rcpaqap.com.au.
Abstract
BACKGROUND: To minimise the risk of patient harm from results, laboratories should establish QC strategies and monitor the performance of assays in line with the analytical and clinical risk. METHODS: Steady state errors were calculated from a distribution normalized for an Analytical Performance Specification expressed as Assay Capability (imprecision) minus Assay Stability (drift). Inverting this error rate gave QC run length containing one error. Multiplying by error detection of a critical shift gave a QC functional run length for stable and unstable situations. Suitability of this technique was examined using laboratory EQA imprecision and drift data against various analytical and clinical performance specifications. RESULTS: Steady state errors and error detection, and hence QC functional run length, were dramatically affected by worsening imprecision, drift or changing performance specifications. For a single analyser type, laboratory steady state errors against RCPAQAP performance specification ranged over five orders of magnitude, with contributions from Assay Capability and Assay Stability varying by laboratory. CONCLUSIONS: Steady state errors accumulate for all assays. Our functional QC run length based on steady state error rate adjusted for error detection of the QC algorithm, amounts to a risk approach using the first two elements of FMEA-like calculation and allows laboratories to examine the suitability of their combinations of QC run length, algorithm, workload and timing of QC challenges. An appropriate common performance specification is critical when assessing and comparing risk.
BACKGROUND: To minimise the risk of patient harm from results, laboratories should establish QC strategies and monitor the performance of assays in line with the analytical and clinical risk. METHODS: Steady state errors were calculated from a distribution normalized for an Analytical Performance Specification expressed as Assay Capability (imprecision) minus Assay Stability (drift). Inverting this error rate gave QC run length containing one error. Multiplying by error detection of a critical shift gave a QC functional run length for stable and unstable situations. Suitability of this technique was examined using laboratory EQA imprecision and drift data against various analytical and clinical performance specifications. RESULTS: Steady state errors and error detection, and hence QC functional run length, were dramatically affected by worsening imprecision, drift or changing performance specifications. For a single analyser type, laboratory steady state errors against RCPAQAP performance specification ranged over five orders of magnitude, with contributions from Assay Capability and Assay Stability varying by laboratory. CONCLUSIONS: Steady state errors accumulate for all assays. Our functional QC run length based on steady state error rate adjusted for error detection of the QC algorithm, amounts to a risk approach using the first two elements of FMEA-like calculation and allows laboratories to examine the suitability of their combinations of QC run length, algorithm, workload and timing of QC challenges. An appropriate common performance specification is critical when assessing and comparing risk.