BACKGROUND: Proficiency testing or external quality assessment schemes (PT/EQASs) are an important method of assessing laboratory performance. As each scheme establishes assigned values and acceptable ranges for the analyte according to its own criteria, monitoring of participant performance varies according to the scheme and can lead to conflicting conclusions. METHODS: Standard deviations (SDs) for PT were derived from Thompson's and biological variation models applied to blood and urine manganese (Mn) robust data from four EQASs from North America and Europe. The fitness for purpose was verified by applying these SDs to individual results. RESULTS: Using Thompson characteristic function the relationship between SD and Mn concentration, expressed in nmol/L was the square root of [19.72+(0.07712×Mn concentration2)] for blood and the square root of [6.772+(0.09852×Mn concentration2)] for urine. While the biological variation model was not suitable for urine, it produced an acceptable range for blood as ±54.4 nmol/L (assigned value ≤320 nmol/L) or 17% (assigned value >320 nmol/L). For blood, individual performance evaluated by the two approaches led to similar conclusions. CONCLUSIONS: The biological variation model can be used to propose quality specifications for blood, however it could not be applied to urine. The Thompson characteristic function model could be applied to derive quality specifications for Mn in urine and, to a lesser extent in blood. The more lenient quality specifications for blood highlight the difficulty of determining Mn in this matrix. Further work is needed to harmonize PT, such as using assigned ranges for the specimens.
BACKGROUND: Proficiency testing or external quality assessment schemes (PT/EQASs) are an important method of assessing laboratory performance. As each scheme establishes assigned values and acceptable ranges for the analyte according to its own criteria, monitoring of participant performance varies according to the scheme and can lead to conflicting conclusions. METHODS: Standard deviations (SDs) for PT were derived from Thompson's and biological variation models applied to blood and urine manganese (Mn) robust data from four EQASs from North America and Europe. The fitness for purpose was verified by applying these SDs to individual results. RESULTS: Using Thompson characteristic function the relationship between SD and Mn concentration, expressed in nmol/L was the square root of [19.72+(0.07712×Mn concentration2)] for blood and the square root of [6.772+(0.09852×Mn concentration2)] for urine. While the biological variation model was not suitable for urine, it produced an acceptable range for blood as ±54.4 nmol/L (assigned value ≤320 nmol/L) or 17% (assigned value >320 nmol/L). For blood, individual performance evaluated by the two approaches led to similar conclusions. CONCLUSIONS: The biological variation model can be used to propose quality specifications for blood, however it could not be applied to urine. The Thompson characteristic function model could be applied to derive quality specifications for Mn in urine and, to a lesser extent in blood. The more lenient quality specifications for blood highlight the difficulty of determining Mn in this matrix. Further work is needed to harmonize PT, such as using assigned ranges for the specimens.