EXTERNAL QUALITY ASSESSMENT
Analyte Na K
Ca Creat
Glucose Urate
Cholesterol HDL
HbA1c
Intervention target Conc.
135 mmol/L 3.5 mmo/L 2.2 mmol/L 90 μmol/L
2.0 / 6.5 mmol/L 360 μmol/L 5.0 mmol/L 1.0 mmol/L
48 mmol/mol TEa (%)
Min Des Opt 0.9 0.6 7.3 3.4
0.3
4.9 2.4 2.3
1.1
11.7 7.8 9.2
6.1
3.9 3.1
19 12.6 6.3 12.5 8.3
4.2
14.9 9.9 5.0 4.7
3.1 1.6
Proposed APS TEa (%)
1.0 hybrid (Best method) Hybrid (2.4 opt + 4.9 des) 3.4 min 7.8 des 6.1 des
Hybrid (6.3 opt + 12.6 des) 8.3 des 9.9 des
5.0 hybrid (min + best method)
Table 1. Proposed analytical performance standards (APS) for nine routine analytes measured in clinical biochemistry, with figures for total allowable error (TEa).
supporting data, as well as the intended clinical application of the test whether for screening, diagnosis or monitoring. Thus, the choice of model often represents a balance between clinical relevance and real-world feasibility. Although outcome- based APS represents the highest level of the Milan hierarchy, their application is limited, and most EQA schemes currently rely on APS derived from biological variation or ‘state of the art’ models.
Current work The aim of the study was to review the strengths and weaknesses of the three models and compare with what was achievable in a real-world environment to establish clinically appropriate APS for several serum chemistry analytes. Models based on the biological variation of the analyte (Model 2) and the highest level of analytical quality achievable (Model 3) were reviewed respectively. Laboratory and POCT method
performance data from Weqas (Cardiff, UK) over the last five years was collected across a wide clinical concentration for common analytes in clinical biochemistry. The data covered 60 distributions, using 240 samples, and assayed by 200 laboratories for a range of analytes, including sodium, potassium, creatinine, calcium, glucose, urate, total- and HDL-cholesterol and HbA1c. Overall precision profiles were calculated for each analyte, and for each of the major methods in use for that analyte. These were represented as standard deviation (SD) and coefficient of variation (CV%) against analyte concentration. The overall and method profiles were compared with the optimal, desirable and minimum APS available from the EFLM Biological Variation Database based on Model 2, and the methods with the best analytical quality established. The strengths and weaknesses of the various models were reviewed and compared to what was achievable in a real-world environment.
Results A summary of proposed APS for nine routine analytes is presented in Table 1.
APS based on biological variation (Model 2) achievable For potassium, a universal APS based on the optimal EFLM total allowable error (TEa = 2.4%) from Model 2 was achievable for all indirect methods above 2.0 mmol/L; with indirect methods achieving the desirable EFLM (TEa = 4.9%) across all concentrations tested (Fig 1). Similarly, for urate, a universal APS based on optimal EFLM total allowable error (TEa = 6.3%) from Model 2 was achievable for all methods above 350 μmol/L; and achieving desirable EFLM (TEa = 12.6%) below 350 μmol/L (Fig 1). For creatinine, a universal APS
based on the desirable Model 2 TEa was achievable at concentrations exceeding 200 μmol/L, and minimum TEa at concentrations exceeding 70 μmol/L for all methods in current use, with optimal Model 2 TEa achieved for all enzymatic creatinine methods as well as two Jaffe methods at concentrations > 100 μmol/L (Fig 2). For glucose, a universal APS based
on desirable Model 2 TEa (6.1%) was achievable at concentrations exceeding 3.0 mmol/L for most methods (Fig 2). Regarding clinical decision limits, desirable APS was achieved at 2.0 mmol/L for Abbot, Roche and Siemens methods; and optimum APS for the Abbot method at 2.5, 4.0 and 7.0 mmol/L.
For total cholesterol, although a
universal APS based on the desirable EFLM TEa (8.3%) from Model 2 was achievable for all methods, performance was influenced by concentration of triglycerides in the sample, with increased CV% noted for samples with triglyceride concentration exceeding 3 mmol/L (Fig 2). For calcium, the universal APS based on minimum Model 2 TEa (3.4%) was achieved at concentrations exceeding 1.8 mmol/L for most methods, with close to linear correlation between concentration and performance (Fig 3). Thus, the proposed APS uses minimum TEa at concentrations above 1.8 mmol/L; and Model 3 below 1.8 mmol/L. Both Abbot Alinity and Beckman AU400 instruments consistently achieved performance between desirable and optimum TEa (2.3% and 1.1% respectively). For HDL-cholesterol, a universal APS
based on the minimum TEa (14.9%) was achieved at concentrations exceeding 1.0 mmol/L (Fig 3). In addition, most methods could achieve desirable performance, with some, including Siemens and Roche instruments frequently achieving APS based on the optimum TEa.
APS based on state of the art (Model 3) achievable Although some improvement in sodium performance over the last five years has been noted, results did not consistently meet the minimum TEa limit (0.9%) for Model 2 across the concentration range tested (Fig 4). Instead, a hybrid model based on the minimum TEa and Model 3, the current ‘best method’ is proposed. For HbA1c, a universal APS based
on Model 2 minimum TEa (4.7%) was not achievable for all methods (Fig 4). However, further examination of individual HbA1c methods showed that most laboratory electrophoresis and ion exchange methods could achieve minimum TEa and a hybrid approach of minimum TEa and Model 3 is proposed. Different APS were proposed for diagnosis and monitoring.
Conclusions Although Model 2 was achievable for many analytes, including calcium, creatinine, glucose, total- and HDL- cholesterol, it was rarely achievable across the full pathological range. The
The choice of model often represents a balance between clinical relevance and real-world feasibility
46
WWW.PATHOLOGYINPRACTICE.COM May 2026
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52
