Tools, Technologies and Training for Healthcare Laboratories

Sigma-metrics Trueness Evaluation of Serum Albumin methods

In 2017, Clinical Chemistry published a study evaluating the trueness of 24 Serum Albumin methods.

Sigma-metrics Trueness Evaluation of 24 Serum Albumin methods

March 2017
Sten Westgard, MS

[Note: This QC application is an extension of the lesson From Method Validation to Six Sigma: Translating Method Performance Claims into Sigma Metrics. This article assumes that you have read that lesson first, and that you are also familiar with the concepts of QC Design, Method Validation, and Six Sigma. If you aren't, follow the link provided.]

State of Harmonization of 24 Serum Albumin Measurement Procedures and Implications for Medical Decisionste of Harmonization of 24 Serum Albumin Measurement Procedures and Implications for Medical Decisions Bachmann LM, Boyd JC, Bruns DE, Miller WG, Clinical Chemistry 63:3 770-779 (2017).

"We examined the current state of harmonization and analytical performance of 24 commercially available albumin measurement procedures using freshly collected, nonfrozen serum and plasma from patients without renal disease and serum from patients receiving hemodialysis."

What we'll add to this study is a graphic analysis of the same instrument data, this time using the CLIA goal for Albumin, as well as a focused graph that just looks at the lower critical level of performance.

The Imprecision and Bias Data

"Residual patient samples were used to prepare (a) serum pools from patients without renal disease (n = 50 pools), (b) heparin plasma pools from patients without renal disease (n = 48 pools), and (c) serum pools from patients with kidney failure collected before hemodialysis (n = 53 pools). "

Bias was calculated by comparisons against a reference material. "The ERM-DA470k/IFCC Proteins in Human Serum certified reference material (RM) from the Institute for Reference Materials and Measurements was prepared centrally at Virginia Commonwealth University....The target values for the RM pools were determined by gavimetric reconstitution.... Bias was estimated by the moving average of 21 consecutive differences centered on the indicated value."

Imprecision was calculated based on 4 replicates of QC samples performed at the beginning, meiddle and end of each run.

 

  TEa Low end % Bias CV Sigma
Immage - chem 10 2.0 0.8 10.0
BN II - chem 10 8.5 1.2 1.3
ARCHITECT c4000 BCG 10 7.8 0.6 3.7
ARCHITECT c8000 BCG 10 4.9 0.4 12.8
ARCHITECT c16000 BCG 10 4.9 0.7 7.3
AU 680 US BCG 10 8.8 0.7 1.7
AU 680 Intr BCG 10 8.3 0.2 8.5
Vitros 5600 BCG 10 8.2 0.8 2.3
Cobas c501 BCG 10 11.2 1.3 -0.9
ADVIA 1200 BCG 10 23.0 2.1 -6.2
ADVIA 2400 BCG 10 23.5 2.0 -6.8
ARCHITECT c4000 BCP 10 5.2 0.3 16.0
ARCHITECT c8000 BCP 10 5.2 0.4 12.0
ARCHITECT c16000 BCP 10 5.8 0.6 7.0
         
DxC 800 BCP 10 0.8 0.3 30.7
DxC 800 BCP cm 10 0.0 0.3 33.3
Cobas c501 BCP 10 3.3 1.4 4.8
ADVIA 1200 BCP 10 2.1 2.0 4.0
ADVIA 1800 BCP 10 1.4 2.0 4.3
ADVIA 2400 BCP 10 0.7 2.1 4.4
ADVIA XPT BCP 10 2.7 2.8 2.6
Dimension RxL BCP 10 1.5 0.4 21.3
Dimension Vista BCP 10 0.1 0.5 19.8

 Yes, that is a whole lot of numbers! What do they all mean? In the absence of context, it's often hard to know.

So let's apply Sigma-metrics and plot the performance visually.

Summary of Albumin Performance by Sigma-metrics Method Decision Chart

We can make a visual assessment of this performance using a Normalized Sigma-metric Method Decision Chart:

2017 Albumin Method Trueness NMEDX
 

 It is clear that the BCP method is superior to BCG. "None of the BCG methods met the minimum performance specifications for bias based on biological variability criteria over a physiologically reasonable range of concentrations. Eight of 12 BCP methods met the minimum performance for bias and those that did not had  generally smaller biases than were observed for BCG methods. Our data show that BCP methods had better selectivity for albumin and had proportional biases; thus, it is likely that harmonization of results can be achieved."

It's also clear that some manufacturers have consistent performance, and others have method performance all over the map.

  Conclusion

The authors stated "In summary, BCG methods have larger biases than BCP methods when compared to the RMP. Furthermore, the bias of BCG methods, but not BCP methods, varies with the concentration of albumin....[S]ingle decision thresholds for albumin concentration are likely inappropriate for patient-care decisions. Standardardization of albumin results using dye-binding methods will require BCP as the preferred reagent."

Based on Sigma-metric analysis, we can see that there are not only bias issues, but there are also imprecision issues. Just switching to the BCP methodology is not a panacea, but it is a step in the right direction.