Sigma Metric Analysis
Sysmex XN 350 in India, multimode analysis
For over a year, we've been looking at the impact of new CLIA and EFLM goals on the assessment of biochemistry instruments. Well, the goals for hematology have changed as well. How does a Sysmex XN 350 fare when judged by the latest goals for CLIA and EFLM
Sysmex XN 350 in India, multimode analysis
January 2023
Sten Westgard, MS
See the other analyses in this series:
- Beckman Coulter DxC 700
- Abbott Alinity
- Siemens Atellica
- Siemens Atellica in Romania
- Siemens Atellica in Spain
- Siemens ADVIA 2120i
- Roche c501 in Turkey
- Roche c501 in Saudi Arabia
- MicroLab RX-50 in India
- Roche cobas 6000 immunoassays in Turkey
- Sysmex XN 350 in India
- Mindray 7500 in China
- Mindray BS 2000M in China
It is a weakness of the Westgard website that we have seem to have a "bias" for biochemistry analysis. There are historical roots for this: primarily, the "Westgard Rules" originally being formulated to govern the monitoring of the then-emerging, then-novel automated multitest instruments. Many if not most of the original Westgard papers published in the journals used biochemistry analytes as examples. There was no intent to malign or ignore hematology parameters, but there it was, all the examples are biochemistry, thus people assumed that they didn't apply to hematology. So, too, has it gone with Sigma-metrics. Most of the initial examples have been in biochemistry, therefore the application of Sigma-metrics in hematology has been slower.
We hope to rectify this with our next set of analyses.
This Sysmex XN 350 data comes from Central Diagnostic Laboratory of Shre Krishna Hospital, Karamsad, India.
Application of Sigma Metrics in Haematology Laboratory, Dr. Monica Gupta, Dr. Mustafa Ranapurwala, Dr. Kevna Kansara, Dr. Mitul Chhatriwala, International Journal of Clinical and Diagnostic Pathology, 2022; 5(2): 21-25.
The study was conducted with IQC data from June 2021 to September 2021.
"IQC data from 8 control run cycles (with a minimum of 20 runs) were collected....[B]ias was obtained as the mean of a peer group available in the kit insert of control material."
There are three levels and two control lots that will be analyzed here. So for 5 parameters there will be 15 data points to plot on the Normalized Method Decision Chart.
Significantly, the study used the CLIA 1992 goals to calculate Sigma metrics. Our new analysis will look at the 3 new sets of goals: CLIA 2024 goals, EFLM desirable specifications, and EFLM minimum specifications. [Unlike our earlier multimode analyses, we have narrowed our own focus to this trio of benchmarks. Increasingly, the other goals are losing their relevance.]
INDIA SYSMEX XN 350 LOT 1 |
||
TEST | % Bias | CV |
Hemoglobin | 3.2 | 1.5 |
1.6 | 1.6 | |
1.2 | 1.0 | |
Hematocrit | 2.2 | 1.2 |
2.0 | 1.9 | |
1.7 | 1.0 | |
Platelets | 15.3 | 9.0 |
1.3 | 3.7 | |
4.7 | 2.2 | |
RBC | 2.6 | 1.0 |
2.0 | 1.5 | |
1.9 | 0.8 | |
WBC | 4.9 | 3.5 |
3.3 | 2.7 | |
2.8 | 2.2 |
INDIA SYSMEX XN 350 LOT2 | ||
TEST | % Bias | CV |
Hemoglobin | 3.6 | 1.5 |
1.7 | 1.2 | |
1.9 | 0.9 | |
Hematocrit | 1.2 | 1.1 |
0.9 | 1.5 | |
0.7 | 0.8 | |
Platelets | 2.2 | 7.9 |
7.9 | 6.2 | |
2.9 | 2.3 | |
RBC | 2.6 | 1.1 |
2.5 | 1.4 | |
2.0 | 0.7 | |
WBC | 3.2 | 2.2 |
0.1 | 1.9 | |
0.9 | 1.5 |
The TEa goals applied can be found on our Consolidated Hematology Performance Specifications page.
Sigma-metrics according to EFLM-derived DESIRABLE performance specifications
The EFLM desirable specifications used to be the de facto global standard, but have fallen out of favor due to their toughness.
The Sysmex XN 350 cannot get more than a pair of data points into the Bull's-Eye using these goals.
Not surprising then, that EFLM recommended lowering the standards.
Sigma-metrics according to EuBIVAS-derived MINIMUM performance specifications
There is a huge improvement, lowering the number of assays considered poor and unacceptable significantly. A few more assays are in the Bull's-Eye, as well.
Here's one of the most interesting new aspects of CLIA's new 2024 goals. Are they more or less demanding than EFLM goals?
Sigma-metrics according to CLIA 2024 performance specifications
Astonishingly, the CLIA 2024 for hematology are far more demanding than EFLM No 6 Sigma assays here, and a majority of the parameters are missing the target for Lot 1. An improvement in Lot 2, but still not looking better than any assessment by EFLM standards.
Conclusion
The new era of goals for hematology appears ominous. If the three dominant benchmarks deliver an unacceptable verdict of a leading hematology instrument, what are we to conclude? Are the instruments fundamentally flawed, or the goals themselves?
Bonus Analysis: How are these assays judged when MU is compared to MAU, pU specifications?
While the original publication didn't intend to assess uncertainty, since imprecision was measured, the simple estimation of uncertainty can be made. So let's take a good look at that. Perhaps the specifications for hematology MU are more forgiving?
INDIA SYSMEX XN 350 LOT2 | |||||
TEST | CV | Expanded MU |
EFLM min MAU |
Panteghini Preferred pU |
Final Verdict? |
Hemoglobin | 1.5 | 3.0 | 4.1 | 2.8 (des) 4.2 (min) |
Passes MAU and pU |
1.2 | 2.4 | 4.1 | 2.8 (des) 4.2 (min) |
Passes MAU and pU | |
0.9 | 1.8 | 4.1 | 2.8 (des) 4.2 (min) |
Passes MAU and pU | |
Hematocrit | 1.1 | 2.2 | 4.2 | -- | Passes MAU |
1.5 | 3.0 | 4.2 | -- | Passes MAU | |
0.8 | 1.6 | 4.2 | -- | Passes MAU | |
Platelets | 7.9 | 15.8 | -- | 4.85 (des) 7.28 (min) |
Passes pU |
6.2 | 12.4 | -- | 4.85 (des) 7.28 (min) |
Passes pU | |
2.3 | 4.6 | -- | 4.85 (des) 7.28 (min) |
Passes pU | |
RBC | 1.1 | 2.2 | 3.9 | 1.55 (des) 2.33 (min) |
Passes MAU and pU |
1.4 | 2.8 | 3.9 | 1.55 (des) 2.33 (min) |
Passes MAU and pU | |
0.7 | 1.4 | 3.9 | 1.55 (des) 2.33 (min) |
Passes MAU and pU | |
WBC | 2.2 | 4.4 | 16.2 | -- | Passes MAU |
1.9 | 3.8 | 16.2 | -- | Passes MAU | |
1.5 | 3.0 | 16.2 | -- | Passes MAU |
If we judge these assays by the various performance specifications set forth by Panteghini et al, there are far more victories than when these parameters are judged by the TEa benchmarks. MAU and pU are more forgiving.