Tools, Technologies and Training for Healthcare Laboratories

Can hematology analyzers hit EFLM goals?

The new EFLM database establishes new desirable specifications for imprecision, bias, and allowable total error. Can any of today's hematology analyzers hit these targets?

 

Can hematology analyzers hit EFLM goals?

Sten Westgard, MS
June 2019

The new EFLM biological database has published updated data which can be used to establish desirable imprecision, bias, and allowable total error. But the next question is, can any analyzers and methods can meet these new specifications?

We cobbled together a number of previous analyses of hematology analyzers. Now it's far from a perfect compilation, since these are different labs, with different analyzers, at different times, with even different types of controls. We don't have all the same analytes measured, nor do we have all the three levels of the controls to examine. So it's not a definitive nor ideal assessment. But in the short run, this is exactly kind of data real world labs have at their fingertips.

Multicenter verification of the Sysmex XN-Series. Schoorl M1, Schoorl M1, Chevallier M1, van der Ploeg T2, van Pelt J1. Int J Lab Hematol. 2017 Oct;39(5):489-496. doi: 10.1111/ijlh.12674. Epub 2017 May 18. https://www.westgard.com/sigma-metric-sysmex-xn.htm

Evaluation of the fully automated hematological analyzer Mindray BC 6800: comparison with Horiba ABX Pentra DX120. Grillone R, Grimaldi E, Scopacasa F, Dente B. Int Jnl Lab Hem2014;36:e55-e58. https://www.westgard.com/mindray-bc-6800.htm

Critical Systematic Errors in the Implementation and Follow-up of Performance in the Hematology Area: A Prospective Study. Critical Systematic Errors in the Implementation and Follow-up of Performance in the Hematology Area: A Prospective Study. Medicina Universitaria 2018;20(1):22-34. Jeel Moya-Salazar, Liz Pio-Davila. https://www.westgard.com/sigma-metric-landwind.htm

Multicenter evaluation of the cobas m 511 integrated hematology analyzer, Int J Lab Hem, 2018; 1-11. Bruegel M, George TI, Feng B et al.  Not perviously published on westgard.com

The first table here shows the levels where the controls were run, and then whether or not the bias was acceptable. The third column states the desirable bias, and then the next columns show the bias observed and you can see whether it was above or below the desirable specification.

TEST TEa Desirable Bias Roche m 511 level Roche m 511 bias LW D3600 level LW D3600 bias XN 2000 level XN 2000 bias BC 6800 level BC 6800 bias
Hemoglobin 3.9 1.70 6.26 0.60 53.46 5.52 64.10 3.50 14.10 1.6
  3.9 1.70 12.31 2.40 121.49 11.89 129.80 3.70    
  3.9 1.70 17.45 3.30 168.28 16.21 172.30 3.70    
Hematocrit 3.7 1.50 18.11 1.10 17.33 2.00 0.19 too high 42.80  
  3.7 1.50 34.95 3.00 36.57 3.50 0.37 too high    
  3.7 1.50 49.24 3.60 49.67 4.80 0.49 too high    
MCV 1.6 1.00 70.77 4.00 71.80 1.60     91.00 6.4
  1.6 1.00 81.47 3.40 83.00 1.50        
  1.6 1.00 88.27 3.20 92.70 2.10        
MCH 1.8 1.10 24.48 3.40 22.10 0.90        
  1.8 1.10 28.68 2.40 27.60 1.00        
  1.8 1.10 31.28 1.90 31.10 1.20        
MCHC 1.3 0.50 34.58 3.00 308.60 13.60        
  1.3 0.50 35.20 3.90 333.00 12.00        
  1.3 0.50 35.43 4.20 337.10 12.60        
RBC 3.8 1.80 2.56 0.00 1.93 0.20 2.38 2.20 4.70 2.8
  3.8 1.80 4.29 0.00 7.04 0.70 4.40 1.40    
  3.8 1.80 5.58 0.00 17.00 1.20 5.38 2.20    
WBC 13.1 4.80 1.64 0.60 2.26 0.30 3.20 5.50 9.31 11.5
  13.1 4.80 8.00 0.70 4.39 0.30 7.05 6.00    
  13.1 4.80 16.93 1.70 5.35 0.40 16.54 6.30    
Lymphocytes 14.1 6.00 0.99 3.00     1.05 3.40 1.80 15.6
  14.1 6.00 2.99 1.00     1.96 3.10    
  14.1 6.00 6.67 0.40     4.03 3.80    
Monocytes 17.3 6.40 0.08 39.60     0.47 0.20 0.71 1.6
  17.3 6.40 0.20 17.10     1.00 0.40    
  17.3 6.40 0.34 10.90     2.14 0.50    
Eosinophils 27.6 15.80 0.02 4.20     0.30 5.60 0.06 too high
  27.6 15.80 0.02 4.20     0.70 5.60    
  27.6 15.80 0.03 4.20     1.80 5.60    
Basophils 16.9 7.00 0.00 35.40     0.15 25.60 0.04 too high
  16.9 7.00 0.01 too high     0.34 24.80    
  16.9 7.00 0.02 too high     0.79 24.50    
Neutrophils 18.6 6.50 1.55 6.30     1.23 8.50 19.69 12.7
  18.6 6.50 4.78 4.10     3.05 8.40    
  18.6 6.50 9.87 3.60     7.77 8.40    
    count 36 36   21   27   9
    #acceptable 17   9   10   2
    #fails   19   12   17   7
    %acceptable 47.2   42.9   37.0   22.2
    %failure   52.8   57.1   63.0   77.8

For each analyzer, a majority of the measurements are in excess of the desirable bias. The worst analyzer is the Mindray BC 6800, where more than three-quarters of the measurements could not meet the EFLM specifications. Hemoglobin, hematocrit, and MCV, MCH, MCHC are particularly difficult assays, it appears.

Table 2 looks at the imprecision from these studies:

TEST Desirable CV Roche m 511 CV LW D3600 CV XN 2000 CV BC 6800 CV
Hemoglobin 1.35 3.7 10.3 1.0 0.8
  1.35 3.4 9.8 0.5  
  1.35 3.0 9.6 0.7  
Hematocrit 1.30 2.6 11.2 1.0 4.7
  1.30 2.3 9.5 1.0  
  1.30 2.0 9.6 0.5  
MCV 0.35 0.8 2.2    
  0.35 0.9 1.8    
  0.35 1.0 2.2    
MCH 0.40 1.0 4.0    
  0.40 1.4 3.6    
  0.40 1.1 3.8    
MCHC 0.50 1.5 4.4    
  0.50 1.4 3.6    
  0.50 1.3 3.8    
RBC 1.20 2.7 8.4 0.6 2.4
  1.20 2.8 10.1 0.5  
  1.20 2.5 7.3 0.5  
WBC 5.00 2.4 11.0 1.7 2.1
  5.00 3.4 6.8 1.0  
  5.00 7.3 8.1 1.9  
Lymphocytes 4.90 10.4 3.4 2.2
  4.90 8.1 3.1  
  4.90 6.8 3.8  
Monocytes 6.55 28.3 0.5 4.2
  6.55 29.5 1.0  
  6.55 30.2 2.1  
Eosinophils 7.20 too high 5.3 16.7
  7.20 too high   6.8  
  7.20 too high   7.2  
Basophils 6.00 too high   2.4 25.0
  6.00 too high   1.8  
  6.00 too high   1.7  
Neutrophils 7.35 7.7   0.5 2.0
  7.35 4.8   1.0  
  7.35 3.4   2.1  
  count 36 21 27 8
  #acceptable 7 3 26 4
  #fails 29 18 1 4
  %acceptable 19.4 14.3 96.3 50.0
  %failure 80.6 85.7 3.7 50.0

Again, we see that half or more of the observed measurements for most of the analyzers do not meet desirable imprecision specifications. However, there is definitely a differentiation here between analyzers, with the Sysmex showing an almost perfect ability to meet the desirable imprecision.

What should we do with this snapshot? In the short run, this means hematology sections of laboratories need to invest more in their operations, and prepare for a lot more work. It looks like these goals, if applied, will force a major change of practice in the laboratory. More analyzers will be considered unacceptable. More QC should be applied. More runs should be considered unacceptable for release to clinicians and patients. The alternative is to determine a different form of analytical performance specifications.