Analytical validation of the new plasma calibrated Accu-Chek Test Strips (Roche Diagnostics®)

(1)

Analytical

Analytical validation of

validation of

the new plasma

the new plasma calibrated

calibrated

Accu

Accu--Chek

Chek Test

Test Strips

Strips

(Roche Diagnostics

(Roche Diagnostics®)

®)

(Roche Diagnostics

(Roche Diagnostics®)

®)

Meex Cécile

Clinical Chemistry Clinical Chemistry C.H.U. Liège, Belgium C.H.U. Liège, Belgium

(2)

CHU Liège: 3 sites

The site of the CHU consists of three

establishments:

• CHU Sart-Tilman

• CHU Notre-Dame des Bruyères

(3)

CHU Liège: 3 sites

± 1000 beds

635 beds CHU S-T 226 beds CHU NDB 226 beds CHU NDB 94 beds CHU OA

Wards: medicine, surgery, paediatrics, maternity, geriatrics, dialysis, emergencies, intensive care, center of the burned, haematology, sterile unity, metabolic unity…

(4)

Point of Care in CHU Liège

• 8 blood gas analyzers.

Before December 2004

• 2 PCx (Abbott) in the metabolic unity under the control of the laboratory. Recovery of the fees.

• A lot of Accu-Chek Sensors (Roche) disributed in the other services without control of the laboratory.

(5)

Point of Care in CHU Liège:

Accu

Accu--Chek Inform.

Chek Inform.

Since December 2004

• 65 Accu-Chek Informs distributed • 65 Accu-Chek Informs distributed

in all the departments of the 3 sites.

Recovery of the results and invoicing via the software DataCarePOC (Roche).

(6)

Accu

Accu--Chek Inform: Why?

Chek Inform: Why?

In Belgium: new legislation for the laboratories.

POC analyzes charged if:

Traçability: operator’s identification, patient’s identification. Daily quality controls.

Daily quality controls.

Maintenances, QC, calibrations under the responsability of the

laboratory.

Recording of the lots of used tigettes. Documented training of the staff.

(7)

Training of the staff

Nurses = Qualificated technicians under the responsability of the clinical biologist.

Importance of the training of the staff.

Sessions of training organized in every ward by the biologists and the responsible technicians.

Printing of an identification bar code on the badge of the users of the Accu-Chek Inform.

Every user is registered to work only on the devices of his own ward.

(8)

Point of Care in CHU Liège:

Accu

Accu--Chek Inform

Chek Inform

Validation of the Accu-Chek Informs

Each devices was validated before its first use: Each devices was validated before its first use: • Repeatability

• Reproducibility • Linearity

(9)

Point of Care in CHU Liège:

Accu

Accu--Chek Inform

Chek Inform

March 2005: Accu-Chek Informs effective in

the wards.

27000 glycaemias realized from March 05 to this day. By extrapolating: ± 120000 glycaemias / year.

The most consumer wards:

(10)

Point of Care in CHU Liège:

Accu

Accu--Chek Inform

Chek Inform

7362 5000 6000 7000 8000 158 945 1974 1593 709 677 737 1599 21222087 842 1978 323 335 450 20 1617 272 405 234 _{81 255 46} 0 1000 2000 3000 4000 5000 +3A Hop Sem +2D Max +4C D Ab do +4A Uro +3B OR L +4B Gas tro -1AB Neu ro -2A B C ardi o -4A B P neu +2B Neu roC hir +1A Car dioC hir -2D Si C oron a -3E Psy cho -2C SIM ed +5A Nep hro +3D Ger ia +2A Orth oChi r +1C SI C +1B SI B +1D SI D Série1

(11)

Plasma calibrated test strips

May 2005: New plasma calibrated test

strips.

Test strips calibrated to deliver results of glycaemia Test strips calibrated to deliver results of glycaemia

(12)

Plasma calibrated test strips

Validation:

Repeatability Reproducibility Linearity

Plasma strips VS Whole blood strips Plasma strips VS Analyzer (on plasma) Plasma strips VS Blood gas analyzer

Capillary whole blood VS Venous whole blood with the

(13)

Repeatability

10-fold-repeated measure of controls low and high level (mg/dl):

Level 1 Level 2 54,4 327 Labo mean 1,65 10,52 3,03 3,22 55 329 56 342 55 329 54 330 54 312 57 321 54 331 51 317 53 316 55 343 Standard deviation CV

(14)

Reproducibility

Daily measure of controls low and high level during 10 days (mg/dl): Level 1 Level 2 55,5 319,2 Labo mean 2,92 9,55 5,26 2,99 55 316 58 330 57 333 59 329 57 316 57 325 49 305 54 308 53 314 56 316 Standard deviation CV

(15)

Linearity

Level Reference Labo

1 39 30

2 62 60

3 131 148

Linearity test kit (Roche)

mg/dl 600 400 V a le u rs _ la b o 4 315 319 5 472 537 6 510 HA Linear regression: Y= -12,4095 + 1.1345 X

Level 6 not considered because out of the range accepted by the Accu-Chek Inform. 0 100 200 300 400 500 Valeurs_de_référence 200 0 V a le u rs _ la b o

(16)

Plasma strips

VS

WB strips

Material

60 venous whole blood (EDTA):

• 20 with normal glucose concentration: 60-100 mg/dl • 20 with low glucose concentration: < 60 mg/dl

• 20 with high glucose concentration: > 200 mg/dl

10 samples of capillary whole blood drawn in premature babies (high haematocrit level).

(17)

Plasma strips

VS

WB strips

Method

Measure of the glycaemia of each of the samples with the plasma and then the WB test strips.

Results Results

Considering the 70 samples together:

Correlation: r = 0,9965

Regression: New strips = 0,2090 + 1,0726 × Former strips Wilcoxon test: p < 0.0001

(18)

Plasma strips

VS

WB strips

Linear regression 600 500 0 100 200 300 400 500 600 Anciennes 400 300 200 100 0 N o u ve ll e s

(19)

Plasma strips

VS

WB strips

Bland-Altman plot 40 30 (N o u ve ll e s - A n ci e n n e s ) / A ve ra g e % +1.96 SD 25,8 0 100 200 300 400 500 600

AVERAGE of Nouvelles and Anciennes 20 10 0 -10 -20 -30 (N o u ve ll e s - A n ci e n n e s ) / A ve ra g e % Mean 7,3 -1.96 SD -11,2 25,8

(20)

Plasma strips

VS

WB strips

The same statistical tests were realized on every group of samples taken separately.

Conclusions

Significant difference between the results obtained with the Significant difference between the results obtained with the plasma and the WB strips:

• considering the 70 samples together

• considering each group separately except for the low concentration samples.

Results of plasma strips on average 7 % superior to those given by the WB strips.

(21)

Plasma strips

VS

Analyzer

Material

54 patients

Simultaneous puncture of a drop of capillary whole blood and a tube of venous whole blood.

tube of venous whole blood. Method

Capillary whole blood directly analyzed by Accu-Chek Inform with plasma calibrated test strips.

Centrifugation of the tubes and measure of glycaemia on plasma (hexokinase method). Delay: < 2 hours.

(22)

Plasma strips

VS

Analyzer

Results

Plasma strips VS Integra

Regression: New test strips = -6.1860 + 1,1766 × Integra Regression: New test strips = -6.1860 + 1,1766 × Integra Mann-Whitney test: p = 0.0351

(23)

Plasma strips

VS

Analyzer

Bland-Altman plot: 30 25 20 (i n fo rm i n te g ra ) / A ve ra g e % +1.96 SD 24,0 50 100 150 200 250 300

AVERAGE of inform and integra 20 15 10 5 0 -5 (i n fo rm i n te g ra ) / A ve ra g e % Mean 10,4 -1.96 SD -3,2

(24)

Plasma strips

VS

Analyzer

CLARK ERROR GRID

500 600

Clark error grid

0 100 200 300 400 500 0 100 200 300 400 500 600 R ef er ence ( mg / d L) Clark Grid A D E D E C C B B B

(25)

New strips

VS

Analyzer

Conclusions

Results of the Accu-Chek Inform about 10% higher than results of the Integra.

Statistically significant difference. Statistically significant difference.

Underestimation of plasma glucose concentration might be

explained by the delay in separating plasma from blood. (A.Y.W. Chan, Clin. Chem. 35/2, 315-317 (1989)).

But no clinically significant difference. Good correlation.

(26)

Plasma strips

VS

Blood

gas analyzer

Material

20 syringes of whole blood (11 arterial et 9 venous) drawn in ICU. Method

Method

Determination of the glycaemia by the electrode of glucose of the blood gas analyzer (Rapidlab 865, Bayer: glucose oxydase

method, amperometric detection) on the syringe.

Right after that, measure of the glycaemia from a drop of blood of this syringe with Accu-Chek Inform and plasma calibrated test strips.

(27)

Plasma strips

VS

Blood

gas analyzer

Results

For all the 20 samples:

Correlation: p = 0,9778

Regression: New strips = -2,7847 + 1,0409 × Bayer 865 Wilcoxon test: p = 0,2579

(28)

Plasma strips

VS

Blood

gas analyzer

Bland-Altman plot 20 15 +1.96 SD 15,1 50 100 150 200 250

AVERAGE of accu and phmetre 10 5 0 -5 -10 -15 (a c cu p h m e tr e ) / A ve ra g e % Mean 1,8 -1.96 SD -11,5 15,1

(29)

Plasma strips

VS

Blood

gas analyzer

Conclusions

No significant difference between the results given by the blood gas analyzer and those given by the Accu-Chek Inform, whatever arterial or venous sample was used.

or venous sample was used. Good correlation.

(30)

Capillary

VS

Venous

Material

20 patients

Simultaneous puncture of a drop of capillary whole blood and an Simultaneous puncture of a drop of capillary whole blood and an EDTA tube of venous whole blood.

Method

Capillary whole blood analyzed directly by Accu-Chek Inform with plasma calibrated test strips.

Simultaneous determination of the glycaemia on a drop of venous whole blood.

(31)

Capillary

VS

Venous

Results

Regression: Capill blood = 3,4657 + 0.9633 × Venous blood Mann-Whitney test: p = 0,7150

(32)

Capillary

VS

Venous

Bland-Altman plot 20 15 10 (i n fo rm e d ta ) / A ve ra g e % +1.96 SD 12,6 80 100 120 140 160 180 200 220

AVERAGE of inform and edta 5 0 -5 -10 -15 -20 (i n fo rm e d ta ) / A ve ra g e % Mean -0,8 -1.96 SD -14,2

(33)

Capillary

VS

Venous

Conclusions

No significant difference between a glycaemia measured on capillary whole blood or venous whole blood with the plasma calibrated Accu-Chek test strips.

calibrated Accu-Chek test strips. Good correlation.

(34)

Conclusions

Global difference between the plasma and WB test strips of 7,3 %.

Not influenced by: Not influenced by:

the glucose concentration in the sample the haematocrit rate

(35)

Conclusions

Comparable values of glycaemia obtained with:

Plasma calibrated strips on capillary whole blood Plasma calibrated strips on venous whole blood Blood gas analyzer with venous or arterial blood.