Transcript 2010 TEQAS_Comparison FESTEM(slides)

Comparison of different
proficiency testing schemes
for Al, Cu, Se, Zn
Josiane Arnaud1, Patrick J Parsons2,
Marina Patriarca3, Andrew Taylor4,
Jean-Philippe Weber5, Cas W Weykamp6
1 – CHU de Grenoble, France; 2 – New York State Department of Health, Albany, USA;
3 – Istituto Superiore di Sanità, Rome, Italy; 4 – University of Surrey, Guildford, UK;
5 – Institut National de Santé Publique, Québec, Canada;
6 – Queen Beatrix Hospital, Winterswijk, The Netherlands;
Thematic Network
Organizers of external quality assessment (EQAS)/
proficiency testing (PT) schemes related to
occupational and environmental laboratory medicine
ISO,
EURACHEM,
EQALM,
ILAC,
IUPAC,
IFCC…
Different possibilities for
organisation,
statistical analyses,
participant monitoring
Conflicting conclusions
at the participant level
Harmonization
Coordonnateur :
Dr Andrew Taylor
(University of Surrey)
Material and methods
Samples and PT schemes
Serum Pool
Native
Sample 157
Spiked with
Al-Cu-Se-Zn
Sample 158
Spiked with
Al-Cu-Se-Zn
Sample 159
Shipment to 8 PT scheme organizers (Dry ice)
- Belgium, Canada, France, Italy, The Netherlands,
Spain, The United Kingdom, USA (Cu, Se, Zn)
- France, Italy, The United Kingdom, USA (Al)
Shipment to participants (Ambient temperature)
Assigned value
 Reference laboratory + reference method
 Tracability from certified reference material
 Consensus value (reference laboratories)
 Consensus value (participants)
Outlier elimination
Mean, Median, Robust mean….
Algorithm A,
ISO / FDIS 13528
Quality specification (QS)
Sample dependent
 Consensus participants’ standard deviation
(after exclusion of outliers)
 Reference laboratory standard deviation
 Metrological laboratory standard deviation
Consensus QS (analyte and medium)
takes into account:
 Analytical possibilities (concentration range)
 Expertise of laboratory
 Biological individual variability
 Overlaps between « normal » and pathological states
 Law recommendations (if any)
 EQAS role: laboratory licensing or education
Quality specification choice
Fraser’s equation:
QS = 0.25(CV2intra+CV2inter)1/2 + z(0.5xCVintra)
CV intra = intra-individual variability
CV inter = inter individual variability
z = 1.65 for a 95% probability level.
Assay
Minimal quality specifications
Serum Al
Serum Cu
Serum Se
± 20% or 0.18 µmol/L
± 12% or 0.84 mol/L
± 12% or 0.07 mol/L
Serum Zn
± 15% or 1.20 mol/L
Whichever
is
the
greatest
Result analyses
Statistical tests
Robust mean
→
Kruskall Wallis
Pm
Robust SD
→
Levene
Pvar
Each scheme,
all participants
Scheme
differences
Results
and
Discussion
Scheme robust means vs
overall robust mean: Al 158
Pm=0.51
Pvar=0.45
Robust SD
QS/2
Robust mean
Scheme robust means vs
overall robust mean: Se 158
Pm=0.77
Pvar=0.001
Robust SD
QS/2
Robust mean
Scheme robust
means vs
overall robust
mean - Zn
Pm=0.10
Pvar=0.002
Pm=0.24
Pvar=0.42
Pm=0.04
Pvar=0.004
Scheme
robust
means
vs
overall
robust
mean
Cu 157
&
Cu 159
Pm=0.41
Pvar=0.07
Pm=0.06
Pvar=0.008
Method robust means vs
overall robust mean
Pm=0.59
Pvar=0.006
Robust SD
QS/2
Robust mean
Conclusion
Factors modulating PT results
→ Number of participants
→ Expertise of participants
→ Aim of the PT scheme
→ Method used by the participants
→ Sample properties (concentrations)
Harmonization
of the procedures
did not eliminate discrepancies
Conclusion
Analytical
performances
Reliable
determinations
Patient’s
diagnosis
and
follow-up
Participation in EQAS
EQAS harmonisation
EQAS gathering
Thank you for your attention
and welcome to additional PT