32 |
Klinisk Biokemi i Norden · 1 2013
Commutability assessment by use of external
quality assessment surveys – A means to reduce
the uncertainty in the commutability decision
Sofie K. Van Houcke
1
, Pål Rustad
2
, Hedwig C.M. Stepman
1
, Thomas H. Røraas
3
, Sverre
Sandberg
3
, Linda M. Thienpont
1
1
Laboratory for Analytical Chemistry, Faculty of Pharmaceutical Sciences, Ghent University,
Ghent, Belgium,
2
Norsk Klinisk-kjemisk Kvalitetssikring (“NKK”), Bergen, Norway,
3
Norsk
kvalitetsforbedring av laboratorievirksomhet utenfor sykehus (“NOKLUS”), Bergen, Norway
Abstract
Background:
Current experimental and statistical
protocols for commutability assessment do often not
allow meaningful decisions in relation to quality spe-
cifications. We investigated the potential to solve that
problem by use of a special External Quality Assess-
ment (EQA) survey with native sera conducted by the
Norsk Klinisk-kjemisk Kvalitetssikring (NKK).
Materials and Methods:
Twenty native sera and two
EQA materials were measured in duplicate for S-Ca,
S-Mg, S-Alb and S-Prot. Five peer groups were iden-
tified: Abbott Architect-, Ortho Vitros-, Roche Modu-
lar-, Roche Cobas- and Siemens Advia. Commutability
was assessed according to the EP14-A2 statistics.
Results:
The 95% prediction intervals (i.e. defined as an
estimate of an interval in which the next future obser-
vation will fall) at the mean of the concentration range
were for S-Ca and S-Prot maximum 0.7% (for all but
one peer group). For S-Mg and S-Alb, they were wider
(up to 1.3%) and for some peer groups even broad (2.0%
- 3.5%). The study also reveals differences between the
peer groups. The EQAmaterials were judged commuta-
ble for S-Ca and S-Prot for the majority of peer groups,
but not for S-Mg and S-Alb.
Conclusion:
The study demonstrates the utility of an
EQA survey for commutability assessment, because the
high number of measurements achieved has the effect
of reducing the prediction intervals to an extent neces-
sary for meaningful decisions about commutability.
Nevertheless, the prediction intervals that remain broad
for some analytes and peer groups show that sample-
related effects may still be the limiting factor and allow
detection only of pronouncedmatrix-effects. Finally, the
partial commutability of the EQAmaterials shows ques-
tionable utility for bias assessment between methods.
Introduction
It is generally accepted that external quality assess-
ment (EQA) schemes that use processed materials
Foto: Henrik Alfthan.
