Klinisk Biokemi i Norden · 1 2013
| 37
limited by the variance caused by sample-related
effects. In this study, this is shown, for example, by
the markedly higher prediction intervals obtained
for some S-Mg and S-Alb assays, which only allows
detection of pronounced matrix effects (>3%, in the
examples). In the case of the Abbott Architect S-Mg
assay, for example, we expected a prediction interval
of approximately 1.5% (calculated from duplicates of
4 laboratories and a median CV of 2%). The observed
prediction interval of 3.5% is thus an indication for
the presence of sample-related effects in the method
comparison. Again, in view of the biological bias spe-
cification for S-Mg of 1.8%, it should be stressed that
a prediction interval in the order of 3.5% is useless
for commutability assessment of an EQA material
intended for trueness assessment, but should be res-
tricted to 0.6%. Contrarily, in the S-Ca example, the
prediction interval of the Siemens Advia peer group
is indicative for the absence of sample-related effects.
Note also the difference between the assays, e.g., for
four of the S-Prot assays the prediction intervals were
maximum 0.7%, which contrasts with the 1.8% of the
fifth assay; the same is observed for the prediction
intervals of maximum 1.0% for two assays for S-Alb,
versus those of the three other assays ranging from
2.1% - 3.2%.
Finally, for what concerns the quality of the investi-
gated EQA samples they clearly are different in terms
of commutability. While material #1 was generally
judged commutable, this was not the case for mate-
rial #2 that showed in particular non-commutability
with certain S-Mg and S-Alb assays. This generally
questions their utility in EQA surveys for bias assess-
ment between methods because, typically, such sera
are intended to cover multiple analytes.
Acknowledgements
The authors are indebted to Dietmar Stöckl (STT-
Consulting) for valuable discussions.
References
1. Thienpont LM, Stöckl D, Friedecký B, Krat-
ochvíla J, Budina M. Trueness verification in
European external quality assessment schemes:
time to care about the quality of the samples.
Scand J Clin Lab Invest 2003;63:195-201.
2. Miller WG. Specimen materials, target values
and commutability for external quality assess-
ment (proficiency testing) schemes. Clin
Chim Acta 2003;327:25-37.
3. Miller WG, Jones GR, Horowitz GL, Wey-
kamp C. Proficiency testing/external quality
assessment: current challenges and future
directions. Clin Chem 2011;57:1670-80.
4. Vesper HW, Miller WG, Myers GL. Reference
materials and commutability (Review). Clin
Biochem Rev 2007;28:139-47.
5. Long T. Statistical power in the detection
of matrix effects. Arch Pathol Lab Med
1993;117:387–92.
6. Stöckl D, Stepman HCM, Van Houcke SK,
Thienpont LM. Importance of sample-rela-
ted effects for commutability testing accor-
ding to the EP14 protocol. Clin Chim Acta
2010;411:1378–9.
7. Clinical and Laboratory Standards Institute.
CLSI/NCCLS approved guideline EP14–A2.
Evaluation of matrix effects. Wayne (PA):
CLSI, Wayne, Pennsylvania: 2005.
8. National Cholesterol Education Program.
Recommendations for improving cholesterol
measurements. US Department of Health and
Human Services publication NIH 90-2964
1990 National Institutes of Health Washing-
ton, DC.
9. Eckfeldt J, Miller WG, Myers GL, Welch M,
Caudill S (project team). NKDEP, NIST, CAP
reference material commutability study: A
preliminary look at the data. Presented at the
NKDEP Manufacturers Forum July 27, 2006.
groups/Eckfeldt_mfg_forum_0706_508.pdf
10. Clinical and Laboratory Standards Institute.
CLSI/NCCLS approved guideline C37–A.
Preparation and validation of commutable
frozen human serum pools as secondary refe-
rence materials for cholesterol measurement
procedures. CLSI, Wayne, Pennsylvania: 1999.
Key words: EP 14-A2, matrix effects, sample-
related effects, sample variance, serum-albu-
min (S-Alb), serum-calcium (S-Ca), serum-
magnesium (S-Mg), serum-protein (S-Prot).
