Instructions for Use. All the AMRs met the claims though Supplemental
Table 2 shows some minor differences, based on variations in testing
and in local implementations. Reference intervals were similarly validated
though again with some differences in local implementation
and within the institution's policies on changes of reference intervals.
Precision data shown in Supplemental Table 1 in general were well
within the claims with minor differences. Apparent exceptions for
high CVs for DBIL and TBIL are actually within the SD claims for low
analyte levels. Other precision exceptions with apparent high CVs for
lithium, acetaminophen, digoxin, and tobramycin are all instances of a
single QC level at a single institution showing a high CV but with an
acceptable SD at a low analyte concentration and that were not reproducible
at other sites. The analytes with well-established and clinically
relevant performance criteria were well accepted by clinicians in all
three institutions.
Conclusions
Four AU5822 analyzers independently evaluated at three separate
institutions provided consistent results for precision, linearity/analytical
measurement range, method comparison and reference ranges such
that the data could be merged for aggregate analysis. The merged data
of 12,064 samples provided a substantially larger data set than those
of previously published studies. A total of 66 analytes, each with at
least one site providing all elements of the performance data, were included
in our study. Precision CVs were b10% for almost all analytes
studied and the majority of linearity/AMRs were within the limits specified
by the manufacturer and established by the individual laboratories
separately. Deming regression analysis on the aggregate data set
showed that 87.9% of the analytes studied had slopes of 0.900–1.100.
Slopes outside this range were found for 6 analytes with changes in
methodology or standardization between the instruments, and for 2
analytes with no change in methodology or standardization. Changes
in methodology required changes in reference ranges at all sites for 2
analytes, amylase and lipase, as was not unexpected. Many analytes,
however, despite methodology change, only required minor or no adjustments.
When there was no methodology change, the DxC and AU
platforms yielded similar results.
This large multi-center study demonstrates that the AU5822 analyzer
offered very consistent results among individual laboratories even in
geographically separated locations with diverse patient populations.
We also concluded that, despite methodology or standardization changes
on the AU5822 analyzer relative to the Synchron DxC, the impact of
changing analyzers in our laboratories created few challenges while
allowing us to update and enhance our analytical and workflow processes.
Since installations, the test results had been well received by
clinicians from all three institutions.
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.clinbiochem.2015.06.010.