requisite combination of performance and false-
rejection rate. A number of multiple-rule proce-
dures have been proposed that perform much better
than single-rule procedures. The multiple-rule
procedure of Westgard
et al.
(1981) has achieved the
greatest popularity. The rules used in the procedure
are listed in Table 2.4. The control sample set for
this procedure consists of a low concentration
control sample and high concentration control
sample. If both control results pass the 1
2s
control
rule using the respective established result distribu-
tions, the batch is considered to be in-control. If one
of the control results exceeds its 2 SD
within-laboratory
limits, evaluate the results using the remaining
control rules. If the results fail to pass any of the
rules, the batch of measurement is rejected. If the
results pass all of the rules, the batch is in-control.
Cumulative sum procedures and moving average
procedures have also been proposed as tools for
monitoring for persistent degradation in method
quality (Nix
et al.
1987, Strike 1996). The perform-
ance of these approaches in the control of method
trueness has been shown to be superior to that the
multiple-rule procedures (Bishop and Nix 1993,
Parvin 1992). These approaches are computational
intensive but can easily be implemented in the
modern computerized laboratory.
Test sample-based quality control.
The use of
control material for internal quality control has
several practical problems. The material is expen-
sive, it may have limited stability, and often its
composition is different from that of the test
samples. This has prompted the development of
alternative procedures for monitoring method preci-
sion and trueness that use test samples rather than
control material. These procedures have not been
accepted as replacements for quality control using
control material but many laboratories use them to
supplement their control material-based internal
quality control program.
To monitor within-run precision, a test sample is
divided into aliquots prior to being assayed and the
aliquots are run in the same batch. The variance of
the replicate results for the test sample is calculated
using the formula,
var
=
(
x
i
−
mean
)
2
n
−
1
where
x
i
is the
i
th replicate result,
mean
is the mean
of the replicates, and
n
is the number of replicates.
If duplicates are used, the formula is,
var
=
(
x
1
−
x
2
)
2
Once 20 to 30 such replicate determinations have
been made, the within-run imprecision is estimated
using the formula,
SD
within
−
run
=
var
j
N
where N is the number of test samples studied. This
estimate is compared to the within-run imprecision
established for the method. When the next test
sample replicate set is run, its variance is added to
the variance data set and the oldest variance value in
the data set is deleted. The within-run imprecision
is recalculated and the updated estimate is compared
to the imprecision standard. Within-laboratory
analytic precision can be evaluated in a similar
fashion by having test sample replicates assayed in
different batches.
To monitor method trueness, the mean value is
calculated for a block of consecutive test sample
results. The block may consist of a specified
number of results, of all of the results in a batch, or
of all of the results for a defined period of time,
usually a day. The block mean is compared to the
population mean established for the method.
Assuming that the mix of patients is similar over
time, the mean value of a block of test sample
results will equal the established population mean.
Truncation of the data to exclude extreme result
values improves the performance of the method.
Improved performance also comes from the use of a
Laboratory Methods
2-12
Table 2.4
Multiple-Rule Control Procedure (SD is SD
within-laboratory
)
Warning rule
1
2s
one control result exceeds mean ± 2 SD
Within batch rules
1
3s
one control result exceeds mean ± 3 SD
R
4s
one control result exceeds mean + 2 SD
one control result exceeds mean – 2 SD
Within and between batch rules
2
2s
two consecutive control results
exceed mean + 2 SD or mean – 2 SD
4
1s
four consecutive control results
exceed mean + 1 SD or mean – 1 SD
10
x¯
ten consecutive control results
fall on one or the other side of the mean
