Examination and processing of human semen - libdoc.who.int ...

194 PART III Quality assurance
vide the standard error of the differences between technicians, which may have
to be computed separately. Substantial differences between technicians should
prompt a review of all procedures to identify how consistency can be improved.
The worked example in Box 7.6 illustrates how to compute the standard error of
the differences among technicians in sperm concentration directly, and assess
whether these are greater than would be expected from chance variation alone.
When performing computations directly from the observations, a suffi cient number
of decimal places must be kept to avoid rounding errors.
7.10.2 Monitoring monthly means
While the primary IQC procedures are based on assessment of differences among
and within technicians, additional information may be obtained by monitoring
trends in results of semen analysis.
The mean values of each variable for all the patients examined over a certain
period (e.g. monthly) can be plotted on an X bar chart, with warning and action
limits 2 and 3 standard errors either side of the mean. The standard error can be
estimated from the standard deviation of the original observations divided by the
square root of the number of semen analyses in each interval, or directly from the
observed distribution of the mean. The control limits should be determined using
at least 6 months’ observations, and should be revised regularly. There should be
at least 20 results for each mean; a small laboratory may have to pool results from
more than 1 month. Refinements to the method include monitoring monthly means
of patients with normal values and the use of cumulative sum (CUSUM) charts for
the rapid detection of any systematic departures from the mean
(Barnett, 1979).
Deviations from the expected values may refl ect different client characteristics
(time-dependent changes in the men being analysed; a change in the number of
repeat tests on the same men; changes in the pattern of referral of men with different
types of infertility) or technical factors (changes in technicians, laboratory
supplies, seasonal temperature variations, etc.).
7.11 External quality control and quality assurance
External quality control (EQC) is an integral part of the complete QC process
(Cekan et al., 1995) that monitors assay results, while external quality assurance
(EQA) monitors all laboratory procedures relating to collecting and reporting data
to ensure that laboratory processes are under control. EQC allows a laboratory to
compare its results with those of others. It permits different methods to be evaluated
and compared on a scale not possible in a single laboratory.
EQC and IQC are complementary processes. EQC may reveal problems with
accuracy that may not be apparent from IQC if control samples are not adequately
masked or selected. EQC has the advantage that it allows a laboratory to monitor
the accuracy and stability of its methods (Plaut & Westgard, 2002). However,
as EQC samples are clearly of external origin, they are liable to be handled in a
special way; this should be guarded against so that they are processed as far as
possible in the same way as routine samples.