At seminars on "Validation of analytical methods" the question
emerges again and again as to how "INTERMEDIATE PRECISION" - as
it is referred to in ICH Guidelines Q2A and Q2B - is to be verified and
expressed.
ICH lays down that Intermediate Precision
is to cover the various influences within a laboratory, i.e. conducting
analyses on two different (or several) days by different laboratory staff
members, with different equipment (if available), etc. This is to examine
in accordance with ICH Guideline Q2A the effects of random events on the
precision of an analytical method. Intermediate Precision therefore gives
a first indication already of the future transferability of an analytical
method, but it must not be confused with robustness since in the testing
of the robustness of an analysis method deliberate changes are made with
an assumed influence on the results.
How can we now determine Intermediate
Precision?
The easiest option is to summarize or
compare without comment the two series of numbers obtained. This procedure
might not always be sufficient when applying for marketing authorization.
Furthermore, the two series of numbers
(e.g. 6 [or 10] measurements each with a mean and relative standard
deviation) can be compared with the aid of statistical tests (F- and
t-test). However, argumentation problems often ensue when the test arrives
at the result that the two series of measurements differ statistically
significantly (particularly frequent and problematical in case of good
performance, i.e. little scatter)! In
this case it is recommended that the validation SOP used be furnished with
an opening clause in case a difference is statistically significant but
analytically irrelevant.
In practice the procedure has proved
successful of calculating the mean and the relative standard deviation
from all results obtained in test for intermediate precision and
prescribing an acceptance criterion for the standard deviation (e.g. 3%
relative standard deviation for intermediate precision in case of 2%
relative standard deviation for the repetition precision).
Furthermore, FDA also expects an
extensive validation protocol now already in which methodology and
acceptance criteria for the individual validation parameters are defined.
The basis of the acceptance criteria for the validation parameters is, of
course, always the specification.
Owing to large demand, our Education
Course "Validation
of Analytical Test Procedures" will be repeated in Vienna on
14-15 May. (Early booking is recommended.) The course will be recognised
for the ECA Certification.
Writer:
Dr Günter Brendelberger
CONCEPT HEIDELBERG
Summary compiled after the CONCEPT
seminar "Validation of analytical methods" held on September
25/26, 2001 in Ladenburg, Germany.
|
