Page 112 - Assurance of Sterility for Sensitive Combination Products and Materials

96 Assurance of sterility for sensitive combination products and materials

its combination cannot be verified at the point of use. In addition to patient
safety requirements, the health-care industry is faced with the need for in-
creased efficiency and productivity in order to drive down costs. Carefully
designed validation approaches along with reliable test methods, and process
control tools enable leveraging of data that can dramatically reduce the costs
of changes and keep them to a minimum.

5.3.2 Package integrity and the limitations of sterility testing

Sterility testing has been used in the past to verify the sterility of products
but it only provides a false sense of security since the conclusions that can
be drawn from sterility test data have significant limitations. Sterility testing
can only detect microorganisms that are viable, but testing is dependent
on culture media and growth conditions. Testing for all types of microor-
ganisms and all conditions is impractical. In addition, the sterility tests are
destructive, take days to complete and are prone to false positives. This often
leads to inconclusive root cause analysis if results are positive for microbial
growth. Most importantly, the low incident rates require extremely large
sample sizes to generate meaningful conclusions for sterility. For example:
Assume the hypothetical case where sterility testing is used for accepting
or rejecting a batch of packaged sterile devices with a sterility assurance
−2
level, SAL =10 . In this case, one item out of hundred from this batch
would potentially be non-sterile, vs the normal objective of one in a mil-
lion. With only one tested sample, the probability to accept that batch is
−2
1–10 or 99%. With two tested samples, the probability to accept the batch
−2
−2
would be (1−10 ) × (1–10 ) = 98%. We can deduct the following formula:
−2
Probability to accept a batch with SAL = 10 with n samples tested for
−2 n
sterility: p = (1–10 ) .
Using this formula for 20 samples, the batch is accepted in 82% of the
cases, with 50 samples accepted in 50% of the cases and with 300 samples
the batch would still be accepted in 5% of the cases. The typical target of
−6
a SAL of 10 would require far more samples. One can easily conclude
−6
−6 n
that with a SAL of 10 , the equation will be: p = (1–10 ) and millions
of samples will be required to achieve similar confidence. Obviously, the
practicalities of manufacturing millions of samples to destructively test them
are prohibitive.
5.3.3 Test methods

Test methods are the cornerstone of all validation. An accurate and reli-
able method to assess the output is required to make informed decisions.