Page 215 - Assurance of Sterility for Sensitive Combination Products and Materials

194 Assurance of sterility for sensitive combination products and materials

the exposure time. The overkill approach provides 95% confidence of a
12 log reduction in the bioburden.

7.3.5.3 Conclusion—Point estimates and confidence bounds—
Terminal sterilization; overkill method
The point estimate for the PNSU* for the Overkill method is zero (shown
−6
−9
as 10 in Fig. 7.11) and the UCB is 1 × 10 , one part in a million. This
method provides a high confidence demonstration that the PNSU* is
−6
<1 × 10 . The point estimate and the UCB are illustrated in Fig. 7.11. It is
important to note that while the UCB of this terminal sterilization method
−6
demonstrates a PNSU* of <10 , it is still in compliance with the entirety
of the standard that assures sterility.

7.3.6 Summary of statistical analysis
The quantifiable aspects of standards that are intended to ensure medical
devices, as used by the customer, have a low probability of being non-sterile
were reviewed from a statistical point of view. This analysis is not an ex-
haustive review of the standards, it was a survey. Most of the testing used to
demonstrate acceptability as per the standards results in pass/fail data. This
data is analyzed using nonparametric statistics and is shown in Fig. 7.12 in
blue. Two of the methods provide data that can be analyzed using paramet-
ric statistics. These are shown in green in the graph. Fig. 7.12 compiles the
data previously reported in Figs. 7.4, 7.5, 7.9, and 7.11.
The UCBs that would result from following the quantifiable aspects
of the standards were calculated. Most of the standards that were exam-
−6
ined require the point estimate of the PNSU* to be <1 × 10 . Only a
single standard that was included in the analysis, for terminal sterilization
using the overkill method, demonstrated an UCB with a PNSU* at this
level. The bioburden-based terminal sterilization method was close to this
range. The statistical UCB of PNSU* that is demonstrated as per packaging
−4
and aseptic processing standards is typically much greater, that is, 10 or
greater. As a group, the quantifiable aspects alone called out in the standards
−6
do not provide statistical evidence that the associated PNSU* is <1 × 10 .

Fig. 7.11 The PNSU* and upper confidence bounds for the ISO 11135:2014, Annex B.