»
MANUFACTURING
» “
Another test to evaluate the diff erence between cleaning validation runs is to use 2 sample t-tests to compare data pop- ulations for 2 runs. Figure 4 shows the result of such tests as produced by statistical package Minitab.
Non-normal is a very typical outcome of a cleaning validation study since the point of ECP is to remove manufacturing and cleaning process residuals completely.
We may inspect a number of parameters presented in the Figure 4 summary. However, the most important one we are looking for is a p-value. The p-value or calculated probability is the estimated probability of rejecting the null hypothesis of a study question when that hypothesis is true. When the p-value is above 0.05 when using it to test null hypothesis that both cleaning runs represent same sample populations (because ECP should be consistent), then we can say that 2 population means are the same with 95% confi dence. In this case, the p-value is 0.454 which well above a 0.05 p-value.
A Confi dence Interval (CI) gives an estimated range of values which is likely to include an unknown population parameter, the estimated range being calculated from a given set of sample data. If independent samples are taken repeatedly from the same population (consistent cleaning process should produce same results), and a CI calculated for each sample, then a certain percentage (confi dence level) of the intervals will include the unknown population para- meter. CIs are typically calculated so that this percentage is 95%, but we can produce 90%, 99%, 99.9%, etc, confi dence intervals for the unknown parameter. Then we can examine tolerance intervals5
and perform
early process capability analyses using normal and non-normal capability analyses (depending on normality of the data sets). When data are found to be non-normal, fi nd the best fi tting model.
The following inter-cleaning run variability analyses can be utilized for cleaning process qualifi cation study examination:
• Individual value plot
• Box-plot • ANOVA
*Default is limit presented for illustration only. Limits for each cleaning process should be determined and evaluated separately. Please refer to Part 1 of this article for information on determining product-specifi c limits.
30 | | January/February 2015 Figure 4. Results of 2 sample t-tests.
• Two-sample t-test •
Two one-sided t-test
• Non-parametric tests (eg, Mann-Whitney test)
We will illustrate some of these tools in the following few examples. Figure 3 shows a Box-plot of TOC results of the cleaning studies of 7 products that utilize the same clean-in- place (CIP) cycle.
As we can see in Figure 3, all of the cleaning studies are quite similar and are well within set default limit of 3 ppm.*
Additionally, two one-sided t-test (TOST) can be used to determine the equivalence of 2 data sets. When a regular t-test is used to conclude there is a substantial diff erence, we must observe a diff erence large enough to conclude it is not due to sampling error, p-value above pre-set α (eg, 0.05). TOST applies with equivalence testing to conclude there is not a substantial diff erence we must observe a diff erence small enough to reject that closeness is not due to sampling error from distributions centered on large eff ects (within 95%, 90%). Figure 5 illustrates the diff erence in the population tail examination between a regular t-test and TOST.
“
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76