Issue link: https://www.e-digitaleditions.com/i/1171020
20 Pharmaceutical Technology REGULATORY SOURCEBOOK SEPTEMBER 2019 P h a r mTe c h . c o m Analytical Procedure Validation is included in the estimate of IP variability (which is a sum of that from IP factors and repeatability factors). Second, varying it in a deliberate manner enables the variability due to cupset to be estimated separately from the variability when the same cup- set is always used (one or two devices were analyzed for each combination). The design chosen has 18 independent runs (shown in Table III) and thus 17 degrees of freedom for intermediate precision. Table IV details an allocation of analysts (3), throat/pre-sep combinations (5), and cupsets (5), which is a slightly improved design (from a within-run perspective) from the one actually used. Originally throats and pre-seps were used in an adhoc manner as they had been assessed to be low risk. However, also delib- erately varying throat and pre-sep levels (in com- bination)in the design provides a useful check on whether they do provide minimal variation without increasing the number of measurements to be made, this is shown in Table IV. For scheduling reasons, only two analysts used each NGI on a day. In the improved design, each throat/pre-sep combination and cupset are used at least once on each day. Note that the order the devices are analyzed within a day should be randomized. Example 3: Automated content method for a tablet An IP exercise was performed for an automated tablet content and identity method. This example details the design that was used, as well as illustrat- ing two adaptations. The risk assessment identified workstation (including all equipment components) as a factor. As the workstation is fully automated in- cluding diluent preparation and bulk sample prepa- ration and introduction of the tablet, analyst was not considered a factor to be incorporated. Typically, this exercise is combined with tech transfer from one laboratory to another. Given a laboratory usu- ally will have no more than two workstations, this combination of activities allows additional worksta- tions to be evaluated to provide a better estimate of intermediate precision. The actual design this example is based on was for a method considered low risk with only four independent runs being performed. If the method had been considered me- dium/high risk, then this design could have been adapted as shown in Tables V and VI. Table V shows an adapted design that has six independent runs where one of the workstations on each site has runs on two separate days so there are six independent runs (five degrees of freedom) for IP. Note, the second runs on workstation 2 and 4 were omitted in the original design. Table VI shows another alternative design to study intermediate precision that could Table VI: Intermediate precision design for Example 3 in a single laboratory. Condition Factor: Workstation 1 2 Day 1 4 5 2 3 6 Independent run 1 2 3 4 5 6 Within run Three batches and two bulk samples (of five tablets) per batch per independent run. Table V: Intermediate precision design including tech transfer requirement for Example 3. Condition Factor: Site/Analyst 1 2 Workstation 1 2 2 3 4 4 Day 1 3 5 4 2 6 Independent run 1 2 3 4 5 6 Within run Three batches and two bulk samples (of five tablets) per batch per independent run.