PAT & QbD SUPPLEMENT


suitable methods and to select the (objectively) most suitable. The latter is always connected with the current situation, which may alter during the lifecycle. Then a new evaluation perhaps with new possibilities added can build upon the existing knowledge. The classification may also change with the scale of production, i.e. number of batch analysis performed (see Example 3).


Figure 3QbD Lifecycle concept for the Mock Example water determination


mainly the random contribution of the accuracy determination. If a true bias must be limited, the random part has to be minimised and an appropriate experimental design must be chosen, as described in the ‘Total Error’ concept13 For the given example of water deter -


.


mination, from a risk perspective a less rigorous approach without numerically defined statistical risks seems to be acceptable. For assay (which can be generalised for determinations at concentrations sufficiently away from the quantitation limit), the relative accuracy corresponds to an acceptable precision14,15


.


For titrations, in the Technical Guide for Elaboration of Monographs of the European Pharmacopoeia, a relative accuracy of twice an acceptable repeatability is established16


.


This is in line with the proposed approach, because in the ATP concept, a higher precision level is addressed.


Analytical lifecycle Method selection Having defined the ATP, next an appropriate method would be identified which can be assumed to meet the requirements. This may be assisted by a priority matrix (see Table 1 on page 18). Here, besides the ATP requirements as primary objective, also business considerations (operational requirements) can be applied. The latter should be preferably aligned with the end users, i.e. the industrial quality control units. Depending on the complexity, such a systematic approach may help to evaluate potentially


EXAMPLE 2 The control of variability is especially important at the lower range limit to avoid the occurrence of the nonhydrated polymorph. Assuming the theoretical water content of 4.19 per cent as the lowest acceptable true product value, the available distribution range for analytical results would be 0.19 per cent. Describing this range by a 95 per cent prediction interval, an overall absolute and relative method variability of 0.10 and 2.5 per cent can be derived (seeEquation 1)11


. 95 per cent analytical prediction interval: Equation 1


t(P,df) = Student-t-Factor for a statistical confidence P of 95 per cent and degrees of freedom df, related to the precision studies the standard deviation was obtained from. For an acceptance limit, a high reliability is required. With increasing df, t approximates a value of 2


sAP


= Standard deviation of the analytical procedure (overall, including calibration and the defined number of determinations)


The defined precision is also acceptable at the upper specification limit. Assuming a normal statistical distribution at 95 per cent confidence level, with such a (true) precision, the maximum true water content leading (in less than five per cent cases) to experimental results exactly at the specification limit of five per cent would be approximately 5.2 per cent.


20


European Pharmaceutical Review Volume 16 | Issue 3 | 2011


analytical procedure. Under airtight conditions, 0.5 millilitres of standardised solution of water was added and the water content was determined. This standard addition procedure was repeated seven times, so that a range up to eight per cent of the water content in the sample was covered. An unweighted linear regression of the cumulative water determined vs. water added was performed and the slope, intercept,


EXAMPLE 3 Assuming a moderate number of batches to be analysed, the Karl-Fischer volumetric titration was initially selected as the most suitable method.


Method development and validation For method development and validation, the general ATP requirements are ‘translated’ into the method-specific performance characteristics. All variables that have a potential impact on method performance criteria should be identified and considered. Risk assessment tools can be applied to identify potentially critical variables. These variables can be further evaluated by experimental design studies. If they need to be controlled, appropriate system suitability tests must be established. By such investigations, the robust method range is defined (method operable design region7


) (see


Figure 1on page 17 andExample 4on page 22). Accuracy of the Karl-Fischer titration was ,


addressed by a standard addition approach16


which also includes linearity. The water content of the sample was determined according to the


Table 2 Calculation of intermediate precision by analysis of variances


Parameter Overall mean


Result 4.35 per cent


Inter-serial variance (repeatability) 0.000131 between-series variance Overall variance


0.000550 0.000681


Pooled repeatability Intermediate precision


0.26 per cent 0.60 per cent


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