Proof of process stability

The EU and FDA guidelines on process validation require that the stability of manufacturing processes must be proven. How can this be achieved?

A simple and effective way to analyze your data with regard to process stability is to display the results to be evaluated in control charts.

In contrast to the Excel charts usually used in process validations, the control chart does not contain any acceptance criteria, but statistical key figures that allow the current production to be compared with previous productions. For example, the plotted mean value of previous results can be used to identify mean value shifts in current production (see Fig. 1), while the display of warning and intervention limits allows the scattering of results to be evaluated (Fig. 2).

The warning and intervention limits in control charts are calculated from the mean value and the standard deviation of the sample, which should comprise at least 30 data points.

This is based on the assumption that, in the case of a nominal distribution, 95 % of the values lie within the range of mean ± 2σσ and 99.7 % of all values lie within the range of mean ± 3σσ (see Fig. 3).

What does that mean?

If 100 values are entered in the control chart, no values outside the intervention limits may occur with an unchanged stable process and only 5 values may be in the range between the warning limit and the intervention limit.

If 1000 values are entered in the control chart, only 3 values will be outside the intervention limits if the process stability remains unchanged.

What else is important:

Setting intervention limits based on the results of only a few batches, e.g. three process validation batches, carries the risk of underestimating the “normal” inter-batch variability, i.e. the differences between the individual batches, so that the calculated intervention limits may not be sufficient. are too narrow. It therefore makes sense in this case to clearly state that the limits determined are provisional limits that will be adjusted in the further course of batch production as soon as further findings on inter-batch variability have been obtained.

By the way: By using control charts, you are using a statistical tool and thus already fulfilling the requirement for statistical analyses of your data!

If you have fewer than 30 data points from the production of previous batches, it makes more sense to carry out a purely visual evaluation of process stability based on diagrams without intervention limits, but with the reported acceptance criteria.

Dipl. Ing. Wolfgang Rudloff

Vita

• 30 Jahre Erfahrung in der pharamzeutischen Industrie

Expertise:

• Geprüfter Sachverständiger für Reinraumtechnik und GMP-Management
• Lead Auditor und Gutachter

Stationen:

• Technischer Leiter
• COO / Gesamt-Herstellungsleiter
• Remidiation-Manager
• Projektleiter Neubau/GMP-Upgrade
• Management-GMP-Beratung
• Fachautor, GMP-Trainer, Mediator