FAQ - Contagem de partículas em laboratórios farmacêuticos ou de ciências biológicas

Pergunta
We have installed and validated a continuous particle monitoring system from Particle Measuring Systems. Now we are struggling with the question how to deal with isolated excursions. We are using IsoAir Plus sensors sampling for 1 minute. What is your experience/recommendation?

Resposta

The whole EC GMP issue is confusing and many times it is confusing to the inspectorate. This is due to some unfortunate wording being chosen when the September 2003 was written. I will try to answer your question as easily as I can.

1. A cleanroom needs to go through 3 phases of activity prior to use, the majority of times though this is condensed down to just two Room Certification (performed to ISO14644-1) and Room Monitoring (risk assessment).
2. If we look at the table 1 in the EC Guide it shows the limits for room CERTIFICATION. If we then look at Note "a" below it states, a1 - use a particle counter, a2 - where there is a high risk monitor continuously and a3 - for routine testing (certification) take a m3 of sample. When it is read as a single line it is confusing as it almost reads that for continuous sampling take a m3 of sample - it does not, there are 3 statements made in Note a and each is a new line and each starts with a capital letter and ends in a period.

a1 Use a particle counter.

a2 Where there is a high risk, monitor continuously.

a3 For routine testing (certification) take a m3 of sample.

So the table is for certification purposes only, what then do we do about monitoring limits.

1. For monitoring you should look to your process. For 0.5 um monitoring the limits in the table can be used as alarm limits, for alert limits you can go to Pharmaceutical Net software and use the SPC graph for a specific period of time when operators are active, this will give a 3sd (95% UCL) value for the activities that normally happen in the cleanroom, this may well be apx 20% of the alarm limit.

For 5.0 um this changes, the frequency of alarm is so low that a 95th percentile would run at 0.2 counts which is impossible to manage in real time, so we must now look at the frequency of events not the value of those events. Say for instance your process runs with 5-10 individual counts throughout a 30 minute interval; this is normal for your process, they occur randomly and cannot be attributed to any occurrence in the room, and product continuously meets the sterility tests. So the alarm limit for that process for 5.0um particles would be alarm if you see 3 events in 10 minutes, and alert if you see 2 events in 10 minutes. The value for these events are in m3 and so would read 35counts/m3; this is okay and is normalization, the key is the number of events not the magnitude of such event. If then the system starts to run not at 5-10 but at 10-12 then this increase in random background counts will be annunciated through the alarm system, as it would yield a rate of grater than 2 (or 3) events per 10 minute period. This also allows you to respond to these events in a timely manner.

This is just an outline, but hopefully the same rationale you apply to managing any process variable is applied to the particle counts in the cleanroom.

Contact us if you need more information or have questions.