Skip to content

Newsroom

News & Insights

Company news from Humiley Group INC and technical notes on cleanroom HVAC, GMP validation, semiconductor environments and energy-storage engineering.

  • 780 articles

Everything

From the Humiley newsroom

87 articles tagged #Manufacturing.

Who answers for a room's qualified state

"Quality is QA's job" is not how the GMP texts read. The 2013 text of EU GMP Chapter 1 makes the quality objective the responsibility of senior management, with staff in many departments taking part. Chapter 2 lists qualification and maintenance of premises and equipment among the duties the heads of Production and Quality Control generally have.

InsightRead

CPV: one abbreviation, two different things

Continued process verification is the FDA's Stage 3: assurance during routine production that the process remains in a state of control. EU GMP Annex 15 (2015) calls it ongoing process verification. Continuous process verification, defined in ICH Q8(R2), is an alternative approach to validation. Write the phrase out and cite its source.

InsightRead

A registration batch, also called an exhibit or primary batch, is made to generate the data filed for registration. ICH Q1A(R2), the stability guideline, asks for data on at least three primary batches made by a process that simulates production. Their data underpins the shelf life, so make them on qualified equipment.

InsightRead

Why the registration batch should not be batch one

Dropping the trial batch makes the first batch a receiving site ever produces the one whose stability data goes into the dossier. ICH Q1A(R2) (2003) says the process used for primary stability batches should simulate the production process. Our advice is to run a trial batch first, compare it with the sending unit's data and fix what it shows.

InsightRead

Cpk 1.33: a convention, and what the GMP texts say instead

A Cpk of 1.33 puts the nearer specification limit 4 standard deviations from the process mean. It is a statistical convention: EU GMP Annex 15 (2015 text) and the FDA guidance of 2011 recommend statistical tools and name no index. With 10 batches the confidence interval is wide, so report the batch count and the interval.

InsightRead