Statistical Review and Cost Reduction Strategy for Final Product Inspection

[MedtechQuality]

I was talking with a friend who works in medical device manufacturing, and he brought up an interesting scenario around sampling plans and cost reduction.

The setup was this:
The company has a documented procedure that requires AQL-based sampling during final inspection. For the past two years, they’ve collected inspection data and found zero defects in the sampled lots. Now there’s a push from management to reduce cost and improve efficiency, so the question becomes:

With two years of clean data, how would you justify reducing or even eliminating that final inspection sampling plan, without compromising quality or compliance?

I would love to gain some traction for the above scenario. Please share your approach/solution/perspective.

Thanks,

 

[John Predmore]

One thought: sampling doesn't have to be all or nothing. There are different acceptance sampling schemes where frequency and sample size are adjusted based on track record. Look up "switching rules" for normal/tightened/reduced inspection.

Another thought: can you be certain your inspection is detecting any defects which might exist? Have you received complaints from any customer{s) of detectives found in your shipments? Best practice is to occasionally challenge your inspection method{s}, with a "red rabbit", for example. If defects are ever discovered and sorted out at earlier inspections steps PRIOR to final inspection, it would be wise to keep some level of final inspection, which becomes a fail-safe to detect failures of earlier inspection/containment or rework loops.

If AQL is defined in the supplier's documented procedure, and there are no overriding customer or regulatory requirements which supersede the supplier's process decisions, then I would say the decision to modify quality control procedures is up to the supplier. A process change such as reducing inspection, should be implemented through your change management process, with appropriate levels of documented justifications, notifications and approvals.

 

[Bev D]

So this is a bit complicated and often flies in the face of ‘generally accepted yet wrong’ - practices.

First John is correct about the need to assure that you are not missing defects AQL inspections tend to be far too small for the low defect rates of most industries that have critical quality requirements like medical, pharma, aerospace, etc. this is why you see a lot of preventive actions and in process inspections like first and last piece and SPC. Final inspection is seen as a last lien of defense that is more a ‘good feeling’. Although at 2 years of no defects found in final inspection it seems that either you weren’t producing defects or your sample size is too small if you have escapes. Hopefully no escapes of defects that could have been caught. This is the real convincing analysis that needs to be done.

if you have no escapes then a summary of the preventive and in-line checks that are used is necessary - are these sufficient and how will you ensure their continued effectiveness? For example have you caught any changes/defects with your in-line checks? Do you need additional preventive and/or in-line checks? This is the second analysis needed.

Remember that ‘no defects’ doesn’t happen thru wishing, hoping and praying…

 

[Ed Panek]

Another thought: can you be certain your inspection is detecting any defects which might exist? Have you received complaints from any customer{s) of detectives found in your shipments?

This is what I would say. Are you receiving no production info because your product is that good or because your screening is that bad? What are customers seeing? Does it align with production screening?>

 

[Kurt Smith]

I am certainly swimming out of my depth here. But with years of no rejections, I would immediately suspect inaccurate reporting by the inspectors. Situations where it is easier to quickly fix something rather than call attention to it.
Is there extra data available besides Pass / Fail that can be checked for trends? Do the people making these decisions have an accurate understanding of what that final inspection process is like? Have the floor staff been included in discussions about findings to rule out issues they know about but haven't shared?
I immediately do not trust any claim of perfect results. In a hypothetical this might be overthinking. But its too big a red flag to even imagine moving forward without checking some things.