What is the value of documenting in-process rework for easily-detectable issues?

[Mark Meer]

In a recent inspection, the inspector was concerned about inadequate documentation of in-process assembly rework.

My position is that for simple rework to correct easily-detectable issues (for which controls are already established to detect), that creating additional records documenting names, dates, investigation assessment, post-rework testing, etc. is of little value.

Example in question:

• All steps include visual inspections, and the final product is subject to a formal (documented) visual inspection.
• In-process, a sub-unit is found to have an aesthetic blemish on a plastic part. This is documented in a log, and the sub-unit must be "reworked" by simply swapping out the plastic part with an unblemished one. The plastic part is logged as NC and scrapped.

The above example is quite common, and I see little value in documenting the rework, impact assessment, and names/dates, etc. for such routine in-process corrections. In particular, even IF the rework was done improperly (the plastic part was swapped for another blemished one), there are controls in place to detect this in subsequent stages.

In this case, could it be argued that evidence that post-rework units are conforming (though not explicit) is documented by our final device inspection records (in which a complete visual inspection is documented)? As for recording names, dates, and other details of rework for such routine corrections...I don't see any value.

Comments?

 

[PaulJSmith]

In a recent inspection, the inspector was concerned about inadequate documentation of in-process assembly rework.
...I don't see any value.

And that's the point, really. Is there value added by doing this?
If you cannot find any value to it, then don't do it. However, don't just arbitrarily dismiss it because you don't see any value. Ask the people doing the work. Ask the managers for those areas. Ask anyone associated with that process if they would see any value to keeping such records. If it's a contracted part, reread your customer's contract to see if they might require it. Ask their SQEs.

On the flip side of that, you may be missing some very helpful data by not collecting it in some manner.

 

[Marc]

Since the defective plastic part is logged as NC it sounds like you're getting the data you need, at least for this specific situation. Reactions would be to the NC data - If acceptably low, no problem. If it increases a red flag should come up for investigation of why and what to do about it.

<snip> In particular, even IF the rework was done improperly (the plastic part was swapped for another blemished one), there are controls in place to detect this in subsequent stages. <snip>

I'd be OK with it. The only potential failure mode of this "control" is it is a cosmetic visual inspection which are always subjective to some degree.

 

[PaulJSmith]

Mark, you noted that the issue was raised during an inspection. Out of curiosity, was it an in-process inspection? Internal audit? External audit? What was the justification for the concern?

 

[Mark Meer]

Out of curiosity, was it an in-process inspection? Internal audit? External audit? What was the justification for the concern?

It was the US FDA. Not documenting the rework, violates 21 CFR 820.90(b), so I guess any position I take as to the value of the documentation he was looking for is irrelevant.

-----
Somewhat related, and to further the discussion:

I find it frustrating that there are no caveats to regulations with respect to what controls (a) actually affect final product; versus
(b) are just implemented for internal efficiency.

Regulations, in my opinion, should deal only with (a).

Here's another example:
(1) Final (documented) testing prior to release includes a test that the sensor is functional. All devices are subject, and records of acceptance are required to release devices.

(2) Nevertheless, we also test the sensor early in the manufacturing process, before it is installed in a device. This is purely a preliminary check, for our own purposes (to prevent possibility that a finished device is found to have a faulty sensor).

As you can see, (2) is purely an internal, practical check and has no effect on device quality (even if it were not there, devices would be unaffected because of the general controls of the final inspections in (1)).

I don't see much value (given our small shop), in keeping meticulous records for (2). But, regulations say otherwise, so we have to do it anyway... (sigh).