Incoming Inspection: Is Pass/Fail OK?

Q: At my company, incoming inspection includes measuring the length of a cable used in our products. The results are recorded as “pass/fail”.

My colleague thinks the actual results (the measurement) should be recorded instead. I can’t recall an actual regulation that requires this, but I suspect it’s a better practice than recording pass/fail.

I’d appreciate any information or advice you could provide.

A: The Quality System Regulation that covers incoming inspection — 21 CFR Part 820.80(b) for Acceptance Activities — does not directly discuss recording of the information, but it does address documenting the results of the activity.

There are also requirements under the CAPA provisions (21 CFR Part 820.110) and FDA’s Guidance for Investigating Out-of-Specification Test Results that imply you always must have the ability to conduct adequate investigations of non-conforming products and product failures.

Furthermore, under FDA’s Quality System Inspection Technique (QSIT), FDA tells its inspectors that:

“The firm should routinely analyze quality data regarding product and quality problems. This analysis should include data and information from all acceptance activities [my emphasis added], complaints, service, and returned product records. Determine if the firm is capturing and analyzing data from acceptance activities relating to component, in-process and finished device testing.”

Here’s why it’s important to capture specific information:

If your specification for cable length has a particular value range — say, for example, 10 cm plus or minus .1 cm — then you should record the actual results. This way, you can better investigate failures that may occur later in production or post-production.

This approach also helps you to verify that the specification you originally developed is indeed appropriate and effective for your products.

If you use different suppliers, you can track which supplier is best meeting your specification.

In addition, if you use different suppliers for the cables, you can track which supplier is best meeting your specification. Are the supplier’s parts at the upper range of the specification or nearer the lower limit most of the time? Do you find a lot of variability? Is there a drift towards just barely passing the specification?

The problem with recording only “pass/fail” at incoming inspection should be evident: such a notation becomes meaningless for the times when you need more detailed information to conduct failure investigations — and also to determine the impact and risks of changing suppliers.

Answered by Janis Olson, Vice President of Regulatory and Quality Services for EduQuest, who served 22 years at the U.S. FDA as an expert field investigator, chaired the PDA task force that developed the Auditing Resource Center, and co-wrote the PDA technical report on computer system supplier auditing. Jan also is the developer and instructor of EduQuest’s training class on Quality Risk Management for FDA/ISO/ICH Compliance.

Share

5 comments for “Incoming Inspection: Is Pass/Fail OK?

  1. David Manalan
    July 20, 2016 at 7:35 pm

    Suppose you discover that your measurement device is out of calibration. You have to review all the acceptance data since the last calibration- but unfortunately it simply says- passed or failed. You cannot decide whether any of these inspections are still acceptable because you have no data. Thus, you have to recall all of it!

    • Martin
      July 21, 2016 at 12:06 pm

      Good point, David — you bring out the critical importance of keeping all measurement and testing equipment calibrated.

      • DAVID MANALAN
        July 21, 2016 at 1:57 pm

        Thanks, but the real point was that if you had the data and discovered that your measurement device was 0.005 inches too high (or low), you could reassess whether the lot still passes your final acceptance criteria. Absent the data, you can’t come to any “accept” conclusion. By the way, it could also be an oven temperature in a process where there’s an operating range.

  2. Dan O'Leary
    September 27, 2016 at 8:55 pm

    There are two questions involved here, and, in my opinion, the response doesn’t address them.

    1) Is there a regulatory requirement to record actual data v. pass/fail?

    2) Is it “better” to record variables vs. attributes data?

    In #1 820.80(b) says, “Acceptance or rejection shall be documented”. Since most incoming inspection is based on sampling plans applied to lots, in practice this means accepting or rejecting the lot. Using attribute sampling plans, a good practice would record the sampling plan, lot size, sample size, and the number of nonconforming products in the sample. There is no requirement to record the actual measurement value. In fact, the QSR preamble says, section 147, “Further, the regulation does not specify quantitative data but simply requires that the results be recorded. FDA believes that it is essential for the manufacturer to maintain records which provide evidence that the product has gone through the defined acceptance activities.”

    Unfortunately, there are about five Warning Letters that cite a nonconformance for failure to record quantitative data. For example, one Warning Letter says, “Specifically, the following deficiencies are noted: Your firm’s inspection reports on file do not include a record of actual measurements for dimensional checks.” I trust that by now FDA has taken corrective action, retrained the offending Investigators, documented training records, and implemented systemic corrective action to ensure Investigators understand QSR and don’t invent “requirements”.

    For #2, there is no basic reason to record variables data when the decision is based on attributes. In addition, attributes are easily analyzed to meet 820.10(a)(1) requirements. The real issue, in my opinion, is the sampling plan used. If the acceptance activities measures values and convert them to attributes, this is a non value added task. A better approach would implement Z1.9, use the data directly, and enjoy the corresponding reduction in sample size and cost.

    • Martin
      September 29, 2016 at 4:50 pm

      Thanks for your comments, Dan. Jan Olson provides this reply:

      FDA expects that the sampling plan, lot size, and sample size be part of the procedure for acceptance activities, and that the actual lot size and number of nonconforming products in the sample be recorded.

      When FDA writes the preambles to the regulations, that is the thinking of FDA at that time. One reason FDA uses cGMP (the QSR is of course the medical device GMPs) is so that current good practices will be used. To see what FDA’s current practice is, it’s illuminating to read Warning Letters. While an FDA-483 is the current thinking of the investigator, the Warning Letters are from the Agency.

      The advantage of writing more than just pass/fail is that companies should, as part of the CAPA system, gather data to be analyzed to determine if the processes are in control or out of control. A record of actual measurements for dimensional checks is necessary to do this trending.

Leave a Reply

Your email address will not be published. Required fields are marked *