Regulatory Open Forum

I have some concerns about a July 11, 2013 Quality and Compliance article entitled, "Medical Device Corrective and Preventive Actions".

The article says, "Evaluation of a defective condition should include a determination of the necessity for a corrective action based on the percentage of defective product and comparison of the percent defective with the average outgoing quality limit (AOQL)." 

While I agree that the need to take corrective action should be determined in each case, I don't agree with the criterion stated, "a comparison of the percent defective with the average outgoing quality limit (AOQL)".

The AOQL is a characteristic of an (attribute) sampling plan after using rectifying inspection. See ANSI/ASQ Z1.4 for example. In rectifying inspection, the manufacturer submits each lot for inspection to a sampling plan. The lots are classified as accept or reject. Rejected lots are screened 100% and nonconforming units replaced with conforming ones. One can calculate the Average Outgoing Quality as a function of the Average Incoming Quality, i.e., the lot percent defective from the process. This curve will have a maximum, called the AOQL.

The argument is that if the process produces nonconforming material below the AOQL, then the process is acceptable and corrective action, elimination of the cause of nonconforming product, is not necessary. Z1.4 Table XI shows that the AOQL can be quite high. For example, Code Letter E, AQL of 1.0, with a sample size of 13 accept on 0, has an AOQL of 1.9%. The implication is that if the process nonconforming average is as high as 1.9%, then corrective action is not indicated. In addition, the AOQL changes with the sampling plan.

The article says, "A production lot containing an amount of defective product below the AOQL may be released without corrective action required if the risk to patients is acceptable." This seems to imply that a manufacturer could knowing release lots of medical devices with know nonconformities.

For risk, the article cites ISO 14971:2012. Unfortunately, there is no such standard. Presumably the article meant EN ISO 14971:2012, which is an EU variant of the international standard and not applicable outside the EU.