Inspections, Compliance, Enforcement, and Criminal Investigations
Guide to Inspections of Medical Device Manufacturers
A determination must be made as to whether there are any instances of release and distribution of lots that failed to meet any specifications.
This could be done by examining device history records and in-process control records for any lots, or portions of lots (including components or raw materials) that have been rejected during either in-process or finished device inspection for failing to meet any or all of the product's specifications. Any distribution of these out-of-specification products should be reported and documented. When the nonconformance was reviewed by a material or engineering review board, evaluate the rationale used to justify continued use of nonconforming raw material, component, in-process product or sub-assembly.
Records must be examined to determine whether any lots that have failed specifications were reworked, and whether this reworking is adequate to assure specifications will be met without affecting the safety or performance of the device.
A high rate of rework would also provide strong evidence that the manufacturing process is not operating in a state-of-control.
A review should be made to determine if all sampling plans for inspection and rework are based on an acceptable statistical rationale and technique. (Mil. Std. ANSI/ASQC, etc.)
Records maintained under 21 CFR 820.90 for nonconforming product and product rework information maintained within device history records should be analyzed for failure trending data.
The inspection of this data for repeat component failures may further connect the problem to a manufacturing step, a work station or to one or more employees.
Determine if the manufacturer has established procedures to "control" nonconforming product. The procedure must "address the identification, documentation, evaluation, segregation, and disposition of nonconforming product." The procedure must also include a "determination of the need for an investigation." The evaluation and any investigation must be documented, including documented explanations for not performing an investigation. Documentation should also include how nonconformances will be trended and/or monitored.
Review all records for the proper disposition of nonconforming products for assurance that use of nonconforming product has not resulted in the distribution of defective devices. The distribution and justification for concessions (allowance to use otherwise nonconforming product, often done through a Material Review Board) must be documented and based on scientific evidence. Concessions should be closely monitored and not become normal practice.
Deficiencies would include a lack of scientific evidence for justification of the concession. If a concession resulted in a change of product specifications (form, fit or function), the change should be evaluated for possible 510(k) submission. At a minimum, a risk analysis should always be considered for any changes.
A determination must be made as to whether the firm's device design and/or process changes are, or may be, contributing to defective devices. There are several ways of making this assessment, for example;
a. Review any changes to the device design and/or process that may have occurred as a result of complaints, servicing/repairs, failure investigations, in-process/finished device inspection, or rework/ reinspection.
Instances of such changes can be determined by examining the change control documentation as well as by revisions to existing procedures, DMRs or deviation reports associated with Device History Records (DHR). If changes have been made to either the device design or the device manufacturing process, determine whether they have been adequately validated, and/or where appropriate, verified.
b. Determine if there are a series of change orders (found by reviewing the DMR and various engineering drawings) for the same device which are intended to sequentially correct the same or similar problems.
This, or complaints received after a design change, may indicate the original change was not effective or was not adequately verified and validated and that subsequent changes have been, or are being, made to correct the problem.
A memo, signed by the Directors of the Office of Regional Operations and Office of Enforcement, dated June 6, 1997, regarding when to cite changes on the FDA 483 for the transition year (June 1, 1997 - May 31, 1998) is included as Attachment B to this guide.
All changes relative to the device design (this includes all changes for software that is part of the device, is the device and some automated design testing software) should be documented on the Design Control Inspectional Strategy (DCIS) Report -see CP 7382.830, Attachment F.