Circuit board tests, compilers figure in Vitalograph audit.

Failure to validate the process for making printed circuit boards (PCB), plus the equipment to test the boards, were key problems noted in last May's inspection of Vitalograph, an Ennis, Ireland, maker of spirometers and peak-flow meters.

Center for Devices investigator Kevin Vogel noted in the EIR that Vitalograph's PCB process was not validated, but the observation did not make the 483 in that manner.

Instead, Vogel wrote in the inspection report that the company tested the PCBs with testing equipment and conducted visual examinations as well. He stated that the firm insisted that "because of this 100% inspection, no formal validation process had been completed." Vogel nonetheless insisted that they needed to "control the process in some manner."

The closest observation on this issue in the 483 was that Vitalograph could not produce documentation for PCB test equipment. An Aug. 6 warning letter (Doc. 7926W) noted that the firm "failed to validate and document a manufacturing process," and went on to refer to the unvalidated test equipment, apparently equating validation of test devices with validation of the production process.

In a June 25 response to the 483, Ger Fennel, QA director at the Ennis, facility, seemed to be on the same page when he referred to the comment regarding documents on test equipment validation. Fennel promised that Vitalograph would make every effort to "acquire validation documentation for existing boards, all future boards introduced will have validation documentation."

However, the warning letter deemed Fennel's response inadequate due to lack of documentation.

The remaining validation issues covered software, opening with the observation that Vitalograph had not documented its coding standards and had conducted no white-box testing. The EIR went into no detail on these observations and the warning letter only commented that the firm had not "adequately validated software for its intended use." Code walk-throughs and stress testing were similarly documented.

During the inspection, Vogel inquired as to the validation status of the firm's compiler. According to the EIR, an unnamed Vitalograph employee insisted that "the compiler is not required to be validated." Vogel disagreed and Fennel's letter promised to validate compilers in the future.

An unspecified computer problem triggered an observation that never made it past the EIR. Vogel wrote that he inquired as to why the firm failed to validate software for "boundaries covering the maximum number of records which could be migrated from the firm's software database."

An unnamed Vitalograph employee responded that the boundary had been tested, but that it was not anticipated that any user would "include the large number of patients...as [a] complainant did." Complaint handling was a major facet of this audit. Vogel recommended notifying users of the "maximum number of records the software can handle."

FDA also deemed deficient Vitalograph's procedures for electrostatic discharge reduction, employee training and soldering for process control deviation, according to the 483.

Vitalograph, Ennis, Ireland, 5/31-6/3/99, Doc. 108739M, $8.50 plus retrieval.

RELATED ARTICLE: The Checklist - Vitalograph, foreign audit

* Software and computer validation

* Cleaning validation (water)