Abbott DD's reversal: VAI in '97, $100 million hit in '99.

Abbott Labs, Abbott Park, IL, Chicago District/CBER

FDA's inspection of Abbott Labs' Diagnostics Division in Abbott Park, IL, was documented on 86 pages in the EIR and was the lynchpin for the $100 million consent decree that the firm entered with FDA in November 1999. The firm responded with a 45-page letter after the inspection, but it did the company little good.

The inspectional team of eight FDAers was led by Mary Concannon and Chad Schmear of the Chicago district. The other investigators were with the Chicago district, Center for Biologics and "Team Biologics" staff.

According to the just-released EIR, the previous comprehensive GMP audit, which was rated VAI, took place in February 1997 followed by an inspection of sponsor/monitor operations in July and August of 1998. The EIR stated that the 1998 sponsor/monitor inspection resulted in a warning letter (Doc. 7049W) issued in November of that year.

The EIR stated that this inspection "identified nine recalls and/or market corrections" of which FDA was previously unaware, and that two additional market corrections were triggered by the inspection.

The firm's manager of manufacturing quality assurance, Marilyn Nyche, evidently told Concannon that the firm does not have an official log of CAPA items. Two members of the firm's quality team showed the FDAers Abbott's documentation for deviation reports, copies of which were retained by a data entry specialist and the originator of the non-conforming material reports (NCMRs).

Concannon detailed five instances in which individual lots of Auszyme conjugate had failed conformance tests and had not been entered into the CAPA system at the time of the inspection. The earliest of these was in August 1997 and the latest in March 1998. The EIR stated that the NCMR paperwork was still in the hepatitis departmental records.

Concannon also documented test failures for seven lots of Auszyme conjugate that did make the CAPA system, but not for periods ranging from 13 to 20 months. This section of the EIR detailed a variety of other such instances.

In his Nov. 25, 1998, letter to FDA, Diagnostics Division Vice President for QA Dr. Robert Doss, Ph.D., cautioned that "approval of the NCMR and its entry into the corrective and preventive action system are independent from material control" and that test failures resulted in the lot in question being placed on hold.

Doss added that the placement of materials on hold status required no authorization or approval of the NCMR and that the lots identified in the first observation of the 483 and EIR "were not kit-packed or distributed."

Doss also pointed out that "failure investigations were ongoing for the 41 NCMRs" referred to in the first observation and that corrective actions for 35 of those had been completed and the files closed.

He commented that the procedures governing the forwarding of NCMRs to the CAPA system had been revised to call for entry into the CAPA system within 14 calendar days. Doss stated that the revision had been issued in November and would be implemented by December of that year.

Despite the firm's efforts, this observation appeared with a number of others under the first entry in the March 17, 1999, warning letter, which cited Abbott for "failure to establish and maintain procedures for implementing corrective and preventive action."

Abbott diverted failed components for reuse

Schmear's review of a number of NCMRs that made the corrective and preventive action (CAPA) system prior to the inspection disclosed that "some products... seemed to have a greater number of non-conformances than other products" and that most of these had been identified as coming from the retrovirus area.

Schmear described these as examples of non-conformance reports that lacked documentation of investigation into the cause of the non-conformance or identification of corrective action.

According to Schmear, the identified corrective action was described in the NCMR as "components to be reconfigured in other master lots per planning." Schmear said Abbott staff stated that "most of the retrovirus assays are older and, at the time they were designed, the specification for each of the components/subs-components was set relatively wide" and that the wide specifications accumulated "as you move up the manufacturing ladder toward finished products."

The EIR noted that Abbott personnel stated that, although each reagent met specifications individually, specification width caused some failures in end lot testing and that the firm expected end lots to flunk between 30-50% of the time.

This section in the EIR added that failed reagents would be reconfigured into different master lots but that failed configurations would not be approved for sale.

In his letter, Doss reiterated Abbott's approach to component/configuration rejection, but promised to revise procedures to call for "determining an expected master lot pass rate using statistical methods, trending of master lot pass rate performance and a review of component test qualification data."

Doss also committed to monitoring the number of NCMRs at the master lot stage and that an investigation would be triggered by pass rates lower than historical levels.

The warning letter did not refer directly to either of the retroviral compounds, the identity of which was not disclosed. The two that were still on the EIR were human T-cell lymphocytes I and II, and Chagas, but did not discernibly refer to any other feature of this observation.

Joan Loreng, a Team Biologics national expert based in Rockville, MD, reviewed Abbott's computer operations and noted that the company had not documented a decision to not send a customer alert regarding a system software defect in the AxSYM immunoassay analyzer. The EIR described the defect as a database defect which "causes incorrect test results to be reported" and stated that the software was written at Abbott's Dallas plant.

Customer complaints regarding the system, which ran back to a redacted date, were still coming in at the time of the inspection and the firm had commenced shipping a fix for the software in November, 1997. The EIR noted that a number of installed machines had not yet received the update and that the problem was not limited to one assay, "but can affect any testing being performed."

The firm received complaints in three categories, including appearing/disappearing results, assay results migrating between patient files and error messages concerning system software. The second of these three was attributed to a file deletion code that only eliminated the patient identification from the file.

Loreng wrote that Abbott concluded that, while the probability of such an event was remote, the severity of the problem involved was "major."

The fix that was implemented in November 1997, and described as software version 3.03, failed to correct problems in machines employing "laboratory information systems" for data collection, prompting another evaluation. At the time of the inspection, the upgrade for machines thus configured had not been written.

Abbott probably won no friends at FDA when investigators discovered that the firm had described the fix to customers as "an enhancement to internal result database management." Software version 3.03 was nonetheless deemed a "mandatory" upgrade and was shipped first to customers who reported problems.

The decision to keep mum about the nature of the upgrade, according to Kevin Billingslea, the manager of the Quality Unit at Abbott, was guided by the low incidence of the problems in question.

In his letter to the agency, Doss insisted that the problem with transient test results occurred only once in four million such operations and was accompanied by the activation of a message indicator light. Doss wrote that the operations manual called for review of logged messages in that event. He nonetheless indicated that such problems in the future would call for management review "to determine the appropriate customer communication strategy."

This observation made the warning letter with no elaboration.

Abbott Labs, Abbott Park, IL, 9/8-11/4/98, Doc. 108825M, $69.50 plus retrieval.