October 14, 1998

ADDRESSES ATTACHED

Dear Sir or Madam,

The purpose of this letter is to request your assistance in relaying to your membership our concerns and expectations regarding the year 2000 (y2k) problem as it affects the pharmaceutical industry. Although much has been written about this subject, we believe its potential impact on pharmaceutical safety, efficacy and availability merits special attention.

Nature of the problem:

When mainframe computers were in the majority and computer memory was at a premium, software frequently represented dates in formats (such as DDMMYY) that used only two digits to represent the year. Date related computations were calculated reliably using this format. For example, if a person was born in 1960 the software might calculate the individual’s age by subtracting the last two digits of the birth year from the last two digits of the current year (e.g., 98-60=38). However, using this method when the current year is 2000 would yield a negative number (00-60=-60), with unpredictable consequences. Date sorting, too, can be erroneous. The years 1965, 1905 and 1966 would, for instance, correctly sort in ascending order as 05, 65 and 66, but adding 2015 to the series would incorrectly yield 05, 15, 65, and 66.

Some computer firmware may also have difficulty with y2k. Basic input and output systems (BIOSes) may, by ignoring the century indicating bit, not accurately read or set some older real-time clock chips. In addition, some firmware reportedly "wraps" back to 1994 from 1999.

Impact on pharmaceutical operations:

The y2k matter can cause a variety of problems in how dates are expressed or computed that could adversely affect automated drug process controls and clinical and non-clinical data integrity. For example, in the context of current good manufacturing practice, erroneous date calculations may result in incorrect expiration dates, equipment maintenance/calibration schedules, and batch performance trend analyses; some computer operated or monitored manufacturing processes may also slip out of control. In the arena of non-clinical and clinical studies the y2k problem may, in the worst case scenario, generate flawed data. In the case of adverse drug experience reporting, the y2k problem could disrupt tracking of 15 day expedited reports and cause erroneous trend analysis. Moreover, time stamps applied to electronic records under the provisions of 21 Code of Federal Regulations Part 11, our electronic records and electronic signatures rule, need to express the year in an accurate and unambiguous manner.

Our expectations:

Y2k is an issue that, if not addressed by your members, could adversely affect the safety and health of the American public. Of special concern are manufacturing processes, which if disrupted by y2k could result in severe shortages of needed pharmaceuticals. We expect the pharmaceutical industry to address the problem as a high priority, to thoroughly assess and test their computer systems and develop appropriate contingency plans, and to complete this task before January 1 of 2000.

At this time, we are not asking firms to make any formal submissions to us regarding their y2k efforts. This information should be available for FDA review, however.

It is important that suppliers to the pharmaceutical industry also have y2k compliant systems because a disruption in the flow of components, packaging materials and equipment, for example, could halt or slow pharmaceutical production, even if a dosage form manufacturer itself has y2k well under control. We therefore urge pharmaceutical producers to work with their suppliers to ensure there will be a minimum of disruption.

Finally, on a more personal note, I appreciate your attention and assistance in making your membership aware of this matter and FDA’s expectations. Because the pharmaceutical industry has proven itself to be highly progressive and innovative, I am confident that it will have little difficulty in managing the y2k issue.

Sincerely,

Janet Woodcock, M.D.
Director
Center for Drug Evaluation and Research

HFD-1
HFD-300
HFD-320 year 2000
p.motise 9/23/98
rev 9/25/98 per CDER/OC
doc id cdrltr.998