Why FDA's OOS Expectations Exist: The Barr Laboratories Case

In short: FDA’s strict rules on out-of-specification (OOS) results trace back to a 1993 US federal court case against Barr Laboratories. The court rejected “testing into compliance”, limited retesting, averaging and outlier tests, and FDA built its inspection approach and later OOS guidance on that ruling. This article explains the background; the step-by-step procedure is in the linked posts.

See also (procedure): FDA OOS Phase I and FDA OOS Phase II. For the UK view, see MHRA Phase Ia and Ib and MHRA Phase II.

The Barr Laboratories case (1993)

FDA’s OOS guidance refers to a decision of the United States District Court for the District of New Jersey in United States v. Barr Laboratories, Inc., et al., Civil Action No. 92-1744, 812 F. Supp. 458 (D.N.J. 1993), decided by Judge Alfred M. Wolin on 4 February 1993.

The United States sued the generic manufacturer Barr Laboratories seeking an injunction for alleged CGMP violations, and a central issue was how the firm’s laboratory handled failing test results. FDA’s 1993 inspection guide, Pharmaceutical Quality Control Labs (7/93), was written around the court’s reasoning and still helps explain why investigators ask the questions they do.

What the court held (as FDA summarises it)

Topic	Position taken in the ruling, as reported by FDA
Terminology	The court spoke of an “out-of-specification” laboratory result, not a “product failure”. An OOS result caused by documented laboratory error is not a product failure.
Retesting	Retesting is acceptable only after a failure investigation has started, must use the same sample (not a different one), and must follow a predetermined plan with a stopping point. It cannot continue indefinitely.
Averaging	Releasing a batch on averaged results can hide variability between individual results. In the case, a content-uniformity batch released on averaged results had to be recalled.
Outlier tests	Not acceptable to reject a failing result in the content-uniformity case. FDA’s inspection guide says they are not for chemical testing and never for statistically based tests such as content uniformity and dissolution.
Resampling	A passing resample result alone cannot invalidate the original OOS result.
Blend sampling	Sample size can influence blend test results; samples should resemble the dosage size.

Source for this table: FDA, Pharmaceutical Quality Control Labs (7/93), which summarises the court’s rulings. We have not read the opinion itself.

“Testing into compliance”

The core idea behind all of this is simple. If a laboratory keeps repeating a test until it passes and then ignores the failing results, the specification means nothing. FDA’s guidance calls this practice of “testing into compliance” unscientific and objectionable under CGMP. Absent an identified laboratory error, the original OOS result stands and must be evaluated with all other data.

From court ruling to FDA guidance

1993: the court decision and FDA’s Pharmaceutical Quality Control Labs inspection guide (7/93).
2006: FDA’s guidance Investigating Out-of-Specification (OOS) Test Results for Pharmaceutical Production, which put the expectations into a formal, structured guidance (Phase I laboratory investigation, Phase II full-scale investigation).
May 2022: Revision 1 of the same guidance. It keeps the structure, still cites the Barr opinion (for outlier testing), and notes that much of the guidance may also help when examining results that are out of trend.

For the main changes introduced in Revision 1, see OOS Phase I: what changed from the 2006 guidance.

Key expectations in the current guidance, in brief:

Limit retests to a number fixed in advance in an SOP, based on sound science, and do not adjust it according to the results.
Averaging is appropriate only where the method itself specifies it; it should not hide variability among individual results.
Decide in advance, in an SOP, whether any outlier test will be used. It should not be used to invalidate a result from a validated chemical test with small variance, because it cannot identify the cause of the extreme value.
All results, passing and failing, should go to the quality unit for evaluation.

Why it matters for your laboratory

Write retest and resample rules into SOPs before a failure occurs, not after.
Do not invalidate a result without an assignable, documented cause.
Keep raw data in controlled notebooks or electronic systems, not on loose paper.
Treat the investigation as a quality-system matter, not just a laboratory task: if the laboratory phase finds no cause, the investigation extends to manufacturing.

Frequently asked questions

Is the Barr opinion part of FDA’s guidance?

No. FDA’s guidance only cites it (for example on outlier testing). The guidance itself is a nonbinding FDA document.

Is the 1993 inspection guide still current?

FDA still lists it as an inspection guide, but the page shows “content current as of 11/13/2014” and the 2022 OOS guidance is the main reference today.

Can we average two retests with the original result?

Not as a way to get a passing result. Averaging is acceptable only when the method specifies it, and it should not hide variability between individual results.

Does FDA’s OOS guidance bind a company?

It is a “nonbinding recommendations” document, but it reflects FDA’s current thinking, and the underlying CGMP regulations (21 CFR Part 211) are legally enforceable.

Do these rules apply in India?

Indian firms that supply the US market are inspected by FDA against this thinking, and the same principles are used in good practice elsewhere.

References

US FDA, Guidance for Industry: Investigating Out-of-Specification (OOS) Test Results for Pharmaceutical Production, Revision 1, May 2022 (supersedes the October 2006 version).
US FDA, Inspection Guide: Pharmaceutical Quality Control Labs (7/93), content current as of 11/13/2014.
United States v. Barr Laboratories, Inc., et al., D.N.J., Civil Action No. 92-1744, Opinion of 4 February 1993, as cited in the FDA guidance (opinion text not reviewed).