Skip To The Main Content
PDA News

PDA News

Inconsistent Expectations Clash with Industry Best Practices for Sterile Products

by Paul Larocque, Acerna Inc | May 27, 2015
When it comes to manufacturing sterile drug product, discrepancies exist in the available regulatory guidances/compendial documents and industry best practices that cause tension in the industry. These divergences often affect my recommendations to clients regarding updating or adding new facilities. As examples, I discuss several of these divergent interpretations as they relate to environmental programs.

When it comes to manufacturing sterile drug product, discrepancies exist in the available regulatory guidances/compendial documents and industry best practices that cause tension in the industry. These divergences often affect my recommendations to clients regarding updating or adding new facilities. As examples, I discuss several of these divergent interpretations as they relate to environmental programs.

Viable Environmental Monitoring

Having spent a year working on a Warning Letter off and on, one message became clear: the USP approach to environmental monitoring recovery rates (1) as applied by some companies is not being accepted by inspectors in the United States and other countries. This is something my clients often struggle with. The problem lies in that USP <1116> uses a relative (percent excursions) approach whereas the U.S. FDA, (2) European authorities under Annex 1 (3), and Health Canada (4) all call for an absolute (recovery) approach. Table 1 shows that the regulators have harmonized on the absolute values. For example, for active air samples in Grade A/ISO 5 areas, the regulators wish to see less than one cfu/m3 generally over all the samples during a shift. Thus, if hourly samples are drawn, seven zeros and one +1 cfu sample meet the regulators’ guideline.

In contrast, USP suggests that less than 1% of samples be positive (in an ISO 5 area)—a relative approach. In this area, if thousands of samples are taken from multiple ISO 5 areas, percentages can mask trouble in a particular room or zone. If the 1% metric is used for each individual zone, however, then 100 of 101 samples in that zone must all be zero to meet the metric, which is perhaps what USP intends but not what all companies do.

USP <1116> goes on to suggest that no individual sample should exceed 15 cfu without investigation. Notwithstanding the rationale put forward in USP <1116>, it is my opinion that upon seeing a +15 cfu/sample result in a Grade A/ISO 5 area during active processing, many inspectors are not likely to let it pass without observation, especially if it is an objectionable organism.

PDA Technical Report No. 13 notes that: “For ISO 5 [Grade A] environments, there is no difference in excursion [relative] rates and recovery [absolute] rates” (5). All major regulators emphasize the absolute approach in another way: they have all harmonized on expectations for media fills (aseptic process simulations). In short, if there is one positive in 10,001 units, an investigation is needed; two positives is a failure. In traditional cleanrooms, the USP approach based on percentages may represent reality, but it is inconsistent with the regulators’ approach. The bar is being raised in favor of isolators and/or robotic designs instead of traditional cleanroom facilities. Isolators and/or robotic designs are the only types of cleanrooms we recommend now to clients planning a new or renovated facility.

Table 1 Active Air Sampling Levels

Alert and Action Levels

Setting Alert and Action levels is also something clients struggle with. Clients who do not adopt meaningful Alert and Action Levels will not be forewarned of a rising level, which often becomes an exponential rise in a short time. Then the company has real—and usually expensive—problems. Many clients adopt the aseptic and sterile processing guidance values, but because most companies’ environments are much better than the guidance levels, they do not normally provide the company with either an “alert” or an “actionable” trend.

For example, if we look at Table 2, some hypothetical raw data are presented. Using this typical data for our calculations, we find that using the two-sigma and three-sigma approach noted by PDA (5), Alert and Action Levels—well below the guidance level—are obtained. Adopting these levels will not result in a flood of deviations for the company adopting them. A three-sigma action level means that only one in a hundred samples will require some action—not a burdensome amount.

Table 2 Setting Alert and Action Levels – nonviables – Grade B/ISO 7

Isolators vs. RABS

I have seen a potential trend of investigators not accepting traditional cleanrooms, even tightening the use of RABS. At an ISPE meeting in 2012, an FDA expert stated: “It is supposed to be a restricted access barrier system [RABS]. That means it is restricted. If it is open, that should be an exceptional occasion,” with documentation on “why it was opened, the extent of the deviation, the time of the deviation, what kind of intervention took place” (6). He also explained that the FDA definition for a closed RABS is that the doors are only open for changeover and cleaning/sanitization; otherwise they are closed throughout the run. If the door is opened, the rest of the batch should be scrubbed.

This appears to be an indication of FDA’s expectations for traditional cleanrooms and RABS filling. It also suggests that the FDA is attempting to lead other regulators and the global industry in the same direction.

Container Closure Integrity and the Sterility Test

The sterility test is still with us because it remains the primary tool for analyzing sterility even though a passing result adds almost no assurance the lot is sterile. Conversely, a failing result must be taken as definitive—short of obvious contamination during the test. The probability of detecting a contaminated batch using the sterility test is expressed by the equation p = n(1 – (1 – c)), where p = probability of detection, c = true fraction contaminated, and n = number of units tested (7).

In a typical lot, we know from media fills that the true fraction contaminated is less than one in 10,000 (c < 0.0001) and the number of units tested is typically 20. Thus, the probability of detecting a contaminated unit in a typical lot is less than 0.002 (0.2%). Whenever a batch release decision hangs on whether the lot is deemed sterile, the result of the sterility test is barely helpful unless it fails. Since the sterility test is so poor, using it as a container closure integrity test during stability studies is unwise and raises a number of challenges. For instance, what does a failure at 24 months mean? Is your product’s package unsound? Must you withdraw product from that batch and recall everything out there? Is it wise to base such important considerations on a test that is outmoded, laborious, time consuming, prone to false positives, and expensive? Clearly, a proper container closure integrity test should be used.

My recommendation to my clients is to make sure their boss, or the Qualified Person in charge of quality, fully understands that this sterility test adds almost no assurance that the lot is actually sterile. Until a reliable test is available, the critical process parameter—ideally a closed isolator—trumps the critical quality attribute, i.e., the sterility test.

And if the client does not have a viable container closure integrity test, I recommend that one be developed and tested before implementing. If not, the client may end up working on an unexpected Warning Letter or receive other bad news from regulators.

In summary, companies should look at their use of percentages in their environmental monitoring program and ensure that, if used, the percentages measure the average trend in one area, and not, for example, all Grade A zones in the building. They should also ensure they have established historically based Alert and Action Levels so there are no surprises. If your company still uses a traditional cleanroom with curtains or Plexiglas shields, consider upgrading to isolators as soon as possible. You and your executive suite should understand the weakness of the sterility test before the next lot is in question for microbiological reasons. If you are using the sterility test as a container closure integrity test during your stability program, develop a more meaningful container closure integrity test and submit it for approval prior to implementation. Do not be lulled into a false sense of compliance by a few good inspections. Those inspectors did not look at everything—and even if they looked at something, their expertise in that subject may be limited.


  1. USP, “USP <1116> Microbiological Control and Monitoring of Aseptic Processing Environments,” USP 35 vol. 1 2012a, 2012: pp. 697-707.
  2. Guidance for Industry: Sterile Drug Products Produced by Aseptic Processing — Current Good Manufacturing Practice, U.S. Food and Drug Administration: September 2004
  3. EudraLex, The Rules Governing Medicinal Products in the European Union, Volume 4: EU Guidelines to Good Manufacturing Practice, Medicinal Products for Human and Veterinary Use, Annex 1, Manufacture of Sterile Medicinal Products, November 2008
  4. Good Manufacturing Practices (GMP) Guidelines - 2009 Edition, Version 2 (GUI-0001), Sterile Products, Health Canada, March 2011
  5. PDA Technical Report No. 13 (Revised): Fundamentals of an Environmental Monitoring Program. PDA: 2014.
  6. Friedman, R. “Barrier Isolation.” Presented at the 2012 ISPE Aseptic Conference, Tampa, FL, February 29
  7. PDA Technical Report No. 30 (Revised): Parametric Release of Pharmaceutical and Medical Device Products Terminally Sterilized by Moist Heat. PDA: 2012.

About the Author

Paul Larocque is the President of Acerna Inc., a biological, pharmaceutical, and medical device consulting firm specializing in aseptic processing and GMP services.