Food Protection Trends - May/June 2020 - 209

useful; it is still an important tool for verifying the effectiveness
of a food safety system (for example, by testing for indicator
organisms). However, no sampling plan can guarantee that a
product is pathogen-free.
The effectiveness of a microbiological sampling is also
dependent on distribution of the contaminant within a lot.
Pathogens often are not randomly distributed. Buchanan gave
an example in which pre-harvest sampling of a produce field
used field irrigation patterns (a likely contamination source)
to illustrate how knowledge of a food production or field environment and its potential contamination sources can improve
sampling strategies.
Thiex reminded the audience that sampling also occurs in
the analytical laboratory; lab workers will sample only part of
the material they receive, which will introduce sampling errors.
Sampling error is typically "huge" compared to the error associated with analytical methods, which are usually smaller and
have been well characterized during the method validation.
Thiex discussed three tools for taking and testing "GOOD"
samples: (1) establishing Sample Quality Criteria, (2) characterizing the material properties and (3) applying the Theory of
Sampling developed by the late Pierre Gy to mitigate sampling
error through collection of sufficient mass and sufficient
number of increments. Thiex advocated increased training on
this set of tools, which is the basis for a systematic approach to
representative sampling of food products detailed in two recent
publications: (1) Guidance on Obtaining Defensible Samples
(GOOD Samples) and (2) Guidance on Obtaining Defensible
Test Portions (GOOD Test Portions) (see "Resources and References"). Thiex emphasized that validation of protocols and
incorporation of quality control are as important for sampling
as for analytical methods.
How to set specifications
Katherine M. J. Swanson (retired consultant), discussed
important factors to consider when setting specifications and
action limits for microbiological test results. Specifications
are often part of a purchasing agreement between a buyer and
the supplier of an ingredient or food. They are also internal
limits used to monitor manufacturing processes. As explained
by Swanson, specifications may be mandatory or advisory
and usually cover much more than just food safety (e.g., most
specifications cover parameters affecting quality). Swanson
discussed the "anatomy" of a microbiological specification.
Testing methods should be specified and describe the number,
type, and size of samples and the frequency of testing. For
every tested organism (whether an indicator or pathogen),
there should be an action limit and an actionable outcome
(rejection, process adjustment, recall, etc.) if that limit is
exceeded. As discussed by several speakers, it may be more
useful and informative for a specification to include test results
for indicator organisms to identify trends rather than simply
looking for the presence or absence of pathogens. Indicator
organism testing is not always useful, however. Swanson used

coliforms as an example: They are useful in assessing milk but
not in evaluating fresh produce, on which coliforms are always
naturally present.
How to set action limits. What actions should be taken
when you get a positive?
Swanson stated that credible sources need to be used to set
action limits for specifications and cited various helpful resources (see "Resources and References"). In some cases, regulatory standards may exist. Trade association guidelines and
international standards such as the Commission Regulation
(EC) No 2073/2005 on microbiological criteria for food standards and Codex Alimentarius Commission's Principles and
Guidelines for the Establishment and Application of Microbiological Criteria Related to Foods (CAC/GL 21-1997) may
also be useful tools in identifying appropriate action limits.
The 2011 ICMSF book Microorganisms in Foods 8: Use of Data
for Assessing Process Control and Product Acceptance is another
useful resource for setting specifications. Although FSMA
requires FDA to establish risk-based, technologically-feasible
performance standards which could be useful for setting specifications, such standards have not yet been released. When performing microbiological testing to verify a preventive control,
a "positive" result (one that does not meet your specification)
requires a corrective action. That action should be part of a
corrective action plan that is developed before problems arise.
If the positive result occurs when performing environmental
monitoring, Akins-Lewenthal recommended verifying the impact of the corrective action by obtaining at least three passing
swabbing events in a row within the affected area.
Attorney Maile Hermida of Hogan Lovells discussed the legal dilemma that food manufacturers face: "Should we conduct
microbiological testing, knowing that some false positives will
occur that could be a liability for my company? Or should we
avoid testing even if that makes our company negligent?"
Ignoring or minimizing microbiological problems ("it's an
outlier," "the finished product tested negative," "no one will get
hurt," etc.) does not make them disappear. False positives are
the exception, not the norm. It's important to identify potential
contamination early and respond to it appropriately, including
identifying the root cause. Criminal investigations associated with foodborne disease outbreaks have often focused on
ignored warning signals. Hermida discussed the statutory and
regulatory framework surrounding testing and pointed out that
FSMA significantly expands FDA's access to records during
inspections. Legal privilege cannot be used to hide testing results, and merely copying a lawyer on testing documents does
not protect them from disclosure.
FDA and state regulators are now routinely conducting
"swab-a-thons" in RTE facilities. During an inspection, inspectors may swab 200 to 300 strategic sites within the facility.
New technology such as whole-genome sequencing can allow
regulators to precisely match patient isolates with product
samples or environmental samples (sometimes collected years

May/June Food Protection Trends

209

Food Protection Trends - May/June 2020

Table of Contents for the Digital Edition of Food Protection Trends - May/June 2020