Microbiological Critetia For Food - Usda
Microbiological Critetia For Food - Usda
Microbiological Critetia For Food - Usda
NACMCF Executive Secretariat, * U. S. Department of Agriculture, Food Safety and Inspection Service,
Office of Public Health Science,
Stop 3777, PP3, 9-210B, 1400 Independence Avenue S W, Washington, DC
20250-3700, USA
*Author for correspondence: NACMCF Executive Secretariat. Tel: 202-690-6537; Fax: 202-690-6364; Email:
james.rogers@fsis.usda.gov
Participating agencies include the U. S. Department of Agriculture, Food Safety and Inspection Service; U.S.
Department of Health and Human Services, Food and Drug Administration, and Centers for Disease Control and
Prevention; U.S. Department of Commerce, National Marine Fisheries Service; and U.S. Department of Defense,
Veterinary Service Activity. This article may be reproduced without prior permission.
Disclaimer: Mention of trade names or commercial products in this publication is solely for the
purpose of providing specific information and does not imply recommendation or endorsement by the
U. S. Department of Agriculture and other participating agencies.
TABLE OF CONTENTS
Executive Summary
Recommendations
Introduction: Statement of Charge to NACMCF and the Rationale for the Approach
to Address the Charge
Specific Charge to the Committee
Public Health Focus
Committee’s Approach to Answering the Charge
Scope of Committee’s Work
General
Background: Department of Defense Procurement
Food Categories
Principles Used in Making the Process Flow Diagrams
Interpreting the Process Flow Diagrams
Intended Use of the Process Flow Diagrams
Manufacturing Processes and Opportunities for Loss of Process Controls
Measuring Insanitary Conditions
Sampling and Testing
Use of Statistical Sampling Plans in the Supply Chain
Finished Product Testing to Aid in the Management and Control of Suppliers
Process Control
Statistical Process Control Limits
Process Capability
SPC Monitoring via Microbiological Testing
Considerations for Finished-Product Testing
Sampling Plans for Screening and Auditing Suppliers
Surveillance at Point of Sale
Microbiological Limits and Criteria
Development of Limits and Criteria
Pathogens Important to Public Health
Indicators that Reflect Loss of Process Control or Insanitary Conditions
Comments on Microbiological Limits for Specific Food Categories
Routine and Non-routine Testing
Plan of Action if Limits are Exceeded
Commodity Specific Comments on Microbiological Limits
Other Indicators of Process Control and Sanitary Conditions
Glossary
Appendices
Bibliography
Response to Questions Posed by the Department of Defense Regarding Microbiological
Criteria as Indicators of Process Control or Insanitary Conditions
EXECUTIVE SUMMARY
To assist DOD with its ability to assess suppliers that do not have well-established food safety
plans, the NACMCF Committee (hereafter the Committee) has provided microbiological limits
for food categories that reflect process control and sanitary manufacturing conditions. These
limits are not microbiological criteria for finished products typically found in a product
specification, but are provided to help DOD assess process control and sanitary conditions in
those suppliers without evidence of a documented and functioning food safety plan. Combined
with process flow diagrams of manufacturing processes, the microbiological limits also provide
guidance to DOD auditors when assisting suppliers with corrective and preventive actions taken
when there is evidence of insanitary conditions and lack of process control. The processes for
statistical analyses of microbiological data for DOD and suppliers are provided to optimize the
use of the data in making decisions affecting process control and sanitation. These limits are
based on expert opinion, industry recommendations, and published finished-product
microbiological criteria from global sources.
RECOMMENDATIONS
• DOD should develop and implement a supplier expectations policy and program to address
supplier programs such as crisis management, environmental monitoring, sanitation
effectiveness monitoring, pest control, Good Manufacturing Practices (GMPs), Hazard
Analysis and Critical Control Point (HACCP) systems, preventive maintenance, the use of
statistical process control (SPC), and verification testing, as appropriate to the individual
operation.
• DOD should share the information contained herein with suppliers who do not have
documented and functioning food safety plans to begin the process of having them develop
SPC charts to demonstrate process control and sanitary conditions. These charts should be
based on microbiological limits provided in Appendix J. Suppliers also should examine
trends in the data from the supplier’s Environmental Monitoring Program (EMP) and
1
sanitation effectiveness monitoring program. A timeline for development and use of these
charts should be set.
• DOD should provide a list of expert consultants that can assist suppliers with development
and implementation of the SPC charts, and EMP.
• DOD should develop purchasing specifications that include microbiological criteria for foods
purchased through the Worldwide Directory as well as for those foods purchased outside of
the Directory. These specifications should be set initially based on consultation with
industry experts, shared as draft specifications with the supplier community, and fully
adopted after feedback and data confirm that the specifications, and the microbiological
criteria imbedded therein, ensure safe and wholesome products, and are realistic and
practical.
• DOD should communicate microbiological standards, specifications and guidelines to all
suppliers and brokers.
• DOD should request that suppliers document their acceptance of the standards,
specifications and guidelines in manufacturing food for DOD.
• DOD should require that their suppliers, even if instructed through brokers, use the sampling
plan, specified limits, and analytical methods specified in the microbiological criteria (when
formally developed and implemented), and maintain documentation for audit purposes.
• DOD should require Certificates of Analysis and consider the use of Certificates of
Compliance with each shipment of product received to verify compliance with the specified
microbiological criteria (when formally developed and implemented).
• If there is a third-party intermediary that is involved in the food supply chain, the
intermediary should be required to receive, maintain and transfer the Certificate of Analysis
or Certificate of Compliance with the products.
• Whenever and wherever possible, meat, poultry and processed egg products should be
purchased from countries with United States Department of Agriculture (USDA)-equivalent
inspection programs, and from manufacturing establishments that meet the requirements of
the inspection system. When this is not possible, the manufacturing facility should meet the
requirements specified by USDA for production of meat, poultry and egg products. The
product specification for fresh (unfrozen) raw meat and poultry should include a maximum
time between slaughter and receipt by DOD.
• DOD should leverage the implementation of the Food Safety Modernization Act (FSMA)
legislation and regulations, requiring all suppliers that would be regulated by the Food and
Drug Administration (FDA) to meet statutory and regulatory requirements as mandated by
FSMA and corresponding regulatory rules.
• DOD should use an information technology solution that requires all suppliers to input key
data such as location, contacts, product identification, code dating and traceability program,
significant hazards, audit scores, regulatory actions (e.g., equivalent to recalls, market
withdrawals, non-compliance records), SPC data, and microbiological test data. DOD
should capture appropriate data in a standardized electronic spreadsheet or database.
• The risk of potential foodborne pathogens should be considered not only for fresh-cut and
frozen fruits and vegetables but also for whole or unprocessed fruits and vegetables.
• The risk of potential foodborne pathogens should be considered not only for processed nuts,
spices and herbs but also for unprocessed nuts, spices and herbs.
• DOD should develop procedures to collect appropriate meta-data associated with assay
results. Meta-data are data about the data, such as, methods, sample size, analytical unit,
point of sampling, and the reason the sample was collected.
• DOD should incorporate evaluation of sampling schemes and SPC into audit procedures for
those suppliers using the microbiological limits to assess process control and sanitary
conditions.
2
• DOD should consider enhancing diagnosis and reporting of foodborne illness, and
integrating this information amongst the Services, to help identify potential problems within
the supply chain.
DOD has specific action levels for various microbiological pathogens (e.g., Salmonella, Listeria
monocytogenes, Escherichia coli O157:H7, and Clostridium perfringens) and microbiological
toxins in certain raw and processed meat, poultry, egg products and other products, such as
fresh fruits and vegetables, procured globally for U.S. military personnel (U.S. Army Public
Health Command (USAPHC), Circular 40-1: Worldwide Directory of Sanitarily Approved Food
Establishments for Armed Forces Procurement, 2012; Appendix O, 2013 (27). Hereafter,
USAPHC Circular 40-1 is referred to as the Worldwide Directory. In addition, there are bacteria
that, when present in higher numbers, may indicate that processing conditions did not
adequately prevent bacterial growth or reduce bacterial contamination of the product. DOD has
encountered circumstances where the presence of potential pathogens or the numbers of non-
pathogenic indicator bacteria have generated concerns about the safety and/or wholesomeness
of products. DOD seeks updated microbiological limits to better evaluate process control and
insanitary 1 conditions at the point of production.
The Committee agreed with the need to establish microbiological limits to help assess process
control and sanitary conditions at DOD suppliers that do not have documented and functioning
food safety plans, including HACCP systems. In time, the testing by these suppliers, and to a
lesser extent by DOD, should assist these suppliers to develop functioning food safety plans
and enable the suppliers to meet the microbiological specifications established by DOD. DOD
also expressed interest in the use of criteria such as Staphylococcus aureus and Bacillus
cereus levels in ready-to-eat (RTE) products, mesophilic aerobic plate count (APC) in raw and
RTE products, and other possible indicators (e.g., generic E. coli, coliforms,
Enterobacteriaceae, enterococci and gas-forming anaerobes) for establishing that food was
manufactured with process controls and under sanitary conditions.
Because of the many questions regarding microbiological limits that might indicate poor process
control or insanitary conditions, the Committee was asked for its guidance to clarify the following
issues.
1
The terms insanitary and unsanitary are considered as one and the same in this document.
Insanitary is a word that has been used in regulatory language. In this document insanitary is
used as this term was provided in the charge to the NACMCF Committee.
3
conditions. Describe how the processes and considerations could differ in other regions
of the world where processing conditions may make certain indicators or levels of
indicators more or less appropriate.
At the point of production, how many Staphylococcus aureus, Bacillus cereus, generic
Escherichia coli, coliforms, Enterobacteriaceae, enterococci and/or gas-forming
anaerobes in RTE finished products might indicate: a) a possible process control
problem or insanitary conditions, or b) potentially hazardous product unfit for
distribution? How might the levels and the applicability of these criteria vary between
different RTE products (e.g., processed meat, poultry, egg products, refrigerated
meat/poultry salads, and bagged leafy green salads)?
At the point of production, what level of mesophilic aerobic plate count in RTE finished
products and in non-intact raw meat and poultry products might indicate a possible
process control problem or insanitary conditions? How might these criteria vary between
different RTE products (e.g., processed meat, poultry and egg products, and refrigerated
meat/poultry salads)? How might these criteria vary between different non-intact raw
products (e.g., beef trimmings versus ground product)? How might these levels be
expected to change during the expected shelf-life of the product?
The Committee notes that the microbiological limits reflecting process control and sanitary
conditions requested by DOD should not be misinterpreted as microbiological criteria
(specifications and guidelines) for finished food products. It is important that persons reading
and using this document do not immediately transfer the limits provided herein to
microbiological criteria for foods. Over time, as suppliers without documented and functioning
food safety plans, including HACCP systems, use the microbiological limits to establish that
their processes are in control and that sanitary conditions exist during manufacturing, they can
complement this testing with their development of food safety plans that will demonstrate and
ensure that the products purchased by DOD meet the microbiological criteria for finished food
products. Once such documented and functioning food safety plans are audited by DOD and
found to be effective, testing using the microbiological limits provided herein will be secondary
and useful when there is evidence that there is a lack of process control or sanitary conditions
and investigative actions are undertaken to determine root causes.
4
With the large number of personnel served by DOD, the wide variety of raw, RTE and fresh
foods procured, and the high number of countries, brokers and suppliers, the implications for
failures in the food safety systems are considerable. While insanitary conditions and process
failures can lead to higher numbers of indicator organisms (or classes of microorganisms such
as coliforms or aerobic bacteria detected by APC; hereafter “indicator organisms”), the greater
risks are failures leading to increased prevalence of pathogens in foods.
Verification testing by DOD, while limited in scope and absolute numbers of tests, should
provide feedback to suppliers to improve controls where necessary. DOD inspection and
auditing staff need to be equipped with tools to assist them in their evaluation of suppliers of a
wide array of products. One tool will be process flow diagrams that illustrate points in the
manufacturing process where loss of control or insanitary conditions can lead to introduction or
growth of microbial contamination.
The Committee leveraged the expertise of the Committee members, additional experts and
published literature and finished-product microbiological criteria to assist in developing
microbiological limits indicative of process control and sanitary conditions for food
manufacturing. The Committee prepared process flow diagrams to reflect the major food
categories purchased by DOD and used these diagrams to predict unit operations that would
lead to an increased prevalence of pathogens and levels of indicator organisms, or growth of
contaminants, based on loss of control or insanitary conditions. The diagrams also indicate
where in the process there are lethality steps.
The Committee focused on major food product categories to address the questions posed by
DOD. DOD purchases food products that include what one would find in a retail supermarket.
It was not in the scope of the Committee to recommend finished-product microbiological criteria
(i.e., product specifications and guidelines with levels of microorganisms describing acceptable,
marginally-acceptable and unacceptable products) for the vast array of products. In addition,
some food items purchased by DOD will no doubt fall outside of the major food categories
included by the Committee. DOD will need to work with food safety experts to address any
foods not covered in the major food categories.
The Committee recognized that a food safety program for DOD requires a farm to table
approach; but the charge did not ask for the Committee to address producer food safety
programs, supplier GMPs, broker responsibilities, management of the microbiological data,
information technology to optimize use of supplier testing and DOD verification testing, or food
service operations managed by DOD or their contractors. All of these components affect food
safety and quality of the food purchased and used by DOD and should be included in its
comprehensive food safety plan.
5
The Committee did not address the variability in food manufacturing around the world. The
Committee chose to recommend microbiological limits that reflect traditional processes that are
in control and running under sanitary conditions. The Committee did not address the
consequences for suppliers whose processes are deemed out-of-control or operating with
insanitary conditions. DOD will determine what steps it will take in the event a supplier is unable
to substantiate their process is in control or that sanitary conditions exist for manufacturing.
This report is intended to assist DOD in meeting mission requirements, particularly when
purchasing from suppliers without documented and functioning food safety plans, including
HACCP systems.
In addressing the charge, the Committee did not focus on establishing microbiological criteria as
part of purchasing specifications, which DOD does not currently use. The Committee does
discuss the use of microbiological limits for both assessment of process control and sanitary
conditions, and the use of the limits, when and where appropriate, as the initial step toward
developing microbiological criteria for lot acceptance.
The Committee did not address the programs and systems for delivering microbiological limits
to suppliers, ensuring suppliers implement testing against the limits, reviewing microbiological
data from suppliers, targeting of suppliers that do not test or do not meet the limits, collecting
and managing data on microbiological quality of the products produced for DOD, and for
selecting new suppliers or terminating existing suppliers.
GENERAL
While sampling and testing of food products are tools to verify compliance with preventive and
pre-requisite programs, process control and sanitary conditions, HACCP systems and
microbiological criteria, the results do not guarantee food safety. For all refrigerated and frozen
products, temperature monitoring should be done throughout storage and distribution channels,
as well as at receipt by DOD. Appropriate organoleptic and visual evaluation of the product and
the means of conveyance in which it was delivered should occur. Where possible, continuous
temperature recording documentation associated with the container delivering these products
should be reviewed before accepting the products.
For food products classified under the jurisdiction of FDA inspection, the facilities supplying
DOD should meet all applicable regulatory requirements, including those promulgated under the
authority of the FSMA with regard to preventive controls and product safety. Meat, poultry and
egg products that would be classified under the jurisdiction of the USDA Food Safety and
Inspection Service (FSIS) should meet the regulatory requirements defined by FSIS for the U.S.
and as equivalent for foreign suppliers.
DOD procures food products from all 50 states, U.S. territories, and over 60 countries. These
food products are made available to active duty and reserve service members and to retirees
6
and eligible family members who choose to purchase from on-post facilities. Clearly the ability to
safeguard these food products and ensure high quality is of paramount importance.
The DOD selection and approval process for new suppliers can take three months. In some
situations where foods are required more rapidly, expedited processes are used to approve
suppliers. All purchases of food for the military whether on bases, remote locations, ships, or
through commissaries or other commercial establishments, should occur using the Worldwide
Directory. Most of the purchases occur through the Defense Logistics Agency, but the Defense
Commissary Agency also purchases food products for grocery-type operations. Ship supply
officers will purchase food products for their ship. There are instances where procurement
occurs outside the Worldwide Directory, especially where fresh foods, including meat and
poultry, are purchased. In many instances, these non-standard situations are corrected when
detected; however, ship supply officers are granted more freedom in buying from unapproved
sources. It is noteworthy, and potentially problematic, that fresh fruits and vegetables are
currently exempt from requirements to purchase from approved suppliers.
Based on the food product and a DOD informal risk ranking, approved suppliers are scheduled
for DOD food protection audits on a quarterly, semi-annual, or annual basis. Food protection
audits encompass an establishment's total food safety and food protection systems and
programs. Those facilities receiving a passing score are then listed in the Worldwide Directory.
The audit scores are based on observations, with major and critical defects noted, and different
ramifications on the approval status for each type of finding. Audit documentation is reviewed
first at one of the 20 districts, then at one of the five regions, and finally at the Army Public
Health Command where new or continued approval is granted. If major or critical failures occur,
a corrective action request with a timeframe for completion is made of the supplier. Follow-up is
scheduled at a time reflective of the seriousness of the failure.
DOD evaluates the supplier’s food safety plan, including HACCP system, to help determine
whether the supplier can provide safe and wholesome food products. This evaluation also
includes a review of verification testing data that supports the efficacy of the supplier’s food
safety plan. In instances where a supplier is needed to meet mission requirements, but does
not have a documented and functioning food safety plan, DOD requires an alternative means to
assess the supplier’s processes and sanitary condition of the production environment.
Microbiological testing is one of the tools that help with this assessment. The microbiological
limits provided herein were requested by DOD to provide guidance on what tests are
appropriate for various foods and production processes, and what test results may be indicative
of process control and sanitary conditions.
Many food manufacturing facilities reference microbiological criteria from various entities or
have established their own criteria to monitor the safety and quality of raw or RTE components
used to manufacture finished products. The Codex Alimentarius defines a microbiological
criterion as consisting of the following components (30, 31):
• The purpose of the microbiological criterion (e.g., lot acceptance or process control);
7
• The food, process or food safety control system to which the microbiological criterion
applies;
• The specified point in the food chain where the microbiological criterion applies;
• The microorganism(s) and the reason for its selection;
• The microbiological limits (e.g., m, M, or other action levels);
• A sampling plan defining the number of sample units to be taken (n), the size of the
analytical unit, and where appropriate, the acceptance number (c);
• Depending on its purpose, an indication of the statistical performance of the sampling
plan; and
• Analytical methods and their performance parameters.
DOD has established their own action levels (not-to-exceed limits) for finished products to assist
auditors in their evaluation of various processing systems and finished products. DOD
procurement requires that food products adhere to U.S. regulatory requirements; however, as
mentioned above, exceptions to this requirement may be granted under limited circumstances.
Laboratory analysis forms an integral part of the overall mission of protecting military personnel
and DOD beneficiary populations from foodborne and waterborne (hereafter foodborne will
include waterborne) illness. The DOD program allows for testing of food products and the
environments in which they are produced. Laboratory testing includes qualitative and
quantitative analyses for pathogenic and nonpathogenic bacteria, respectively, as well as
verifying other wholesomeness and quality parameters. Food testing equipment is located
within each DOD deployable veterinary detachment to provide presumptive (considered Level 1
testing by DOD) microbiological testing results, with the staff of each detachment responsible for
animal care, food protection, and review of area facilities that supply food. Testing by a food
manufacturing facility using an accredited laboratory (e.g., ISO 17025) is required for DOD
procurement. Currently, DOD uses microbiological test results in combination with audit findings
to determine the status of an establishment regarding initial and on-going approval, or whether
product that has been procured is safe and wholesome for military personnel.
Appropriate organoleptic evaluation of food products may be useful to assess quality. While
organoleptic examination has its value, it is inherently subjective and dependent upon sensory
capabilities that vary from analyst to analyst. Numbers of indicator bacteria such as APC might
be more effective for determining quality of products that may have been stored for a significant
period of time. However, fresh produce may have appropriate quality for use while also
containing substantial comparatively high concentrations of aerobic bacteria.
Food processors, including those who supply DOD with RTE multi-component products (e.g.,
meals, sandwiches), should be responsible for evaluating individual components (e.g.,
processed meats, cheese, poultry, egg products and spices) received at their establishments.
In many cases, these components may be included as ingredients in the final product without
further processing to inactivate biological hazards. The supplier establishments should perform
microbiological testing on these raw materials, require microbiological test results from the
secondary suppliers on a Certificate of Analysis, or require the listing of microbiological criteria
as elements of a Certificate of Conformance that accompanies the raw materials.
8
A variety of analytes (e.g., aerobic bacteria, E. coli, Enterobacteriaceae, coliforms, enterococci)
currently are monitored on a limited basis by DOD to suggest potential insanitary conditions or
poor process control. This report recommends that this testing should be done by suppliers
without documented and functioning food safety plans, including HACCP systems, using the
microbiological limits provided herein to demonstrate process control and sanitary conditions.
Currently, there is no consensus in the U.S. on acceptable microbiological limits for indicator
bacteria to indicate a process is in control. Such limits may vary by facility, process and food,
and may be best determined through the use of SPC as described herein.
FOOD CATEGORIES
Because of the vast array of food products purchased by DOD, categorization is complex. It is
beyond the scope of this document to list or cover all foods purchased by DOD. The major food
categories and the subcategories covered herein include:
Beverages
Bottled water
Ice, packaged
Juices and drinks, pasteurized, refrigerated
Shelf stable
Dairy
Butter, margarine
Cheese, hard
Cheese, soft, semi-soft, surface ripened
Cultured, pH<4.8
Cultured, pH>4.8 and < 5.4
Dried products (does not include dairy ingredients used to make infant formula)
Frozen desserts
Milk and milk products (fluid)
Processed cheese
Egg Products
Pasteurized, processed
Shell eggs, raw
Grain-based Products
RTE, baked items, refrigerated or temperature/time controlled for safety (TCS)
RTE, baked items, shelf stable or non-TCS
RTE, cereals
RTE, cold pressed bars
Non-RTE, Dry flour-based mixes
Non-RTE, Pasta, dried or refrigerated
9
Meat, Pork, Poultry Products
Non-RTE, beef and pork, raw, intact and non-intact
Non-RTE, poultry, raw
RTE, cooked, perishable
RTE, fermented, dried
Produce
Fruits and vegetables, cut, frozen or refrigerated, minimally processed
Fruits and vegetables, whole
Mushrooms
Packaged salads and leafy greens
Vegetable sprouts
Seafood
Non-RTE, raw
RTE, fish, cold smoked
RTE, cooked or hot smoked
RTE, raw molluscan shellfish
The generic process flow diagrams for these food categories are included (Appendix A) to
identify for DOD auditors the steps in the manufacturing process where microbiological counts
could potentially increase with loss of process control or development of insanitary conditions.
In addition, the flow charts illustrate where there are lethality steps that reduce numbers of
indicator organisms and pathogens.
Steps for receiving and storing packaging materials were omitted to simplify the creation and
use of the process flow diagrams. It is expected that a DOD-approved food processing plant
would have appropriate control and documentation of these functions, either as part of product-
specific preventive controls or HACCP system, or as preventive and pre-requisite programs
such as Standard Operating Procedures (SOPs) for receiving and storage. It was recognized
that a finished food product could move through many storage and distribution facilities as part
of the supply chain. Moreover, it is possible that a finished product of one production system
could be an input for another production system. The final two steps were denoted “store
finished product” and “distribute finished product” to simplify the creation and use of the process
flow diagrams.
For several types of food, there are many different possible combinations of manufacturing
steps. Rather than try to show all multiple combinations and step sequences, the steps that
10
could be used in the relevant portion of the manufacturing process were listed collectively. For
example, in the process flow diagram for yogurt, the “add culture” step also includes the
information “(may be preceded by concentration)” and the “process” step also includes “filter,
heat, separate, concentrate, stir (optional)”. In the coffee process flow diagram the “process
raw coffee cherries” step lists the component steps of a wet method and a dry method to
process the coffee cherries. The Committee assumes that DOD personnel will be able to
recognize the specific steps observed at a food processing plant from among the general
manufacturing steps shown on the process flow diagrams.
The name of a processing step may be followed by any of the following designations:
C, a step at which significant contamination may occur when adequate process controls are not
in place, G, a step in the process where growth of microorganisms can occur, K, a step where
there is a pathogen kill step, and S, a point where sampling and testing by the supplier are
recommended for verification or investigation.
The effectiveness of the expected process controls at preventing contamination may differ
considerably from step-to-step and product-to-product. For example, there would be a greater
likelihood of contamination during the harvesting of coffee cherries than during the packaging of
ground roasted coffee beans. Similarly, less contamination might be expected during yogurt
packaging than during the packaging of raw, non-RTE seafood.
Programs for minimizing contamination at the identified steps include Good Agricultural
Practices (GAPs), Sanitation Standard Operating Procedures (SSOPs), GMPs, SOPs for
specific steps, and purchasing specifications. Steps denoted as potential contamination points
may occur before or after a step causing significant reductions in the numbers of
microorganisms present in the food. For example, there may be a high level of concern about
L. monocytogenes contamination of RTE foods during the “package” step and this step will be
labeled with a “C.”
DOD personnel should use the process flow diagrams to review the general steps to
manufacture the food product under evaluation. From the process flow diagram, DOD
personnel should determine the step(s) at which sampling should be done by the supplier
without a documented and functioning food safety plan to demonstrate process control and
sanitary conditions. When microbiological or organoleptic analyses indicate that any supplier
may have shortcomings in process or sanitation controls, DOD personnel should use the
process flow diagram to determine steps at which contamination could occur or steps at which a
failure to achieve the expected destruction of bacteria may be occurring. It shall be important
that DOD consider that test results or organoleptic assessments for finished products at the
point of use (e.g., commissaries) may not reflect loss of process control or insanitary conditions
at the supplier since factors such as temperature control during storage and distribution can
11
affect microbiological results and organoleptic properties, and should be taken into account
when deriving conclusions about a supplier’s manufacturing processes.
The designation of food categories and subcategories is based on criteria such as the food
description itself, type and extent of processing, RTE status, and chemical characteristics of the
food. For each subcategory a general process flow diagram depicts the manufacturing process
for the foods in that subcategory. If DOD investigates a process following the review of
verification test data or as part of an on-site audit, the process flow diagrams provide insights
into where in the manufacturing process the investigator or auditor could focus their attention.
The Committee believes that the best assessment of insanitary conditions is not necessarily
finished product testing, but is typically best achieved through strategic evaluation of the
production inputs, cleaning and sanitation practices and their efficacy, and the environmental
monitoring and sanitation effectiveness monitoring data generated by the supplier facility as part
of their preventive controls program.
There are various reasons for sampling and testing by DOD itself. While relying primarily on
supplier testing, DOD may sample food products at locations such as distribution centers, field
locations or commissaries to determine the microbiological quality of the food product at a
particular point in the supply chain. The test results from analysis of these samples can provide
insights into supplier compliance with specified microbiological limits; although, as pointed out
above, the results would be affected by the warehousing, distribution and handling processes
and conditions in the supply chain from the time of manufacturing to the point of sampling. For
example, the results can provide indirect information regarding temperature control during
warehousing and its impact on the shelf life of the food product.
DOD also may take samples during supplier audits. If finished products are sampled, these
samples represent verification samples; the test results provide some indication of the ability of
the supplier to manufacture safe and wholesome food products and provide an incentive to
establish and maintain process control and sanitary conditions. The allocation of verification
testing resources should include consideration of the potential presence of biological, chemical
and physical hazards, type of food, supplier characteristics and where the supplier is located,
audit results, shelf life, the distribution system and likelihood of temperature abuse, as well as
the cost of sampling and testing. DOD has an informal risk ranking process that has been used
to define audit frequencies. A more systematic and analytical approach to risk ranking of foods
and suppliers by DOD considering the factors specified above would enhance controls over
food safety and quality, as well as resource allocation.
12
The DOD process of evaluating suppliers with documented and functioning food safety plans
should rely more on the documented evidence supporting effective food safety plans, including
verification testing results (EMP, sanitation effectiveness monitoring, and finished product
testing where appropriate) generated by the suppliers, with the DOD sampling and testing used
only for periodic verification. For those suppliers without documented and functioning food
safety plans, DOD should ensure the suppliers are conducting sufficient sampling and testing to
demonstrate process control and to establish that their manufacturing is occurring under
sanitary conditions, using the guidance provided in this report. When deemed necessary, more
finished product verification testing by the supplier and DOD may be appropriate for these
suppliers until they develop functioning food safety plans.
Currently, DOD, through the USAPHC, maintains the Worldwide Directory but does not stipulate
purchase specifications, such as microbiological criteria including sampling plans,
microbiological limits, and reference methods for specific microorganism-commodity
combinations. This section addressing sampling plans is not intended to provide guidance to
DOD (or any other entity) for elaborating microbiological specifications for foods. Instead, the
aim is to provide some contextual and statistical background for DOD to consider when
evaluating food suppliers, their microbiological data, and the extent to which their manufacturing
process is in control.
In some instances (e.g., immediate need by DOD for a supplier without a documented and
functioning food safety system, rapid development and implementation of HACCP systems and
preventive control programs by a supplier may not be possible in the short term. In such
instances, use of the microbiological limits provided in this report may be useful for suppliers
and DOD to evaluate the food safety and quality performance of the manufacturing process.
Furthermore, analysis of the data may help identify improvement opportunities. The Committee
recommends that a long-term goal be that all approved suppliers develop and implement
effective food safety plans, including HACCP systems, preventive control and prerequisite
programs. In doing so, suppliers and DOD can rely less on the use of the microbiological limits
described herein and finished-product testing, and more on data associated with the food safety
plan that demonstrate the manufacturing process is stable and capable, and sanitary conditions
are maintained continuously.
13
SPC methods are a powerful tool to evaluate process capability and monitor the extent of
control within a manufacturing process. In particular, SPC can be used to identify an out-of-
control process and consequently flag events warranting investigation for an assignable cause,
corrective action and potential preventive action. In this document, we focus on sampling
schemes that allow the use of SPC to assess process control and sanitary conditions,
particularly, but not exclusively, for suppliers without a documented and functioning food safety
plan. Some approaches described herein also may be suitable for a variety of other qualitatively
or quantitatively measurable observations such as those identifying chemical hazards or
physicochemical measurements; but control of these food process characteristics is beyond the
scope of this report.
As mentioned previously, the microbiological limits provided in this report are not microbiological
criteria for finished products; although as data generated for SPC accumulate over time, they
may help define realistic finished-product criteria that reflect wholesomeness, safety, process
control and sanitary conditions. Finished-product testing does have a role for verification that
food is manufactured under sanitary conditions with processes that are under control.
As used herein, finished-products refer broadly to food products or ingredients that have
completed a manufacturing process by a supplier. It does not necessarily imply a RTE product.
For example, beef trim may be considered a finished product from the perspective of a
slaughter plant supplying trim to a customer (e.g., a producer of ground beef). Consequently, a
finished product of one process may be an input of another.
In order to ensure the integrity of its food supply, DOD should assess a supplier’s product as the
output of a process that should be under control and delivers wholesome and safe product. This
assessment is achieved through reviewing data supporting the supplier’s food safety plan,
supplier microbiological test data, surveillance of food products at receiving or in distribution,
monitoring of process control at the supplier, and supplier audits, among other activities. In
what follows, the elements of process control are reviewed, and guidelines are given for
statistically-based activities of surveillance and process control monitoring that help ensure
process control, sanitary conditions and high-quality finished products. It is important to
understand that assessing process control can take many forms including measurement and
documentation of critical processing parameters such as time, temperature and pressure,
documentation of employee compliance to personnel requirements, verification and monitoring
programs for SOPs and SSOPs, and evaluation of microbiological, chemical and physical
characteristics of food before, during and after processing.
Process Control
In simple terms as it relates to food manufacturing, storage and distribution systems, process
control can be defined as maintaining the output of a specific process within a desired range.
Control of a process (or management of a process in general) requires accomplishment of six
basic steps:
14
1. The output of the process must be sampled and quantified on key attributes. Even
limited information (e.g., above or below target) can be used to establish control, if the
sampling rate is high enough. The higher the information content of the measurement
(e.g., enumeration vs. presence/absence), generally the lower the minimum required
sampling rate for control.
2. There must be predefined relevant process control performance limits and targets
traceable to the basic requirements for acceptable outputs (e.g., specifications) and the
history of the process.
3. The actual sample output results must be compared to the relevant process control
limits.
4. There must be a predetermined plan of action (POA, such as a corrective action plan)
based on the size and frequency of deviation from relevant limits. This POA should
include the conditions under which ‘take no action’ is the proper response to a deviation
from control limits. For example, a typical set of POA choices might be: take no action,
move to tightened inspection with increased sampling frequency or sample size, conduct
a pre-determined internal or external audit of the process that is typical for out-of-control
variability, or identify an assignable cause through root-cause analysis and take
corrective and preventive actions. The corrective actions specified must be validated to
ensure they do help to prevent future deviations.
5. The proper action must be decided upon based on the observed deviation.
6. The proper action must be promptly taken to adjust the process. Failure to be prompt is
equivalent to lowering sampling frequency and reduces the ability to control the process.
• Failure to execute any of these steps will obstruct control of the process.
A process is considered under statistical control when its output varies as expected within a
standard operating range (SOR) of variation (Appendix B). This refers to common cause
variation and represents the random variation inherent in a process. When a process becomes
out-of-control, its average shifts, variation increases beyond the SOR, or both. This loss of
control is typically is due to the introduction of a disturbance generated by an assignable cause.
SPC limits bracket the SOR, and indicate the boundary between controlled and out-of-control
operations. The SPC limits may be supplemented by additional statistical rules, such as run
tests (i.e., a rule defining loss of control based on a run of sequential observations, such as
seven measurements over the center line).
15
The first way is difficult to carry out successfully, particularly for microbiological data. The third
method is typical for non-microbiological applications. However, all three may be useful options
for establishing SPC limits in various settings.
There is a trade-off involved in the choice of the quantiles used to establish the SOR. If the
upper control limit (UCL) is too low (or the lower control limit, LCL, is too high), the
corresponding false alarm rate (FAR) will be too high, and will monopolize resources in
performing corrective actions and searching for assignable causes when actually the process is
under statistical control. For example, if the UCL is chosen at the 90th percentile, then 10% of
testing can be expected to result in false alarms. If the chosen percentile is too high, the FAR
will be too low, and the process may drift out of control too far before it is discovered, or the
sampling rate would need to be increased to counteract this effect. Similar arguments apply to
the LCL used, if any.
Typical quantiles used for the upper control limit in SPC are 95%, 99%, 99.7% or 99.9%.
Choice of the quantile is related to FAR, production lots defined in part by time (e.g., hours,
days, or months), and the amount of resources budgeted for dealing with exceptions. Absent
other information, a reasonable rule of thumb might be to use 95% or 99% limits if the sampling
rate is low (e.g., weekly), so there is no more than one or two expected false alarms per year of
production; otherwise it is conventional to use 99.7% or 99.9% limits.
It is important to note that there is a difference between a process being in statistical control and
meeting specifications. A process is considered under statistical control if it is stable over time
and the observed variation is due to common, chance causes inherent to the process (e.g.,
background noise due to normal variation in ambient temperature and humidity) and there is no
between-lot variation. A food manufacturing process being under statistical process control
does not imply its capability with respect to meeting microbiological specifications. The ideal
situation is when a process is both under statistical control and is capable of manufacturing
products that meet specifications. However, a process can be in statistical control and not
capable of satisfying specifications. For example, the process consistently generates
substandard product. Alternatively, a process can be out of statistical control but capable of
satisfying specifications. For example, the process is designed to be robust in regard to
deviations from the norm, such that it meets specifications despite high variability. Given
seasonal and other sources of variability beyond a supplier’s control, the latter situation may be
particularly relevant to food production processes.
Process Capability
Observations that fall within the SPC limits indicate the SOR of production at a facility that is
under control. They indicate the typical range of results on product (in-process or finished
product samples) produced when the process is under control. Specification limits are different
in that they indicate the range of results that indicate company or customer requirements.
The degree by which the SPC limits fall within the specification limits reflects the process
capability to meet specifications when the process is in control. If the process UCL exceeds the
16
upper specification limit (USL) or the LCL is less than the lower specification limit (LSL), a
fraction of the product produced under normal conditions will not meet the specification, even
though the process is in control.
Process capability is traditionally quantified by a Process Capability Index (Cp, Appendix B).
Typically a recommendation for a new process is Cp = 1.45, or for an established process Cp =
1.25. Equivalent nonparametric rules would be that the USL corresponds to the 99.999
percentile for a new process or the 99.99 percentile for an established process. In both
instances, the USL is higher than the UCL.
Microbiological testing presents some unique features not present in other applications where
SPC is used. Unless a chemical or physical surrogate variable is used, microbiological testing
typically results in a discrete count, not a continuous result. The count may be 0 or 1 (i.e.,
presence/absence testing) or a plate count, or the result of a sequence of serial dilutions. A
zero count represents a concentration below the limit of quantification or detection (e.g., <10/ml
or negative in 325 g) for the particular method and test portion size involved.
Because of the discrete count nature of microbiological testing, test results are governed
typically by one or more of three distributions:
Examples of control charts (that illustrate statistical analysis of microbiological test results)
based on DOD data are provided in Appendices C, D, E, F, G and H. In addition, other
distributions that characterize microbiological populations include the Poisson lognormal
distribution. This distribution is a generalization of the Poisson that assumes that the mean
concentration varies log-normally rather than remaining constant throughout the product.
Furthermore, the combination of low prevalence and a range of concentrations when the analyte
is detected results in a zero-inflated distribution that complicates analysis. Zero-inflated refers
to a higher frequency of zero counts than expected under a parametric distribution. For
example, if the microbiological counts in a product follow a simple Poisson distribution with a
mean concentration of 0.04 CFU/g, zero counts in 25 g portions are expected with a frequency
of 37%. If a higher frequency of zero counts is observed, the distribution may be a
heterogeneous mixture in which the microorganism is completely absent from some proportion
17
of the product and present and Poisson-distributed in the remainder. The result would be a
zero-inflated Poisson distribution.
The microbiological limits provided in this report for DOD are useful to establish process control
and sanitary conditions. If suppliers or DOD test finished products, the results may be useful in
assessing the microbiological quality of the product. However, to determine finished-product
acceptability, additional samples may be required (n>1), a three-class plan may be more
appropriate, and microbiological criteria for a food category (and not provided in this report)
shall be required. Considerations for finished-product testing are discussed herein to provide
insights and guidance as the suppliers without a documented and functioning food safety plan
move from establishing process control and sanitary conditions using microbiological limits to
collaborating with DOD to implement microbiological criteria for product acceptance.
Determining the beginning and endpoint of a clearly defined product lot, and delineating it
microbiologically from other lots is critical. A product lot may be defined using a number of
criteria, such as:
The process of defining lots involves thoughtful balancing of various (and sometimes
competing) factors such as sampling costs, the likelihood that a lot is rejected by a customer,
and the cost of lot rejection. The International Organization for Standardization (ISO) observes
that from the point of view of the cost of sampling inspection, there is an advantage in large lots,
provided the same frequency distribution is maintained as lot size increases (12). However,
there are a number of reasons for limiting the lot size including: large lots might result in
inclusion of widely varying quality (i.e., heterogeneity of assignable causes), storage and
handling might preclude the formation of large lots, and the economic consequences of rejecting
or recalling large lots might be unacceptably large. In process control, therefore, there are
tradeoffs between the increased resolution of frequent testing (e.g., every shift or daily) and the
costs of sampling and laboratory analysis. While general rules are available for lot size,
frequency of lot sampling, and number of samples per lot, a sampling scheme can be devised to
optimize control subject to cost constraints (14) .
Lot definition also has implications for SPC when used for assessing the acceptability of a lot.
For purposes of SPC, an important consideration is that a lot is produced under reasonably
constant conditions so that a lot is a homogeneous volume of contemporaneous production.
Statistically, a volume of production is considered homogenous relative to a given characteristic
(e.g., concentration of the microorganism) if the characteristic follows the same probability
distribution throughout the volume (e.g., lognormal with fixed mean µ and fixed standard
deviation σ). It does not mean that the characteristic is the same throughout the volume (2).
18
That is, the conditions result in a homogenous frequency distribution that may or may not
produce a spatially uniform distribution within a lot.
Selection of the appropriate microorganisms when deploying SPC is critical. Typically the best
organisms are either a) those that are predictably present within the sample matrix at some
quantifiable concentration; or b) those that are neither exceptionally rare (i.e., approaching 0%
prevalence) nor ubiquitous (i.e., approaching 100% prevalence) when detected with qualitative
assays. In some instances, microorganisms present at low prevalence may be useful for SPC
(Appendices D and E).
Sampling Frequency
Although DOD currently conducts some sampling and testing during screening, auditing, and
surveillance, to develop fully the use of SPC, suppliers would need to do sampling and testing at
a frequency described above. As such, the supplier needs to have access to a competent
laboratory, have the technical ability to collect the appropriate samples, have the financial
resources to pay for the program, and have the knowledge of SPC necessary to interpret and
use the data.
Even under ideal conditions, a large quantity of data may be required before stable, precise
estimates are obtained for process parameters (e.g., mean, variance, prevalence). Shewhart
(16) cautioned that assignable causes of variation are almost always present in the early stages
of process control and that a long data sequence (e.g., a total sample size not less than 1000)
19
may be required to demonstrate that a process is in statistical control . However, acquiring
additional data is subject to diminishing returns, and requiring a very long sequence of data may
not be economically or technically feasible under operational conditions (Appendix I). For
example, the only suppliers of perishable foodstuffs required to support DOD operations in
austere areas may be small facilities without long production histories. Also, attainment of
process control is often a gradual, stepwise process. Therefore, in practice, a pragmatic
compromise is often warranted. As a general rule, Shewhart suggested a data sequence of not
less than twenty five samples of size four (e.g., sampling 25 lots at 4 samples per lot for a total
of 100 samples) is the minimum requirement for concluding that a process is in a state of
statistical control (16) . Similarly, the ICMSF (ICMSF, 2011) recommends that a minimum of 30
lots should be examined; but cautions that it may be necessary to conduct an initial process
control study for longer periods or in phases.
The first step in screening a new supplier is to have the supplier conduct a self-audit against
DOD supplier expectations (currently a pre-audit checklist). With the self-audit, or upon an
initial visit, DOD should request that the supplier provide microbiological data that demonstrates
that their production process is under control and occurs under sanitary conditions. The
supplier could be asked for verification data supporting its food safety plan, or for those
suppliers without a documented and functioning food safety plan, SPC charts that help to
demonstrate their level of control (although it is unlikely such suppliers will have these charts
and will need to be provided direction, such as that given in this report). If either type of supplier
does not have the information, DOD should consider whether the supplier is willing to begin the
process of demonstrating that their process is under control and is operating under sanitary
conditions by collecting verification data to support their food safety plan or by using the
microbiological limits provided herein to support that their process is in control and production is
occurring under sanitary conditions. Suppliers might be accepted under a probationary status.
During the probationary period, finished product testing may be required to assess the
acceptability of the supplier’s product.
When a potential problem has been identified (e.g., failure to achieve a microbiological criterion,
prematurely spoiled product, or an outbreak of illnesses associated with consumption of a
product), sampling is frequently required to determine the extent and source of the problem. The
ICMSF (ICMSF, 2002) refers to investigational sampling, which includes sampling for this
objective. While the sampling conducted in the course of for-cause auditing would typically
require more extensive sampling than normal sampling, it differs from tightened inspection in
that there are no conventional sampling plans specifically designed for determining the extent of
a problem and identifying the underlying cause. The success of such sampling depends greatly
on knowledge of the process, product, and microorganism. The process flow diagrams
20
presented in Appendix A should be a useful resource for guiding the sampling conducted during
for-cause auditing.
DOD performs intermittent point of sale surveillance of finished products at locations such as at
commissaries. The accumulated data are valuable for various purposes such as assessing not
only the suppliers’ products and processes, but also the potential for contamination or abuse
during transportation, and storage and handling practices throughout the supply chain and at
the commissaries themselves. Various sampling plans are appropriate for surveillance
purposes including that sampling and testing being performed currently by DOD. However,
improvements in standardization of sampling plans and associated meta-data (characterization
of the data and the methods used) are warranted.
ICMSF defines three types of microbiological criteria: standards, specifications, and guidelines.
Standards are mandatory criteria incorporated into a law or ordinance (normally pathogen
oriented). Specifications are part of a purchasing agreement between a buyer and a supplier of
a food and may be advisory or mandatory according to use. Guidelines are advisory criteria
used to inform food operators and others of the microbiological content that can be expected in
a food when best practices are applied (ICMSF, 2002).
Regardless of where food products are manufactured in the world, the finished-product
microbiological criteria indicating safe, wholesome products for DOD would be the same. This
presents challenges for DOD because manufacturers around the world do not have the same
facility design requirements and standards, processing equipment and technology, sampling
and testing programs, regulatory requirements, preventive and pre-requisite programs,
oversight and auditing, customer expectations and food safety culture. Further complicating the
development of microbiological criteria for finished products purchased by DOD is the large
number and variety of products and suppliers.
21
sanitary and insanitary conditions or lack of process control. This requires a site-specific
assessment for each product individually to gain an accurate assessment of these data; this
resource-intensive effort is not commonly done at manufacturing locations. Setting uniform
microbiological limits for process control, while purposeful, may not accurately reflect individual
processes and products within that general category. Thus, the suggested microbiological limits
(Appendix J) described herein should be considered guidance to DOD representing a
provisional starting point for developing empirically based microbiological data and a basis for
discussion of DOD expectations with suppliers that do not have documented and functional food
safety plans.
Microbiological analyses and comparison of the test results to microbiological limits, for the
purpose defined herein, or finished product microbiological criteria, yet to be fully defined by
DOD for the products they purchase, may be used to verify that a supplier’s control programs
for controlling microbiological contamination are effectively designed and implemented. When
there is evidence that the supplier’s controls are poorly designed or implemented, it may be
prudent to increase the frequency of microbiological testing; this testing may include testing
against microbiological limits provided herein, finished product testing, environmental
monitoring, and sanitation effectiveness monitoring. It seems reasonable to expect that
appropriate food safety and quality programs are more likely under the following conditions:
• the food safety regulatory program in the supplier’s country has been deemed equivalent
to its U.S. counterpart,
• the supplier has developed, implemented, and documented appropriate preventive and
pre-requisite food safety programs such as ensuring a safe and properly plumbed water
supply, GAPs, GMPs, and SSOPs,
• the supplier has developed, implemented, and documented a process-oriented risk-
based preventive food safety plan, including a HACCP system, that substantially
complies with risk-based preventive controls regulations authorized by FSMA, and
• the supplier’s food safety system has achieved third-party certification against standards
fulfilling the requirements such as those specified in the Global Food Safety Initiative
Guidance Document.
The Committee considered where pathogens are reasonably likely to occur for each category of
food. The pathogens may have resulted from process control failures (e.g., contaminated raw
materials and ingredients, inadequate processing conditions and insufficient interventions,
failures in pre-requisite programs and preventive programs) or insanitary conditions (e.g., failure
in cleaning and sanitation, inferior facility and equipment design, poor personal hygiene).
Combining these analyses with summaries on the causative agents of foodborne outbreaks
allowed the Committee to prepare the microbiological limits for pathogens for the major food
categories that may reflect loss of process control or insanitary conditions (3).
22
Indicators that Reflect Loss of Process Control or Insanitary Conditions
Indicator organisms typically used to reflect process control or insanitary conditions include
those familiar to food manufacturers, e.g., APC, coliforms, E. coli, Enterobacteriaceae, S.
aureus, pseudomonads, and yeasts and molds. The levels of indicator organisms which
indicate loss of process control or insanitary conditions during processing are dependent upon
factors such as the cleaning and sanitation procedures and products, the types of processes
used, the sanitary design of equipment and the facility, and the food being manufactured.
One of the more difficult microbiological limits to establish to reflect loss of process control or
insanitary conditions is that for Gram-negative bacteria, whether coliforms, fecal coliforms,
Enterobacteriaceae or E. coli. Kornacki and others (13) provide an historical evaluation of these
criteria for foods and their utility based on current knowledge. None of these Gram-negative
bacteria accurately and consistently reflect fecal contamination of raw and processed foods nor
are they useful or reliable as index organisms predicting the presence of pathogens. These
criteria may be useful indicators of insanitary conditions and loss of process control; however
these uses are dependent upon many factors such as the type of food, the extent and type of
processing, the relationship between bacterial numbers and food quality, and the length of time
between production and sampling and testing. Kornacki et al. also reviewed the testing
methods and the many variables that affect the accuracy and utility of the results. For these
reasons, whichever indicator microorganisms are used, they are generally considered
guidelines for use. Based on this current review, in general, the indicator microorganisms of
most value would be Enterobacteriaceae, followed by E. coli, coliforms and fecal coliforms.
DOD is at a disadvantage without data from suppliers defining their normal cleaning and
sanitation practices, and their sanitation effectiveness monitoring program, as well as process
control data measured by manufacturers throughout their production runs. Setting arbitrary
quantitative limits for indicator organisms for a category of food products is guidance at best and
may or may not be reflective of insanitary conditions or lack of process control. For this reason,
the microbiological limits provided herein to DOD should be considered guidelines and a starting
point for suppliers and DOD to evaluate the process controls and sanitary conditions under
which the products were manufactured. The process flow diagrams indicating where bacterial
numbers may increase during manufacturing provide some guidance to DOD on questions to
ask of suppliers regarding where samples are taken, or process control measurements made,
during processing and what corrective actions might be taken based on the results of such
sampling and testing.
23
considered provisional starting points toward more formally designed microbiological limits for
process control that are updated and revised over time as additional data are acquired.
The tables (Appendix J) presented in this document are intended to provide guidance on
microbiological limits, proposed primarily for use by DOD for suppliers without documented and
functioning food safety plans, that reflect effective process controls and sanitary conditions used
to produce food products using good quality ingredients, validated pathogen intervention
strategies and lethality steps, GMPs and GAPs. Microbiological populations in raw commodities
are expected to be higher and more diverse than those in foods produced using a validated
lethality process. The limits identified are on a “per gram” or “per ml” basis and typically assume
a 25 g analytical unit unless otherwise described.
The microbiological limits are intended to help identify when a process is not in control so the
manufacturer can investigate causes and implement corrective actions. The limits reported for
indicator organism testing are not lot acceptance criteria. In some cases, the action to be taken
after exceeding the limit may be to increase sampling to determine the source of contamination
or to test for pathogens or other indicators of insanitary conditions. In cases where any
microorganism or class of indicator organisms exceed regulatory limits, then the lot should be
evaluated appropriately, and typically destroyed or diverted for reconditioning if appropriate. As
an example, the FDA Dairy Compliance Policy Guide 527.300 (23) considers cheese made with
pasteurized milk to be adulterated if the cheese contains 104 CFU/g S. aureus or B. cereus or
100 CFU/g E. coli; these lots should be rejected and additional investigation conducted. If
enterotoxins produced by S. aureus or B. cereus are detected, the product also should be
destroyed.
Assaying for APC to assess process control and sanitary conditions may be relevant for some
RTE foods but not others. APC values used to assess process control and sanitary conditions
during production should be low in RTE foods in which all components of the food have
received a lethality step (e.g., pasteurization, cooking, roasting). When RTE foods contain
some components that have received a lethality step, but then were further handled (e.g.,
sliced, assembled or mixed) before preparation of the final food product, APC levels would be
expected to be moderately higher. In contrast, using APC to assess process control and
sanitary conditions during the production of foods such as fresh fruits and vegetables,
fermented or cultured foods and foods incorporating these, has little value as these foods would
have an inherently high APC because of the normal microbiota present.
24
The presence of E. coli in RTE foods is undesirable because it represents poor hygienic
(insanitary) conditions or inadequate heat treatment (lack of process control). Thus, E. coli
should not be detected in RTE foods; generally, when microbiological specifications are
established, a microbiological limit of <10/g or <3 MPN/g (the limit of detection of usual test
methods) is typical for this microorganism. Levels exceeding 100/g are typically interpreted as a
level of contamination that may be associated with the introduction of pathogens or conditions
that allowed pathogen survival.
The Committee concurs with the common practices for environmental monitoring, i.e., testing for
Listeria spp. in wet, RTE-food processing environments, particularly for foods that support
growth of Listeria, and for Salmonella in dry, RTE-food processing environments. Salmonella
monitoring in warm, wet, RTE-food processing environments also may be appropriate
depending upon the product and facility. If product contact surfaces (Zone 1) are tested,
finished product should be held until results are confirmed negative; if testing demonstrates that
the product contact surfaces are positive for the pathogen, investigational testing in finished
product and corrective action is indicated. As of 2014, the U.S. maintains a standard of non-
detectable L. monocytogenes for all RTE food products. Other countries may allow up to 100
CFU/g for L. monocytogenes in RTE foods that do not support growth (e.g., frozen foods, those
with pH <4.4, water activity (aw) < 0.92, or pH < 5 and aw < 0.94) (5, 6).
All dairy food categories listed below are presumed to be made with pasteurized milk to
eliminate common vegetative bacterial pathogens. Therefore, the presence of any pathogens
when testing for process control or sanitary conditions represents post-process contamination.
Salmonella, E. coli O157:H7 and L. monocytogenes are considered adulterants in RTE dairy
products. In the U.S., these dairy products are either regulated under the PMO Pasteurized
Milk Ordinance (24) or microbiological standards are identified in the Dairy Compliance
Guidelines (23). Other resources for microbiological specifications and guidelines include the
Compendium of Methods for the Microbiological Examination of Foods (Milk and Milk Products
(1) and Standard Methods for the Examination of Dairy Products (29) . Alkaline phosphatase
level in pasteurized fluid bovine milk is limited to less than 2.0 micrograms phenol equivalent per
gram in one or more subsamples whereas cheeses may have higher limits. Actionable limits for
S. aureus and B. cereus are set to 104 CFU/g whereas limits for E. coli or coliforms are product
specific.
The general recommendation for DOD procurement of any beef, pork or poultry product,
whether raw or RTE, is to identify an establishment in the country which is authorized to ship
that product to the U.S. and procure product from that establishment. This will ensure the
establishment meets current FSIS performance standards and/or regulatory requirements. If
such an establishment cannot be identified, the testing recommended in Appendix J may be
used to determine the level of process control and sanitary conditions for establishments not
currently authorized to ship the product to the U.S.
Microbiological testing of finished products that receive a lethality step, such as baking or
cooking, may not be a good indicator of improper storage temperatures and hold times (process
25
controls) of ingredients or blends before the lethality step (such as extended runs between clean
up). Certain ingredients or foods may support microbiological growth and production of heat
stable toxins, such as those produced by S. aureus or B. cereus. Thermal treatments may
inactivate the vegetative cells in the final product but the toxin may remain. As a result, the
process must have validated microbiological control steps throughout the production to minimize
the risk of toxin being present in the finished product.
In setting the microbiological limits to be used by suppliers that do not have documented and
functioning food safety plans, including HACCP systems, the Committee defined the
recommended testing frequency as routine and non-routine. Specific time intervals cannot be
set for each indicator organism, class of indicator organisms, pathogen, environmental
monitoring, or in some instances, chemical hazard (e.g., mycotoxin). The frequency of routine
and non-routine testing will be dependent upon numerous factors such as the production
process, the product being produced, the sanitary design of the facility and the equipment used
at the facility, the historical data generated by the supplier, the organism or class of indicator
organisms, and the investigative reason for testing. General guidance on the definition of these
frequencies is as follows.
The microbiological limits provided in Appendix J are useful to assess process control and
insanitary conditions. The action taken by a supplier if indicator organisms in samples taken at
the supplier location exceed the specified limits should be to investigate the cause of the high
counts, implement corrective and preventive actions, and reevaluate the effectiveness of the
actions after implementation. In the cases of a pathogen detected when there has been no
additional lethality step, an evaluation of the finished product associated with the sample tested
should occur to determine if the product should be rejected or, if appropriate, reworked or
diverted for processing that will inactivate the pathogen. Products contaminated with heat-
stable toxins typically will be destroyed as reconditioning likely will not eliminate the hazard.
26
If levels of indicator bacteria in samples assayed during distribution or at the point of sale
exceed the limits provided in Appendix J, a more thorough investigation should be taken by
DOD and the supplier to identify the cause of the higher counts. The investigation should note if
the food was at the end of the marked shelf-life, is considered perishable, if the packaging was
intact, and if the chill-chain was maintained during storage and distribution. Growth of spoilage
microbes is expected to occur during extended storage of perishable items. The higher counts
may have resulted from normal growth of spoilage microorganisms or temperature abuse rather
than the lack of process control or sanitary conditions during manufacture.
Beverages – Bottled water (artisan, mineral, purified, sparkling, spring) – Appendix A, Flow
Diagram A.1, Appendix J, Table J.1
The Committee recommends routine coliform testing for bottled water and ice to assess process
control and sanitary conditions. In countries where additional microbiological regulations apply,
testing for those organisms may be done periodically. A 2013 WHO Draft Report on regulations
and standards for drinking water quality recommends routine testing for E. coli or thermotolerant
coliforms to provide evidence that these microorganisms are undetectable in a 100-ml sample
(WHO, 2013). Other indicators also were reviewed in the WHO Draft Report and the following
recommendations were made. The presence of total coliforms immediately after treatment
indicates inadequate treatment. C. perfringens (undetectable in 100 ml) can be used an
indicator of the effectiveness of filtration process to eliminate enteric viruses or protozoan
oocysts (WHO, 2013). Enterococci (undetectable in 100 ml) may survive longer than E coli and
can be used as an indicator instead of E. coli. Total heterotrophic bacteria (limit of 100 CFU/ml
at 22 or 20 CFU/ml at 37°C) can be used for operational monitoring of treatment and
disinfection and assessing cleanliness of the distribution system. Pseudomonas aeruginosa,
parasites and enteric viruses were not considered in the WHO report; although they may be
required by individual country regulations.
Beverages – Ice, packaged – Appendix A, Flow Diagram A.2, Appendix J, Table J.2
Microbiological testing and limits will be similar to those for bottled water. In countries where
additional microbiological regulations apply, periodic testing for the organisms listed in those
regulations is appropriate.
Beverages – Juices and drinks, pasteurized, refrigerated – Appendix A, Flow Diagram A.3,
Appendix J, Table J.3
The Committee recommends routine coliform testing for process control purposes. Fruit juices
in the U.S. are subject to FDA regulations mandating HACCP and achievement of lethality
against pathogens of significance (E. coli O157:H7, Salmonella spp.); thus, periodic testing for
pathogens may be indicated (28) . This category also includes low acid drinks such as bottled
coffees, teas, and vegetable juices. For low-acid juices and drinks, the food safety plan should
address the control of pathogenic sporeformers, such as C. botulinum. For products that
27
support the growth of pathogenic sporeformers and where cold-chain management cannot be
guaranteed, alternative safety measures could be the inclusion of ingredients that inhibit growth
(e.g., blending with acidic juice to reduce pH) or alternative processing such as ultra-high
temperature processing to destroy spores. High levels of patulin can be produced in decaying or
moldy apples, and thermal processing does not destroy the mycotoxin. Therefore, apple juice
products should be tested for patulin (21) (U. S. Department of Health and Human Services,
2005)
Beverages – Shelf stable – Appendix A, Flow Diagram A.4, Appendix J, Table J.4
Dairy – Butter, margarine – Appendix A, Flow Diagram A.5, Appendix J, Table J.5
Although whipped butter held under unrefrigerated conditions has been associated with
outbreaks of S. aureus intoxication, the low moisture and high salt content, or lactic acid levels
of many of these products, generally preclude microbiological growth. However, routine
monitoring of sanitation and process control using indicators such as coliforms should be done.
Products containing added seasonings, herbs, or spices may have additional testing
requirements as the inclusion of unsafe adjunct ingredients has been linked to foodborne
illness. Testing for S. aureus, Enterobacteriaceae, and yeast and molds is useful under special
circumstances, such as the investigation of out-of-specification results. Due to listeriosis
outbreaks linked to contaminated butter, routine environmental testing of Zone 2 and 3 surfaces
for Listeria spp. should be done. Although not routinely tested, if Zone 1 environmental samples
are found to be positive for Listeria spp., investigational testing of finished product should be
undertaken.
Dairy – Cheese (hard) – Appendix A, Flow Diagram A.6, Appendix J, Table J.6
Although reported cases of foodborne illness have been linked to foods in this category,
microbiological safety issues in hard cheeses made with pasteurized milk and active starter
cultures are extremely rare. The presence of active cultures in these products makes the use of
routine microbiological testing for APC impractical as a tool for evaluation of process controls
and sanitary conditions. In contrast, routine testing for coliforms as an indication of sanitary
conditions should be conducted. Testing for S. aureus or E. coli is useful under special
circumstances such as validation, verification and investigation when production has occurred
without adequate process control. Finally, routine environmental testing of the food production
28
environment for the presence of Listeria spp. is recommended as a verification step for
sanitation programs.
Dairy – Cheese (soft, semi-soft, surface ripened) – Appendix A, Flow Diagram A.7, Appendix J,
Table J.7
This category represents a broad range of cheeses. Routine environmental monitoring for
Listeria spp. in the environment and coliforms in finished product should occur for all products in
this category. For products in this category which support the growth of L. monocytogenes and
have been implicated in illness such as soft cheeses with high pH values, in-plant monitoring for
this pathogen may be appropriate (15) . Testing for S. aureus and E. coli may be used when
processing or insanitary conditions indicate a potential increased microbiological risk.
Dairy – Cultured, pH<4.8 – Appendix A, Flow Diagrams A.8a and 8b, Appendix J, Table J.8
Rapid acidification and low final pH of these products precludes growth of bacterial pathogens.
The presence of active cultures in cultured dairy products make the use of most routine
microbiological testing impractical as a tool for evaluation of process controls and sanitary
conditions. Routine testing by suppliers for coliforms is recommended to assure compliance
with pertinent U.S. regulations and guidance (24). Non-routine testing for S. aureus is advisable
under limited conditions such as evaluating the impact of a slow fermentation processes. Mold
and yeast testing may be applicable when producing cultured products without mold inhibitors or
when products contain inclusions such as fruit puree that are known to carry spores. Finally,
routine environmental testing of the food production environment for the presence of Listeria
spp. is recommended as a verification step for sanitation programs.
Dairy – Cultured, pH>4.8 and < 5.4 – Appendix A, Flow Diagram A.9, Appendix J, Table J.9
The active starter culture and acid content present in these fermented products reduces the
growth rate of bacterial pathogens; but because the pH is higher than the aforementioned
cultured products with pH <4.8, prevention of post-pasteurization contamination is more critical.
The presence of active cultures in these products makes the use of most routine microbiological
testing impractical as a tool for evaluation of process controls or insanitary conditions.
However, routine testing by suppliers for coliforms is recommended to assure compliance with
pertinent US regulations and guidance (24) and routine environmental testing of the food
production environment for the presence of Listeria spp. is recommended as a verification step
for sanitation programs. Although typically not done, if Zone 1 environmental samples are
positive for Listeria spp., finished product testing for L. monocytogenes should occur. Testing for
S. aureus, psychrotrophic microorganisms, yeasts, and molds is useful under the special
circumstances described above for Dairy – Cultured, pH<4.8, when investigating results
exceeding microbiological limits, or during validation and verification efforts.
Dairy – Dried products (does not include dairy ingredients used to make infant formula) –
Appendix A, Flow Diagram A.10, Appendix J, Table J.10
29
The low moisture content of dried dairy product precludes microbiological growth. However,
routine monitoring of sanitation using coliforms and APC should occur. Furthermore, routine
testing for Salmonella by suppliers should occur as these products have been implicated in
cases of salmonellosis. Non-routine testing for S. aureus and B. cereus should be done under
special circumstances such as during investigation of possible mishandling prior to drying,
validation or verification efforts, or an investigation done in response to results indicative of
process failures or insanitary conditions.
Dairy – Frozen desserts, Appendix A, Flow Diagram A.11, Appendix J, Table J.11
Dairy ingredients used in a dessert mix are pasteurized and will have low microbiological
counts; frozen storage will control microbiological growth. Routine testing for coliforms by
suppliers should occur to establish process control and monitor sanitation. Although APC can
be used to monitor process control, inclusions, such as nuts, cookie dough and fruits, may result
in higher populations than the base mix. Periodic testing for Salmonella may be indicated under
special circumstances such as when lack of process control is suspected, the supplier is using
inclusions which have been previously associated with outbreaks, or during validation or
verification efforts.
Dairy – Milk and milk products (fluid) – Appendix A, Flow Diagram A.12, Appendix J, Table J.12
Fluid milk in the U.S. is produced under the PMO (24) which provides microbiological limits;
when done, such as when there is a pasteurization issue, alkaline phosphatase must be <2.0
micrograms phenol equivalent per gram as an indicator of adequate pasteurization. Routine
testing of APC and coliforms by suppliers should occur to ensure regulatory compliance, to help
establish process control, and to assist with evaluating sanitary conditions. Routine
environmental monitoring of Zone 2 and 3 surfaces for Listeria spp. is recommended.
Dairy – Processed Cheese – Appendix A, Flow Diagram A.13, Appendix J, Table J.13
This product is manufactured by heating cheese with water, emulsifier and other ingredients to
kill vegetative pathogens; molten cheese may then be hot-filled into loaves or blocks and chilled
and cut into individual slices for use; these cheeses are intended to be stored refrigerated.
Shelf-stable hot-filled cheese spreads or cheese sauces must be formulated for safety to inhibit
Clostridium botulinum. Cooling process cheese on casting belts or chill rolls may involve a
relatively high degree of environmental exposure of the product. The presence of non-
sporeforming microorganisms is indicative of post-process environmental contamination. Low
levels of such contamination are inevitable in these cases. Consequently, process cheese
producing facilities need to have robust environmental sampling and control plans for Listeria
spp. and Salmonella spp. Formulae with low levels of salt in the moisture phase could
potentially allow growth of enterotoxin producing Staphylococcus spp., principally S. aureus;
likely originating from human contact. The presence of generic E. coli on process cheese is
reflective of production in an insanitary environment.
Egg Products – Pasteurized, processed – Appendix A, Flow Diagram A.14, Appendix J, Table
J.14
30
Pasteurized egg products and pasteurized shell eggs receive a lethality treatment during
processing and may be used in dishes which are uncooked or lightly cooked. These products
may be recontaminated during packaging, handling and storage. These products should be
tested by suppliers routinely for S. aureus, coliforms, APC and Salmonella to verify process
control. Periodically, suppliers may test these products for B. cereus and Enterobacteriaceae.
Routine environmental testing for Listeria spp. and Salmonella is useful to evaluate sanitary
conditions. If samples exceed the microbiological limits, further investigation and correction
action should occur. Environmental monitoring of Zone 2 and 3 surfaces for Listeria spp. is
recommended; if Listeria spp. are found, it may lead to testing of Zone 1 surfaces for Listeria
spp. Finished product testing should occur for L. monocytogenes if Listeria spp. are detected
on Zone 1 surfaces (indicative of insanitary conditions) or suspected illnesses are reported.
Egg Products – Shell eggs, raw – Appendix A, Flow Diagram A.15, Appendix J, Table J.15
Raw shell eggs are not pasteurized and are not intended for consumption without an additional
lethality step, such as thorough cooking. Regulations in the U.S. require that high-volume
producers (>50,000 laying hens) test for Salmonella serotype Enteritidis to verify non-detection
of this pathogen in the shell eggs (22). High-volume producers supplying shell eggs to DOD
should test for S. Enteritidis. For other producers, the Committee recommends only periodic or
investigational testing of raw shell eggs and no microbiological limits are provided. Testing for
E. coli, coliforms or Enterobacteriaceae by suppliers may be useful to assess sanitary
conditions or establish process control.
Grain-based Products – RTE, baked items, refrigerated or temperature/time controlled for safety
(TCS) – Appendix A, Flow Diagram A.16, Appendix J, Table J.16
These products are prepared with a lethality step to eliminate pathogens; but the potential of
recontamination during handling and the pH-aw range (that can support microbiological growth
during extended out-of-refrigeration storage) warrants microbiological testing. Routine
monitoring of coliforms by suppliers should assess insanitary conditions (including post-process
contamination). APC testing should not be conducted if the products include ingredients which
are prepared using starter cultures (e.g., cheese, salami).
Grain-based Products – RTE, baked items, shelf stable or non-TCS – Appendix A, Flow
Diagram A.17, Appendix J, Table J.17
When manufacturing these products, the dough or batter goes through a baking step which
provides lethality against pathogens and pathogen growth is unlikely during storage due to
reduced water activity. While routine microbiological testing by suppliers generally is
unnecessary, environmental monitoring and in-process sample testing may be appropriate
under special circumstances that may increase the microbiological risk (e.g., excessive water
due to condensate or roof leaks) or when ingredients are added after the lethality step (e.g.,
dusting of bread surface with flour).
31
Grain-based Products – RTE, cereals – Appendix A, Flow Diagram A.18, Appendix J, Table
J.18
RTE cereals are made from grains that go through a lethality step sufficient to eliminate
pathogens of concern. Mycotoxin surveillance testing should be completed on incoming grains
to ensure the grains meet the individual country’s regulations. These RTE grain-based products
do not support the growth of microorganisms due to the very low aw. Routine microbiological
testing of finished product by suppliers is not recommended; but routine environmental testing
for Salmonella is useful to assess sanitary conditions. Non-routine testing for coliforms,
Enterobacteriaceae, APC and Salmonella by suppliers is appropriate for verification purposes,
qualifying lines, or when events occur during processing that may increase the microbiological
risk (e.g., excessive water due to condensate or roof leaks). If vitamin-containing or other such
solutions are sprayed atop cereals after heat-processing, and depending on the source and
processing of these solutions, sampling and testing of these solutions may be a useful measure
of process control.
Grain-based Products – RTE, cold pressed bars – Appendix A, Flow Diagram A.19, Appendix J,
Table J.19
Cold-pressed bars are made from cooked grains, carbohydrate-based binders, and inclusions
such as fruit, nuts and chocolate. Verification of the microbiological quality of ingredients used
in the cold-pressed bar formula is important since the bars will not receive a validated lethality
step during manufacturing. Recommendations for finished product and environmental testing by
suppliers are the same as for RTE cereals above.
Grain-based Products – Non-RTE, dry, flour-based mixes – Appendix A, Flow Diagram A.20,
Appendix J, Table J.20
These Non-RTE grain-based products harbor a complex and extensive microbiota and routine
microbiological testing by suppliers does not provide useful data to indicate process control and
sanitation (17). Flour is a minimally-processed commodity that is ground and sifted without any
lethality step. These products should receive a lethality step to eliminate pathogens before
consumption.
Pasta is produced by combining flour and water and sometimes other minor ingredients. The
microbiological profile may be similar to that of flour and routine testing by suppliers is not
particularly useful. However, the manufacturing process must be controlled to minimize
proliferation of naturally occurring microbiota after the introduction of moisture. Non-routine
testing of in-process samples by suppliers may be useful in special circumstances (e.g.,
evaluation of potential growth and enterotoxin production by S. aureus during extended down
time prior to drying or refrigeration). Although most of these products are intended to be cooked
by consumers before consumption, some varieties, such as instant noodles, may be prepared
with limited heating. Cooking of refrigerated pasta filled with meat or cheese may be sufficient to
32
cook the outer pasta, but not sufficient to provide a validated lethality step in the product interior.
Verification testing of raw materials (to support the Certificate of Analysis) and periodic testing of
product by suppliers for Salmonella may be appropriate; and environmental testing for Listeria
spp. or Salmonella should occur to verify sanitary conditions.
Meals and Entrees – Non-RTE, Ready-To-Cook (RTC) meals, includes raw ingredients –
Appendix A, Flow Diagram A.22, Appendix J, Table J.22
This category includes a wide range of multi-component (some raw), frozen or refrigerated food
products which are expected to be cooked by the consumer or food service operation. Routine
testing of these meals is not recommended; however manufacturers should be aware of the
following points. Suppliers should assess the pathogens and indicator organisms associated
with their products and sample and test if there is a reason to do so. Some of these meals and
entrees may be improperly prepared by the consumer using conventional or microwave ovens
and not undergo a validated lethality step. Pathogens of concern may vary depending on the
specific food. For example, meals prepared with cooked rice may pose a greater risk for B.
cereus; E. coli O157:H7 may be of concern for foods including raw, non-intact beef, and poultry
products may contain Salmonella. Histamine testing may be appropriate when scombroid
species are present.
Meals and Entrees – RTE, deli salads, sandwiches, heat-eat meals, sushi – Appendix A, Flow
Diagram A.23, Appendix J, Table J.23
This category includes a wide range of multi-component, short shelf-life, refrigerated food
products. They are expected to have diverse microbiological populations depending on the
ingredients used, may include ingredients which are raw, such as fresh produce, and are
frequently subjected to multiple handling steps which can introduce contamination. Routine
testing by suppliers of in-process or finished products for E. coli and environmental testing for
Listeria spp. and in some instances, Salmonella spp., should occur to assess process control
and sanitary conditions. As with the non-RTE, RTC meals, other non-routine testing of indicator
organisms and pathogens may be appropriate depending on the ingredients used and the type
of finished product. Although not routinely done, if Listeria spp. is found in Zone 1 environmental
samples, investigational testing for L. monocytogenes may be indicated.
Meals and Entrees – RTE sous vide, cook and chill – Appendix A, Flow Diagram A.24,
Appendix J, Table J.24
Sous vide products are prepared with raw or partially cooked foods, which are vacuum
packaged in an impermeable bag, cooked in the bag, rapidly chilled, and refrigerated with time-
temperature combinations that inhibit pathogen growth. If the cook process does not provide at
least a validated 6-log10 reduction of non-proteolytic C. botulinum spores (7), validation data
should be provided by the supplier to demonstrate that the process eliminates vegetative
pathogens. Because of the lack of inhibitory barriers in typical sous-vide products and the
concern for potential outgrowth of botulinum spores, strict adherence to refrigerated storage
after treatment is extremely important. If a validated cook step is used and verified, no routine
testing is recommended. In the absence of a validated cook process, testing for vegetative
33
microorganisms should be done by the supplier on post-cook samples to verify the thermal
process. Testing for E. coli can serve as a verification of thermal processing; periodic testing of
coliforms, Enterobacteriaceae and APC are useful for verification purposes. If cooling deviates
from prescribed requirements such as those given in USDA Appendix B (18) , testing for C.
perfringens may be useful as a part of the supporting documentation for safety. Routine testing
for C. perfringens typically is not done.
Meat, Pork, Poultry Products –Non-RTE, beef and pork, raw, intact and non-intact – Appendix
A, Flow Diagram A.25, Appendix J, Table J.25
These products include both intact (e.g., non-tenderized steaks, chops) and non-intact (e.g.,
whole muscle destined for ground product, trim, ground product, needle-tenderized steaks) raw
beef and pork products. Under normal operating conditions, no routine testing is recommended.
When it is necessary to meet a regulatory or customer requirement to confirm production is
occurring with process control and sanitary conditions, suppliers should test for E. coli (typical
for the U.S.) or Enterobacteriaceae (typical for the European Union). Those manufacturers
supplying DOD with non-intact product should request that their suppliers (secondary suppliers)
provide a Certificate of Analysis demonstrating that the raw materials have tested negative for
E. coli O157:H7 and other STEC, if appropriate. Suppliers to DOD also may test for Salmonella
to meet regulatory requirements or to provide evidence that they are meeting performance
standards that indicate production has occurred under sanitary conditions; this testing may
typically be done only for ground products.
Meat, Pork, Poultry Products –Non-RTE, poultry, raw – Appendix A, Flow Diagram A.26,
Appendix J, Table J.26
These products include both intact (e.g., non-injected whole birds, non-injected parts) and non-
intact (e.g., injected or “enhanced” or vacuum-tumbled poultry parts, ground poultry) raw poultry
products. Under normal operating conditions, no routine testing is recommended. Production of
these foods should include appropriate process controls to reduce pathogens to acceptable
levels and to prevent pathogen growth. When it is necessary to meet a regulatory or customer
requirement to confirm production is occurring with process controls and sanitary conditions, or
under specific circumstances when an investigation is underway, suppliers may test for
Salmonella and Campylobacter to verify process control and that pathogens are being reduced
to acceptable levels. In this case, testing should be performed on the relevant product type
such as raw poultry parts if they are the product type purchased. Testing for indicator
organisms or classes of organisms such as generic E. coli, coliforms, Enterobacteriaceae, or
APC, could provide additional information regarding maintenance of process control and
sanitary conditions.
Meat, Pork, Poultry Products – RTE, cooked, perishable – Appendix A, Flow Diagram A.27,
Appendix J, Table J.27
This group includes a spectrum of cooked beef, pork and poultry products which require strict
refrigeration for shelf life and safety (e.g., deli meats, hot dogs). While process control is often
monitored through routine testing of E. coli, potential contamination of L. monocytogenes is a
34
major concern and should be addressed by the supplier through routine environmental
monitoring of Zone 2 and 3 surfaces for Listeria spp. Although not routinely tested, if Zone 1
environmental samples are positive, finished product testing for L. monocytogenes may be
indicated. Non-routine testing of coliforms or Enterobacteriaceae, APC, Salmonella, and C.
perfringens may be useful for additional verification of sanitary conditions, adequate cooling, or
as periodic verification of process control.
Meat, Pork, and Poultry Products – RTE, fermented, dried – Appendix A, Flow Diagram A.28,
Appendix J, Table J.28
These products (e.g., jerky, dried pepperoni, meat sticks) are characterized by having
chemical/physical characteristics (e.g., aw and pH) that ensure the products will not spoil or
become unsafe when stored out of refrigeration throughout the manufacturer’s specified shelf-
life. However, it is essential that production of these foods include appropriate process steps to
reduce pathogens to acceptable levels and prevent growth of pathogens or the formation of
their toxins (e.g., cooking jerky with adequate humidity to prevent surface drying, active
fermentation to inhibit growth of S. aureus, and a lethality step to eliminate low-infectious dose
pathogens such as Salmonella and E. coli O157:H7) (8). Suppliers should use E. coli for
routine monitoring; coliforms and Enterobacteriaceae may be appropriate for verification
monitoring. Testing of products for bacteria, such as Salmonella, E. coli O157:H7 and S.
aureus may be appropriate when process controls are suspect, e.g., failed fermentation or
extended drying times.
Nuts and Nut Butters –RTE, not processed for lethality – Appendix A, Flow Diagram A.29,
Appendix J, Table J.29
Raw nuts (not processed for lethality) may be contaminated with microbiota from orchards, the
ground, or equipment and personnel during harvesting, shipping, processing, and handling.
Because consumption of raw nuts has been associated with illness, suppliers should test in-
process samples and finished products routinely for Salmonella and implement an environment
testing program that includes testing for Salmonella. For certain nuts (e.g., peanuts, pistachios,
Brazil nuts), routine testing for aflatoxin B1 should be done. Non-routine testing for E. coli and
aflatoxin B1 (for those not tested routinely for aflatoxin B1) may be done to assess sanitary
storage and production, and the quality of the raw nuts.
Nuts and Nut Butters – RTE, processed for lethality – Appendix A, Flow Diagram A.30,
Appendix J, Table J.30
In this category, peanuts and tree nuts are processed for lethality (e.g., by dry roasting, oil
roasting, or steam processing). Because nuts and nut butters have been associated with illness,
routine environmental testing, testing in-process samples, and finished product testing for
Salmonella should be done. For certain nuts (e.g., peanuts, pistachios, Brazil nuts), routine
testing for aflatoxin B1 should be done. Non-routine testing for E. coli and aflatoxin B1 (for
those not tested routinely) may be conducted to help assess sanitary storage and production,
and the quality of the raw nuts used in manufacturing.
35
Produce –Fruits and vegetables, cut, frozen or refrigerated, minimally processed – Appendix A,
Flow Diagram A.31, Appendix J, Table J. 31
Further processing of fresh fruits and vegetables may increase or decrease microbiological
populations depending on GMPs, sanitary design of equipment, washing, blanching, or the use
of antimicrobials. Routine testing by suppliers of product for E. coli and the environment for
Listeria spp. should be done to assess process control and sanitary conditions. Periodic testing
by suppliers of in-process or finished products for Salmonella or E. coli O157:H7 (or other
appropriate STEC) may be pertinent depending on the commodity, geographic location and use
of GAPs.
Produce –Fruits and vegetables, whole – Appendix A, Flow Diagram A.32, Appendix J, Table
J.32
Fruits and vegetables are expected to have microbiota associated with them. Whole fruits and
vegetables may be washed before introduction to commerce, but undergo no other lethality
step. Environmental testing in the packing house for Listeria spp. and Salmonella should be
done by the supplier to assess sanitary conditions, with the frequency dependent upon factors
such as the commodity, geographic location and use of GAPs. Although not listed in Table J.32
nor routinely done, the DOD may consider testing (by the supplier or DOD) for Cyclospora
cavetanensis, Cryptosporidium parvum, enteric viruses, or Shigella spp. as appropriate when
there is knowledge or suspicion high risk farming and handling practices (e.g., where evidence
of previous contamination exists, water contamination is likely, or contaminated fertilizer is
used).
Mushrooms are generally commercially produced indoors on composted substrate. They are
grown, harvested, sorted, graded, and packaged, and may or may not be sliced. No routine
testing of product is typically conducted because populations of indigenous microbiota likely will
be high. Routine monitoring and testing of the environment s by suppliers for Listeria spp. may
be deemed appropriate by DOD to assess sanitary conditions and process control. Such
testing would depend on factors such as the type of compost used, the water used, the
harvesting techniques, the storage and handling conditions, and the intended end use.
Produce – Packaged salads and leafy greens – Appendix A, Flow Diagram A.34, Appendix J,
Table J.34
Salad greens are expected to have microbiota that can originate from numerous sources such
as irrigation water, insects, birds, animals, and post-harvest handling and processing. When
salad greens are washed, particularly when a chemical such as chlorine is added to the wash
water, some microorganisms can be physically washed off; however, the washing process also
can contribute to cross contamination. Packaged salads and leafy greens generally have a
limited shelf life. Suppliers can use testing for E. coli to assess process control and sanitary
conditions. Environmental testing for Listeria spp. in processing facilities should be conducted
to monitor sanitary conditions.
36
Produce – Vegetable sprouts – Appendix A, Flow Diagram A.35, Appendix J, Table J.35
These are sprouted vegetable seeds before true leaves emerge that may be consumed raw or
cooked. Routine testing of in-process and finished products by suppliers for E. coli should be
done as an indicator of process control and sanitary production. Appropriate testing of spent
irrigation water for Salmonella and E. coli O157:H7 should be conducted to assess potential
product contamination. Routine environmental monitoring for Listeria spp. also should occur to
assess sanitary conditions.
Seafood – RTE, fish, cold smoked – Appendix A, Flow Diagram A.37, Appendix J, Table J.37
Suppliers should conduct routine environmental testing for Listeria spp. to demonstrate that
production is occurring under sanitary conditions. The supplier also should test in-process and
finished products periodically for L. monocytogenes and Salmonella to demonstrate that the
product is produced under sanitary conditions. The pH of pickled herring should be verified
periodically. Scombroid species may contain histamine and products made from these species
should be tested to verify that proper temperature control was maintained.
Seafood –RTE, cooked or hot smoked – Appendix A, Flow Diagram A.38, Appendix J, Table
J.38
The supplier should apply a validated process that results in at least a 6-log10 reduction of L.
monocytogenes. When such a validated process is used, routine sampling of in-process and
finished product for S. aureus and the environment for Listeria spp. should occur to verify that
controls are in place to prevent recontamination. If required to further demonstrate that
production is occurring under process control and sanitary conditions, the supplier could also
test in-process and finished products for coliforms, APC, Salmonella and L. monocytogenes. If
it is apparent that there is a potential for recontamination through mechanical or manual
handling, testing finished products for Salmonella and L. monocytogenes should be done
routinely. Scombroid species may contain histamine if temperature abused and fish
decompose; finished products should be tested for histamine per FDA’s guidance documents
(26).
Seafood – RTE, raw molluscan shellfish – Appendix A, Flow Diagram A.39, Appendix J, Table
J.39
37
Suppliers must demonstrate traceability that establishes that the product was harvested from
approved waters in the U.S. or in countries (Canada, Mexico, New Zealand, South Korea) that
have a Memorandum of Understanding with the U.S. Under these conditions, no routine
microbiological testing of products is necessary by the supplier. Where the supplier is unable to
prove the status of the harvest waters, or where contamination is suspected, the DOD should
not accept the product. Non-routine in-process and finished product testing by suppliers on
RTE, raw molluscan shellfish from approved waters to demonstrate process control and sanitary
conditions may include analyses for APC, fecal coliforms, and Vibrio paraheamolyticus (or other
Vibrio spp. if warranted). In addition, Vibrio control plans as outlined in the National Shellfish
Sanitation Program (25) may be required if conditions warrant.
Spices and Herbs, Coffee and Tea – Appendix A, Flow Diagrams A.40.a, A.40.b, and A.40.c,
Appendix J, Table 40
Harvested spices are expected to have a varied microbiota associated with them, including
spore-forming bacteria and fungi. Also, when a dehydration process is performed outdoors
there is the potential to acquire additional contamination. Suppliers should test in-process and
finished products routinely for APC and Salmonella to assess process controls and sanitary
conditions. The suppliers also should routinely test the environment for Salmonella. Non-
routine testing of finished products by suppliers, when deemed necessary, to assess process
control and sanitary conditions may include testing for B. cereus (or other toxigenic Bacillus
spp.), E. coli, coliforms, mold and yeasts, and E. coli O157:H7 (or other STEC as appropriate).
• Histamine in scombroid fish at high levels indicates possible temperature abuse, lack of
sanitary conditions, and decomposition of these fish.
• The presence of non-microbiological alkaline phosphatase in milk is an indication that the
milk has been inadequately pasteurized. Under these conditions microbiological pathogens
endemic to raw milk may survive and result in milk-borne illness.
• Peroxidase testing is used to indicate that blanching of fresh vegetables has been adequate.
Typical blanching temperatures (195 – 205°F for 3 minutes) would likely be sufficient to
provide a lethality step eliminating vegetative pathogens.
• The presence of aflatoxin or other mycotoxins is indicative of significant growth of molds.
The presence of aflatoxin or other mycotoxins may render the food unacceptable for human
consumption or for use in further food processing.
• Gas formation causing swollen product containers would be indicative of spoilage and
potential pathogen growth. Similarly, slime formation, visible mold growth, discoloration and
product leakage from a container would be indicative of spoilage or potential growth of
pathogens. Changes in product viscosity may be indicative of microbiological proteolysis or
38
starch hydrolysis; such activity may be the result of post-processing contamination and
temperature abuse, or under processing.
• Peroxide values and concentrations of free fatty acids in nuts exceeding tolerance limits
would be indicative of poor storage conditions, extended age or temperature abuse. In such
situations, these changes would not indicate microbiological spoilage or growth, but
oxidation that impacts quality.
• When free fatty acid concentrations in milk exceed tolerances, this is indicative of hydrolytic
rancidity associated with poor raw material control and potential post-process
contamination.
• Any signs of pests or pest infestation indicate contaminated packaging materials, poor
storage conditions within a plant or distribution center, pest contamination within a transport
container or at the location of sampling. These products should be considered
compromised and unacceptable.
• Development of acidity (measured by pH or titration) is critical to the safe production of
many fermented products such as cheeses, and fermented sausages. Fermentation of
these products by harmless starter organisms retards or prevents the growth of pathogenic
bacteria like E. coli, Salmonella and L. monocytogenes. However, in other products acid
development is undesirable, e.g., flat sour defect in canned food resulting from undesirable
microbiological growth. Undesirable fermentation can result in expression of purge in RTE
meat products.
GLOSSARY
Acronym
Term Definition
/Symbol
Indicates the maximum number of non-conforming analytical
Acceptance units (two-class sampling plans) or marginally acceptable
C
number analytical units (three-class sampling plans) that can result
in lot acceptance.
39
A single unit of food, from which a predetermined analytical
portion is removed and tested for microorganisms. All or
Analytical unit part of the sample unit may be used as the analytical unit, or
multiple sample units may be composited into a single
analytical unit for presence/absence testing.
Attributes sampling plans are used when the measured
characteristics are qualitative or categorical. Microbial
Attributes
presence/absence data and quantitative concentration data
sampling plans
categorized into numerical ranges are classified as
attributes.
40
The prevalence that the sample is designed to detect with a
Design specified probability. May or may not be the assumed
prevalence prevalence of an attribute in a population from which
samples are drawn.
Empirical
The cumulative distribution function associated with the
cumulative
ECDF empirical (observed) measure of a sample. The non-
distribution
parametric estimator of the CDF.
function
Empirical
distribution EDF Synonymous with empirical cumulative distribution function
function
Exponentially
A curve smoothing technique applied to time series data
weighted EWMA
that exponentially down weights older observations.
moving average
41
Statistically, a volume of production is considered
homogenous relative to a given characteristic (e.g.,
concentration of the microorganism) if the characteristic
Homogeneous
follows the same probability distribution throughout the
(statistical)
volume (e.g., lognormal with fixed mean µ and fixed
standard deviation σ). In contrast to a homogeneous
(uniform) spatial distribution.
Individuals Chart
Control chart for individual measurements
(i-chart)
42
Microbiological
Marks the limit beyond which the level of contamination is
limit for
M hazardous or unacceptable Used in 2- and 3-class sampling
unacceptable
plans
concentration
Operating
Describes the probability of accepting a lot as a function of
characteristic
lot quality
curve
43
When the target organism is detected in the analytical unit,
Positive
then the analytical unit is commonly referred to as "positive."
Process
Cp The ability of a process to meet specification limits.
capability
Maintaining the output of a specific process (e.g., food
Process control manufacturing, storage and distribution system) within a
desired range.
The probability of rejecting a conforming lot. A false positive
Producer's risk Α
or type I error.
The value associated with a percentile of the cumulative
Quantile distribution function. If p(X≤A) = B, A is the quantile value
and B is the percentile of the CDF.
Range Chart used to monitor process variability for
R-Chart
continuous numerical data.
Routine testing is defined as testing done at pre-determined
intervals at sufficient frequency to establish process control
or sanitary conditions. The sampling interval may be on a
Routine physical lot basis (e.g., 2,000 lb. combos for ground beef) or
temporal basis (e.g., per shift, daily, weekly, monthly). The
frequency of testing should be determined based on
potential risks and performance of the system.
Ready-to-eat Food that is in a form that may be safely eaten without
RTE
food additional preparation to achieve food safety
A subset of units from the lot or production process,
Sample
selected in some predetermined manner.
44
Specification
LSL and
limits, lower and Boundaries that define acceptable product
USL
upper
Specifications are part of a purchasing agreement between
Specifications a buyer and a supplier of a food and may be advisory or
mandatory according to use.
45
A measurement between 0.00 and 1.00 defining the amount
of moisture available for microbiological or chemical activity.
Water activity aw
Deionized water has an aw of 1.00 under standard
conditions. Microbes are not known to grow below aw 0.60.
Worldwide Worldwide Directory of Sanitarily Approved Food
Directory Establishments for Armed Forces Procurement, 2012
REFERENCES
1. Bradley, R. L., K. S. Houck, and M. Smukowski. 2013. Milk and milk products. p. 1-16.
In, S. Doores, Y. Salfinger and M. L. Tortorello (ed.), Compendium of Methods for
Microbiological Examination of Foods, 5th Ed. American Public Health Association. doi:
10.21005/MBEF.0222.052
2. Codex (Codex Alimentarius Commission). 2004. General Guidelines on Sampling, CAC/GL
50-2004. www.codexalimentarius.org/download/standards/10141/CXG_050e.pdf Accessed
February 5, 2015.
3. Dey, M., J. A. Mayo, D. Saville, C. Wolyniak, and K. C. Klontz. 2013. Recalls of foods due
to microbiological contamination classified by the U.S. Food and Drug Administration, fiscal
years 2003 through 2011. Journal of food protection 76: 932-938.
4. Elliott, P. H., and A. Kataoka. 2013. Canned Foods - Test for Commercial Sterility. In, S.
Doores, Y. Salfinger and M. L. Tortorello (ed.), Compendium of Methods for
Microbiological Examination of Foods, 5th Ed. American Public Health Association,
Washington, D.C. doi: 10.21005/MBEF.0222.052
5. Food Standards Australia New Zealand. 2012. Standard 1.6.1 Microbiological Limits for
Food - F2012C00862 http://www.comlaw.gov.au/Details/F2012C00862 Accessed February
5, 2015
6. Food Standards Australia New Zealand. 2014. Criteria for Listeria monocytogenes in ready-
to-eat foods. http://www.foodstandards.gov.au/code/microbiollimits/Pages/Criteria-for-
Listeria-monocytogenes-in-ready-to-eat-foods.aspx Accessed February 5, 2015.
7. Hyytia-Trees, E., E. Skytta, M. Mokkila, A. Kinnunen, M. Lindstrom, L. La Hteenma Ki, R.
Ahvenainen, and H. Kornkeala. 2000. Safety evaluation of sous vide-processed products
with respect to nonproteolytic Clostridium botulinum by use of challenge studies and
predictive microbiological models. Appl. Environ. Microbiol. 66: 223-229.
8. Ingham, S. 2008. Critical Limit Summary: Validating Shelf-Stability and Prevention of
Staphylococcus aureus Growth on RTE Meat Products.
http://www.meathaccp.wisc.edu/validation/assets/CL_Staph_JFP_68.pdf Accessed
September 19 2014.
9. ILIS -Europe, . 2010. Impact of Microbial Distributions on Food Safety.
http://www.ilsi.org/Europe/Publications/Microbial%20Distribution%202010.pdf Accessed
February 5, 2015
46
10. International Commission for the Microbiological Specifications for Foods (ICMSF). 2011.
Microorganisms in Foods 8: Use of Data for Assessing Process Control and Product
Acceptance. Springer, New York.
11. International Commission for the Microbiological Specifications of Foods (ICMSF). 2002.
Microorganisms in Foods 7: Microbiological Testing in Food Safety Management. Springer,
New York.
12. International Organization for Standardization. 2007. Guidance on the selection and usage of
acceptance sampling systems for inspection of discrete items in lots — Part 1: Acceptance
sampling. ISO, Geneva.
13. Kornacki, J. L.., J. B. Gurtler, and B. A. Stawick. 2013. Enterobacteriaceae, Coliforms, and
Escherichia coli as Quality and Safety Indicators. . In, S. Doores, Y. Salfinger and M. L.
Tortorello (ed.), Compendium of Methods for Microbiological Examination of Foods, 5th Ed.
American Public Health Association. Washington, D.C. doi: 10.21005/MBEF.0222.052
14. Powell, M. R. 2014. Optimal Food Safety Sampling Under a Budget Constraint. Risk
Analysis 34: 93-100. doi: 10.1111/risa.12054
15. Ryser, E. T., and E. H. Marth. 2007. Listeria, Listeriosis, and Food Safety. 3rd ed. CRC
Press, Taylor and Francis Group, Boca Raton, FL.
16. Shewhart, W. A. 1986. Statistical Method from the Viewpoint of Quality Control. Dover
Publications, Inc. New York
17. Sperber, W. H., and North American Millers' Association Microbiology Working Group.
2007. Role of microbiological guidelines in the production and commercial use of milled
cereal grains: a practical approach for the 21st century. Journal of food protection 70: 1041-
1053.
18. U.S. Department of Agriculture. 1999. Appendix B. Compliance Guidelines for Cooling
Heat-Treated Meat and Poultry Products (Stabilization). Food Safety Inspection Service.
Washington D. C. http://www.fsis.usda.gov/OPPDE/rdad/FRPubs/95-033F/95-
033F_Appendix%20B.htm. Accessed February 5, 2015.
19. U.S. Department of Agriculture. 2005. Risk Assessment of the Impact of Lethality Standards
on Salmonellosis from Ready-to-Eat Meat and Poultry Products. Food Safety Inspection
Service. http://www.fsis.usda.gov/wps/wcm/connect/210d3d19-96bc-4b1e-8829-
325e25e63e0b/Salm_RTE_Risk_Assess_Sep2005.pdf?MOD=AJPERES Accessed February
5, 2015.
20. U. S. Department of Agriculture. 2014. FSIS Compliance Guideline for Meat and Poultry
Jerky Produced by Small and Very Small Establishments.
http://www.fsis.usda.gov/wps/wcm/connect/5fd4a01d-a381-4134-8b91-
99617e56a90a/Compliance-Guideline-Jerky-2014.pdf?MOD=AJPERES Accessed February
5, 2015.
21. U. S. Department of Health and Human Services.2005. Apple Juice, Apple Juice
Concentrates, and Apple Juice Products - Adulteration with Patulin CPG Sec.510.150.
http://www.fda.gov/ICECI/ComplianceManuals/CompliancePolicyGuidanceManual/ucm074
427.htm Accessed February 5, 2015.
22. U. S. Department of Health and Human Services. 2009. Prevention of Salmonella Enteritidis
in Shell Eggs During Production, Storage, and Transportation: Final Rule. 21 CFR Parts 16
and 118, Federal Register 74(130).
47
23. U. S. Department of Health and Human Services. 2010. Compliance Policy Guide Sec.
527.300 Dairy Products - Microbial Contaminants and Alkaline Phosphatase Activity.
http://www.fda.gov/downloads/ICECI/ComplianceManuals/CompliancePolicyGuidanceMan
ual/UCM238465.pdf Accessed February 5, 2015.
24. U. S. Department of Health and Human Services. 2011. Grade “A” Pasteurized Milk
Ordinance, Including Provisions from the Grade “A” Condensed and Dry Milk Products and
Condensed and Dry Whey--Supplement I to the Grade “A” Pasteurized Milk Ordinance.
http://www.fda.gov/downloads/Food/GuidanceRegulation/UCM291757.pdf Accessed
February 5, 2015.
25. U. S. Department of Health and Human Services. 2013. National Shellfish Sanitation
Program (NSSP).
http://www.fda.gov/Food/GuidanceRegulation/FederalStateFoodPrograms/ucm2006754.htm
Accessed February 5, 2015.
26. U. S. Department of Health and Human Services. 2014. Fish and Fishery Products Hazards
and Controls Guidance - Fourth Edition.
http://www.fda.gov/food/guidanceregulation/guidancedocumentsregulatoryinformation/seafo
od/ucm2018426.htm. Accessed February 5, 2015.
27. U.S. Department of Defense. 2013. Department of Defense Food Safety and Quality
Assurance Laboratory Action Levels. USAPHC Circular 40-1, Appendix O. In, Worldwide
Directory of Sanitarily Approved Food Establishments for Armed Forces Procurement, 2012.
28. U.S. Department of Health and Human Services, Food and Drug Administration. 2004.
Guidance for Industry Juice HACCP Hazards and Controls Guidance.
http://www.fda.gov/Food/GuidanceRegulation/GuidanceDocumentsRegulatoryInformation/J
uice/ucm072557.htm Accessed February 5, 2015.
29. Wehr, H. M., and J. F. Frank. 2004. Standard Methods for the Examination of Dairy
Products, 17th ed. American Public Health Association, Washington, DC.
30. World Health Organization. 2013. Water Sanitation Health – Drinking Water Quality.
http://www.who.int/entity/water_sanitation_health/draft_regscan_may_2014.pdf Accessed
August 14 2014.
31. World Health Organization, and Food and Agricultural Organization. 2013. Proposed Draft
Principles and Guidelines for the Establishment and Application of Microbiological Criteria
Related to Foods (at steps 5/8) Appendix III (Pages 43 to 49). Joint FAO/WHO Food
Standards Programme CODEX Alimentarius Commission Thrity-sixth Session Rome, Italy,
1-5 July 2013 and Report of the Forty-Fourth Sessionof the CODEX Committee on Food
Hygiene: 27-56.
48
Appendix A. Schematic Flow Diagrams of Production of Various Food Categories
and Bottled Water
These generic process flow charts are intended to provide DOD auditors with potential
steps in the manufacturing process where microbiological counts could increase with
loss of process control or development of insanitary conditions. In addition, the flow
charts illustrate where there are lethality steps that reduce numbers of indicator
organisms and pathogens if present.
Steps for receiving and storing packaging materials were omitted to simplify the creation
– and use – of the process flow diagrams. It is expected that a DOD-approved food
processing plant would have appropriate control and documentation of these functions,
either as part of product-specific HACCP plans, or as preventive and pre-requisite
programs such as Standard Operating Procedures for receiving and storage. It was
recognized that a finished food product could move through many storage and
distribution facilities as part of the supply chain. The final two steps were denoted “store
finished product” and “distribute finished product” to simplify the creation and use of the
process flow diagrams.
The intent was to include those steps relevant of the manufacturing process relevant to
microbial aspects of food rather than to include all possible aspects or combinations of
receipt, processing, storage, and distribution of production. The Committee assumes
that DOD personnel will be able to recognize the specific steps observed at a food
processing plant from among the general manufacturing steps shown on the process
flow diagrams.
Programs for minimizing contamination at the identified steps include Good Agricultural
Practices, Sanitation Standard Operating Procedures, Good Manufacturing Practices,
and purchasing specifications. DOD personnel should use the process flow diagrams to
review the general steps to manufacture the food product under evaluation. From the
process flow diagram, DOD personnel should determine the step(s) at which verification
sampling should be done by the supplier. When analysis of verification samples
indicates that the supplier may have shortcomings in process or sanitation control, DOD
personnel should use the process flow diagram to determine steps at which
contamination might occur or steps at which a failure to achieve the expected
destruction of bacteria may be occurring.
1
Flow Diagram A.1. BEVERAGES – BOTTLED
WATER (ARTESIAN, MINERAL, PURIFIED,
SPARKLING AND SPRING WATER)
Package (C, S)
2
Flow Diagram A.2. BEVERAGES – ICE, PACKAGED
Freeze
Store
Package (C, S)
3
Flow Diagram A.3. BEVERAGES – JUICES AND DRINKS, PASTEURIZED,
REFRIGERATED
Harvest fruits or vegetables (C)
Transport
Storage (optional, G)
Wash
Cool/Chill
Package (C, S)
4
Flow Diagram A.4. BEVERAGES – SHELF STABLE
Package
5
Flow Diagram A.5. DAIRY – BUTTER, MARGARINES
Butter Margarine
Package (C, S)
6
Flow Diagram A.6. DAIRY – CHEESE (HARD)
Press (C)
7
Flow Diagram A.7. DAIRY – CHEESE (SOFT, SEMI-SOFT AND SURFACE-
RIPENED)
Press (C)
8
Flow Diagram A.8.a. DAIRY PRODUCTS Cultured pH<4.8 (Example – Yogurt)
Cool
Package (C, S)
9
Flow Diagram A.8.b. DAIRY – CULTURED, pH<4.8 (Example – Sour Cream
Buttermilk, etc.)
Add culture
Cool
10
Flow Diagram A.9. DAIRY – CULTURED, pH>4.8 AND <5.4 (Example – Cottage
Cheese)
Package (C, S)
11
Flow Diagram A.10. DAIRY – DRIED PRODUCTS
(does not include dairy ingredients used to make infant formula)
Package (C, S)
12
Flow Diagram A.11. DAIRY – FROZEN DESSERTS
Freeze
Package (C, S)
Hard-freeze (optional)
13
Flow Diagram A.12. DAIRY – MILK AND MILK PRODUCTS (Fluid)
Package (C, S)
14
Flow Diagram A.13. DAIRY – PROCESS CHEESE
Cook(K)
Package (C, S)
15
Flow Diagram A.14. EGG PRODUCTS – PASTEURIZED, PROCESSED
Pasteurize/cook (K)
Cool
Package (S, C)
16
Flow Diagram A.15. EGG PRODUCTS – SHELL EGGS RAW
Candle
In-shell pasteurization
(optional; K)
17
Flow Diagram A.16. GRAIN BASED PRODUCTS – BAKED ITEMS, RTE,
REFRIGERATED OR TCS
Mix ingredients
Form dough
Proof
Bake (K)
Cool (C)
Slice (optional; C)
Package (S, C)
18
Flow Diagram A.17. GRAIN BASED PRODUCTS – BAKED ITEMS, RTE, SHELF
STABLE, NON-TCS
Mix ingredients
Form dough
Proof
Bake (K)
Cool (C)
Slice (optional; C)
Package (S, C)
19
Flow Diagram A.18. GRAIN BASED PRODUCTS – RTE, CEREALS
Cook (K)
Extrude
Puff/toast (K)
Dry (C)
Package (S)
20
Flow Diagram A.19. GRAIN BASED PRODUCTS – RTE, COLD PRESSED BARS
Press/form (C)
Enrobe (optional; C)
Cool (C)
Package (S)
21
Flow Diagram A.20. GRAIN BASED PRODUCTS – NON-RTE, DRY FLOUR BASED
MIXES
Package (S)
22
Flow Diagram A.21. GRAIN BASED PRODUCTS – NON RTE, PASTA, DRIED OR
REFRIGERATED
Form dough
Extrude
Cool (S, C, G)
Dry/dewater (C)
23
Flow Diagram A.22. MEALS AND ENTRÉES – NON-RTE, READY TO COOK (RTC)
MEALS, INCLUDES RAW INGREDIENTS
Package (S)
Cool/freeze (G, C)
24
Flow Diagram A.23 MEALS AND ENTRÉES – RTE, DELI SALADS, SANDWICHES
HEAT-EAT MEALS, SUSHI
Cool (S, G)
Package (C, S)
25
Flow Diagram A.24. MEALS AND ENTRÉES – SOUS VIDE, COOK AND CHILL
Pasteurize/cook (K)
Cool (S, G)
26
Flow Diagram A.25. MEAT, PORK AND POULTRY PRODUCTS – NON-RTE, BEEF
AND PORK RAW, INTACT AND NON-INTACT
Apply steam or wash with organic acid and/or hot water (optional –
step varies with species and country; pathogen reduction may
occur; C, S)
Cool (G)
Apply steam or wash with organic acid and/or hot water (optional –
step varies with species and country; pathogen reduction may
occur; C, S)
Cut/”fabricate” (G, C)
Grind (optional; G, C)
Package (S)
27
Flow Diagram A.26. MEAT, PORK AND POULTRY PRODUCTS – NON-RTE,
POULTRY, RAW
Wash
Process: cut-up,
debone or further
process (G, C)
Distribute finished
product
28
Flow Diagram A.27. MEAT, PORK AND POULTRY PRODUCTS – RTE, COOKED
PERISHABLE
Cook (K)
Cool (S, C, G)
Package (G, C, S)
29
Flow Diagram A.28. MEAT, PORK AND POULTRY PRODUCTS – RTE
FERMENTED AND DRIED, DRIED
Ferment (optional; G)
Heat (K)
Cool (S, G)
Package (C, S)
30
Flow Diagram A.29. NUTS AND NUT BUTTERS – NUTS, RTE,
NOT PROCESSED FOR LETHALITY
Transport
Package (C, S)
Distribute finished
product
31
Flow Diagram A.30. NUTS AND NUT BUTTERS – RTE, PROCESSED FOR
LETHALITY
Store
Shell (C)
Package (C, S)
32
Flow Diagram A.31. PRODUCE – FRUITS AND VEGETABLES, CUT
FROZEN, OR REFRIGERATED, MINIMALLY PROCESSED
Transport
Process options: inspect, sort, cull, trim, wash, de-water, shell, chop, cut,
slice, shred, grade, blend (C, G)
33
Flow Diagram A.32. PRODUCE – FRUITS AND VEGETABLES, WHOLE
Transport
Pre-cool (optional, C)
Pre-cool (optional, C)
Wash (optional; C)
34
Flow Diagram A.33. PRODUCE – MUSHROOMS – FRESH OR FROZEN,
WHOLE, SLICED, NOT CANNED OR MARINATED
Incubate (G)
Harvest (C)
35
Flow Diagram A.34. PRODUCE – PACKAGED SALADS AND LEAFY
GREENS
Transport
Package (C, S)
36
Flow Diagram A.35. PRODUCE – VEGETABLE SPROUTS
Package (C, S)
37
Flow Diagram A.36a. SEAFOOD – NON-RTE, RAW
Harvest (G)
Ice and pack (chilled Off-load (unless processed on- Store live
storage) (C) ship; G)
Store under
refrigeration or on
ice (G)
38
Flow Diagram A.36b. SEAFOOD –RAW
Wild Salmon Shashimi
Finished Product
Refrigerated storage (C, S, G)
39
Flow Diagram A.36c. SEAFOOD –RAW
Wild Salmon Sushi
Cut rolls
Finished Product
Refrigerated Storage
(C, S, G)
40
Flow Diagram A.36d. SEAFOOD –RAW
Ceviche
Refrigerate (C,G)
41
Flow Diagram A.36e. SEAFOOD –RAW
Receive
(G,S)
Freeze
Thaw
Dry Salt
Rinse
Cure
(C,S)
Drain
Finished Product
Refrigerated Storage
(C, G)
42
Flow Diagram A.37. SEAFOOD – RTE FISH, COLD
SMOKED
Smoke (G, C)
Cool (S, C, G)
Package (C)
43
Flow Diagram A.38. SEAFOOD – RTE, FISH OR CRUSTACEAN, COOKED OR
HOT SMOKED
Cut/portion (C)
Rinse
Smoke/dry (K)
Cool (G, C, S)
Package (G, C, S)
44
Flow Diagram A.39. SEAFOOD – RTE, RAW MOLLUSCAN
SHELLFISH
Re-pack/shuck (C, G)
45
Flow Diagram A.40a. SPICES AND HERBS
Harvest (C)
Dry (C)
Pack (bulk; C)
Package (S)
46
Flow Diagram A.40b. BEVERAGES – COFFEE
Harvest (C)
Brew (K)
Remove/discard
extracted grounds
Freeze- or spray-dry
(C)
Package instant
coffee (C, S) Distribute finished product
47
Flow Diagram A.40c. BEVERAGES –
Freeze-dry or spray-dry
(C)l Store finished product
48
Appendix B: Statistical Process Control (SPC) Charts
Control charts are plots of process data collected over time used to determine if a
process is in statistical control. It is important to note that there is a difference
between a process being in statistical control and meeting specifications. A
process is considered under statistical control if it is stable over time and the
observed variation is due to common, chance causes inherent to the process
(e.g., background noise due to normal variation in ambient temperature and
humidity) and there is no between-lot variation. A process is considered out of
statistical control if shifts in the process central tendency (e.g., mean), variability,
or both result from uncommon sources associated with special or assignable
causes (e.g., equipment malfunction, a change in raw materials, or failure of a
laboratory procedure). A food process being under statistical process control
does not imply its capability with respect to meeting microbiological
specifications. The ideal situation is when a process is both under statistical
control and is capable of manufacturing products that meet specifications.
However, a process can be in statistical control and not capable of satisfying
specifications. For example, the process consistently generates substandard
product. Alternatively, a process can be out of statistical control but capable of
satisfying specifications. For example, the process is designed to be robust to
deviations from the norm, such that it meets specifications despite high
variability. Given seasonal and other sources of variability beyond a supplier’s
control, the latter situation may be particularly relevant to food production
processes.
SPC charts can be classified as control charts for variables (e.g., average and
range charts) or control charts for attributes (e.g., p charts). Microbiological food
safety characteristics can be classified as variables or attributes. Microbiological
concentration data expressed on a continuous numerical scale are classified as
variable data. Microbiological presence/absence data or concentration data
classified into numerical ranges (e.g., m < x ≤ M) are classified as attribute data.
Montgomery (2) cautions: “It is not possible to give an exact solution to the
problem of control chart design, unless the analyst has detailed information about
both the statistical characteristics of the control chart tests and the economic
factors that affect the problem. A complete solution would require knowledge of
the costs of investigating and possibly correcting the process in response to out-
of-control signals, and the costs associated with producing a product that does
not meet specifications. Given this kind of information, an economic decision
model could be constructed to allow economically optimum control.” However,
such detailed information is not generally available for even a small subset of
food production processes, and the available information is subject to
considerable uncertainty, variability, and disagreement (e.g., regarding consumer
health impacts). Therefore, this discussion is limited to some general guidelines
that will aid in SPC chart design rather than identifying optimal designs.
1
Control Charts for Variables
Suppose that the microbiological concentration data (y) from a food process are
lognormally distributed such that the log-transformed data (x = log10(y)) are
normally distributed with mean µ (log10 cfu/g) and standard deviation σ (log10
cfu/g). Estimates of µ and σ are based on an initial process capability study
conducted when the process is considered under statistical control. Let k =
number of lots (subgroups) sampled and n = number of samples per lot
(subgroup). As a rule of thumb, Shewhart (3) suggested that “a sequence of not
less than twenty five samples of size four” is the minimum requirement for
concluding that a process is in a state of statistical control (e.g., 4 samples per lot
from 25 lots).
Let 𝑥𝑥̅1 , 𝑥𝑥̅2 , … , 𝑥𝑥̅𝑘𝑘 be the geometric (log10) sample means from each lot (subgroup).
The estimate of process average (µ) is the grand mean (𝑥𝑥̿ ):
∑𝑘𝑘𝑗𝑗=1 𝑥𝑥̅𝑗𝑗
𝑥𝑥̿ =
𝑘𝑘
The sample range (R) is the difference between the largest and smallest
observations within each lot (subgroup): 𝑅𝑅 = 𝑥𝑥𝑚𝑚𝑚𝑚𝑚𝑚 − 𝑥𝑥𝑚𝑚𝑚𝑚𝑚𝑚 . The average sample
range (𝑅𝑅� ) provides an estimate of the process standard deviation (σ):
∑𝑘𝑘𝑗𝑗=1 𝑅𝑅𝑗𝑗
�
𝑅𝑅 =
𝑘𝑘
𝑅𝑅�
𝜎𝜎� =
𝑑𝑑2
Where d2 is the expected mean of R/σ. Table B.1 provides calculated d2 values
for n=2-25.
2
Table B.1. Factors for Control Charts for Variables Assuming a Normal
Distribution
n d2 d3 A2 D3 D4
2 1.128 0.853 1.880 0.000 3.267
3 1.693 0.888 1.023 0.000 2.575
4 2.059 0.880 0.729 0.000 2.282
5 2.326 0.864 0.577 0.000 2.114
6 2.534 0.848 0.483 0.000 2.004
7 2.704 0.833 0.419 0.076 1.924
8 2.847 0.820 0.373 0.136 1.864
9 2.970 0.808 0.337 0.184 1.816
10 3.078 0.797 0.308 0.223 1.777
11 3.173 0.787 0.285 0.256 1.744
12 3.258 0.778 0.266 0.283 1.717
13 3.336 0.770 0.249 0.307 1.693
14 3.407 0.763 0.235 0.328 1.672
15 3.472 0.756 0.223 0.347 1.653
16 3.532 0.750 0.212 0.363 1.637
17 3.588 0.744 0.203 0.378 1.622
18 3.640 0.739 0.194 0.391 1.609
19 3.689 0.733 0.187 0.404 1.596
20 3.735 0.729 0.180 0.415 1.585
21 3.778 0.724 0.173 0.425 1.575
22 3.819 0.720 0.167 0.435 1.565
23 3.858 0.716 0.162 0.443 1.557
24 3.895 0.712 0.157 0.452 1.548
25 3.931 0.708 0.153 0.459 1.541
The 𝑥𝑥̅ chart monitors between-lot variability in the process mean. The equations
for constructing 3σ upper and lower control limits on the 𝑥𝑥̅ chart are as follows:
3
variability (𝜎𝜎𝑥𝑥̅ ) as indicating a lack of control, rather than a source of variation that
may be intrinsic to the process. Achieving negligible between-lot variability may
not be feasible in some food production processes. Even for relatively small
sample sizes, the sampling distribution of the sample mean is approximately
normal even if the underlying data are not; although the limit of quantitation
presents a potential complication for microbiological data if the proportion of
negative results is large (1). Montgomery (2) addresses monitoring processes
with a high proportion of data such as those that fall outside the detection limit or
are too numerous to count.
The data used to construct 𝑥𝑥̅ and R charts also provide information about
process capability. The two-tailed process capability index (Cp) is defined in
terms of the upper and lower specification limits (USL and LSL):
𝑈𝑈𝑈𝑈𝑈𝑈 − 𝐿𝐿𝐿𝐿𝐿𝐿
𝐶𝐶𝑝𝑝 =
6𝜎𝜎
Note, however, that the equation for Cp only considers process variability. It
implicitly assumes that the process is centered on a mean of (USL-LSL)/2.
Compare to an upper-tail 𝐶𝐶𝑝𝑝 :
Cp = (USL – μ) /(3 σ)
Further details about the statistical basis for control charts for variables are
available in standard texts (e.g., (2) ). On-line calculators for control charts for
variables are available from a variety of sources (e.g.,
http://www.sqconline.com/).
4
process under statistical control. However, extreme values may simply be
random outliers, and identifying an assignable cause for each extreme value may
not be possible. Similarly, apparent patterns in small datasets (e.g., a sequence
of extreme values or trends) may be simply due to random variation. If the initial
data indicate that process variability is not in statistical control, then the control
limits on the 𝑥𝑥̅ chart may not be meaningful. Therefore, beginning the analysis
with the R chart can be useful. It is customary to treat the control limits obtained
in the initial phase as provisional and to update and revise the control limits over
time as additional information is acquired (Appendix I) and the process matures.
BIBLIOGRAPHY
5
Appendix C: Process Control for Attributes (p charts)
There are a variety of process control charts for attributes. The p chart is widely
used; it charts the fraction of non-conforming analytical units over a sampling
sequence. The p chart is based on the binomial distribution, which assumes that
there are only two possible outcomes for each observation (conforming or non-
conforming), the proportion of non-conforming analytical units (p) is constant, the
samples are independent (e.g., defects do not cluster), and a fixed sample size
(n). The sample proportion non-conforming (𝑝𝑝̂ ) is the ratio of the number of non-
conforming analytical units (d) observed in a sample of size n:
𝑑𝑑
𝑝𝑝̂ = 𝑛𝑛 (eq. C.1)
For the binomial distribution, the mean and variance of the sampling distribution
of 𝑝𝑝̂ are:
𝜇𝜇𝑝𝑝� = 𝑝𝑝 (eq. C.2)
and
𝑝𝑝(1−𝑝𝑝)
𝜎𝜎𝑝𝑝2� = (eq. C.3)
𝑛𝑛
respectively.
Note that equations C.5 and C.6 assume that 𝑝𝑝̅ represents the desired target
value of the proportion non-conforming for the process. In food safety, concern
would normally focus on exceeding the upper control limit (UCL); however,
observations below the lower control limit (LCL) could indicate problems with
sampling and analytical procedures, or it could represent an opportunity on how
to improve process quality. On-line calculators are available for computing
conventional 3 sigma control limits for p charts (e.g.,
http://www.sqconline.com/control-chart-calculator-attributes-discrete-data).
It should be noted that the further the target value of p is from 0.5, the larger the
sample size required for the normal approximation to be reasonable. As a
general rule, the normal approximation is reasonable if np ≥ 5 and n(1-p) ≥ 5. In
many food safety applications where the target value for p is substantially less
1
than 0.5, the sample size required for the normal approximation would be costly.
More generally, even if an exact binomial method is used to calculate control
limits, practical application of p charts is limited to cases where the target value
for p is not very small (Table C.1). The sample size should be large enough to
provide a reasonably high degree of confidence of observing at least one non-
conforming unit (1). For example, if the target value for p = 0.01 and n = 5, the
conventional upper control limit is 0.14 (eq. C.5). Consequently, observing a
single non-conforming unit in the sample (𝑝𝑝̂ = 1/5 = 0.2) would suggest a lack of
process control.
p target n1 n2 n3
0.01 500 299 230
0.02 250 149 114
0.03 167 99 76
0.04 125 74 57
0.05 100 59 45
0.10 50 29 22
0.20 25 14 11
0.30 17 9 7
0.40 13 6 5
0.50 10 5 4
n1 = minimum sample size required for normal approximation
n2 = sample size required for 95% confidence of observing at least one non-conforming unit
n3 = sample size required for 90% confidence of observing at least one non-conforming unit
BIBLIOGRAPHY
1. Montgomery, D. C. 2005. Introduction to Quality Control. 5th ed. John Wiley &
Sons, Hoboken, NJ.
2
Appendix D. High-Event Period Process Control
Suppose that the prevalence (p) is constant such that the number of positive test
results (x) out of n independent samples follows a binomial distribution. Then we
can determine combinations of x and n that are unlikely to occur by chance if the
true prevalence is no more than the design prevalence. A sampling period would
proceed until the testing results indicate a high-event period, or non-conformance
with the design prevalence. After appropriate action is taken in response to the
non-conformance, a new sampling period begins.
Tables D.1 and D.2 present the limits of conforming sample results for a false
alarm rate (FAR) of 5% and 1%, respectively. From Table D.1, for example, if the
number of positive test results observed is x = 4 out of n < 198, then there is less
than a 5% chance of observing the data if the true prevalence is 1%. A 5% FAR
might be appropriate for cases of low sampling frequency because a long
sampling period may elapse before a producer receives an indication that the
process is out of control.
Similarly from Table D.2, if the number of positive test results observed is x = 4
out of n < 129, then there is less than a 1% chance of observing the data if the
1
true prevalence is 1%. A 1% FAR might be appropriate for cases of high
sampling frequency because a sampling period of limited duration would elapse
before a producer receives an indication that the process is out of control.
2
Appendix E. Control Charts for Very Low Prevalence
Here ‘very low prevalence’ is taken to mean that less than 2% of the samples
taken are found positive for the analyte. The testing is typically limited to
presence/absence methods, and the finding of a positive is sufficient to implicate
the underlying lot.
The statistic of interest for this scenario is the prevalence proportion of positive
results, or, equivalently, the ‘mean time between positives’ (‘MTBP’).
Lots of ground beef are tested for E. coli O157:H7. While positives may result in
rejected lots, this testing can also be used for process control.
The normal operations are modeled as a Poisson process, with the MTBP
following an exponential distribution, which has a standard deviation equal to the
mean. Hypothetical data for 325-gram samples from several years indicate a
process MTBP of 690 sampling units with a standard deviation of 730, not much
different from 690, supporting the use of an exponential model.
A parametric ‘g-Chart’ is shown in Fig. E.1, with an upper control limit calculated
as:
= 2762
In the example, the LCL is zero. The factor ‘3’ is based on the 3 sigma
convention and corresponds to the 99.9th percentile (Alternatively, the exact
exponential distribution quantiles could be used).
Also plotted is the exponentially weighted moving average (EWMA) of the MTBP
data, with a smoothing constant of 0.1 and starting value EWMA0 = 690:
1
The EWMA smooths the rough curve and is helpful in visualizing the drift of the
MTBP average.
The g-Chart helps define operational conditions that lead to relatively stable low
prevalence. However, it has the limitation that it is an individual data trend chart
and the LCL is absent. If the data line moves above the UCL, this suggests
maintenance of this state of operations would result in lower prevalence, and
should be investigated to see how this lower prevalence could be sustained.
Also, if the data line exhibits a strong non-random pattern, this suggests a
systematic cause, which should be investigated. Figure E.1. shows a saw-tooth
appearance (‘up’ followed by ‘down’), indicating negative autocorrelation.
Finally, we can supplement the control limits with a ‘runs test’, e.g., if a run of 11
or more MTBP data consecutively fall below the mean MTBP, then ‘abnormal’
operations are detected, and an assignable cause should be sought (for the
exponential distribution, the mean is the 63rd percentile, so 0.6311 = 0.6% FAR).
The longest run observed below the mean MTBP is 7 samples here, which falls
within expectations for normal operations. As an alternative, a ‘center-line’ at
0.693 MTBP could be added, for which results under statistical control are
equally likely to fall on either side. A run of 7 or more results on the either side of
this center line represents detection of an abnormal change in the process that
should be investigated for an assignable cause, i.e., for the exponential
distribution, the median = 0.693 x mean, so 0.57 = 0.8% FAR.
2
3000
10/5/2011
2500
Interval Between Positives
3/22/2012
6/3/2013
2000
1500
1000 9/21/2012
500
0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45
Sequence number of Positive found
3
Appendix F. Control Chart for Low Prevalence with Quantification
The g-Chart will still be useful for maintaining normal prevalence (MTBP).
If the observed counts (not concentration) in quantitation are below 100 and a
single dilution is used, the counts may be modeled as Poisson distributed. If the
observed counts are 10 or more typically, or multiple dilutions or a most probable
number (MPN) technique is used, the counts (or estimates in the case of MPN)
may be modeled as normally distributed after a logarithmic transformation.
where ‘x’ is the concentration estimated for the positive result, ‘d’ is the
concentration corresponding to a single count result, and ‘y’ is the logarithmic
metamer. For example, if the analytical portion is 1 ml, and there is a one
decimal dilution, then a single count would result in an estimated concentration of
1
10 CFU/ml. Single count measurements (i.e., =10 CFU/ml) would be
transformed as 1.11 = log10(10+0.3 x 10).
Lots of soft cheese (e.g., Brie) are sampled and tested for coliform bacteria.
Specifications require each lot should not exceed 1,000 CFU/25g test portion.
History is comprised of 702 samples, of which 28 were positive, for a prevalence
(p) of 3.99%. The MTBP was 24.4 samples with a standard deviation of 27.5
samples, close to the MTBP, supporting the exponential distribution assumption.
Figure F.1. shows the g-Chart, with an UCL exception at sample #309, and a run
of 9 values below the MTBP line. The exception indicates better control is
possible in normal operations, and this should be explored. The run of 9 below
the MTBP line, although not an exception, is suggestive of a problem with control
in this range of samples.
Note: All quantiles are for log10 transformed positive data using eq.(F.2).
The nonparametric quantiles are derived from the EDF of the 702-sample data.
As a rule, these are very imprecise for probabilities greater than 701.5 / 702 =
99.9%.
2
In Table F.1, the ‘Adjusted Quantile Probability’ column is the desired quantile
probability ‘P’ (given in the first column of the table) adjusted for prevalence p.
The adjusted quantile probability P* is given by
Finally the normal distribution quantiles are the quantiles associated with P*.
Note that the nonparametric and normal quantiles are in excellent agreement
here, supporting the lognormal assumption. The probability of obtaining less than
a single count was about 0.1%, indicating no problem with bias at the low end.
Figure F.2. shows the individuals’ chart (‘i-Chart’) with associated UCLs for some
soft cheese data. The SPC UCL line corresponds to the mean 1.87 plus 2.575
times the standard deviation of 0.783. This SPC UCL line represents the 99.5%
quantile (normal vs. abnormal division) of the lognormally distributed positive
result data, given that a positive result occurs. The Quantile UCL line represents
the nonparametric 99.5% quantile across all results, including those which
correspond to zero counts. The point at sample #12 exceeds both UCLs,
indicating the point is unusual for observation as a result, and also unusual from
the baseline normal distribution point of view. Sample #12 represents an
unexpected shift in operations. The sample #24 result exceeds the Quantile UCL
line, which means it is a rarity in sampling, but does not exceed the SPC UCL
line, indicating it is not that unexpected from a positive data distribution point of
view, so still represents normal operations (same distribution of positive results).
This difference in interpretation between the nonparametric and parametric
approaches shows another advantage (besides allowing the estimation of high
probability quantiles using small samples) of the latter.
3
5.0
4.0
Concentration (log10)
3.0
2.0
1.0
0.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
Sample number
Figure F.1. g-Chart for coliforms in soft cheese. Note UCL exception around
sample #309 and run of 9 values below the MTBP.
4
140
Observed Time Between Positives
120
100
80
60
40
20
0
53 77 128 189 312 416 442 460 462 502 553 590 664 678
Sample number
Figure F.2. Individuals chart (i-Chart) for positive samples observed for coliforms
in Soft Cheese.
5
Appendix G. Control Chart for Moderate to High Prevalence with
Quantitation
This example corresponds to prevalence in the range 10% to 95%. The sampling
can be interpreted as the output of a Bernoulli process, and the prevalence
estimated and controlled. It is also assumed that positive results are routinely
quantitated.
Because of the moderate to high prevalence, rational subgroups (i.e., lots that
represent test units belonging to a homogeneous population with the same
constant population parameters) of samples may be combined to increase
normality and provide better tools for prevalence SPC. If grouping is to be done
by time period, equal sampling for each period is advised, and the sample size
large enough to achieve an expected 5 positive results or more, and similarly for
negative results. For example, if the mean prevalence is 20%, the sample size
should be at least 5 / 0.2 = 25. If the mean prevalence is 80%, the sample size
would also be 25.
SPC may be carried out by a ‘p-Chart’, where the control limits are given by
𝑝𝑝̅ (1−𝑝𝑝̅ )
𝑈𝑈𝑈𝑈𝑈𝑈 = 𝑝𝑝̅ + 3� (eq. G.1.)
𝑛𝑛
𝑝𝑝̅ (1−𝑝𝑝̅ )
𝐿𝐿𝐿𝐿𝐿𝐿 = 𝑝𝑝̅ − 3� (eq. G.2.)
𝑛𝑛
Although there is fixed sample size during a time period (in this case, 40 samples
per quarter), the number of positive results is a random variable, so a standard
X-bar chart cannot be used.
1
Example: Aerobic Plate Counts in Ground Beef
Lots of ground beef in cold storage are sampled and tested for aerobic plate
counts (APC). Assume the microbiological guidelines require that the APC
should not exceed 10,000,000 CFU/g (i.e., 7.0 on a log10(CFU) scale).
Data consist of 455 samples, of which 393 were positive (86.4% prevalence).
The positive samples had average log10-transformed concentration of 5.19 with
standard deviation of 1.34.
The nonparametric quantiles from the expected distribution function (EDF) are
precise up to 99.5%. The agreement between nonparametric and normal-based
quantiles is good, but not perfect. The probability of getting less than one count
under the normal distribution model is 2.2%, a small error which should be
adjusted out by using a truncated normal distribution, but which we will ignore
here.
Given a prevalence of 86%, the minimum sample size required for the normal
approximation to be valid is 36. Based on this, sampling was carried out by
quarter of the year, with 40 samples taken randomly each quarter. For each
quarter, the proportion of the 40 samples with enumerative results >0 was
calculated via eq. F.1 (Appendix F).
Figure G.1. shows a p-Chart for hypothetical data. Note the exceptions at periods
#7 and #22, either of which would indicate a drop in prevalence of samples with
enumerative results >0.
Figure G.2. shows an i-Chart for the samples with enumerative results >0 with
control limits based on + 3 σ. The data are under control with no exceptions.
2
1.00
0.95
0.90
Proportion Positive
0.85
0.80
0.75
0.70
0.65
0.60
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Period
Figure G.1. p-Chart for aerobic plate count in ground beef. Note the exceptions
at periods #7 and #22 which imply a drop in the proportion of samples with
enumerative results >0 from that expected. There is also a general tendency to
fall under the average.
3
11.0
9.0
Concentratrion (log10)
7.0
5.0
3.0
1.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Positive #
Figure G.2. Individuals’ chart for aerobic plate count in ground beef.
4
Appendix H. Control Chart for Very High Prevalence with Quantitation
This example corresponds to prevalence in the range 95% to 100%. For this
case, the number of samples with enumerative results equal to 0 is low enough
to apply eq. F.2. (Appendix F).
Small rational subgroups (i.e., lots) are now possible, corresponding to shift, day,
week, month or quarter, within which operations are expected to be consistent.
The use of such subgroups allows control not only of the mean, but also the
spread of the process.
where the center line is μ (determined from historical data), n is the subgroup
sample size, A 2 is a numerical factor available from standard control chart tables
(Appendix B) and R ave is the average range from the same historical dataset.
The range R is the difference between the maximum and minimum values in a
subgroup.
where D 4 is obtained from the same table as A 2 . For most purposes, n should
be between 2 and 6, with 4 or 5 preferred. The Range Chart responds to
changes in within subgroup variation.
Table H.1. shows various scenarios for using X-bar-R- control charts for
microbial testing.
1
Table H.1. Average-Range statistical process control chart characteristics for
different sampling time frames
2
Operation Take single unit from each production day during week, across
shifts. The collection of results is the sample. Plot average and
range of each sample.
Average chart Range chart
Purpose Average Chart PURPOSE: Range Chart PURPOSE:
Are the sample averages Is the variation observed
consistent across weeks, free consistent within and across
of trends and disturbances? different weeks' production?
Assignable causes ASSIGNABLE CAUSES: ASSIGNABLE CAUSES:
1. Personnel changes 1. New employees within
between weeks. week.
2. Introduction of new raw 2. New management within
material lots. week.
3. Management changes 3. New equipment or
between weeks. procedures within week.
4. Staffing and volume issues 4. Day of week volume or
between weeks. procedure effects.
5. New equipment or
procedures between days
IV. Sample Within Production 'Month', n = 4 (loosened inspection)
Operation Take single unit from each production week, across days and
shifts. The collection of results is the sample. Plot average and
range of each sample
Average chart Range chart
Purpose Are the sample averages Is the variation observed
consistent across months, free consistent within and across
of trends and disturbances? different months' production?
Assignable causes 1. Personnel changes 1. New employees within
between months. month.
2. Introduction of new vendors. 2. New management within
3. Management changes month.
between months. 3. New equipment or
4. Staffing and volume issues procedures within month.
between months 4. New equipment or
5. New equipment or procedures within month
procedures between months.
6. Seasonal changes in raw
materials and production
volume
3
Example: Aerobic Plate Count in Bagged Salad
Received lots of bagged salad mixed are tested for APC. Five samples are
taken per lot. Based on prior test data, the long-term average APC log 10
concentration is 5.19 with standard deviation of 1.34 and an average range of
3.12.
Figure H.1. shows the Average Chart for recent data. No abnormal behavior is
apparent.
Figure H.2. shows the Range Chart for the same data. Two exceptions at the
UCL are prominent.
8.00
7.00
6.00
5.00
Sample Averages
4.00
3.00
1.00
0 10 20 30 40 50 60
Lot
4
9.00
8.00
7.00
6.00
Sample Ranges
5.00
4.00
3.00
2.00
0.00
0 10 20 30 40 50 60
Lot
Figure H.2. Range Chart for APC in bagged salad mix. Note the two out-of-
control points.
5
Appendix I. Number of Samples and Statistical Uncertainty about Setting
Control Limits for X Bar Charts
There are no firm rules for how much data are needed to develop control charts.
As the number of lots used to develop a control chart increases, the uncertainty
about setting the control limits (too high or too low) decreases. However, there
are diminishing returns to using more data.
Figure I.1. Uncertainty about 3 Sigma Control Limits for Mean (µ = 3 log10
CFU/g, σ = 1 log10 CFU/g)
The uncertainty about the control limits depends on the number of samples per
lot (n) as well as the number of lots (subgroups) used to develop the limits.
Assuming a lognormal distribution with geometric mean = 3 log10 CFU/g (1,000
CFU/g) and a standard deviation = 1 log10 CFU/g, Figure I.2 compares the
1
relationship between the uncertainty about average chart control limits and the
number of lots used to develop the limits for n = 5 and n = 3 samples per lot (i.e.,
rational subgroup). Both cases show an initial rapid decrease in uncertainty in
control limits followed by diminishing returns from additional lots. However, the
control limit uncertainty for n = 3 samples per lot (subgroup) starts from a
substantially higher level relative to n = 5 samples per lot (subgroup).
Figure I.2. 90% Confidence Range of 3 Sigma Control Limits for Mean (µ = 3
log10 CFU/g, σ= 1 log10 CFU/g)
It should be remembered that all data included in the dataset used to compute
the control limits should be consistent with a tenable assumption of a time period
of unchanging conditions (e.g., the same season of raw materials, the same
production process, the same equipment).
2
Appendix J. Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for United
States Department of Defense (DoD) use.
Introduction
The microbiological and chemical limits provided in the following tables are useful for suppliers and DOD to assess process control and sanitary
conditions associated with the production of various foods. The limits and sample size or procedures described in these tables are not regulatory
limits, although in certain cases, they may reference regulatory limits.
The food categories correspond to those listed in Appendix A where flow diagrams for manufacturing of the foods are provided. The
microbiological data for the various microorganisms or classes of microorganisms can be used to develop statistical process control (SPC) charts,
as well as to gain an understanding of the microbiological quality and safety associated with the various products.
The environmental monitoring program (EMP) data have less utility for development of SPC charts because of the potential high number of
monitoring sites, and thus, the longer time frame required for sufficient data for SPC charts. However, the EMP data have been correlated with
food-product contamination (Kornacki, 2014) and have usefulness for assessing process control, cleaning and sanitation practices, targeting
supplier and DOD resources, and for trending EMP data over time to assess continuous improvement.
Each table in Appendix J includes the microorganisms that are useful for assessing process control and sanitary conditions during production of
foods within the given food category. The microbiological limits for these microorganisms are provided, as well as recommended actions to be
taken if the limits are exceeded. In many instances, the actions include investigating to determine a root cause, developing and implementing
corrective and preventive actions, and conducting follow-up sampling and testing to determine if the corrective and preventive actions have been
effective. In all tables, where applicable, these actions are identified as “Investigate” and “Implement Corrective Actions” to simplify the actions
listed. The investigative and corrective action processes likely will be unique to each situation.
Samples of the food may be taken at numerous points throughout production; these are considered as in-process samples. Samples taken at the
end of the production line also may be tested, with results compared against the microbiological limits. In some cases, these finished product data
may be useful to assist in the development of finished product microbiological criteria; however, initially, these data should be used to assess
process control and sanitary conditions, and compared against the limits provided for each criterion. When samples are taken at the end of the
production line and tested, and results exceed the limits, the recommended action may be to reject the lot of food represented by the sample. This
will be especially true when the microorganism detected is a pathogen and the food will not receive further processing using a validated kill step.
The number of in-process, finished product, or environmental samples to take and test may not be given for all criteria in the tables. In general,
taking more samples is better; and larger numbers of samples taken for pathogens can increase the confidence of detecting pathogens present at
a low prevalence. Analytical unit weights for testing should be a minimum of 25 grams; for pathogen testing, the analytical unit (usually a
composite weight) in the table may specify a particular weight (e.g., 325 or 375 grams) and provide the weights for the individual samples
contributing to the composite sample (e.g., 15 X 25-gram samples to result in a 375-gram analytical unit). The body of the report and Appendix I
discuss how sample numbers affect the design of SPC charts.
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.1. Microbiological Limits for Bottled Water
Notes: The bottled water category includes bottled water described as Artesian, Mineral, Purified, Sparkling or Spring.
a
(Code of Federal Regulations, 2014c; European Communities, 2007; World Health Organization (WHO), 2013, 2014)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J. 2. Microbiological Limits for Ice, Packaged
b
(Code of Federal Regulations, 2014c; European Communities, 2007; World Health Organization (WHO), 2013, 2014)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J. 3. Microbiological and Chemical Limits for Juices and Drinks, Pasteurized and Refrigerated
Notes: Examples of these products are orange juice, carrot juice, and some tea beverages. These products are pasteurized but must be kept
refrigerated to prevent spoilage. Raw citrus juices sold in the U.S. will require additional testing (Subpart B, Juice HACCP regulations). Juices with a
pH>4.6 should address control of Clostridium botulinum.
c
(Code of Federal Regulations, 2014a; U. S. Department of Health and Human Services, 2004)
d
(R.B. Tompkin, 2002; R. B. Tompkin, Scott, Bernard, Sveum, & Gombas, 1999)
e
(U. S. Department of Health and Human Services, 2005e)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.4. Microbiological and Chemical Limits for Shelf-stable Beverages
Notes: Examples of these products are carbonated beverages, commercial sterility/ultra-high temperature/aseptic beverages, and some juice
drinks. Microbiological control is accomplished by one or more of the following: low pH, pasteurization (UHT), and carbonation
There are no microbiological limits set for shelf-stable beverages as these products are considered commercially-sterile (i.e., stable at room
temperature under normal handling and storage conditions). Suppliers should be verifying the raw materials used in the formulation of these
products before the process providing commercial sterility. Shelf-stable liquid products should be examined by means other than routine
microbiological testing; if inspection finds bulging containers, pH changes, odors, etc., then further investigation is warranted.
f
(U. S. Department of Health and Human Services, 2005e)
g
(Elliott & Kataoka, 2013)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.5. Microbiological Limits for Dairy- Butter, margarine
Notes: Either formulated with sufficient salt or lactic acid (for unsalted butter) to prevent growth or refrigerated; products containing added
seasoning/herbs/spices may have additional requirements
h
(National Academies of Science, 2003)
i
Coliforms or Enterobacteriaceae are acceptable for routine testing. (Kornacki, Gurtler, & Stawick, 2013)
j
(Craven, Eyles, & Davey, 2003)
k
(R.B. Tompkin, 2002; R. B. Tompkin et al., 1999)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.6. Microbiological Limits for Dairy-Cheese, Hard
Notes: Ex. Parmesan, Cheddar, aw<0.95 and pH<5.6. All cheeses are made with pasteurized milk.
l
(U.S. Department of Health and Human Services, 2010)
m
(R.B. Tompkin, 2002; R. B. Tompkin et al., 1999)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.7. Microbiological Limits for Dairy-Cheese, Soft, Semi-Soft, Surface-Ripened
Notes: Ex. Brie, Fresh Mozzarella, aw>0.95 and pH>5.4. All cheeses are made with pasteurized milk.
n
(Australia New Zealand Food Authority, 2014; U.S. Department of Health and Human Services, 2010)
o
(R.B. Tompkin, 2002; R. B. Tompkin et al., 1999)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.8. Microbiological Limits for Dairy-Cultured, pH <4.8
p
(U. S. Department of Health and Human Services, 2011j; U.S. Department of Health and Human Services, 2010)
q
(R.B. Tompkin, 2002; R. B. Tompkin et al., 1999)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.9. Microbiological Limits for Dairy-Cultured, pH >4.8 and < 5.4
Notes: Ex. Cottage cheese, cream cheese, moisture >50%; active pH control required
r
(Bradley et al., 2013; U. S. Department of Health and Human Services, 2011j)
s
(R.B. Tompkin, 2002; R. B. Tompkin et al., 1999)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.10. Microbiological Limits for Dairy-Dried Products
Notes: Ex. NFDM, whey powder. This does not cover dried dairy ingredients used in infant formula; those requirements are more stringent.
t
(International Commission for the Microbiological Specifications for Foods (ICMSF), 2011; U. S. Department of Health and Human Services,
2011j; U.S. Department of Health and Human Services, 2010)
u
Coliforms or Enterobacteriaceae are acceptable for routine testing. (Kornacki et al., 2013)
v
(R.B. Tompkin, 2002; R. B. Tompkin et al., 1999)
w
(Chen et al., 2009; Grocery Manufacturers Association, 2009)
x
Recommend 1500 g per lot when high volumes of product are produced per lot (or production day).
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.11. Microbiological Limits for Dairy-Frozen Desserts
Notes:
y
(International Commission for the Microbiological Specifications for Foods (ICMSF), 2011)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.12. Microbiological Limits for Dairy-Milk and Milk Products (Fluid)
Notes: Ex. Fluid milk, cream; Pasteurized, refrigerated; alkaline phosphatase negative (less than 2.0 micrograms phenol equivalent per g)
z
(Bradley et al., 2013; U. S. Department of Health and Human Services, 2011j)
aa
Coliforms or Enterobacteriaceae are acceptable for routine testing. (Kornacki et al., 2013)
bb
(R.B. Tompkin, 2002; R. B. Tompkin et al., 1999)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.13. Microbiological Limits for Dairy- Processed Cheese
Notes: Manufactured by heating cheese with water, emulsifier and other ingredients to kill vegetative pathogens; molten cheese may then be hot-filled
into loaves or blocks and chilled or cut into individual slices for use; these cheeses are intended to be stored refrigerated. Shelf-stable hot-filled cheese
spreads or cheese sauces must be formulated for safety to inhibit Clostridium botulinum.
Criteria & EMP Microbiological Limit Recommended Action if Comments
Target Microorganism Routine Non-Routine Limit is Exceeded
Test for products that are not hot-
filled directly into final container.
APC limit may be adjusted subject
3 Investigate and implement corrective to control chart associated with
APC/SPC 10 /g
action. Statistical Process Control.
Populations are predominantly
sporeformers or heat-stable
spoilage microorganisms.
Test for products that are not hot-
Coliforms <10/g Investigate, implement corrective action
filled directly into final container
Investigate and implement corrective Test for products that are not hot-
E. coli (generic) <10/g
action filled directly into final container
Investigate, consider Zone 1 and
cc Negative for
Listeria spp. (EMP) finished product testing, implement
Zone 2 or 3
corrective action
Investigate, implement corrective
action. Test for products that are not hot-
S. aureus <100/g 4
if >10 /g, reject lot due to potential for filled directly into final container
enterotoxin production
Investigate, consider Zone 1 and Test EMP in areas where products
dd Negative for
Salmonella (EMP) finished product testing, implement are not hot-filled directly into final
Zone 2 or 3
corrective action container
Divert for reprocessing, if appropriate,
Test for products that are not hot-
Salmonella (product) Negative in 125 g or reject. Investigate and implement
filled directly into final container
corrective action
cc
(R.B. Tompkin, 2002; R. B. Tompkin et al., 1999)
dd
(Chen et al., 2009; Grocery Manufacturers Association, 2009)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.14. Microbiological Limits for Egg Products-Pasteurized, Processed
Notes:
ee
(Australia New Zealand Food Authority, 2014; European Commission, 2005; International Commission for the Microbiological Specifications for
Foods (ICMSF), 2011; U. S. Department of Agriculture Food Safety Inspection Service, 2014c)
ff
(R.B. Tompkin, 2002; R. B. Tompkin et al., 1999)
gg
(Chen et al., 2009; Grocery Manufacturers Association, 2009)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.15. Microbiological Limits for Egg Products-Shell Eggs, Raw
Notes:
hh
(U. S. Department of Health and Human Services, 2009)
ii
Coliforms, E. coli or Enterobacteriaceae are acceptable for routine testing. (Kornacki et al., 2013)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.16. Microbiological Limits for Grain-based products-RTE, baked items, refrigerated or TCS
Notes: Examples: focaccia, custard or cream-filled pastries, pies. Qualifying information: APC counts may be high due to containing ingredients
prepared with starter culture
jj
(R.B. Tompkin, 2002; R. B. Tompkin et al., 1999)
kk
(Andrews, Jacobson, & Hammack, 2014)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.17. Grain-based products-RTE, baked items, shelf stable, non-TCS
Notes: Examples: bread. If raw ingredients added after baking step, additional risks should be considered. ICMSF 8 does not recommend routine
testing.
:
ll
(International Commission for the Microbiological Specifications for Foods (ICMSF), 2011)
mm
Coliforms or Enterobacteriaceae are acceptable for routine testing. (Kornacki et al., 2013)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.18. Microbiological Limits for Grain Based Products, RTE, cereals
Notes: Examples: breakfast cereals. Grain based product undergoes a lethality step; mycotoxin surveillance testing completed on incoming
grains as pre-requisite program with limits based on individual country’s regulations
nn
Coliforms or Enterobacteriaceae are acceptable for routine testing. (Kornacki et al., 2013)
oo
(Chen et al., 2009; Grocery Manufacturers Association, 2009)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.19. Microbiological Limits for Grain Based Products – RTE, cold-pressed bars
Notes; Ex. granola bars; Qualifying information: ingredients will undergo mycotoxin surveillance testing as appropriate; shelf-stable, aw <0.85
pp
Either coliform or Enterobacteriaceae testing is appropriate. (Kornacki et al., 2013)
qq
(Chen et al., 2009; Grocery Manufacturers Association, 2009)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.20. Microbiological Limits for Grain Based Products non-RTE, dry, flour based mixes
rr
Notes: Flour can contain pathogens occasionally and should be subjected to a lethality step prior to consumption
rr
(Sperber & North American Millers' Association Microbiology Working Group, 2007)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.21. Microbiological Limits for Grain Based Products – Non-RTE, pasta, dried or refrigerated
ss
(R.B. Tompkin, 2002; R. B. Tompkin et al., 1999)
tt
(Chen et al., 2009; Grocery Manufacturers Association, 2009)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.22. Microbiological and Chemical Limits for Meals and Entrees—Non-RTE, ready-to-cook meals, includes raw ingredients
Notes: This category includes a wide variety of products and processes that will influence appropriate testing choices. Depending on their
ingredients, foods that are likely to be prepared by microwave heating or those that are not labeled with validated cooking instructions may require
more stringent testing (e.g. Raw poultry, beef or flour ingredients may necessitate routine, instead of non-routine, testing for Salmonella or E. coli
(O157:H7 or other STEC).
uu
Coliforms or Enterobacteriaceae are acceptable for routine testing. (Kornacki et al., 2013)
vv
(Chen et al., 2009; Grocery Manufacturers Association, 2009)
ww
(U. S. Department of Health and Human Services, 2011a)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Notes: Survey data indicates a wide range in microbial populations depending on specific food. Items may include ingredients that are raw.
Criteria & EMP Microbiological Limit Recommended Action if Comments
Target Microorganism Routine Non-Routine Limit is Exceeded
Investigate, implement corrective
Test if food contains components
3 action.
B. cereus 10 /g 4 that are high risk for B. cereus, such
If >10 /g, reject lot due to potential for
as cooked rice
enterotoxin production
xx 4
Coliforms 10 /g Investigate, implement corrective action
E. coli (generic) 100/g Investigate, implement corrective action
4
Enterobacteriaceae 10 /g Investigate, implement corrective action
Investigate, consider Zone 1 and
yy Negative for
Listeria spp. (EMP) finished product testing, implement
Zone 2 or 3
corrective action
Investigate, implement corrective
3 action.
S. aureus 10 /g 4
If >10 /g, reject lot due to potential for
enterotoxin production
Investigate, consider Zone 1 and
zz Negative for
Salmonella (EMP) finished product testing, implement
Zone 2 or 3
corrective action
Reject lot; Investigate, implement 375-g analytical unit composed of
Salmonella (product) Negative in 375 g
corrective action 15 x 25-g samples
Histamine testing appropriate only
when scombroid species are
aaa Reject lot; Investigate, implement
Histamine 50 ppm in 250 g present; The FDA Hazards and
corrective action
Controls Guide lists a defect action
level of 50ppm.
xx
Coliforms, E. coli, or Enterobacteriaceae are acceptable for routine testing. (Kornacki et al., 2013)
yy
(R.B. Tompkin, 2002; R. B. Tompkin et al., 1999)
zz
(Chen et al., 2009; Grocery Manufacturers Association, 2009)
aaa
(U. S. Department of Health and Human Services, 2011a)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.24. Microbiological Limits for Meals and Entrees—RTE, sous-vide, cook and chill
Notes: These products receive a lethality treatment; presence of vegetative microbes represents post-process contamination. If not using validated
sous-vide process for a 6-log reduction of non-proteolytic Clostridium botulinum, testing of vegetative microorganisms is warranted.
bbb
Coliforms or Enterobacteriaceae are acceptable for routine testing. (Kornacki et al., 2013)
ccc
(U. S. Department of Agriculture Food Safety Inspection Service, 1999; U.S. Department of Agriculture Food Safety Inspection Service, 1999)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.25. Microbiological Limits for Meat—Beef and Pork, Non-RTE, raw (intact, non-intact)
ddd
(National Advisory Committee on Microbiological Criteria for Foods, 2013; U.S. Department of Agriculture Agricultural Marketing Service (AMS),
2015)
eee
(U. S. Department of Agriculture, 2011)
fff
(U. S. Department of Agriculture Food Safety Inspection Service, 2014e)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.26. Microbiological Limits for Meat—Poultry, Non-RTE, raw (intact, non-intact)
ggg
Coliforms, generic E. coli, Enterobacteriaceae are acceptable for testing. (Kornacki et al., 2013)
hhh
(U. S. Department of Agriculture Food Safety Inspection Service, 2014e); Campylobacter is also proposed In the USDA-FSIS Performance
Standards.
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.27. Microbiological Limits for Meat— RTE cooked, perishable
Notes: Ex. Includes beef, pork and poultry products, deli meats, frankfurters
iii
Either coliforms or Enterobacteriaceae are acceptable for testing. (Kornacki et al., 2013)
jjj
(R.B. Tompkin, 2002; R. B. Tompkin et al., 1999)
kkk
(Australia New Zealand Food Authority, 2014; Canada Food Inspection Agency, 2012; U. S. Department of Agriculture Food Safety Inspection
Service, 2014a; U.S. Department of Health and Human Services, 2008)
lll
(U. S. Department of Agriculture Food Safety Inspection Service, 1999)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.28. Microbiological Limits for Meat— RTE fermented, dried
Notes: Includes beef, pork and poultry products, Jerky, dried fermented sausage, dried acidified meat sticks; e.g. aw <0.85 or pH <5.3 and aw <0.92 for
vacuum-packaged meat sticks. Products should be manufactured with a validated kill step for E. coli O157:H7 (beef) or Salmonella (pork, poultry) as
appropriate for the given meat matrix.
mmm
(U. S. Department of Agriculture Food Safety Inspection Service, 2014a)
nnn
Either coliforms or Enterobacteriaceae are acceptable for testing. (Kornacki et al., 2013)
ooo
(American Meat Institute Foundation, 1997)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.29. Microbiological and Chemical Limits for Nuts and Nut Butters – RTE, Not processed for lethality
Notes: Ex.: include peanuts, tree nuts (e.g., walnuts, almonds, pecans, pistachios, macadamia)
ppp
(Chen et al., 2009; Grocery Manufacturers Association, 2009)
qqq
(Andrews & Hammack, 2003; U.S. Food and Drug Administration, 2015)
rrr
(U. S. Department of Health and Human Services, 2005a, 2005c, 2005g, 2005i)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.30. Microbiological and Chemical Limits for Nuts and Nut Butters – RTE, processed for lethality
Notes: Ex.: peanut butter, almond butter, roasted nuts
sss
(Chen et al., 2009; Grocery Manufacturers Association, 2009)
ttt
(Andrews & Hammack, 2003; U.S. Food and Drug Administration, 2015)
uuu
(U. S. Department of Health and Human Services, 2005a, 2005c, 2005g, 2005i)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.31. Microbiological Limits for Produce—Fruits and Vegetables, Cut, Frozen or Refrigerated
vvv
(European Commission, 2005; Health Canada, 2008)
www
(R.B. Tompkin, 2002; R. B. Tompkin et al., 1999)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.32. Microbiological Limits for Produce—Fruits and Vegetables, Whole
Notes: Ex. products customarily consumed without cooking, tomatoes, cantaloupes, avocado, mangoes, apples, celery, carrots, berries, whole lettuce
xxx
(European Commission, 2005; Health Canada, 2008; International Commission for the Microbiological Specifications for Foods (ICMSF), 2011)
yyy
(R.B. Tompkin, 2002; R. B. Tompkin et al., 1999)
zzz
(Chen et al., 2009; Grocery Manufacturers Association, 2009)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.33. Microbiological Limits for Produce—Produce, Mushrooms
aaaa
(European Commission, 2005; Health Canada, 2008)
bbbb
(R.B. Tompkin, 2002; R. B. Tompkin et al., 1999)
cccc
(Chen et al., 2009; Grocery Manufacturers Association, 2009)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.34. Microbiological Limits for Produce—Packaged Salads and Leafy Greens
dddd
(International Commission for the Microbiological Specifications for Foods (ICMSF), 2011)
eeee
(R.B. Tompkin, 2002; R. B. Tompkin et al., 1999)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.35. Microbiological Limits for Produce— vegetable sprouts
ffff
(Health Canada, 2006; U.S. Food and Drug Administration, 1999)
gggg
Sampling spent irrigation water; collect the total of 1-liter of spent irrigation water from various trays of growing sprouts. Two x 375-ml
subsamples are used for Salmonella detection and 2 x 100-ml subsamples are used for detection of E. coli O157:H7.
hhhh
(R.B. Tompkin, 2002; R. B. Tompkin et al., 1999)
iiii
(Australia New Zealand Food Authority, 2014; Hitchens & Jinneman, 2013)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.36. Microbiological and Chemical Limits for Seafood, Raw
Notes: Ex.: fish, shrimp, crabs. Verification testing for histamine in scombroid species only. The FDA Hazards and Controls Guide lists a defect action
level of 50ppm. See Table J.37, J.38 and J. 39 if raw seafood may be used for applications without full cook such as for sushi or ceviche, additional
testing may be appropriate.
jjjj
(Ahmed & Institute of Medicine (U.S.). Committee on Evaluation of the Safety of Fishery Products., 1991; Australia New Zealand Food Authority,
2001, 2014; International Commission for the Microbiological Specifications for Foods (ICMSF), 2011; U. S. Department of Health and Human
Services, 2011a)
kkkk
Sample size may vary depending on intended usage, e.g. for sushi or ceviche without full cook (Andrews & Hammack, 2003)
llll
(U. S. Department of Health and Human Services, 2011a)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.37. Microbiological and Chemical Limits for Seafood--RTE, Fish, Cold Smoked
Notes: Verification testing for histamine in scombroid species only. The FDA Hazards and Controls Guide lists a defect action level of 50ppm
mmmm
(International Commission for the Microbiological Specifications for Foods (ICMSF), 2011; U. S. Department of Health and Human Services,
2011a)
nnnn
(Scott et al., 2005; R.B. Tompkin, 2002; R. B. Tompkin et al., 1999)
oooo
(U. S. Department of Health and Human Services, 2011a)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.38. Microbiological and Chemical Limits for Seafood-- RTE, cooked or hot smoked
Notes: Ex.: includes cooked and hot smoked products, cooked crabmeat, lobster meat, shrimp, crayfish, surimi, seafood salads, hot-smoked fish.
Histamine testing recommended for scombroid species only with Defect Action Level 50ppm
Criteria & EMP Microbiological Limit Recommended Action if Comments
Target Microorganism pppp Routine Non-Routine Limit is Exceeded
Investigate Source of Post Cook Handling
APC 105/g
and Storage Contamination
qqqq For cooked seafood other than
Coliforms 100/g Investigate; implement corrective action
shrimp and crabmeat
nnnn 3 For cooked crabmeat- fresh (Handled
Coliforms 5x10 /g Investigate; implement corrective action
after final cook)
nnnn 3 For cooked shrimp (Handled after
Coliforms 10 /g Investigate; implement corrective action
final cook)
Investigate, consider Zone 1 and finished
Negative for Zone
Listeria spp. (EMP) rrrr product testing, implement corrective
2 or 3
action
Negative in 5 Divert for reprocessing, if appropriate, or
Listeria monocytogenes
individual 25-g reject. Investigate and implement
(product)
analytical units corrective action
Investigate, implement corrective action. Routine testing is recommended
S. aureus 103/g If >104 /g, reject lot due to potential for especially for products that are
enterotoxin production handled after the final cook (kill) step
pppp
(Australia New Zealand Food Authority, 2014; International Commission for the Microbiological Specifications for Foods (ICMSF), 1986; U. S.
Department of Health and Human Services, 2011a)
qqqq
(Buchanan, 1991)
rrrr
(Scott et al., 2005; R.B. Tompkin, 2002; R. B. Tompkin et al., 1999)
ssss
(U. S. Department of Health and Human Services, 2011a)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.39. Microbiological Limits for Seafood--RTE, Raw Molluscan Shellfish
Notes: Ex.: molluscan shellfish such as oysters, clams, mussels, scallops intended to be eaten without a full cook. Shellfish must be from approved
harvest waters from countries with MOU (I.e., New Zealand, Mexico, Korea, Canada) with the United States. Shellfish from any other source should not
be accepted by DOD. Investigational testing for aquatic toxins. While V. vulnificus and parahaemolyticus may be a concern in RTE, raw molluscan
shellfish, no limits can be recommended at this time.
tttt
(National Advisory Committee on Microbiological Criteria for Foods, 1992; U. S. Department of Health and Human Services, 2011a)
Each analytical unit is comprised of 10-12 individual shellfish composited into one unit. See text FDA Fish and Fishery Products Hazards and
Controls Guidance for sampling procedure.
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Table J.40. Microbiological Limits for Spices, Herbs, Coffee and Tea
uuuu
(Health Canada, 2008; Sagoo et al., 2009)
vvvv
(Chen et al., 2009; Grocery Manufacturers Association, 2009)
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
References:
1. Ahmed, F. E., & Institute of Medicine (U.S.). Committee on Evaluation of the Safety of Fishery Products. (1991).
Seafood safety Retrieved from http://www.nap.edu/catalog/1612.html
2. American Meat Institute Foundation. (1997). Good Manufacturing Practices for Fermented Dry & Semi-Dry
Sausage Products. Retrieved February 18, 2015, from
http://www.meathaccp.wisc.edu/Model_Haccp_Plans/assets/GMP%20Dry%20Sausage.pdf
3. Andrews, W. H., & Hammack, T. S. (2003). BAM: Chapter 1. Food Sampling/Preparation of Sample Homogenate.
In Bacteriological Analytical Manual, Edition 8, Revision A, 1998. Retrieved April 13, 2014, from
http://www.fda.gov/Food/FoodScienceResearch/LaboratoryMethods/ucm063335.htm
4. Andrews, W. H., Jacobson, A., & Hammack, T. (2014). BAM: Salmonella, Chapter 5 FDA, Bacteriological Analytical
Manual (Vol. 2014).
5. Australia New Zealand Food Authority. (2001). Guidelines for the microbiological examination of ready-to-eat
foods. Retrieved September 23, 2014, from
http://www.foodstandards.gov.au/publications/pages/guidelinesformicrobi1306.aspx
6. Australia New Zealand Food Authority. (2014). Criteria for Listeria monocytogenes – Microbiological Limits for
Foods.
7. Bradley, R. L., Houck, K. S., Smukowski, M., Bradley, R. L., Houck, K., & Smukowski, M. (2013). Milk and Milk
Products. In S. Doores, Y. Salfinger, & M. L. Tortorello (Eds.), Compendium of Methods for Microbiological
Examination of Foods (5th ed.): American Public Health Association.
8. Buchanan, R. L. (1991). Microbiological criteria for cooked, ready-to-eat shrimp and crabmeat. Journal of Food
Technology, 45(4), 157-160.
9. Canada Food Inspection Agency. (2012). Implementation of the 2011 Health Canada Policy on Listeria
monocytogenes in Ready-to-Eat Foods. Retrieved September 19, 2014, from
http://www.inspection.gc.ca/food/fish-and-seafood/product-inspection/fish-
listeria/eng/1338572158747/1338572678417
10. Chen, Y. H., Scott, V. N., Freier, T. A., Kuehm, J., Moorman, M., Meyer, J., . . . Banks, J. (2009). Control of
Salmonella in low-moisture foods II: Hygiene practices to minimize Salmonella contamination and growth. Food
Protection Trends, 29, 435-445.
11. Code of Federal Regulations. (2014a). 21 CFR 120. Hazard Analysis and Critical Control Point (HACCP) Systems.
Retrieved February 19, 2015, from
http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfCFR/CFRSearch.cfm?CFRPart=120&showFR=1
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
12. Code of Federal Regulations. (2014c). 21 CFR 165. Requirements for Specific Standardized Beverages.
Retrieved February 3, 2014, from http://www.ecfr.gov/cgi-
bin/retrieveECFR?gp=&SID=d26e287c93ddfdc1dcc06d9b896b5d5f&r=PART&n=21y2.0.1.1.38
13. Craven, H. M., Eyles, M. J., & Davey, J. A. (2003). Enteric indicator organisms in foods. In A. D. Hocking (Ed.),
Foodborne Microorganisms of Public Health Significance, 6th ed. Tempe, Australia: Australian Institute of Food
Science and Technology Inc.
14. Elliott, P. H., & Kataoka, A. (2013). Canned Foods - Test for Commercial Sterility. In S. Doores, Y. Salfinger, & M.
L. Tortorello (Eds.), Compendium of Methods for Microbiological Examination of Foods (5th ed.): American Public
Health Association.
15. European Commission. (2005). Microbiological Criteria for Foodstuff. Commission Regulation (EC) No. 2073/2005.
Retrieved September 23, 2014, from http://www.fsai.ie/uploadedFiles/Reg2073_2005(1).pdf
16. European Communities. (2007). Drinking Water Regulations. Retrieved September 23, 2014, from
http://www.irishstatutebook.ie/pdf/2007/en.si.2007.0278.pdf
17. Grocery Manufacturers Association. (2009). Control of Salmonella in low-moisture foods. Retrieved August 14,
2014, from http://www.gmaonline.org/downloads/technical-guidance-and-tools/SalmonellaControlGuidance.pdf
18. Health Canada. (2006). Health Products and Food Branch (HPFB) Guidance for Industry: Sample Collection and
Testing for Sprouts and Spent Irrigation Water (pp. 1-18).
19. Health Canada. (2008). Standards and Guidelines for Microbiological Safety of Food – An Interpretative Summary.
Retrieved September 2014, 2014, from http://www.hc-sc.gc.ca/fn-an/res-rech/analy-meth/microbio/volume1-
eng.php
20. Hitchens, A. D., & Jinneman, K. (2013). BAM: Chapter 10. Detection and Enumeration of Listeria monocytogenes,
Revised 2013 FDA, Bacteriological Analytical Manual 8th Edition, Revision A, 1998. . Retrieved April 13, 2014,
from http://www.fda.gov/food/foodscienceresearch/laboratorymethods/ucm071400.htm
21. International Commission for the Microbiological Specifications for Foods (ICMSF). (1986). Microorganisms in
Foods 2: Sampling for Microbiological Analysis: Principles and Specific Applications (2nd ed.). University of
Toronto, Canada: University Toronto Press.
22. International Commission for the Microbiological Specifications for Foods (ICMSF). (2011). Microorganisms in
Foods 8: Use of Data for Assessing Process Control and Product Acceptance. New York: Springer.
23. Kornacki, J. I. (2014). An Environmental Sampling Approach to Product Risk Assessment. Food Safety Magazine.
24. Kornacki, J. I., Gurtler, J. B., & Stawick, B. A. (2013). Enterobacteriaceae, Coliforms, and Escherichia coli as
Quality and Safety Indicators, Updated September 2013. In Y. S. S. Doores, M. L.Tortorello (Ed.), Compendium of
Methods for the Microbiological Examination of Foods (5th ed.).
http://ajph.aphapublications.org/doi/book/10.2105/MBEF.0222: American Public Health Association.
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
25. National Academies of Science. (2003). Scientific Criteria and Performance Standards to Control Hazards in Dairy
Scientific Criteria to Ensure Safe Foods (pp. 225-247). Washington, D.C.: Institute of Medicine. National Research
Council of the National Academies.
26. National Advisory Committee on Microbiological Criteria for Foods. (1992). Microbiological criteria for raw
molluscan shellfish. J. Food Prot, 55(6), 463-480.
27. National Advisory Committee on Microbiological Criteria for Foods. (2013). Expedited response to the questions
posed by the USDA agricultural marketing service to support ground beef purchase for the federal food and
nutrition assistance programs. J. Food Prot, 76, 523-5237.
28. Sagoo, S. K., Little, C. L., Greenwood, M., Mithani, V., Grant, K. A., McLauchlin, J., . . . Threlfall, E. J. (2009).
Assessment of the microbiological safety of dried spices and herbs from production and retail premises in the
United Kingdom. Food Microbiol, 26(1), 39-43. doi: 10.1016/j.fm.2008.07.005
29. Scott, V. N., Wiedmann, M., Hicks, D., Collette, R., Jahncke, M. L., & Gall, K. (2005). Guidelines for Listeria testing
of environmental, raw product and finished product samples in smoked seafood processing facilities. Food
Protection Trends, 25(1), 23-34.
30. Sperber, W. H., & North American Millers' Association Microbiology Working Group. (2007). Role of microbiological
guidelines in the production and commercial use of milled cereal grains: a practical approach for the 21st century.
J. Food Prot, 70(4), 1041-1053.
31. Tompkin, R. B. (2002). Control of Listeria monocytogenes in the food-processing environment. J. Food Prot, 65(4),
709-725.
32. Tompkin, R. B., Scott, V. N., Bernard, D. T., Sveum, W. H., & Gombas, K. S. (1999). Guidelines to prevent post-
processing contamination from Listeria monocytogenes. Dairy, Food and Environmental Sanitation, 19(8), 551-562.
33. U. S. Department of Agriculture. (2011). Shiga-toxing Producing Escherichia coli in Certarin Raw Beef Products.
Federal Register, 76(182), 58157-58165.
34. U. S. Department of Agriculture Food Safety Inspection Service. (1999). Appendix B: Compliance Guidelines for
Cooling Heat-Treated Meat and Poultry Products (Stabilization). June 1999.
http://www.fsis.usda.gov/OPPDE/rdad/frpubs/95-033F/95-033F_Appendix%20B.htm. Retrieved December 12,
2012
35. U. S. Department of Agriculture Food Safety Inspection Service. (2014a). Compliance guidelines to control Listeria
monocytogenes in post-lethality exposed ready-to-eat meat and poultry products. Updated January 2014.
Retrieved August 14, 2014, from http://www.fsis.usda.gov/wps/wcm/connect/d3373299-50e6-47d6-a577-
e74a1e549fde/Controlling_LM_RTE_Guideline_0912?MOD=AJPERES
36. U. S. Department of Agriculture Food Safety Inspection Service. (2014c). FSIS Microbiological Testing Program for
Salmonella in Pasteurized Egg Products, 1995–2013. Retrieved August 14, 2014, from
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
http://www.fsis.usda.gov/wps/portal/fsis/topics/data-collection-and-reports/microbiology/testing-program-for-rte-
meat-and-poultry-products/microbiological-testing-program-for-pasteurized-egg-products/pasteurized-egg-products
37. U. S. Department of Agriculture Food Safety Inspection Service. (2014e). Pathogen Reduction – Salmonella and
Campylobacter Performance Standards Verification Testing. Retrieved January 23, 2015, from
http://www.fsis.usda.gov/wps/wcm/connect/b0790997-2e74-48bf-9799-
85814bac9ceb/28_IM_PR_Sal_Campy.pdf?MOD=AJPERES
38. U. S. Department of Health and Human Services. (2004). Guidance for Industry Juice HACCP Hazards and
Controls Guidance.
39. U. S. Department of Health and Human Services. (2005a). Aflatoxin in Peanuts and Peanut Products. CPG Sec.
570.375.
40. U. S. Department of Health and Human Services. (2005c). Brazil Nuts - Adulteration with Aflatoxin. CPG Sec.
570.200.
41. U. S. Department of Health and Human Services. (2005e). Patulin in Apple Juice, Apple Juice Concentrates and
Apple Juice.
42. U. S. Department of Health and Human Services. (2005g). Pistachio Nuts - Aflatoxin Adulteration. CPG Sec.
570.500.
43. U. S. Department of Health and Human Services. (2005i). Tree Nuts - Adulteration with Filth, Involving the
Presence of the Organism Escherichia coli. CPG Sec. 570.450.
44. U. S. Department of Health and Human Services. (2009). Prevention of Salmonella Enteritidis in Shell Eggs During
Production, Storage, and Transportation: Final Rule. 21 CFR Parts 16 and 118, Federal Register 74(130).
45. U. S. Department of Health and Human Services. (2011a). Fish and Fishery Products Hazards and Controls
Guidance - Fourth Edition. Retrieved August 14, 2014, from
http://www.fda.gov/food/guidanceregulation/guidancedocumentsregulatoryinformation/seafood/ucm2018426.htm.
46. U. S. Department of Health and Human Services. (2011j). Grade “A” Pasteurized Milk Ordinance, Including
Provisions from the Grade “A” Condensed and Dry Milk Products and Condensed and Dry Whey--Supplement I to
the Grade “A” Pasteurized Milk Ordinance.
47. U.S. Department of Agriculture Agricultural Marketing Service (AMS). (2015). Microbiological Testing of AMS
Purchased Meat, Poultry and Egg Commodities. . Retrieved February 22, 2015, from
http://www.ams.usda.gov/AMSv1.0/ams.fetchTemplateData.do?template=TemplateA&navID=MicrobialTestingofCo
mmodities&rightNav1=MicrobialTestingofCommodities&topNav=&leftNav=&page=FPPMicroDataReports&resultTy
pe=&acct=lsstd
48. U.S. Department of Agriculture Food Safety Inspection Service. (1999). Performance standards for the production
of certain meat and poultry products. Retrieved October 18, 2014, from
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
https://www.federalregister.gov/articles/1999/01/06/99-32/performance-standards-for-the-production-of-certain-
meat-and-poultry-products
49. U.S. Department of Health and Human Services. (2008). Guidance for Industry: Control of Listeria monocytogenes
in Refrigerated or Frozen Ready-To-Eat Foods; Draft Guidance. Retrieved January 22, 2015, from
http://www.fda.gov/Food/GuidanceRegulation/GuidanceDocumentsRegulatoryInformation/FoodProcessingHACCP/
ucm073110.htm
50. U.S. Department of Health and Human Services. (2010). Compliance Policy Guide Sec. 527.300 Dairy Products -
Microbial Contaminants and Alkaline Phosphatase Activity.
51. U.S. Food and Drug Administration. (1999). Guidance for industry: Sampling and microbial testing of spent
irrigation water during sprout production. Federal Register, 64, 57896-57902.
52. U.S. Food and Drug Administration. (2015). Investigations Operations Manual 2015, Ch. 4 Sampling. Retrieved
June 13, 2015, from http://www.fda.gov/downloads/ICECI/Inspections/IOM/UCM123507.pdf
53. World Health Organization (WHO). (2013). Report on regulations and standards for drinking water quality - draft 12.
Retrieved April 14, 2014, from http://www.who.int/water_sanitation_health/regulations_and_standards_for_drinking-
water_quality/en/
54. World Health Organization (WHO). (2014). Drinking- water quality. Retrieved September 19, 2014, from
http://www.who.int/water_sanitation_health/dwq/en/
Microbiological and chemical limits for foods that are useful to assess process control and insanitary conditions for DoD use.
Appendix K. Sampling for Lot Acceptance or Rejection
In lot acceptance sampling, a sample of n units is drawn from a lot, and a characteristic
of interest (e.g., presence of a pathogen or concentration of an indicator organism) of
the sample units is analyzed. The test portion or analytical unit may represent an entire
sample unit (e.g., enrichment of a 25-g sample), a composite of multiple sample units
(e.g., enrichment of sixty 6.25-g sample units into an approximate 375 g composite test
portion), or a portion or aliquot of a sample unit (e.g., enumeration of 1 ml of diluted
homogenate prepared from a 25-g sample unit). Based on the sample results and the
lot acceptance criteria, a lot disposition decision is made to either accept or reject the
lot.
For lot acceptance sampling to have direct impact, lots must vary with respect to the
analyte of interest. If the analyte is homogeneous among lots, sampling will accept
some lots and reject others by chance alone, and the accepted lots are no better than
the rejected lots. Lot acceptance sampling schemes may involve normal, tightened, or
reduced inspection levels. Lot acceptance sampling may be applied lot-by-lot, or skip-lot
sampling may be applied to less than 100% of lots. Conventional rules for switching
between inspection levels depend on the history of supplier conformance to
specifications and are available as guidelines for frequency and intensity of lot
acceptance sampling schemes (e.g., MIL-STD-1916 [available at
http://guidebook.dcma.mil/34/milstd1916(15).pdf], ISO 2859-1, ISO 2859-3, /ASQ Z1.4
[available at www.asq.org]).
Current microbiological lot acceptance sampling schemes are based on lot acceptance
sampling for attributes. Both microbial presence/absence data obtained from enriched
samples and quantitative concentration data divided into numerical ranges are classified
as attributes. Two-class sampling plans are applicable where product quality is divided
into two attribute classes. For sampling based on detection methods, the classes are
presence or absence. For sampling based on enumeration methods, the classes are
x≤m or x>m, where x is a measure of concentration (e.g., CFU/g, CFU/ml, CFU/cm2, as
appropriate), and m is the microbiological limit separating acceptable from unacceptable
concentrations. Three-class sampling plans are applicable where product quality is
measured by enumeration methods and results are divided into three attribute classes:
x≤m, m<x≤M, or x>M; where m is the limit for a marginally acceptable concentration
(note the change in meaning from the 2-class plan), and M is the limit for an
unacceptable concentration (i.e., similar in meaning to m in the 2-class plan). Two- and
three-class sampling plans also may be used for process control (3), but in the context
of lot acceptance sampling, a consequence of non-conformance with a microbiological
criterion is rejection of the lot represented by the samples.
1
Performance Characteristics of the Sampling Plans
Two-Class Plans
For two-class sampling plans based on a maximum limit (m), the probability of lot
acceptance (pa) can be calculated directly from the proportion of non-conforming
analytical units (p). For presence/absence sampling plans, p refers to the proportion of
test-positive analytical units. For concentration-based sampling plans, p refers to the
proportion of analytical units with CFU/g ≥ m. A two-class sampling plan is defined by
the sample size n and the maximum number of non-conforming analytical units allowed
(acceptance number) c. In general, the probability of lot acceptance (pa) is the
probability that the number of non-conforming independent sample units in the sample
of n is less than or equal to c:
𝑖𝑖=𝑐𝑐 𝑛𝑛
𝑝𝑝𝑎𝑎 = ∑𝑖𝑖=0 𝐶𝐶𝑖𝑖 (𝑝𝑝)𝑖𝑖 (1 − 𝑝𝑝)𝑛𝑛−𝑖𝑖 (eq. K.1.)
𝑛𝑛!
where 𝐶𝐶𝑖𝑖𝑛𝑛 = 𝑖𝑖!(𝑛𝑛−𝑖𝑖)! (the binomial coefficient, or combination of n things taken i at a time),
p is the proportion of non-conforming analytical units, n is the number of samples drawn
from the lot, and c is the acceptance number.
Most two-class sampling plans for pathogens specify c = 0. In this case, pa simplifies to:
For presence/absence sampling plans, it is important to note that the proportion of test
positives depends on the size of the analytical unit (test portion). If analytical units of
different sizes are examined from the same food lot, the proportion of test positives will
be lower for smaller analytical units. For example, for a given concentration, a 25-g
sample is less likely to contain at least one microorganism than a 100-g sample.
Therefore, prevalence needs to be referenced to the size of the analytical unit (e.g.,
2
prevalence in 25 grams). If the detection method is less than 100% sensitive, the
apparent prevalence (proportion of test positives) is less than the true prevalence
(actual proportion of positives). For concentration-based sampling plans, equations K.1.
and K.2. assume that measurement error is negligible (~100% recovery) and that false
positive results are very unlikely (~ 100% specificity). If measurement error is
substantial, the probability of lot acceptance may be affected by the microbial
distribution (not just the proportion of non-conforming sample units) because
measurement error can result in misclassification of sample units above or below the
limit (m).
Figure K.1. presents the operating characteristic curves for c = 0 two-class sampling
plans over a range of sample sizes (n).
Figure K.1. Operating Characteristic Curves for Two-Class Sampling Plans with c = 0.
n 5
c 0
m absence in 25 g
pa p
0.95 0.0102
0.50 0.1294
0.05 0.4507
3
a lot. Assuming that the average concentration is log normally distributed and that the
number of CFU in an analytical unit varies randomly according to the Poisson
distribution (a Poisson-Lognormal distribution), Table K.2. summarizes the performance
of an n = 5, c = 0 two-class sampling plan with m = absence in 25 g. The values
(probability of acceptance) shown in Table K.2. assume a perfect detection method
(100% sensitivity and 100% specificity). The calculations can be performed using on-
line resources (http://www.icmsf.org/publications/sampling_plans.html;
http://www.fstools.org/sampling). For further details about the Poisson-Lognormal used
by these on-line calculators, see (7) .
It should be noted that in terms of consumer health risk, the arithmetic mean (the
average, or expected value) concentration is more relevant than the geometric mean
(median) concentration (5) . Table K.3. presents the arithmetic mean concentration for
the 5 percent probability of acceptance distributions in Table K.2.
4
As an example interpretation of Table K.3, if the arithmetic mean concentration is
0.0396 CFU/g, a 25-g serving would contain an average of 1 CFU. Note that there is no
direct correspondence between the probability of lot acceptance and the level of risk
indicated by the arithmetic mean concentration. This illustrates that evaluating the food
safety impact of sampling plans is not as straightforward as calculating their statistical
operating characteristics.
Three-Class Plans
Three-class sampling plans are based on a marginal limit (m) and a maximum limit (M).
A three-class sampling plan is defined by the sample size n and c the maximum number
of marginal analytical units allowed, or acceptance number. (The acceptance number
for analytical units exceeding M is zero.) For three-class sampling plans, the probability
of lot acceptance (pa) can be calculated directly from the proportion of marginally
acceptable (pm) and unacceptable (pd) analytical units (1):
𝑖𝑖=𝑐𝑐 𝑛𝑛
𝑝𝑝𝑎𝑎 = ∑𝑖𝑖=0 𝐶𝐶𝑖𝑖 (𝑝𝑝𝑚𝑚 )𝑖𝑖 (1 − 𝑝𝑝𝑑𝑑 − 𝑝𝑝𝑚𝑚 )𝑛𝑛−𝑖𝑖 (eq. K.3)
𝑛𝑛!
where 𝐶𝐶𝑖𝑖𝑛𝑛 = 𝑖𝑖!(𝑛𝑛−𝑖𝑖)! (the binomial coefficient), pm is the proportion of marginally
acceptable analytical units (with m < x ≤ M), pd is the proportion of analytical units with x
> M, n is the number of samples drawn from the lot, and c is the acceptable number of
marginal units.
5
Table K.4. Performance Characteristics for n = 5, c = 2 Three-Class Sampling Plan:
Probabilities of Acceptance (pa) for Lots Containing Indicated Proportions pd and pm
Note that some combinations of pm and pd presented in Figure K.2. and Table K.4. may
not be plausible for certain applications. For example, consider a three-class sampling
6
plan for mesophilic aerobic bacteria in dried milk with m = 104 CFU/g, and M = 106
CFU/g. Based on Table K.3, pa = 0.05 for the combination pm = 0.05 and pd = 0.45.
However, if the distribution is lognormal, this combination of pm and pd values would
imply a lot of dried milk with a geometric mean aerobic plate count of 4 log10 CFU/g
(10,000 CFU/g), a standard deviation of 16 log10 CFU/g, and a maximum concentration
in excess of 40 log10 CFU/g.
As noted above, the impact of lot acceptance sampling plans depends on variability
among lots. The Food and Agriculture Organization/World Health Organization jointly
7
supported development of a web-based analytical tool that can analyze the direct
impact of lot acceptance sampling plans based on user-specified distributions of
contamination between and within lots (http://www.fstools.org/sampling).
BIBLIOGRAPHY
8
Appendix L. Design of 2-class and 3-class Sampling Plans
Two- and three-class sampling plans are attribute sampling plans where
quantitative microbiological concentration data are divided into two or three
classes, respectively. The key parameters of these plans are: ‘n’, the sample
size; ‘cm’, the acceptance number for sample units which exceed m in
concentration; and ‘cM’, the acceptance number for sample units which exceed M
in concentration. The sample of n units is presumed to be a ‘rational subgroup’,
such as units chosen from the same lot or time period of production.
Some additional notation that will be useful in the discussion of 2- and 3-class
sampling plans are:
1
1–β = P[(# > M) > cM] (L.1.)
Power Power
n c Target Actual
4 0 95.00% 93.75%
5 0 95.00% 96.88%
6 0 95.00% 98.44%
10 1 95.00% 98.93%
20 5 95.00% 97.93%
10 1 99.00% 98.93%
20 4 99.00% 99.41%
10 0 99.90% 99.90%
20 3 99.90% 99.87%
Note that this type of plan is based on specifications, so is not generally useful
for SPC.
This approach is used in designing sampling plans for process control, and the
focus is on the false positive fraction (‘producer’s risk’ or ‘type I error’ or ‘α’ or
‘FAR’).
2
Some possible plans are:
α Prob. α
n c Target <M Actual
1 0 5.00% 95.00% 5.00%
2 1 5.00% 95.00% 0.25%
3 1 5.00% 95.00% 0.72%
5 1 5.00% 95.00% 2.26%
10 2 5.00% 95.00% 1.15%
1 0 1.00% 99.00% 1.00%
2 1 1.00% 99.00% 0.01%
3 1 1.00% 99.00% 0.03%
5 1 1.00% 99.00% 0.10%
10 1 1.00% 99.00% 0.43%
1 0 0.10% 99.90% 0.10%
2 1 0.10% 99.90% 0.00%
3 1 0.10% 99.90% 0.00%
5 1 0.10% 99.90% 0.00%
10 1 0.10% 99.90% 0.00%
3
Some possible plans are:
α Prob. α
n c Target <= M Actual
1 0 5.00% 95.00% 5.00%
2 1 5.00% 77.65% 5.00%
3 1 5.00% 86.46% 5.00%
5 1 5.00% 92.36% 5.00%
10 2 5.00% 91.28% 5.00%
1 0 1.00% 99.00% 1.00%
2 1 1.00% 90.00% 1.00%
3 1 1.00% 94.10% 1.00%
5 1 1.00% 96.73% 1.00%
10 1 1.00% 98.45% 1.00%
1 0 0.10% 99.90% 0.10%
2 1 0.10% 96.80% 0.10%
3 1 0.10% 98.20% 0.10%
5 1 0.10% 99.00% 0.10%
10 1 0.10% 99.52% 0.10%
4
3-Class Sampling Plans
This approach once again is used for acceptance testing, and the focus is on the
false negative fraction (‘consumer’s risk’ or ‘type II error’ or ‘β’).
Here ‘M’ again denotes the median concentration that, if exceeded, represents
‘unacceptable’ product that should be rejected with high reliability by the plan.
The quantile ‘m’ denotes the median concentration that separates ‘acceptable’
from ‘marginal’ product. Presumably ‘acceptable’ product should be rejected
infrequently.
5
Some possible plans are:
M M m m
m M Power Power Power Power
n c c Target Actual Target Actual
4 1 0 95.00% 93.75% 50.00% 68.75%
5 2 0 95.00% 96.88% 50.00% 50.00%
6 2 0 95.00% 98.44% 50.00% 65.63%
7 3 1 95.00% 93.75% 50.00% 50.00%
10 4 1 95.00% 98.93% 50.00% 62.30%
7 3 0 99.00% 99.22% 50.00% 50.00%
10 4 1 99.00% 98.93% 50.00% 62.30%
10 4 0 99.90% 99.90% 50.00% 62.30%
This approach is used primarily in designing sampling plans for process control,
and the focus is on the false positive fraction (‘producer’s risk’ or ‘type I error’ or
‘α’ or ‘FAR’).
The probability of a randomly chosen lot from normal production failing from both
rejection rules for the special case that cM = 0 is
𝑐𝑐
1-α = � �𝑛𝑛𝑘𝑘�𝑢𝑢𝑘𝑘 𝑝𝑝𝑛𝑛−𝑘𝑘 (eq. L.4.)
𝑘𝑘=0
An additional design rule for 3-class plans is to make the rejection rules for m
and M roughly of equal impact in the determination of the total false positive
fraction. This can be achieved by determining PM for the 2-class plan with sample
size n and acceptance number cM = 0 and half the false positive fraction α, and
then choosing the parameters related to m. We assume therefore that PM has
been determined from
6
α/2 = P[(# > M) > 0] (eq. L.5.)
Exact
α Probability Probability α
n c Target <m <M m+M
5 1 0.50% 98.50% 99.95% 0.454%
5 1 1.00% 97.50% 99.90% 1.047%
5 1 5.00% 94.50% 99.49% 4.720%
5 1 10.00% 91.50% 98.98% 9.650%
5 2 0.50% 93.50% 99.95% 0.493%
5 2 1.00% 91.50% 99.90% 1.019%
5 2 5.00% 85.00% 99.49% 4.894%
5 2 10.00% 80.00% 98.98% 9.919%
5 3 0.50% 85.00% 99.95% 0.470%
5 3 1.00% 81.00% 99.90% 1.041%
5 3 5.00% 71.00% 99.49% 5.028%
5 3 10.00% 64.50% 98.98% 10.045%
7
As in the section titled ‘Given Pm, PM and error fraction target, find n, cm and cM’,
we assume cM = 0 to simplify the equations, and choose PM to satisfy eq.(L.5.).
The implicit eq.(L.4.) is then solved for Pm.
8
Some possible plans are:
9
Approach 4: Finding m and M from Pm and PM
If the entire mixture ECDF (including zero results) is the desired basis for
quantiles, compute adjusted quantiles for Pm and PM as
Find ym and yM as the quantiles of the normal distribution with mean m and
standard deviation s corresponding to Pm’ and PM’. Then compute m and M as
the antilogs of ym and yM. Note that observed zero results are still used in
applying the rejection rules.
10
Appendix M. Implied False Alarm Rate in a 3-class Sampling Plan Given m,
M, n, c, and σ.
Table M.1. False alarm rates implied by existing 3-class sampling plans (M
percentile)
M
n c log(M/m) σlog10 FAR(%) FARM(%) FARm(%)
percentile
5 2 1 0.25 7.7 99.6 2.5 97.1
0.50 71.8 15.8 2.5 13.3
0.80 90.8 3.1 2.5 0.6
1.20 95.9 2.5 2.5 0.0
5 2 2 0.25 0.0 100.0 2.5 97.5
0.50 7.7 99.6 2.5 97.1
0.80 53.0 45.2 2.5 42.8
1.20 81.8 6.6 2.5 4.1
FARM criterion: Second, assume that M is chosen to achieve a desired FARM for
a given sampling plan, such as the conventional 0.3%. For a given sampling
plan, pd = 1-(1-FARM)1/n. For an n = 5 sampling plan, pd = 0.05% (M = 99.95th
percentile). Based on the same approach used to generate Table M.1, Table M.2
presents implied FAR.
1
Table M.2. False alarm rates implied by existing 3-class sampling plans (FARM)
m
σlog10 FAR(%) FARM(%) FARm(%)
n c log(M/m) percentile
The take away message from Tables M.1 and M.2 is that designing 3-class
sampling plans based on n, c, m, and M without considering process variability
(σ) can result in highly inconsistent false alarm rates. On the other hand, the
analysis in this section assumes that limits are based on reliable data from an
ideal process that is under statistical control and capable of meeting
microbiological specifications. If the m and M limit values are chosen instead on
the basis of strong observed associations with certain concentrations being
exceeded (e.g., measurements exceeding M are associated with observed
pasteurization process failures, measurements exceeding m and associated with
significantly reduced shelf life), then frequent occurrences where the
microbiological criteria were exceeded may indicate unsatisfactory process
quality. Under such circumstances, the events would be mischaracterized as
false alarms because the exceedances indicate a lack of process capability to
meet microbiological specifications.