[Federal Register: July 25, 1996 (Rules and Regulations)]
[Page 38956-38989]
From the Federal Register Online via GPO Access [wais.access.gpo.gov]
[DOCID:fr25jy96-22]
 
[[pp. 38956-38989]] Pathogen Reduction; Hazard Analysis and Critical Control Point 
(HACCP) Systems

[[Continued from page 38955]]

[[Page 38956]]

    The estimated annual industry costs (not discounted) are summarized 
in Table 2. These annual costs vary over the first four years as the 
new HACCP-based program is undergoing its implementation phase. After 
the initial four years, the recurring costs are estimated at a constant 
$99.6 to $119.8 million per year. The present value of all industry 
costs summarized in Table 2 for the 20-year time period is $968 to 
$1,156 million as shown earlier in Table 1. This total of $968 to 
$1,156 million ($0.97 to $1.16 billion) is the total industry cost for 
the rule as shown in Table 3.

                                              TABLE 2.--Summary of Annual Industry Costs--All Requirements                                              
                                                                      [$ Thousands]                                                                     
--------------------------------------------------------------------------------------------------------------------------------------------------------
                      Cost Category                              Year 1             Year 2             Year 3             Year 4            Year 5+     
--------------------------------------------------------------------------------------------------------------------------------------------------------
I. Sanitation SOP's:                                                                                                                                    
    Plans and Training...................................              2,992                                                                            
    Observation and Recording............................              8,345             16,691             16,691             16,691             16,691
II. E. coli Sampling:                                                                                                                                   
    Plans and Training...................................              2,627                                                                            
    Collection and Analysis..............................              8,716             16,122             16,122             16,122             16,122
    Record Review........................................                406                752                752                752                752
III. Compliance with Salmonella Standards................  .................       5,472-16,899       5,353-25,753       5,811-25,956       5,811-26,079
    Compliance with Generic E. coli Criteria.............  .................              (\1\)              (\1\)              (\1\)              (\1\)
IV. HACCP:                                                                                                                                              
    Plan Development.....................................  .................              3,769             27,755             35,464  .................
    Annual Plan Reassessment.............................  .................  .................                 69                448              1,179
    Initial Training.....................................  .................              1,270              8,284             18,435  .................
    Recurring Training...................................  .................                 64                542              1,877              2,799
    Recordkeeping (Recording, Reviewing and Storing Data)  .................              3,050             18,479             42,478             54,097
V. Additional Overtime...................................  .................                189                837              1,711              2,125
                                                          ----------------------------------------------------------------------------------------------
      Total..............................................             23,086      47,379-58,806     94,884-115,284    139,789-159,934    99,576-119,844 
--------------------------------------------------------------------------------------------------------------------------------------------------------
\1\ Not Separately Estimated.                                                                                                                           


         Table 3.-- Present Value of 20-Year Costs and Benefits         
                              [$ Billions]                              
------------------------------------------------------------------------
                                               Public health            
 Effectiveness in reducing pathogens in the      benefits      Industry 
       manufacturing sector (percent)        ----------------    costs  
                                                Low    High             
------------------------------------------------------------------------
10..........................................    0.71    2.66   0.97-1.16
20..........................................    1.43    5.32   0.97-1.16
30..........................................    2.14    7.98   0.97-1.16
40..........................................    2.85   10.64   0.97-1.16
50..........................................    3.57   13.30   0.97-1.16
60..........................................    4.28   15.96   0.97-1.16
70..........................................    4.99   18.61   0.97-1.16
80..........................................    5.71   21.27   0.97-1.16
90..........................................    6.42   23.93   0.97-1.16
100.........................................    7.13   26.59   0.97-1.16
------------------------------------------------------------------------

    Note: Analysis assumes zero benefits until year 5. All elements 
of the HACCP-based program will be in place 42 months after 
publication of the final rule.

    The public health benefits of this rule are discussed in detail in 
Section IV. The benefits are based on reducing the risk of foodborne 
illness due to Campylobacter jejuni/coli, Escherichia coli 0157:H7, 
Listeria monocytogenes and Salmonella. Section IV concludes that these 
four pathogens are the cause of 1.4 to 4.2 million cases of foodborne 
illness per year. FSIS has estimated that 90 percent of these cases are 
caused by contamination occurring at the manufacturing stage that can 
be addressed by improved process control. This addressable foodborne 
illness costs society from $0.99 to $3.69 billion, annually. The high 
and low range occurs because of the current uncertainty in the 
estimates of the number of cases of foodborne illness and death 
attributable to the four pathogens. Being without the knowledge to 
predict the effectiveness of the requirements in the rule to reduce 
foodborne illness, the Department has calculated projected health 
benefits for a range of effectiveness levels, where effectiveness 
refers to the percentage of pathogens eliminated at the manufacturing 
stage. The link between effectiveness and health benefits is the 
proportionate reduction assumption which is explained in Section IV. 
Because of the wide range in estimates for the cost of foodborne 
illness, each effectiveness level will have a low and high estimate for 
public health benefits. These estimates of public health benefits are 
shown in Table 2, as the present value of a 20-year benefit stream.
    The analysis assumes that benefits will begin to accrue in year 
five. The five year lag leads to conservative benefit estimates since 
the new HACCP-based inspection program will be fully implemented in 42 
months, and benefits should accrue during those 42 months as well as in 
the 1\1/2\ years that follow. Limiting the benefit estimates to four 
pathogens also leads to conservative cost estimates. To the extent that 
the proportionate reduction estimate may overestimate benefits, these 
other factors provide conservative balance.
    Net benefits exist for every cost and benefit combination 
illustrated in Table 2 except for the case of 10 percent effectiveness 
using the low benefit estimate. If the low benefit estimate is correct, 
the new HACCP-based regulatory program would have to reduce pathogens 
by 14 to 17 percent to cover the projected 20-year industry costs of 
$968 to $1,156 million. For the high benefit estimate net benefits 
begin to occur at an effectiveness level of 4 to 5 percent.
    The costs summarized in Tables 1 and 2 have not been reduced to 
account for firms that already have existing HACCP programs. FSIS does 
not have a good estimate of the number of such firms.

C. Impact on ``Smaller'' Businesses

    The final rule provides regulatory flexibility for smaller firms 
consistent with the Regulatory Flexibility Act. For the slaughter 
facilities, the generic E. coli sampling requirements vary depending on 
the number of birds or animals slaughtered annually. This will 
significantly reduce the microbial

[[Page 38957]]

testing costs for smaller establishments which, under the proposed 
rule, would have been required to test every species or kind they 
slaughter every day on which slaughter of that species or kind occurs. 
Under the final rule, the impact on smaller establishments is mitigated 
by the change to base generic E. coli sampling requirements on annual 
production and by a change to no longer require that every species or 
kind be sampled. The costs to small establishments are also reduced 
because the proposed carcass cooling and antimicrobial near term 
requirements have been eliminated from the final rule and training 
requirements are more flexible. The requirement to sample each variety 
of raw ground product, which caused a heavier burden on small 
establishments, has also been eliminated.
    The regulatory burden on small establishments is eased by the 
provisions which extend the time small establishments have to meet the 
HACCP system requirements. The detailed cost analysis in Section V 
outlines the methodology used in developing cost estimates and varying 
regulatory requirements for the purpose of regulatory flexibility for 
small establishments.

D. Effect on Retail Price

    The preliminary analysis included an estimate that the total four-
year implementation costs represented only $0.0024 per pound of fresh 
meat and poultry. This type of estimate helps put overall cost figures 
into perspective in terms of the potential increase in food prices. A 
large number of smaller processors responded very emotionally to the 
low figure of $0.0024 per pound on the basis that the lack of economies 
of scale in their businesses means their potential unit cost increases 
would be far higher. This ``cost-per-pound'' analysis was not meant to 
imply that the cost impact on all business would be the same. In a 
competitive industry, the impact on overall retail price is, however, 
an important indicator of net societal benefits. The four-year 
implementation costs for the final rule represent $0.0011 to $0.0013 
per pound based on 1993 production of 67.15 billion pounds (66.4 
billion pounds federally inspected and 748 million state inspected) of 
meat and poultry on a carcass weight basis. The annual recurring cost 
of $99.6 to $119.8 million represents $0.0015 to $0.0018 per pound 
based on 1993 production.

E. Impact on International Trade

    The final rule will have an impact on countries and the 
establishments in those countries that export meat and poultry products 
to the United States. The inspection statutes require that imported 
product be produced under an inspection system that is equivalent to 
the U.S. inspection system. The equivalence of a country's system must 
be established by the United States before product can be exported to 
the United States. The notion of equivalence has been clarified under 
the World Trade Organization (WTO) Agreement on Sanitary and 
Phytosanitary measures. Under the WTO, all members have an obligation 
to apply the principle of equivalence on importing countries. 
Equivalence determinations are based on scientific evidence and risk 
assessment methodologies.
    In light of the WTO emphasis on the use of science to determine 
equivalence, a number of countries are moving toward implementation of 
HACCP systems. The preliminary analysis noted that a large portion of 
the eligible exporting establishments are in countries that are 
themselves in the process of implementing HACCP and complying with 
their own country's HACCP requirements may achieve equivalence with the 
requirements of this rule.
    As of January 1, 1995 there were 1,395 establishments in 36 
different countries certified to export meat or poultry products to the 
United States. Canada (599 establishments), Denmark (125 
establishments), Australia (111 establishments) and New Zealand (94 
establishments) accounted for two-thirds of the 1,395 establishments. 
These four countries were the source of 85 percent of the 2.6 billion 
pounds of product imported during 1994. These four countries are 
currently developing HACCP systems for their respective inspection 
programs.
    Half (18) of the 36 countries have fewer than 10 establishments 
approved to export products to the U.S. These 18 countries represent a 
total of 77 establishments, 5 percent of the total. Meeting the 
equivalency requirements may present a problem for some of these 
countries in the near term. Their inspection programs will have to meet 
equivalency requirements for HACCP according to the implementation 
schedule for domestic establishments, i.e., 18 months for large 
establishments, 30 months for small establishments and 42 months for 
very small establishments. This schedule should lessen the burden on 
smaller establishments.
    There are other factors that will affect the burden on foreign 
establishments. As HACCP becomes the international norm, these 
establishments will be required to implement changes to meet the 
requirements of other countries implementing HACCP. Thus, their costs 
may not be solely associated with U.S. requirements. Establishing 
impact is further complicated because the U.S. requirements apply only 
when they are preparing product that is to be exported to the U.S. This 
product may represent only a small portion of total establishment 
production.
    Upon implementation of these regulations, FSIS will review other 
countries' meat and poultry systems to ensure that exporting countries 
have adopted comparable measures, which would entitle them to continue 
exporting product to the United States. As other countries improve 
their regulations by adopting provisions comparable to those contained 
in this rule, it is expected that U.S. exports will similarly be 
affected, i.e., the receiving countries will be closely reviewing 
domestic exporting establishments to assure that they are meeting the 
requirements of the importing country.
    FSIS will continue to carry out its import inspection 
responsibilities with a two-stage approach. The first stage is system 
review, which consists of an evaluation of the laws, policies, and 
administration of the inspection system in each eligible country. This 
overall evaluation will include an assessment of the implementation of 
HACCP supplemented by on-site reviews of individual establishments, 
laboratories, and other facilities within the foreign system. The 
``equivalency'' of foreign requirements will be determined at this 
stage.
    The second level of review involves port-of-entry inspection by 
FSIS inspectors to verify the effectiveness of foreign inspection 
systems. Using statistical sampling plans based on the foreign 
establishment's history and the nature of the product, FSIS will 
continue to give greater scrutiny to shipments posing the highest risk. 
Products that do not meet U.S. requirements, which includes having been 
produced under a HACCP or HACCP-equivalent system, will be refused 
entry. FSIS has concluded that requiring HACCP systems in combination 
with the two-stage inspection approach will better ensure the safety of 
imported meat and poultry products.
    All countries exporting raw products to the U.S. must develop and 
implement performance standards that are equivalent to the pathogen 
reduction performance standards for Salmonella. They must also be able 
to demonstrate that they have systems in place to assure

[[Page 38958]]

compliance with the standards. As with any other type of standard, FSIS 
could choose to test imported product for Salmonella at port-of-entry 
to verify the effectiveness of the foreign inspection system.
    With respect to the specific requirements for sampling generic E. 
coli to validate control of slaughter and sanitary dressing procedures, 
it will be necessary for all foreign countries to demonstrate that they 
have an equivalent procedure to verify that they are controlling their 
slaughter and sanitary dressing processes.
    There were several comments related to trade issues. Most of the 
comments concerning the impact on exports dealt with the proposed 
requirement for antimicrobial treatment of U.S. product. That proposed 
requirement raised particular concerns because the European Union 
member states and Canada restrict the use of certain antimicrobials on 
meat and poultry carcasses. The concerns raised in the comments are no 
longer an issue because the final rule does not require the use of 
antimicrobials. The final rule will affect exports only if a company 
has difficulty meeting the microbial performance criteria without using 
an antimicrobial. One option discussed in the proposed rule was that 
hot water would be considered to be an acceptable antimicrobial 
treatment, and that would be acceptable to Canada and the members of 
the European Union. The public comments also indicated that Trisodium 
Phosphate (TSP) is approved for use in Canada and the United Kingdom 
and is being considered by the European Union, Australia, and New 
Zealand.
    Comments related to imports were concerned about the procedures 
FSIS would use to determine equivalence with the new U.S. requirements. 
As a condition of the NAFTA Treaty and the GATT Treaty, the United 
States has agreed to allow imports from countries that have systems of 
inspection equivalent to that of the United States. FSIS is considering 
alternative methods for determining that a foreign country's system of 
inspection can assure that the establishments within that system are 
using a process control system equivalent to the HACCP-based inspection 
system outlined in the final rule.

F. Impact on Agency Costs

    Implementation of this rule will lead to both one-time nonrecurring 
costs and recurring costs for FSIS. There are three categories of one-
time nonrecurring costs: (1) Training, (2) in-establishment 
demonstration projects, and (3) laboratory renovation. In order to 
implement the rule, FSIS will provide training to in-establishment 
personnel in two segments. The first training segment will cover issues 
related to sanitation standard operating procedures and generic E. coli 
sampling and testing requirements. The estimated costs for this 
activity is $3.6 million in the first year of implementation. The 
second training segment will cover issues related to the implementation 
of HACCP and is estimated the cost $3.6 million spread over the second 
and third year of implementation. FSIS will utilize the train-the-
trainer approach to minimize the costs of these initiatives. FSIS is 
also committed to working with States and industry to sponsor HACCP 
demonstration projects for small businesses. Pursuant to implementation 
of the HACCP rule, microbiological sampling and testing will increase 
dramatically. In the period from 1990 to 1995, FSIS averaged 
approximately 33,000 analyses for microbiology per year. This is 
estimated to increase to 125,000 analyses per year after HACCP 
implementation. In order to accommodate this increase, FSIS will 
renovate its field laboratory facilities to expand their capacity, 
improve ability to test for a broader range of pathogens, and purchase 
new equipment. FSIS estimates that the planned renovation will cost 
$1.5 million.
    By implementing this rule, FSIS will incur recurring costs 
associated with increased microbiological testing and upgraded 
inspector salaries. FSIS estimates that microtesting costs will 
increase approximately $3.0 million annually. Of this amount $2.0 
million is needed for equipment, supplies, and shipping costs to 
conduct Salmonella testing, $0.5 million for microtesting conducted to 
verify HACCP systems, and $0.5 million for personnel necessary to 
handle the increased workload. Under HACCP-based inspection, FSIS 
personnel will be required to assume greater responsbility for more 
complex food inspection tasks. Slaughter inspectors will be required to 
perform health and safety tasks, such as taking microbiological 
samples, and verifying HACCP systems. Processing inspectors' roles will 
take them out of the establishment and put them into retail and market 
place settings to take microbiological samples, and to ensure meat and 
poultry products are handled in a manner to that minimizes the growth 
of pathogenic organisms. FSIS estimates that compensating inspectors 
for assuming more complex food safety tasks will cost $1.6 million per 
year.

G. Impact on State Programs

    Comments stated that FSIS failed to adequately consider the cost of 
the changes to State programs and that FSIS was increasing the resource 
demands for State programs without providing adequate funding. The 
preliminary analysis did not address the impact on State programs. 
However, FSIS recognizes that the 26 States operating their own meat 
and poultry inspection programs will likely have to substantially 
modify their programs after the HACCP/Pathogen Reduction regulation is 
finalized to remain ``at least equal to'' Federal inspection programs 
as required by the FMIA and PPIA. During the regulation's 
implementation period, FSIS will be using the Agency's State-Federal 
Program staff to assist the States in bringing the necessary changes to 
the State inspection programs. Although FSIS has requested some 
additional funds to implement this rule, FSIS has also acknowledged 
that implementation of this rule will require eliminating some tasks, 
conducting other tasks differently and streamlining the organization in 
order to free up resources to fully address the new requirements. FSIS 
believes that the same type of restructuring or reprogramming will take 
place within the State programs. This does guarantee, however, that all 
States with inspection programs will be able to implement the necessary 
program changes without additional funds. FSIS believes, however, that 
with FSIS assistance and with the flexibility provided under the 
``equal to'' provisions, most of the States should be able to modify 
their programs with minimal additional funding. To the extent that 
there are any additional costs, the State inspection programs are 
eligible to receive up to 50 percent Federal matching funds.

H. Consumer Welfare Analysis

    It is likely that at least some of the costs of the new HACCP-based 
regulatory program will be passed on to consumers in the form of higher 
prices. Even if costs are fully reflected in retail prices, the impact 
on consumers and consumption will be small. Retail costs are not 
expected to increase more than 0.02 percent. Retail demand for meat and 
poultry is inelastic. A likely range is -0.25 to -0.75. This suggests 
changes in quantity demanded of less than 0.02 percent. Given that 
annual per capita meat and poultry consumption is about 211 pounds, 
retail weight, the impact on individual consumption will be less than 
\1/10\th of a pound per year. In aggregate, with a high impact

[[Page 38959]]

scenario, consumption would decrease by about 50 million pounds. These 
impacts may be overstated if meat and poultry producers pass some costs 
back to livestock and poultry producers. Improved consumer confidence 
in the safety of meat and poultry could offset price driven decreases 
in consumption.

IV. Analysis of Public Health Benefits

A. Introduction

    This section addresses the methodology used to develop the 
estimates for public health benefits that, for the purpose of this 
final Regulatory Impact Assessment, have been defined as the reduction 
in the cost of foodborne illness attributable to pathogens that 
contaminate meat and poultry products at the manufacturing stage. This 
section is organized around the Agency's responses to the public 
comments related to benefits. The first part of this section addresses 
the general comments related to risk assessment. The Agency has 
responded to these general requirements by providing an overall summary 
of the current state-of-the-art with respect to risk assessment for 
foodborne pathogens. The second part of the discussion (see subsection 
titled ``Analysis of Comments on Public Health Benefits'') addresses 
the more specific comments on the methodology used to estimate benefits 
in the preliminary analysis.
    Several comments suggested that FSIS has not conducted an adequate 
risk assessment and/or should conduct a thorough risk assessment before 
proceeding with the current rulemaking. More focused comments assert 
that the relationship between pathogen reduction at the manufacturing 
stage and foodborne illness reduction is unknown. Those comments 
suggest that establishing that relationship requires a quantitative 
risk assessment, i.e., an estimate of the probability of adverse health 
effects (foodborne illness) given a particular level of a hazard 
(pathogens at manufacturing stage).
    The preliminary analysis and this final RIA recognize that the 
relationship is unknown and acknowledge that there are significant data 
gaps regarding both likelihood and magnitude of illness and numbers of 
foodborne pathogens. These data gaps mean that multiple assumptions 
must be made in order to calculate the probabilities of risk, and FSIS 
is concerned with this tremendous uncertainty. However, the agency is 
developing quantitative assessments and believes that these will become 
the basis on which to make future regulatory decisions. In this 
rulemaking, FSIS estimates of the risk of foodborne disease linked to 
specific pathogens are based upon the best judgement of nationally 
recognized experts in infectious disease, epidemiology, microbiology, 
and veterinary medicine. FSIS is also relying on a qualitative 
estimation of risk as expressed in publications and summary reports 
from the CDC, other public health agencies, and special panels, such as 
the National Advisory Committee on Microbiological Criteria in Foods 
and those established by the NAS. Based on this sizable body of 
information and scientific judgement, FSIS is proceeding to develop 
benefit estimates using the assumption that a reduction in pathogens 
leads to a proportionate reduction in illness and death. The benefits 
analysis could have used a more conservative relationship estimate, 
e.g., a reduction in pathogens leads to a reduction in illness that is 
less than proportional. However, given the current level of knowledge, 
FSIS views the proportional assumption as most appropriate at present.
    The Department has initiatives in place that will begin to relate 
pathogen levels at inspected establishments to incidence of human 
illness and support quantitative risk assessment (see Section IV-D on 
FSIS Data Initiatives). The present paucity of data to support a risk 
model for the major foodborne pathogens causing human disease limits 
the usefulness of quantitative risk assessment in the regulatory arena 
of meat and poultry inspection. It is unlikely that any single 
numerical constant will adequately describe the dose-response 
relationships for all pathogens associated with all of the products 
that FSIS regulates, given the complexity of possible interactions of 
factors associated with the host, the pathogenic strain, the diet, and 
the environment (CAST, 1994).
    The Federal Crop Insurance Reform and Department of Agriculture 
Reorganization Act of 1994 (P.L. 103-354) now requires that for each 
proposed major regulation (i.e. economic effects of at least $100 
million a year and effects on human health, safety, or the environment) 
the Department publish an analysis of the risks addressed by the 
regulation. While this statute does not apply to this final rule, FSIS 
is providing a qualitative estimation of risk (Tables 4 and 5) and a 
recommendation to manage risk using HACCP in meat and poultry 
inspection programs. Concurrently, scientists from FSIS and USDA's 
Agricultural Research Service (ARS), Economic Research Service (ERS), 
and modelers from academia and industry continue to develop risk models 
which blend failure analysis, predictive microbiology, and other models 
into the framework described by the NAS (NRC, 1983). FSIS believes this 
approach is flexible and responsive to new data necessary to fully 
document risks of foodborne diseases.

B. FSIS Risk Assessment

    Following the publication of the 1985 National Academy of Sciences 
(NAS) study on the scientific basis for meat and poultry inspection, 
FSIS requested that the National Research Council of NAS conduct a 
follow-up study that included the objective of developing a risk 
assessment model for the poultry production system. The subsequent 
report, ``Poultry Inspection: The Basis for a Risk-Assessment 
Approach'' was published by the National Academy Press in 1987. The 
1987 study concluded that the present system of inspection provides 
little opportunity to detect or control the most significant health 
risks presented by microbial agents that are pathogenic to humans. The 
study also concluded that current databases can serve as the basis for 
a comprehensive, quantitative risk assessment only for certain well-
characterized chemical residues.
    The committee conducting the study also concluded that their report 
did constitute a qualitative risk assessment that could be useful for 
many purposes, including the evaluation of inspection strategies. That 
assessment found: ``There is evidence linking disease in humans to the 
presence of pathogens on chickens. For example, epidemiological studies 
indicate that approximately 48% of Campylobacter infections are 
attributable to chicken. Data also suggest that chicken is probably an 
important source of salmonellosis in the United States.'' Based on 
these and other findings, the committee recommended that FSIS ``modify 
the existing system so that it more directly addresses public health 
concerns.'' FSIS believes that the implementation of HACCP programs at 
slaughter for meat and poultry is such a ``modification'' of the food 
safety system which will address human health hazards, particularly 
foodborne diseases.

C. Risk Assessment Framework

    The National Research Council (1983) presented a framework for risk 
assessment that has become a standard paradigm to organize risk 
assessments for chemical and microbial hazards. The framework, 
consisting of hazard identification, dose-response assessment, exposure 
assessment, and risk characterization, is flexible and can accommodate 
many different modeling strategies. The major distinction

[[Page 38960]]

between foodborne microbial risk assessments and chemical risk 
assessments may be the additional uncertainties of microbial growth and 
survival in food prior to consumption. Survival of pathogens present in 
a raw food and after cooking can be modeled using predictive 
microbiology methods. These models can also address the growth of 
pathogens with time and temperature abuse of raw and cooked foods.
    One of the first U.S. publications on the application of predictive 
microbiology to microbial risk assessment (Buchanan & Whiting, 1996) 
included estimations of risk of salmonellosis for several ``what-if 
scenarios'' as examples of potential time and temperature abuses of 
partially cooked food. The predictive microbiology model was linked to 
a published dose-response model for salmonellosis (Haas, 1983) to 
calculate a risk estimate. The dose-response model was developed by 
empirically fitting data from human feeding studies conducted at high-
dose challenges with a number of pathogenic strains of Salmonella to 
the ``beta poisson'' model (Haas, 1983). The authors generated risk 
estimates for selected cooking and abuse scenarios, but recognized that 
the risk of illness is zero when the pathogen is not present in the 
sample even with unsafe food handling. HACCP programs at slaughter are 
expected to affect pathogen presence and levels before potential time 
and temperature abuses can occur. Therefore, changes at slaughter, in 
the duration of cooking, and final storage conditions of the food exert 
a tremendous impact upon the model outcomes.
    An unpublished draft risk model is in development as a research 
endeavor by Agriculture and Agri-Food Canada and Health Canada. A 
variety of modeling approaches were organized within the 1983 NRC 
framework to estimate risk of human illness from E. coli 0157:H7 in 
ground beef. The draft risk model includes many stochastic variables to 
account for the variability and uncertainty associated with the inputs 
and assumptions of the model. The authors are developing the model to 
identify current limitations to the construction of quantitative models 
which accurately describe the risk of foodborne disease along the farm 
to fork continuum.
    These recent quantitative risk assessment efforts are an 
encouraging beginning and serve to illustrate the tremendous 
uncertainties created by insufficient data describing processes 
throughout the farm to table continuum that contribute to risk. 
Additional uncertainties surround assumptions based on epidemiologic 
data for human illness. For example, recent data in the U.S. indicates 
a growing number of outbreaks of E. coli 0157:H7 disease linked to 
sources other than ground beef. The ecology of the organism on the 
farm, in the bovine gastrointestinal tract, and in irrigation, 
recreational, and drinking waters is largely unknown. Additionally, the 
primary sources of E. coli 0157:H7 causing sporadic disease may remain 
undercooked hamburger and may differ from vehicles causing outbreaks, 
as has been documented for Campylobacter (CDC, 1988). Outbreaks of 
campylobacteriosis have been caused primarily by unpasteurized milk and 
contaminated water, yet the overwhelming majority of infections are 
sporadic and have been linked to undercooked chicken. Control 
strategies to reduce both outbreak and sporadic case numbers for both 
of these pathogens may require greater understanding of vehicles of 
disease and more information than is currently available.
    FSIS concludes that risk models for foodborne illnesses are 
necessarily based largely on assumptions because scientific data 
describing key foodborne disease processes have not been developed. The 
models are extremely useful to identify basic research needs that might 
reduce the uncertainty associated with the inputs and assumptions of 
the models. The agency is proposing initiatives to generate data which 
may reduce uncertainties associated with modeling the risk of foodborne 
diseases. However, application of microbial risk assessment models to 
regulatory decision-making appears premature at this time. The 
following is a summary of the availability and limitations of data 
supporting risk assessment for foodborne pathogens:
1. Hazard Identification
    The Agency selected from the pathogens listed in Tables 4 and 5 the 
three most common enteric pathogens of animal origin: Campylobacter 
jejuni/coli, E. coli 0157:H7, Salmonella and one environmental pathogen 
Listeria monocytogenes for consideration in risk assessment. FSIS 
believes that these four pathogens may contaminate meat and poultry 
food vehicles at slaughter and can be reduced through improved process 
control in the manufacturing sector. Available data on estimated human 
disease incidence are summarized in Table 4. Data on human disease 
attributable to proven as well as epidemiologically linked pathogens 
and food vehicles are presented in Table 5. Additional and more precise 
information for this section regarding estimated national disease 
incidence and disease severity and duration is expected on these 
pathogens from the sentinel site surveillance initiative.
2. Exposure Assessment
    Rarely can actual exposure to a specific strain of foodborne 
pathogen be quantified with certainty in foodborne disease outbreaks. 
Microbes in food are known to be non-homogeneously distributed, 
imposing additional uncertainty due to sampling error upon the 
analytical variability of the methods for detection and quantification 
of microbes in foods. The outbreak strain may or may not be detected in 
the feces of diarrheal cases or in leftovers or companion samples from 
suspected lots. The levels detected in leftovers or companion samples 
from the same lot of food may or may not be representative of the 
serving that was prepared and consumed since the microbial numbers vary 
with time and temperature conditions and the initial microbial 
populations. The amount of the serving consumed may not be known.
    The FSIS baseline studies provide data on occurrence of pathogens 
(likelihood) and levels (magnitude) in uncooked meat and poultry 
products at slaughter and raw ground processing. Data for likelihood 
and magnitude of pathogens in the distribution, preparation, and 
consumption phases of the farm-to-fork continuum of food production are 
sparse. Predictive microbiology models may be the most cost-effective 
method to deduce possible exposure scenarios in meat and poultry beyond 
the slaughter phase that may result in foodborne illness. The 
likelihood that the selected scenarios of improper cooking and abuse 
actually occur among U.S. consumers may not be measurable, but the 
scenarios may be useful in modification of behaviors that pose 
increased risk to consumers.
3. Dose-Response Assessment
    The relationship between the dose of a pathogen and response in the 
host, when known, can vary greatly for foodborne pathogens. Human 
feeding studies with foodborne pathogens were largely conducted several 
decades ago with small numbers of healthy adult males. One study 
reported both ill and asymptomatic volunteers who had consumed up to 
1,000,000,000 pathogenic Salmonella. Outbreak data for other Salmonella 
serotypes in food vehicles suggest a range of infective doses from one 
cell to 1,000,000,000,000 cells (Blaser & Newman, 1982). Fatty food 
vehicles, including some meat and

[[Page 38961]]

poultry products, are thought to protect enteropathogens from stomach 
acids and digestive enzymes that might otherwise reduce the dose to the 
intestinal tract and reduce the likelihood of disease. The effects of 
competition of the pathogen with the large indigenous microbial 
populations in food (ICMSF, 1980) and in the human gastrointestinal 
tract (Rolfe, 1991) may reduce the likelihood and/or the severity of 
foodborne disease.
    Even carefully controlled volunteer feeding experiments at doses up 
to one billion organisms per volunteer have shown variability in the 
infectious dose of one pathogen for individuals within a group of 
seemingly healthy, young adults. Extrapolation of empirical models of 
effects at high doses to low doses typical of properly handled food may 
or may not be appropriate. The dose-response curve for healthy adult 
males may not be useful in estimating dose-response relationships for 
the general population or sensitive sub-populations. The data available 
from human feeding studies were generated from very few species and 
strains of bacterial pathogens, excluding E. coli 0157:H7. Dose-
response modeling is crucial to microbial and chemical risk 
assessments. FSIS believes that application of dose-response models in 
food safety regulation requires careful examination of the validity of 
the assumptions and inputs of the model and of the plausibility of the 
model as a descriptor of foodborne disease processes.
4. Risk Characterization
    The integration of exposure and dose-response models is expected in 
risk characterization, along with sensitivity and uncertainty analyses 
(Burmaster & Anderson, 1995) for the risk model. Perhaps of greater 
significance than the numerical estimate of risk is the uncertainty 
associated with the estimate. A fully developed risk characterization 
would include risk estimates and sensitivity/uncertainty analyses for 
alternative models and assumptions. FSIS is collaborating with 
scientists in academia, the Agricultural Research Service, the Animal & 
Plant Health Inspection Service, the Economic Research Service, and the 
Office of Risk Assessment and Cost Benefit Analysis to develop and 
validate a risk assessment model for a single pathogen in a single meat 
product. This model may be modified for other specific pathogens of 
concern. The expectation of a generic model for all foodborne disease 
agents in all products does not appear promising based on differences 
in pathogenesis of bacterial species and strains and in human 
sensitivity and pathology.
    FSIS continues to evaluate new information on foodborne pathogens 
and on risk assessment methods and tools in accordance with the FSIS 
public health mission. The NAS Report, the CAST Report and the 1995 
Conference recognize HACCP as a system to reduce the likelihood of 
foodborne illness. The CAST Task Force also concluded that ``the 
efficacy of a HACCP system depends on the rigor and consistency with 
which it is designed and implemented and the use of (a) critical 
control point(s) that will control pathogens.''

D. FSIS Data Initiatives

     The 1994 report, ``Foodborne Pathogens: Risks and Consequences, 
CAST Task Force Report No. 122, September 1994'' concluded that ``a 
comprehensive system of assessing the risks of human illness from 
microbial pathogens in the food supply has yet to be devised.'' They 
cited the limitations of the current food safety information database 
and the difficulty in accumulating dose response and minimum infective 
dose data. A recent multidisciplinary conference, ``Tracking Foodborne 
Pathogens from Farm-to-Table, Data Needs to Evaluate Control Options'', 
carefully reviewed current databases and confirmed limitations outlined 
in the CAST Task Force report.
    FSIS has established initiatives to improve the quality and 
quantity of data in two major areas. First, FSIS is working with the 
Food and Drug Administration (FDA) and the Centers for Disease Control 
and Prevention (CDC) to establish an active sentinel site surveillance 
system for the major causes of foodborne illness. This project is 
designed to accumulate data on the incidence of foodborne illness by 
pathogen and by food.
    Second, the Agency has been developing baseline data for pathogen 
levels on major food animal species at the time of slaughter. The 
baseline data will allow the Agency to detect changes in the overall 
nation-wide pathogen levels. The National Baseline program was 
initiated in 1992 to provide information on the type and level of 
microbiological contamination on raw products under Federal inspection. 
Each sample collected is analyzed for nine microorganisms or groups of 
organisms. Microbiological baseline data are now available for steers 
and heifers, cows and bulls, and broiler chickens.
    If sufficient data on both pathogen levels and foodborne disease 
epidemiology result from current and future initiatives, FSIS should be 
able to develop models showing how these two variables are related for 
different pathogens. These models should then permit/facilitate a 
quantitative estimate of risk. Such data are essential for FSIS to 
evaluate the effect of control measures on both pathogens levels and on 
foodborne illness.

E. ARS Food Safety Research Program

    The Agricultural Research Service (ARS) administers a food safety 
research program that is currently funded at approximately $45 million 
per year. This program addresses problems in four different areas; 
pathogen reduction, mycotoxins, residues, and natural toxins. The 
reduction of microbial pathogens in food products of animal origin is 
the most pressing food safety problem today. Consequently, the pathogen 
reduction component is the largest of the four areas and is currently 
funded at $18.2 million annually. The ARS research in pathogen 
reduction addresses both preharvest and animal production, and post 
harvest problem areas, with approximately equal funding for each.
    Ongoing ARS research will help FSIS improve its capability for 
performing quantitative risk assessment in the area of foodborne 
pathogens or improve the ability to predict the effectiveness of new 
pathogen reduction technologies. Ongoing projects include the modeling 
of bacterial growth or thermal death times which will help set 
standards for meat and poultry products. Ongoing projects will also 
provide new laboratory screening or confirmatory methods. Other 
projects provide and/or evaluate technology and management methods 
which can help producers achieve lower contamination levels in animals 
presented for slaughter, such as vaccines or competitive bacterial 
cultures to prevent pathogens in live animals. There are also 
technology and management methods for use in slaughter and processing 
establishments, such as, organic acids for use in carcass sanitation, 
improvements to the feather picking operation for poultry, washing of 
trailers to reduce microbiological contamination, and establishment of 
guidelines on the microbiological safety of recycling cooling solutions 
for ready-to-cook meat and poultry products. In many cases the research 
may provide the scientific basis for developing and improving 
technology, for example, the nature of bacterial attachment to various 
meat surfaces.
    FSIS can and does forward very specific research requests to ARS. 
In preparation for this final rule, FSIS requested that ARS compare the 
results

[[Page 38962]]

from different microbial sample collection techniques, sponging versus 
excision at one versus three carcass sites. These studies are currently 
being conducted on both cow/bull and market hog carcasses. There are 
other specific ARS projects that will help provide the scientific basis 
for HACCP through risk assessment, predictive microbiology, and 
pathogen reduction interventions for several different bacterial 
pathogens which must be controlled to assure the safety of meat and 
poultry.
    These projects include: (1) Development of models to predict the 
growth rates, survival times, and thermal death rates for microbial 
pathogens potentially present in foods, including meat and meat 
products. (Microbiological modeling is time consuming and expensive 
because it requires that the data be quantified, that is, that numbers 
of bacteria are obtained, rather just the knowledge of the presence or 
absence of a pathogen under the conditions of the test.) The 
microorganisms being studied include E. coli O157:H7, Listeria 
monocytogenes, and Salmonella. These models are written into personal 
computer software that gives FSIS a readily useable tool to help 
evaluate proposed meat processes and assess out-of-process events. 
Refining predictive models has the goal of linking an entire process 
from raw ingredients to distribution of finished product. A specific 
project is to model the survival of E. coli O157:H7 during the 
manufacture of uncooked, fermented meat products. Using the information 
obtained, ARS will closely collaborate with other USDA agencies to 
develop strategies for risk reduction using the various processing 
techniques, and to create risk assessment models.
    (2) Modeling studies to predict the thermal inactivation of spore-
forming and non-spore-forming bacterial pathogens of both cooked and 
ready-to-eat products. These studies will be extended to the cooling of 
these products to ensure that there is no potential for growth of 
Clostridium botulinum and C. perfringens.
    (3) Determination of the long-term effects (21 days of storage at 
refrigerated temperatures) of organic acid treatment of red meat on 
some key pathogens (E. coli O157:H7, Listeria, and Clostridium), as 
well as on spoilage bacteria (mesophilic aerobes, lactic acid bacteria, 
and pseudomonads).
    (4) Delineation of the parameters affecting the antibacterial 
activity of organic acids. These include tissue type (pre-rigor, post-
rigor, frozen post rigor), inoculum type (pure culture or inoculated 
feces), inoculum level and the temperature of spray wash at meat 
surface. These results should clarify inconsistent reports on 
antibacterial activity of organic acids and also define optimum 
conditions to maximize the antibacterial activity of organic acids.
    (5) The correlation of the Campylobacter levels in broilers from 
the chill tank with their Campylobacter levels during production.

F. Analysis of Comments on Public Health Benefits

    There were many comments on the methodology used to estimate public 
health benefits in the preliminary analysis. This methodology used a 
series of estimates or assumptions based on incomplete data related to 
the six following areas:
    <bullet> Incidence of foodborne illness
    <bullet> Cost of foodborne illness
    <bullet> Percentage of foodborne illness and cost of foodborne 
illness attributable to meat and poultry products
    <bullet> Pathogens addressed by the rule
    <bullet> Effectiveness of rule in reducing pathogens
    <bullet> Estimated reduction in cost of foodborne illness related 
to reduction of pathogens
    To facilitate discussion of the issues raised in comments, the 
issues are addressed organized by these six areas.
1. Incidence of Foodborne Illness
    Table 4 presents the most recent estimates on the incidence of 
illness and death for selected pathogens along with the latest 
estimates on the percentage of illness and death which is foodborne. As 
discussed in the preliminary RIA, Table 4 includes the ``best 
estimates'' when precise data are not available. Many of these 
estimates are based on the landmark CDC study by Bennett, Holmberg, 
Rogers, and Solomon, published in 1987, which used CDC surveillance and 
outbreak data, published reports, and expert opinion to estimate the 
overall incidence and case- fatality ratio for all infectious and 
parasitic diseases. Estimates on the foodborne percentage of illness 
and death for bacteria in Table 4 are all based on CDC data. The 
resulting estimates for the number of foodborne cases and deaths are 
presented in the second and third columns of Table 5.
    The benefits for the preliminary analysis and this final RIA are 
calculated for the three most common enteric pathogens of animal 
origin: Campylobacter jejuni/coli, E. coli O157:H7, Salmonella and one 
environmental pathogen Listeria monocytogenes. FSIS believes that these 
four pathogens can be reduced through improved process control in the 
manufacturing sector.
    Although Clostridium perfringens and Staphylococcus aureus also 
cause a significant number of foodborne illnesses, they are not 
included in the benefits analysis because it is not clear that the 
HACCP-based regulatory program, which focuses on federally inspected 
processing, will significantly affect the incidence of disease caused 
by these organisms. Staphylococcus aureus usually enters the food chain 
through food handlers in restaurants and other commercial kitchens. 
Although C. perfringens may enter the food chain through the slaughter 
process, it is so ubiquitous in the environment that FSIS will not 
assume that controls at slaughter will be effective against this 
pathogen.
    One commenter questioned why the Agency has not addressed the 
public health problem of toxoplasmosis given the Table 5 estimate of 
$2.7 billion in annual costs. FSIS believes that while process control 
may help decrease the spread of cysts during boning and cutting 
operations, most of the Toxoplasma gondi cysts are internal to 
infective muscle tissues and are not addressable by process control. 
Therefore, FSIS is making the more conservative assumption to exclude 
this pathogen in the benefits estimate of disease averted.
    Many comments suggested that the large range in the illness 
incidence estimates demonstrates that there are insufficient data on 
which to base a new regulatory program. Historically, the lack of 
quantitative data on benefits and specific health risks have meant that 
health and safety regulations have required decisionmaking under 
uncertainty and have required the decisionmaker to balance the need to 
act with the need for additional or improved data. Compared to such 
issues as whether a chemical is a potential human carcinogen or whether 
low levels of air pollutants cause adverse health effects, the health 
effects of enteric pathogens are relatively well documented. If the 
pathogens enter the food supply, they do, under certain conditions, 
cause foodborne illness. If their presence can be prevented, no amount 
of temperature abuse, mishandling or undercooking can lead to foodborne 
illness.
    The Agency believes that the existing estimates on foodborne 
illness are adequate to conclude that a substantial and intolerable 
public health problem exists. Furthermore, existing estimates are 
appropriate for developing estimates on the cost of foodborne illness 
attributable to meat and poultry. The

[[Page 38963]]

Agency notes that similar estimates on the incidence of foodborne 
illness have been published by scientists from ERS in peer-reviewed 
journal articles (see footnotes to Table 5) and by the 1994 CAST Task 
Force.
    The above statement that Table 4 includes the most recent estimates 
of the incidence of illness and death requires further explanation in 
the case of Listeria monocytogenes. The estimates of 1,795-1,860 cases 
of listeriosis and 445-510 deaths are the ones used in the latest cost 
of illness study conducted by ERS. ERS is in the process of publishing 
a comprehensive documentation for the estimates of cost of illness for 
1993. In their draft document they acknowledge that the estimate for 
listeriosis cases originates from an extrapolation to the U.S. 
population of incidence data from a CDC-conducted surveillance study of 
six geographic regions in 1986 and 1987 (Gellin et al. 1987). They also 
note that (Tappero et al. 1995) found that the incidence of listeriosis 
has decreased since the 1960's and that projections from the 
surveillance data suggest that there were 1,092 listeriosis cases and 
248 deaths in 1993. ERS did not modify their cost of illness estimates 
because Tappero et al., was published after their analysis was 
concluded.
    FSIS considered modifying the cost of illness estimates for this 
final analysis but decided to use the estimates in Tables 4 and 5 
because (1) They are the figures that will appear in the upcoming ERS 
publication and, (2) updating the listeriosis estimates would have 
minimal impact on the overall cost of illness estimates. Considering 
the overall range and uncertainties involved in the cost of illness 
estimates, the change in listeriosis estimates has negligible impact on 
the regulatory analysis information conveyed through the potential 
benefits estimate.
    The Agency also recognizes that in using the 1993 estimates for 
incidence of foodborne illness, the benefits analysis has not accounted 
for possible reductions in foodborne illness attributable to the rule 
that mandated safe handling statements on labeling of raw meat and 
poultry products. The rule mandating safe handling instructions became 
effective on May 27, 1994. Thus, it can be argued that the incidence of 
foodborne illness for 1994 through the present should reflect the 
effectiveness of the 1994 labeling requirement in reducing the 
incidence of illness.
    FSIS is not aware of any quantitative evaluation of the 
effectiveness of safe handling labeling. Two recent surveys indicate a 
high level of awareness, but these surveys do not contain findings that 
can be translated into changes in consumer behavior. A recent 
Associated Press poll found that 9 in 10 Americans say they follow the 
safe-handling instructions. This poll, conducted in April 1996, 
included 1,019 randomly selected adults. This was a telephone survey 
conducted by ICR Survey Research Group. A November 1995 survey 
conducted by Wegman Food Markets in Buffalo, Rochester, and Syracuse 
found that 67.9 percent of respondents indicated they had read the safe 
handling information. The Wegman's survey found that most household 
meat preparers rely on color of meat or clarity of juices rather than 
temperature to determine when meat has been cooked thoroughly.
    In this analysis, FSIS has not attempted to adjust the 1993 
baseline to account for safe handling labeling. The potential effect of 
the 1994 regulation is one of many factors that could be affecting the 
current incidence or cost of illness. A May 1996 GAO study on foodborne 
illness notes that food safety and public health officials believe that 
the risk of foodborne illness is increasing. If they are correct, the 
1994 labeling rule may be slowing the growth rather than reducing the 
absolute level.
    There are many other factors that could have been incorporated into 
the baseline for the analysis such as population growth and increases 
in the cost of medical care. FSIS believes that attempts to adjust the 
cost of illness baseline to account for factors such as inflation, 
possible increases in foodborne illness due to behavior change or 
population increases, and possible decreases due to inventions such as 
safe handling labels are more likely to be misleading than informative 
given the level of uncertainly and wide range in existing estimates.
2. Cost of Foodborne Illness
    The fourth column of Table 5 shows that the 1993 estimated cost of 
foodborne illness by pathogen or parasite was between $5.6 and $9.4 
billion. These cost of illness estimates have been developed by ERS in 
conjunction with CDC over the past 15 years. As indicated in footnotes 
to Table 5, the results of that work have been frequently published in 
peer-reviewed journals.
    There were only a few public comments on the proposed rule which 
addressed the methodology used for estimating the cost of foodborne 
illness. Some comments argued that the public health benefit estimates 
are low because of the low value-of-life factor used in the estimates 
for the cost of foodborne illness.
    ERS intentionally used a conservative method to estimate the value 
of a statistical life (VOSL) acknowledging the controversy over valuing 
lives. ERS used Landefeld and Seskin's VOSL estimates and recognizes 
that the cost of illness estimates would be substantially higher if 
they used alternative methods. For example, Viscusi (1993) summarized 
the results of 24 principal labor market studies and found that the 
majority of the VOSL estimates lie between $3 million and $7 million 
per life. A survey of the wage-risk premium literature on the 
willingness to pay to prevent death concluded that reasonably 
consistent estimates of the value of a statistical life range from $1.6 
million to $6.5 million dollars (1986 dollars) (Fisher et al. 1989). 
Updated to 1993 dollars using the change in average weekly earnings, 
Viscusi's range becomes $3.2 million to $7.6 million per VOSL and 
Fisher's range becomes $2.0 million to $10.4 million dollars for each 
statistical-life lost. Viscusi and the Fisher estimates are greater 
than the highest Landefeld-Seskin (LS) VOSL estimate of $1,584,605 in 
1993 dollars (estimate for a 22 year old).

                                                 Table 4.--Sources of Data for Selected Pathogens, 1993                                                 
--------------------------------------------------------------------------------------------------------------------------------------------------------
                                                            Estimated                                                                                   
             Pathogen                Estimated number of    number of       Source(s) for case and death       Percent                Source            
                                            cases            deaths                   estimates               foodborne                                 
--------------------------------------------------------------------------------------------------------------------------------------------------------
Bacteria:                                                                                                                                               
    Campylobacter jejuni or coli..   2,500,000            200-730       Tauxe...............................   55-70     Tauxe et al.                   
    Clostridium perfringens.......      10,000               100        Bennett et al.......................     100     Bennett et al.                 
    Escherichia coli O157:H7......  10,000-20,000         200-500       AGA Conference......................      80     AGA Conf./CDC.                 
    Listeria monocytogenes........  1,795-1,860           445-510       Roberts and Pinner..................   85-95     Schuchat.                      

[[Page 38964]]

                                                                                                                                                        
    Salmonella....................  800,000-4,000,000     800-4,000     Helmick et al./Bennett et al.          87-96     Bennett et al./Tauxe & Blake.  
    Staphylococcus aureus.........   8,900,000             7,120        Bennett et al.......................      17     Bennett et al                  
Parasite:                                                                                                                                               
    Toxoplasma gondii.............       4,111                82        Roberts et al.......................      50     Roberts et al.                 
--------------------------------------------------------------------------------------------------------------------------------------------------------
Sources: American Gastroenterological Association Consensus Conference on E. coli O157:H7, Washington, DC, July 11-13, 1994. Bennett, J.V., S.D.        
  Holmberg, M.F. Rogers, and S.L. Solomon. 1987. ``Infectious and Parasitic Diseases,'' In R.W. Amler and H.B. Dull (Eds.) Closing the Gap: The Burden  
  of Unnecessary Illness.  Oxford University Press, New York. Helmick, C.G., P.M. Griffin, D.G. Addiss, R.V. Tauxe, and D.D. Juranek. 1994. ``Infectious
  Diarrheas.'' In: Everheart, JE, ed. Digestive Diseases in the United States: Epidemiology and Impact. USDHHS, NIH, NIDDKD, NIH Pub. No. 94-1447, pp.  
  85-123, Wash, DC: USGPO.                                                                                                                              
Roberts, T., K.D. Murrell, and S. Marks. 1994. ``Economic Losses Caused by Foodborne Parasitic Diseases,'' Parasitology Today. vol. 10, no. 11: 419-423.
                                                                                                                                                        
Schuchat, Anne, CDC, personal communication with T. Roberts at the FDA Science Forum on Regulatory Sciences, Washington, DC, September 29, 1994.        
Tauxe, R.V., ``Epidemiology of Campylobacter jejuni infections in the United States and other Industrialized Nations.'' In Nachamkin, Blaser, Tompkins, 
  ed. Campylobacter jejuni: Current Status and Future Trends, 1994, chapter 2, pages 9-19. Tauxe, R.V. and P.A. Blake, 1992. ``Salmonellosis'' Chap. 12.
  In: Public Health & Preventative Medicine, 13th ed. (Eds: Last JM: Wallace RB; Barrett-Conner E) Appleton & Lange, Norwalk, Connecticut, 266-268.     
Tauxe, R.V., N. Hargrett-Bean, C.M. Patton, and I.K. Wachsmuth. 1988. ``Campylobacter Isolates in the United States, 1982-1986,'' Morbidity and         
  Mortality Weekly Report, vol 31, no. SS-2: pages 1-14.                                                                                                



                            Table 5.--Medical Costs and Productivity Losses Estimated for Selected Foodborne Pathogens, 1993                            
--------------------------------------------------------------------------------------------------------------------------------------------------------
                                                                     Foodborne illness                   Percent      Meat/poultry related        Total 
                                                              ------------------------------  Foodborne    from  ------------------------------  costs *
                           Pathogen                                                            * costs    meat/                                   meat/ 
                                                                 Est. No. of     Est. No.      (bil $)   poultry    Est. No. of     Est. No.     poultry
                                                                    cases         deaths                   (%)         cases         deaths      (bil $)
--------------------------------------------------------------------------------------------------------------------------------------------------------
Bacteria:                                                                                                                                               
    Campylobacter jejuni or coli.............................  1,375,000-1,75                                                                           
                                                                        0,000  110-511       0.6-1.0       75     1,031,250-1,31                        
                                                                                                                           2,500  83-383        0.5-0.8 
    Clostridium perfringens **...............................          10,000     100          0.1         50              5,000      50         0.1    
    Escherichia coli 0157:H7.................................    8,000-16,000  160-400       0.2-0.6       75       6,000-12,000  120-300       0.2-0.5 
    Listeria monocytogenes...................................     1,526-1,767  378-485       0.2-0.3       50            763-884  189-243       0.1-0.2 
    Salmonella...............................................  696,000-3,840,                                                                           
                                                                          000  696-3,840     0.6-3.5     50-75    348,000-2,880,                        
                                                                                                                             000  348-2,880     0.3-2.6 
    Staphylococcus aureus **.................................       1,513,000   1,210          1.2         50            756,500     605         0.6    
                                                              ------------------------------------------------------------------------------------------
        Subtotal.............................................  3,603,526-7,13                                                                           
                                                                        0,767  2,654-6,546   2.9-6.7      N/A     2,147,513-4,96                        
                                                                                                                           6,884  1,395-4,461   1.8-4.8 
                                                              ------------------------------------------------------------------------------------------
Parasite:                                                                                                                                               
    Toxoplasma gondii........................................           2,056      41          2.7        100              2,056      41         2.7    
                                                              ------------------------------------------------------------------------------------------
        Total................................................  3,605,582-7,13                                                                           
                                                                        2,823  2,695-6,587   5.6-9.4      N/A     2,149,569-4,96                        
                                                                                                                           8,940  1,436-4,502   4.5-7.5 
--------------------------------------------------------------------------------------------------------------------------------------------------------
Source: ERS, 1993                                                                                                                                       
* Column rounded to one decimal place.                                                                                                                  
** Roberts' rough approximation of costs in ``Human Illness Costs of Foodborne Bacteria'', Amer. J. of Agricultural Economics, vol. 71, no. 2 (May 1989)
  pp. 468-474 were updated to 1993 dollars using the Consumer Price Index (all items, annual average). Cost estimates for other pathogens are more      
  detailed, see the following for a discussion of the methodology:                                                                                      
listeriosis--Roberts, Tanya and Robert Pinner, ``Economic Impact of Disease Caused by Listeria monocytogenes'' in Foodborne Listeriosis ed. by A.J.     
  Miller, J.L. Smith, and G.A. Somkuti. Elsevier Science: Amsterdam, The Netherlands, 1990, pp. 137-149,                                                
E. coli O157:H7--Roberts, T. and Marks, S., ``E. coli O157:H7 Ranks as the Fourth Most Costly Foodborne Disease,'' FoodReview, USDA/ERS, Sept-Dec 1993, 
  pp. 51-59.                                                                                                                                            
salmonellosis--Roberts, Tanya, ``Salmonellosis Control: Estimated Economic Costs,'' Poultry Science. Vol. 67 (June 1988) pp. 936-943,                   
campylobacteriosis--Morrison, Rosanna Mentzer, Tanya Roberts, and Lawrence Witucki, ``Irradiation of U.S. Poultry--Benefits, Costs, and Export          
  Potential, FoodReview, Vol. 15, No. 3, October-December 1992, pp. 16-21, congenital toxoplasmosis--Roberts, T., K.D. Murrell, and S. Marks. 1944.     
  ``Economic Losses Caused by Foodborne Parasitic Diseases,'' Parasitology Today. vol. 10, no. 11: 419-423; and Roberts, Tanya and J.K. Frenkel,        
  ``Estimating Income Losses and Other Preventable Costs Caused by Congenital Toxoplasmosis in People in the United States,'' J. of the Amer. Veterinary
  Medical Assoc., vol. 196, no. 2 (January 15, 1990) pages 249-256.                                                                                     
N/A indicates item is not-applicable.                                                                                                                   

    ERS is currently working on a sensitivity analysis for their cost 
of illness estimates for foodborne illness. The sensitivity analysis 
replaces the LS VOSL estimates with estimates found in the literature 
on wage-risk studies. Preliminary findings show that the estimates of 
the total cost of foodborne illness will increase greatly when these 
higher VOSL estimates are used.
    FSIS considers that the existing conservative estimates are 
appropriate considering the controversy and uncertainty. The 
conservative estimates are more than sufficient to justify the

[[Page 38965]]

final rule implementing a new HACCP-based regulatory program for meat 
and poultry. This final RIA uses the cost of illness estimates shown in 
Table 5.
    Another comment stated that the cost of illness estimates are low 
because they do not account for increases in productivity. In response, 
the Agency notes that ERS used Landefeld and Seskin's estimates for the 
value of a statistical life, and those estimates do include an 
estimated 1% annual increase in productivity.
    One commenter suggested that a methodology based on earning power 
may overestimate the value of life where many deaths from foodborne 
illness are the very elderly, the immunocompromised and the terminally 
ill. This commenter also noted that while all deaths are tragic, from a 
strictly economic standpoint many of these tragic cases have little or 
no productivity left and in fact are utilizing resources at the rate of 
$3,000 to $12,000 or more dollars per month of maintenance.
    The cost of illness methodology used by ERS does account for the 
fact that older individuals have lower remaining earning power than 
younger individuals. This difference was taken into account when 
estimating the costs of lost productivity for salmonellosis patients. 
Different Landefeld and Seskin estimates of the values of statistical 
life were used for the different age categories. The methodology used 
U.S. death certificate data to estimate that the average age for 
patients who die from salmonellosis is over 65 years. The concept of a 
statistical value of life accounts for the fact that older individuals 
may continue to work or be retired or be patients under long term 
health care.
3. Percentage of Foodborne Illness and Cost of Foodborne Illness 
Attributable to Meat and Poultry
    The fifth column of Table 5 includes estimates on the percentage of 
foodborne illness attributable to meat and poultry products. A separate 
estimate has been developed for each pathogen. These estimates are 
based on outbreak data reported under the CDC Foodborne Disease 
Outbreak Surveillance System and on data from community-based and other 
epidemiologic studies. Major data sources are cited in the preamble to 
the final rule. An assumption is made in this analysis that the source 
of foodborne pathogens, i.e., meat and poultry versus dairy products, 
seafood, vegetable, etc., has no effect on the cost of illness. The 
Department is not aware of any data indicating that the severity of 
foodborne illness cases varies by source of pathogens.
    Comments noted that the Department had increased the percentage of 
foodborne illness attributable to meat and poultry from the earlier 
rulemaking for safe handling labels. One commenter stated that the 
Department has not revealed any new information which would support 
such an increase.
    At this time, data on incidence of foodborne illnesses and the 
percentage of cases attributable to different food items are limited. 
Estimates by pathogen have been made by experts at CDC and USDA, based 
on a variety of studies. However, these are, indeed, estimates: FSIS 
does not have exact numbers. The estimates in the 1993 Federal Register 
document were relatively crude, assuming that 100% of Campylobacter and 
E. coli O157:H7 cases, 96% of Salmonella cases, and 85% of Listeria 
cases were foodborne, and that, for all bacterial pathogens, a flat 50% 
of foodborne cases were attributable to meat and poultry. The 1995 
document looked at the numbers in a somewhat more sophisticated way, 
evaluating each pathogen individually and, where appropriate, giving 
ranges for, first, percentage of cases which were foodborne, and, 
secondly, percentage of cases which were attributable to meat and 
poultry. Nonetheless, when all of the various percentages are 
multiplied out, estimates of total cases attributable to meat and 
poultry were remarkably similar, as shown below in Table 6.

                   Table 6.--Percentage of Foodborne Illness Attributable to Meat and Poultry                   
----------------------------------------------------------------------------------------------------------------
                                               Percentage   Percentage                                          
                                                of total     of total                                           
                                                 cases         cases                                            
                                               attributed   attributed    Estimated total      Estimated total  
                  Pathogen                    to meat and   to meat and     cases, 1993          cases, 1995    
                                               poultry <SUP>a     poultry,                                           
                                                  1993         1995                                             
                                               (percent)     (percent)                                          
----------------------------------------------------------------------------------------------------------------
Campylobacter...............................           50      41-53      1,050,000         1,031,250-1,312,500 
Salmonella..................................           48      43-72        921,600         348,000-2,880,000   
E. coli O157:H7.............................           50         60     3,834-10,22        46,000-12,000       
Listeria....................................           43      43-48        649-672           763-884           
----------------------------------------------------------------------------------------------------------------
<SUP>a Reflects percentage of foodborne multiplied by percentage attributable to meat and poultry.                   

    Most other comments related to the estimates on the percentage of 
foodborne illness attributable to poultry. Comments questioned the high 
incidence of poultry-related foodborne illness when even, as a 
commenter asserted, public health authorities tell consumers that the 
problem with poultry meat is not due to consumption because poultry is 
cooked. Comments questioned whether cross-contamination in the kitchens 
could possibly generate such high levels of foodborne illness. Related 
comments suggested that if cross-contamination was such a serious 
problem, the data would show more outbreaks and fewer single cases. 
Other comments suggested that the cost of salmonellosis attributed to 
poultry was high because of the high incidence of Salmonella 
enteritidis in eggs and requested that the Agency exclude any foodborne 
illness costs associated with eggs, because those issues are outside 
the scope of this rulemaking. Another comment cited an Australian 
finding that the Campylobacter strains that infect chickens are not the 
strains that primarily infect humans.
    The Department agrees that undercooked poultry is not a primary 
cause of foodborne illness. The preamble to the proposal stated that 
the majority of salmonellosis results from cross-contamination. The 
best available estimates for foodborne illness do suggest that a high 
incidence of illness is attributable to cross-contamination in 
kitchens--both household kitchens and food-service establishments.
    The comment suggesting that cross-contamination would have led to 
more outbreaks makes sense, if the available estimates on incidence 
were heavily

[[Page 38966]]

based on outbreak data. However, as mentioned in the proposal, it is 
widely recognized that CDC outbreak data do not provide accurate 
estimates of foodborne disease incidence. The outbreak data are more 
useful in identifying factors that lead to illness and have been used 
to estimate proportions of illness attributable to specific food 
groups. They do not play a major role in the overall incidence 
estimates. The existing incidence estimates are for total cases 
including both individual cases and multiple cases. The methodology 
used does not distinguish between outbreaks and single cases. Just as 
there are unreported individual cases of foodborne illness, there are 
unreported cases where entire households or portions of households 
experience foodborne illness due to cross-contamination in household 
kitchens. As discussed above, the estimates of foodborne illness were 
derived from both CDC outbreak data and community-based epidemiologic 
studies.
    The outbreak data (two or more individuals ill from the same 
source) are compiled by CDC from reports that are voluntarily submitted 
from state and local health authorities. The laboratory reporting 
system for Salmonella only captures information on those cases where a 
patient sees a doctor, the doctor collects a stool culture and sends 
the culture to a participating laboratory and the laboratory can 
perform the specific diagnostic test. The estimates for overall disease 
incidence are derived using both databases plus data collected from 
population-based studies in specific geographic areas. The current 
(initiative) collaborative surveillance project should improve the 
estimates in the future.
    The comment referring to the Australian finding is referring to an 
article by Korolik, et al, published in the May 1995 issue of the 
Journal of Clinical Microbiology, entitled, ``Differentiation of 
Campylobacter jejuni and Campylobacter coli strains by Using 
Restriction Endonuclease DNA Profiles and DNA Fragment Polymorphisms.'' 
The study was undertaken to determine if DNA fingerprinting 
technologies could identify strains of Campylobacter in chickens that 
cause disease in humans.
    FSIS reviewed the article and concluded that the study did not 
refute U.S. epidemiologic studies showing that approximately 50% of 
human Campylobacter infections are due to poultry. To confirm FSIS's 
interpretation of the study, a staff member contacted the author, Dr. 
Victoria Korolik, in Australia. She confirmed that her study does not 
shed doubt on the role of poultry in human Campylobacter infections.
4. Pathogens Addressed by the Rule
    While the proposed rule indicated that HACCP systems will be 
designed to control all public health hazards, the preliminary benefits 
analysis assumed that the primary benefits will come from controlling 
the three most common enteric pathogens of animal origin: Campylobacter 
jejuni/coli, E. coli O157:H7, Salmonella and one environmental pathogen 
Listeria monocytogenes. Two other pathogens--Clostridium perfringens 
and Staphylococcus aureus primarily become or create hazards in meat 
and poultry products as prepared in restaurants, other commercial 
kitchens, and in homes. Consequently, the proposed regulatory program, 
which focuses on the manufacturing sector, will not significantly 
affect the presence of these organisms on meat and poultry products.
    The public comments did not address the assumption that the 
proposed rule would have the most impact on the four pathogens 
identified above and that benefits would be most appropriately 
discussed in terms of reducing the level of these pathogens. This final 
RIA will continue to assume that the HACCP-based regulatory program 
will have the most impact on the four pathogens identified in the 
preliminary analysis.
    The preliminary benefits analysis also included an assumption 
concerning the percentage of the four pathogens that contaminate the 
meat and poultry supply at inspected establishments or grow from 
contamination that occurs at inspected locations. Based on the expert 
judgment of FSIS microbiologists, the preliminary benefit analysis 
assumed that 90 percent of the four pathogens result from contamination 
that occurs at inspected establishments.
    The public comments did not directly address the estimate that 
slaughter and processing establishments are the source of 90 percent of 
enteric pathogen contamination. There were, however, a large number of 
comments that cited studies or estimates that show or indicate that the 
majority of foodborne illness can be attributed to improper cooking, 
recontamination and other mishandling and abuse in the food service and 
home environment. Many comments cited data presented in the 1994 CAST 
Report which ``demonstrated'' that only 6.9 percent of outbreaks were 
``attributable'' to the food processing establishments. Other comments 
referred to ``a well-recognized fact that 97 percent of the problems 
with foodborne illness occur outside the realm of state and federal 
inspection.'' Other comments attributed the 97 percent figure to a 
Special Report by the American Association of Meat Processors. These 
types of comments were presented in a manner indicating that the 
commenters believe that the data attributing ``cause'' to the food 
service or home environment directly contradicts the Agency's estimate 
that inspected establishments are the source of 90 percent of the four 
pathogens addressed by this rule.
    In response, the Agency points out that the studies cited by 
commenters concluding that high percentages of foodborne illness are 
attributable to factors such as temperature abuse and mishandling do 
not conflict with either the assumption that slaughter and processing 
establishments are the source of 90 percent of enteric pathogen 
contamination or the assumption discussed later concerning the 
effectiveness of HACCP in reducing that contamination. Occurrence of 
foodborne disease is a multi-step process. The first, and critical, 
step is the introduction of a pathogen into or onto the raw product. If 
a pathogen is present, then subsequent temperature abuse or mishandling 
may permit bacterial counts to increase to levels which increase the 
likelihood that illness will occur; mishandling may result in cross-
contamination of other foods which are not cooked before being eaten; 
or improper cooking may not kill all pathogenic bacteria present in the 
product. In these instances, it may be said that the illness was 
``caused'' by improper handling. However, disease would not have 
occurred if the pathogen had not been present on the raw product in the 
first place.
    The CAST study included a table showing factors contributing to the 
occurrence of 1,080 outbreaks occurring from 1973 to 1982. That table 
consisted of data from the CDC national foodborne disease surveillance 
system that was published in an article in the Journal of Food 
Protection by Frank L. Bryan in 1988. The CAST study and journal 
articles use terminology like ``factors that contribute'' and address 
the location or type of employee/consumer where any mishandling or 
mistreatment of food occurred. The focus of these studies is to enhance 
our understanding of the sequences of events and behaviors that lead to 
foodborne illness since behavioral modification for the food preparer 
and consumer at the end of the food chain may have the greatest impact 
on the incidence of foodborne disease. Many of the comments are written 
in a manner that blurs the distinction

[[Page 38967]]

between factors in the kitchen that may permit an outbreak to occur 
from slaughter-origin contamination and those that would have caused an 
outbreak despite the absence of contamination of the raw ingredients.
    The comments referring to the CAST study or directly to CDC 
estimates have not interpreted the Foodborne Disease Outbreak 
Surveillance Data correctly. The standard CDC foodborne disease 
outbreak report form does not include a question about whether the food 
processing industry was involved, and while many foodborne outbreaks 
have a chain of causation, investigators may differ in their assessment 
of the point or points in the chain to which primary responsibility for 
occurrence of the outbreak should be assigned.
    The Bryan article used for the CAST study had the following summary 
concerning the role of food processing establishments: ``Many of the 
animals that enter abattoirs are either infected or contaminated with 
foodborne pathogens and further spread occurs during processing. Hence, 
abattoirs and raw-product processing establishments must accept some of 
the blame of spreading salmonellae and other pathogens to many 
carcasses and pieces of meat. These products are major sources of 
pathogens for food-service establishments and homes where further abuse 
(e.g., inadequate cooking or cross contamination) leads to outbreaks of 
foodborne illness.''
    The comments have not provided any basis for changing the expert 
judgment of FSIS microbiologists that inspected establishments are the 
source of 90 percent of the four pathogens addressed by the final rule. 
This final benefits analysis is based on this assumption.
5. Effectiveness of the Rule in Reducing Pathogens
    In accordance with the assumption that meat and poultry 
establishments are the source of 90 percent of the four pathogens 
addressed by the rule, the preliminary analysis calculated the benefits 
under a scenario where the proposed rule would eliminate essentially 
100 percent of those pathogens that enter the meat and poultry supply 
at inspected processing establishments. In other words, for the 
preliminary analysis, FSIS calculated an estimate of maximum benefits 
by assuming the rule would eliminate 100 percent of the 90 percent.
    By assuming this scenario, FSIS was not predicting that it believed 
that the rule would result in elimination of 100 percent of those 
pathogens in the manufacturing sector. Rather, the Agency was 
acknowledging that it has responsibility for having a food safety 
objective that recognizes the scope of the problem and attempts to 
reduce pathogens in that sector as much as possible, since without 
pathogens, no amount of subsequent abuse would result in foodborne 
illness.
    By presenting a sensitivity analysis in the proposal, FSIS intended 
to clarify that the benefit estimates were a maximum and not a 
prediction of what is likely to happen. The distinction was unclear to 
many commenters who expressed doubt that the proposed HACCP program 
would result in a 90 percent reduction in pathogens. A large number of 
comments on the potential effectiveness of HACCP programs contrasted 
the FSIS estimates with those contained in the recent study by the 
Institute of Food Science and Engineering, Texas A&M University, titled 
``Reforming Meat and Poultry Inspection: Impacts of Policy Options,'' 
(hereafter referred to as the IFSE study). Both FSIS and IFSE estimates 
are useful as assumptions rather than as quantitative predictions of 
potential effectiveness of HACCP.
    The ISFE study examined four policy options for addressing 
pathogens in the meat and poultry supply. One option called for 
mandatory HACCP for inspected slaughter and processing establishments 
and estimated that mandatory HACCP in inspected establishments would 
produce a 20 percent reduction in pathogens. The difference in the FSIS 
and IFSE estimates is not based on data but on assumptions for 
different ``HACCP'' scenarios.
    The HACCP program scenario considered in the IFSE study did not 
assume a mandatory pathogen reduction performance standard. Requiring 
process control without a standard could lead to processes that are 
well controlled at unacceptable pathogen levels. The Agency would agree 
that such a situation would result in less pathogen reduction. FSIS 
believes that a standard is necessary to encourage innovation and 
provide the impetus for continuing improvement and increasing 
effectiveness. In estimating effectiveness, the IFSE study noted that 
``with experience and additional research, it is possible that higher 
levels of reduction in pathogens could be achieved * * *''.
    Another major difference between the two program scenarios is that 
the IFSE program does not include a prerequisite requirement for SOP's. 
SOP's could cover potential sources of enteric and environmental 
pathogens that are not be covered under a HACCP plan. However, as 
discussed in Section I, this analysis discusses benefits of SOP's in 
terms of increased productivity for inspection resources and clarity of 
responsibilities.
    Several comments refer to the IFSE estimates as being more 
objective or ``scientific'' than those in the Agency's analysis. The 
IFSE authors characterize their own effectiveness estimates as ``the 
consensus judgment of the task force'' or ``the most reasonable 
expectation.'' The IFSE estimates are judgments, as are the Agency's 
estimates.
    A general comment related to the effectiveness issue stated that 
while HACCP remains an interesting theoretical concept, it is still 
only a concept that has never been tested on a meaningful scale under 
actual meat establishment conditions, and never proven to significantly 
improve the microbial quality of the finished product. Although HACCP 
has been tested in food processing establishments to the satisfaction 
of scientists, food technologists, and industry management to produce 
safe food, the Agency recognizes that the potential effectiveness of 
HACCP in reducing pathogens within a regulatory framework is unknown at 
the present time. FSIS conducted a pilot HACCP study in nine 
establishments from 1991 to 1993. Findings regarding pathogen reduction 
effectiveness were inconclusive. FSIS did not receive any data during 
the comment period from establishments currently operating HACCP 
systems. Rather than select an arbitrary effectiveness estimate, or use 
the maximum potential 100 percent estimate from the preliminary 
analysis, this RIA will present a range of effectiveness estimates and 
show the minimum level necessary to generate net benefits.
6. Estimated Reduction in Cost of Foodborne Illness
    Several comments focused on the issue that the relationship between 
pathogen reductions at the manufacturing stage and foodborne illness 
reductions is unknown. The comments recognize that the proposal did 
acknowledge that little data exist on the relationship between pathogen 
levels and incidence of illness. One comment pointed out that FSIS 
recognized that the pathogen testing requirements that are part of the 
proposal will help to elucidate the relationship between pathogen 
contamination and foodborne disease. The commenter concluded that it 
did not seem reasonable for the Agency to rely on an assumption, whose 
very validity can only be tested by the implementation of the proposal 
under examination, to justify the proposal.

[[Page 38968]]

Other commenters concluded that the Agency needed to develop better 
data or complete a thorough risk assessment that would establish the 
public health benefits of pathogen reduction before proceeding.
    The comments asking for better data or requesting a thorough risk 
assessment are not comments on the cost-benefits analysis. These 
comments imply there is insufficient evidence to support new pathogen 
reduction efforts. This issue is addressed in the preamble to the final 
rule. The comments have made a policy judgment with which the 
Department does not agree.
    For the benefits analysis included with the proposed rule, FSIS 
assumed that a reduction in pathogens will lead to a corresponding 
proportional reduction in foodborne illness. The Department notes that 
the IFSE study referred to favorably by many commenters used the same 
method for estimating public health benefits as did FSIS, i.e., a 
reduction in pathogens leads to a proportionate reduction in illness 
and death. The Agency is aware that the proportionate reduction method 
is an assumption that has not been tested or validated. However, the 
Agency also recognizes that research methodology for relating pathogen 
levels at establishments to incidence of illness is in its early 
developmental stages. Risk models for foodborne pathogens are likely to 
develop as the basis for regulatory decision-making in the future. The 
Agency believes the implementation of mandatory HACCP will improve food 
safety and protect public health while research in modeling risk 
associated with foodborne pathogens continues.
    The Agency has and continues to support any effort to improve the 
quality of data and methodology available for risk assessment of 
illness caused by foodborne biological agents. FSIS, FDA, CDC, and 
local public health departments are collaborating with state health 
departments and local investigators at five locations nationwide to 
identify more accurately the incidence of foodborne illness, especially 
illness caused by Salmonella and E. coli O157:H7.

G. Summary

    The final rule addresses four pathogens that are estimated to cause 
from $1.1 to $4.1 billion in annual illness and death costs 
attributable to meat and poultry products. The rule addresses 90 
percent of that cost of illness or from $0.99 to $3.69 billion 
annually. FSIS recognizes that the actual effectiveness of the final 
requirements in reducing pathogens is unknown, and presents a range of 
benefits based on reducing varying percentages of the $0.99 to $3.69 
billion in annual cost of foodborne illness addressed by this rule.

References

    International Commission on Microbiological Specifications for 
Foods (ICMSF). 1980. Microbial Ecology of Foods: Factors Affecting 
Life and Death of Microorganisms. Volume I. Academic Press, New 
York. Pp. 215-231.
    Rolfe, R.D. 1991. Population dynamics of the intestinal tract. 
In: Colonization Control of Human Bacterial Enteropathogens in 
Poultry. Ed. L.C. Blankenshipp, J.S. Bailey, N.A. Cox, S.E. Craven, 
R.J. Meinersmann, N.J. Stern. Academic Press, Inc., New York. Pp. 
59-76.
    Centers for Disease Control (CDC). 1988. Campylobacter isolates 
in the United States, 1982-1986. In: CDC Surveillance Summaries. 
June 1988. MMWR 1988; 37 (No. SS-2:1-13).
    Blaser, M.J. & L.S. Newman. 1982. A review of human 
salmonellosis: I. Infective dose. Reviews of Infectious Disease. 
4(6):1096-1106.
    Buchanan, R.L. & R.C. Whiting. 1996. Risk assessment and 
predictive microbiology. J. Food Protection. 31-36.
    Burmaster, D.E. & P.D. Anderson. 1994. Principles of good 
practice for the use of Monte Carlo techniques in human health and 
ecological risk assessments. Risk Analysis. 14(4):477-481.
    Council for Agricultural Science and Technology (CAST). 1994. 
Foodborne pathogens: Risks and consequences, Report No. 122, p. 87.
    Haas, C.N. 1983. Estimation of risk due to low doses of 
microorganisms: A comparison of alternative methodologies. Am. J. 
Epidemiol. 118:573-582.
    National Research Council. 1983. Committee on Institutional 
Means for Assessment of Risks to Public Health. Risk Assessment in 
the Federal Government: Managing the Process. National Academy 
Press, Washington, DC p. 191.
    Tappero, J.W. et al. 1995. Reduction in the Incidence of Human 
Listeriosis in the United States; Effectiveness of Prevention 
Efforts? Journal of the American Medical Association. 273(14):1118-
1122.
    Gellin, B., C.V. Broome, R. Weaver and A.W. Hightower. 
[Listeriosis Study Group.] Geographic Differences in Listeriosis in 
the U.S. (Abstract). In Program and Abstracts of the twenty-seventh 
International Conference on Antimicrobial Agents and Chemotherapy. 
p. 155. Washington, DC: American Society for Microbiology, 1987.

V. Cost Analysis

A. Introduction

    The final HACCP rule includes several regulatory components all 
directed at improving process control in meat and poultry operations in 
order to reduce the risk of foodborne illness associated with meat and 
poultry products. The requirements of the final rule are organized 
around the following three sections:
    <bullet> Requirements that all inspected establishments develop and 
implement sanitation Standard Operating Procedures (SOP's) within 6 
months.
    <bullet> Requirements that all inspected establishments develop and 
implement HACCP programs within the 18 to 42 month time period 
following publication. Scheduling will be based on establishment size.
    <bullet> Requirements that (1) all establishments slaughtering 
cattle, swine, chickens, or turkeys, or producing a raw ground product 
from beef, pork, chicken or turkey comply with new pathogen reduction 
performance standards for Salmonella and (2) all establishments 
slaughtering cattle, swine, chicken or turkeys implement microbial 
testing programs using generic E. coli within 6 months. Compliance with 
the pathogen reduction performance standards for Salmonella will be 
required at the time the establishment is required to implement HACCP.
    This cost analysis is presented in three sections. The first 
section describes the methodology used in generating cost estimates. 
The next section addresses the regulatory flexibility designed to 
reduce the burden on small business. The last section presents the cost 
estimates for each regulatory requirement. For each broad requirement, 
the discussion of the cost estimates is organized using the following 
five topics:
    <bullet> Summary of the requirements in the final rule identifying 
any changes from the proposal.
    <bullet> Review of the cost estimates from the preliminary RIA.
    <bullet> Summary of the comments related to the preliminary cost 
estimates.
    <bullet> Response to the comments.
    <bullet> Final cost estimates.

B. Methodology for Cost Analysis

    The final pathogen reduction/HACCP rule includes regulatory 
requirements that are directed at improving the control over food 
processing operations. In general, compliance with these requirements 
requires expenditures of time, i.e., employee hours to develop plans, 
monitor critical control points, record findings and collect and 
analyze samples. This final RIA is based on time required by four 
categories of employees that were defined in the supplemental cost 
analysis. These include the following:
    <bullet> Quality Control manager earning $25.60 per hour.
    <bullet> Supervisors or QC technicians that review findings and 
records at $18.13 per hour.

[[Page 38969]]

    <bullet> Laboratory technicians earning $18.13 per hour.
    <bullet> Establishment employees/production workers that would 
monitor sanitation and HACCP programs or collect samples at $12.87 per 
hour.
    The four categories of wages are based on 1993 data adjusted for 
1994 dollar inflation from the Bureau of Labor Statistics and Meat and 
Poultry Magazine and include a 33 percent overhead requirement for 
benefits such as health insurance and retirement contributions. Unless 
otherwise noted, the analysis assumes that all establishments and 
employees work a standard 52 week, 260 day, 2080 hour work year.
    This final cost discussion is based on retracing the steps and/or 
calculations of the preliminary analysis and discussing related public 
comments in the appropriate sections. Other comments that are related 
to the analysis but do not reflect directly on the methodology are 
summarized at the end of the analysis in Appendix A.
    This analysis makes frequent references to the Enhanced Economic 
Database. In 1994, the Research Triangle Institute (RTI) took a 
compilation of existing FSIS databases containing establishment 
production or inspection data and added data on annual sales and 
employment from sources that included Dun and Bradstreet and American 
Business List databases. Actual estimates for annual sales and number 
of employees were available for approximately 80 percent of the 
establishments. In other cases, estimates for sales and number of 
employees were developed using the employment/sales data for 
establishments producing the same type and volume of product.
    The enhanced database includes production data (number of head 
slaughtered, pounds of product produced) from 1993 for all federally-
inspected establishments in operation as of August 1994. The 
preliminary analysis and this final RIA combine 1993 production data 
with the population of federally and state- inspected establishments 
that were in operation as of August 1994. As of August 1994, there were 
6,186 federally inspected and 2,893 state inspected establishments. 
These 9,079 establishments include a total of 11,719 ``operations''--
2,597 red meat slaughter operations, 364 poultry slaughter operations 
and 8,758 further processing operations.
    This final analysis assumes a constant level of 9,079 inspected 
establishments. The analysis does not attempt to account for costs 
associated with exits from or entries into the marketplace. For 
operations that are entirely new, or include a new processing 
operation, the requirements for HACCP plans and sanitation SOPs will 
increase the one-time, up-front cost of entering the market. If 
marketplace entry involves the purchase of an existing business, the 
business will already have an existing HACCP plan and sanitation SOP. 
In these cases, the acquisition cost of the business would include the 
value of the existing HACCP plan and SOP.
    There should be minimal additional cost for HACCP and SOP plan 
development for new construction that expands a firm by replicating an 
existing operation in a new location. This type of new establishment 
can apply HACCP and SOP plans that have been developed for a similar 
existing establishment. This analysis has assumed that each 
establishment is independent and has not reduced cost estimates to 
account for firms that operate several similar establishments.
    The preliminary analysis developed cost estimates for three sizes 
of manufacturing establishments. Most of the costs that involve 
employee time are influenced by a number of factors including the 
physical size of the establishment, the volume of production, the type 
of production practices and the number or production lines. The 
preliminary analysis used the data on annual sales developed by RTI 
because the sales data correlated reasonably well with size and 
production volume data and the Agency had an estimate of sales for 
6,186 federally inspected establishments.
    For the preliminary analysis the Agency defined a large 
establishment as one with over $50 million in annual sales, a medium 
establishment as one with between $2.5 and $50 million and a small 
establishment as one with less than $2.5 million in annual sales. For 
calculating costs, the Agency collected data from the field based on 
these three size categories. Public comments provided good reason to 
change size definitions for implementation (regulatory flexibility) 
purposes and the Agency has done so for the final rule. This does not 
affect the accuracy of proposed or current cost estimates based on 
previously collected data. The final analysis uses the old categories 
for presenting cost data to facilitate comparisons and minimize 
confusion. To summarize, this cost analysis uses the terms high, medium 
and low volume producers for cost presentation that involves average 
establishment costs and uses the terms large, small and very small 
business for discussing regulatory flexibility. The cost and 
flexibility principles do not overlap in this analysis.
    Commenters pointed out that in comparing total costs with the value 
of current production, the preliminary analysis did not address impacts 
on producers, i.e., the costs that would be passed back to livestock 
producers. FSIS recognizes that some costs will be passed back to 
producers in terms of lower prices for live animals and other costs 
will be passed forward in terms of higher consumer prices. Other costs 
may have to be absorbed by slaughter and processing establishments. 
Because the necessary knowledge of empirical cost structures and supply 
and demand elasticities is inadequate, FSIS does not offer any 
quantitative estimates of the distribution of costs of this rule on 
various sectors of the production and marketing chain. The aggregate 
cost estimate establishes an upper bound on the costs any sector might 
ultimately bear.
    There are two types of potential costs that were not addressed in 
the preliminary cost analysis. The first type of cost is the cost of 
taking corrective action when routine monitoring of a CCP finds a 
deviation from a critical limit. The critical limit could be associated 
with assuring compliance with existing regulatory requirements or it 
could be a limit set to assure compliance with the new pathogen 
reduction standards for Salmonella or the criteria established for 
generic E. coli. Corrective action would also occur when FSIS would 
find a problem with either a HACCP plan or a sanitation SOP.
    The second type of potential cost is related to the question of 
whether existing processing methods are adequate to meet the pathogen 
reduction performance standards for Salmonella and the criteria for 
generic E. coli. It is expected that some establishments will have to 
make permanent changes to their existing production practices to have a 
HACCP-based program that assures compliance with the new standards and 
criteria. The final rule raises a third type of potential cost when it 
outlines the Agency's plans for using the results of its own Salmonella 
testing program for regulatory purposes. Whether or not this testing 
leads to industry testing costs depends on whether the government 
testing indirectly forces an establishment to regularly conduct its own 
testing.
    The preliminary analysis did address a fourth category of potential 
costs that includes the cost of necessary materials, such as 
thermometers and test kits, that establishments will need to

[[Page 38970]]

systematically monitor their processes. Recognizing that the rule does 
not make any equipment obsolete, the preliminary analysis suggested 
costs of from $10 to $20 per establishment. These costs were not 
included in the overall cost summary.
    Potential costs are addressed in this final analysis under Section 
V.D.2., Costs of Meeting Pathogen Reduction and Microbial Sampling 
Requirements.

C. Regulatory Flexibility

    The Regulatory Flexibility Act (P.L. 96-354) requires analyzing 
options for regulatory relief for small businesses. This section 
reviews the regulatory relief provided in the proposal, responds to 
comments related to the definition of small business used in the 
proposal and summarizes the regulatory relief for small business 
provided for in the final rule. In Section II, this analysis addressed 
the option of providing an exemption for small business noting that 
comments on an exemption were mixed with a substantial number of 
comments from small businesses strongly opposing an exemption.
    The proposed rule intended to spread the implementation of HACCP 
over a three year period. To minimize the burden on small 
establishments, they would be given a maximum time of 36 months to 
develop and implement their HACCP plans. A small establishment was 
defined as one with annual sales of less than $2.5 million.
    The decision to use the above definition generated a large number 
of comments. ``Very small'' establishments commented that they could 
not compete with a relatively ``large'' business with annual sales of 
$2.5 million. For example one commenter stated that: ``calling an 
establishment, small, that produces $2,500,000 worth of product 
annually is not fair to those establishments producing far less.'' 
Other comments suggested that by defining small at the $2.5 million 
level, the Agency demonstrated that it does not understand what a small 
business is. Comments from businesses with annual sales of $2.5 to 
$10.0 million or even $25.0 million stated that they should also be 
considered small businesses. Commenters also pointed out that other 
Federal agencies use different definitions. For example, one commenter 
noted that OSHA uses 50 employees as their criterion for a ``small 
business.'' Others commented that FSIS should or must use the existing 
definition of fewer than 500 employees published by the Small Business 
Administration (SBA).
    Several comments promoted a set of requirements distinguishing 
``small'' from ``very small'' establishments. ``Very small'' 
establishments would only be required to implement the proposed 
provisions on sanitation standard operating procedures, antimicrobial 
treatment of carcasses, and time and temperature provisions. They would 
be exempt from routine microbial testing and long-term provisions of 
HACCP as long as annual sales do not exceed $1 million (not counting 
``pass through''). The establishments would still be subject to 
incidental sampling for microbial testing as determined by the 
Administrator. Required implementation of the three near-term 
initiatives would be 12 months after publication of the final rule.
    The ``small'' establishments (between $1.0 and $2.5 million) would 
be required to implement SOPs, antimicrobial treatment, time and 
temperature provisions, and limited routine sampling, in proportion to 
the number of slaughtered animals and/or poundage of processed 
products. The establishments would still be subject to incidental 
sampling for microbial testing as determined by the Administrator. They 
would be exempt from long-term provisions of HACCP as long as annual 
sales, as defined above, do not exceed $2.5 million. The required 
implementation of all near-term initiatives would be six months.
    There were other comments that suggested variations on the above 
definitions and requirements for ``small'' and ``very small'' 
establishments. For example, one State department of agriculture 
recommended the same requirements for ``small'' and ``very small'' 
establishments but suggested that size criteria based on head 
slaughtered or pounds produced would be more practical. Another State 
department of agriculture recommended that a ``every small'' plant be 
defined based on the number of employees (no more than 20 full-time), 
slaughter volume (no more than 2,500 animals per year), or processing 
volume (100,000 pounds of meat and/or poultry products per year). The 
recommendation suggested that a plant in this category would be 
required to implement the provisions of the proposed rule pertaining to 
sanitation SOP's and time-temperature requirements. Antimicrobial 
treatment of carcasses would be voluntary, and such a plant would be 
exempted from microbial testing as proposed. Implementation of a HACCP 
program would be initially voluntary, and phased in with considerations 
in the areas of documentation and record-keeping for the limited work 
force.
    FSIS has considered the above regulatory framework for ``small'' 
and ``very small'' establishments. Some of the suggestions are no 
longer applicable because major provisions of the proposed rule have 
been dropped. FSIS believes it has addressed the other concerns in more 
appropriate ways.
    FSIS was aware of SBA Size Standards during the development of the 
proposed rule. If FSIS used the size standard for meat and poultry 
``manufacturing'' firms, over 94 percent of the federally inspected 
establishments would meet the criterion of having fewer than 500 
employees. FSIS is also aware that there are six different SBA size 
standards that apply to the 6,415 FSIS official establishments. FSIS 
determined the SBA size standards by themselves are not appropriate for 
meeting FSIS's need to sequence HACCP implementation.
    Table 7 shows the distribution of 6,415 official establishments by 
Standard Industrial Classification (SIC) code. The SIC codes were 
developed to promote the comparability of statistics describing various 
facets of the Nation's economy. The SIC codes were used as part of the 
Enhanced Economic Analysis Database developed by Research Triangle 
Institute to represent all FSIS inspected establishments. As can be 
seen from Table 7, a significant portion of official establishments are 
not in an SIC Code for manufacturing. Food manufacturing establishments 
have a 4-digit SIC Code beginning with 20. The Census of Manufacturers 
published by the Department of Commerce characterizes the meat and 
poultry manufacturing industry by summarizing data for SIC Code 2011--
Meat Packing Establishments, SIC Code 2013--Sausages and Other Prepared 
Meats, and SIC Code 2015--Poultry Slaughtering and Processing. The SBA 
Size Standards in Table 7 are published in the Code of Federal 
Regulations--13 CFR, Chapter 1, Section 121.601.
    In a written comment, the Office of Advocacy, Small Business 
Administration claimed that FSIS was wrong in concluding that one-third 
of federally inspected establishments would have the maximum time for 
compliance with HACCP requirements using the criterion of $2.5 million 
in annual sales. In supporting their claim, they cited U.S. Census 
Bureau data. However, Census data do not accurately describe the 
federally inspected meat and poultry industry. As shown in Table 7, the 
problem is that less than half of the firms are classified in the three 
4-digit SIC Codes identified above that define meat and poultry 
manufacturing. FSIS addressed this data

[[Page 38971]]

problem by contracting with RTI to develop a more accurate economic 
profile of federally inspected meat and poultry establishments.

                           Table 7.--Establishments Standard Industrial Classification                          
----------------------------------------------------------------------------------------------------------------
                                                                Cumulative                                      
     SIC code          Standard industrial       Number of       number of             SBA size standard        
                         classification       establishments  establishments                                    
----------------------------------------------------------------------------------------------------------------
2011..............  Meat packing                    1,503             1,503   500 employees.                    
                     establishments.                                                                            
5147..............  Meats and meat products.        1,312             2,815   100 employees.                    
2013..............  Sausages and other                939             3,754   500 employees.                    
                     prepared meats.                                                                            
2015..............  Poultry slaughtering and          438             4,192   500 employees.                    
                     processing.                                                                                
4222..............  Refrigerated warehousing          356             4,548   $18,500,000.                      
                     and storage.                                                                               
5421..............  Meat and fish markets...          309             4,857   $5,000,000.                       
5144..............  Poultry and poultry               268             5,125   100 employees.                    
                     products.                                                                                  
5141..............  Groceries, general line.          238             5,363   100 employees.                    
5812..............  Eating places...........          156             5,519   $5,000,000.                       
2038..............  Frozen specialities, nec          139             5,658   500 employees.                    
5142..............  Packaged frozen foods...          130             5,788   100 employees.                    
5411..............  Grocery stores..........           95             5,883   $20,000,000.                      
5149..............  Groceries and related              65             5,948   100 employees.                    
                     products, nec.                                                                             
9999..............  Not applicable..........           63             6,011                                     
2032..............  Canned specialities.....           61             6,072   1,000 employees.                  
2099..............  Food preparations, nec..           55             6,127   500 employees.                    
Other.............  All other SIC codes.....          288            6,415                                      
----------------------------------------------------------------------------------------------------------------
Note: The Enhanced Economic Analysis Database uses the number of active establishments as of August, 1994 and   
  identified 6,415 establishments as active official establishments. Of these 6,415, a total of 229 were        
  identified as cold storage/ID warehouses, universities or churches. From the 6,415 total, 6,186 federal       
  establishments were classified as processing, slaughter or combination operations. nec--(Not Elsewhere        
  Classified).                                                                                                  

    The final rule provides for sequencing HACCP implementation by 
establishment size, using the SBA definition of a small manufacturing 
business, i.e., a small business is an establishment with fewer than 
500 employees. Those establishments with 500 or more employees will be 
referred to as large establishments. In addition, in response to 
comments that there are hundreds of ``very small'' or ``micro'' 
establishments, the Agency will classify an establishment as ``very 
small'' if it has either fewer than 10 employees or annual sales of 
less than $2.5 million.
    This sequencing of HACCP responds to a large number of comments 
requesting that small businesses be given a longer period of time to 
implement HACCP requirements. Many small businesses stated they did not 
want to be exempt, but asked for more flexibility in implementing 
HACCP. Some commenters specifically requested five, eight or 10 years 
to implement HACCP.
    While the final rule does not provide for longer periods of five, 
eight or 10 years, it does substantially extend the implementation 
period for hundreds of small and very small establishments.
     To illustrate, the proposed rule would have required HACCP plans 
in over 2,100 establishments producing raw ground product within 12 
months. Under the final rule, over 1,800 of those establishments will 
have either 30 or 42 months to implement HACCP. The smallest 5,127 
establishments (2,893 state and 2,234 federal) will have an additional 
six months. The proposed rule called for implementation of a HACCP 
system in all ``small'' establishments by 36 months; the final rule 
allows 42 months for the newly defined ``very small'' category.
    Table 8 illustrates the distribution of 6,186 federally-inspected 
slaughter, processing, and combination establishments used for the 
sequencing of HACCP implementation in the proposed rule and in the 
final rule. There are 496 more establishments in the two smaller 
categories than there were in the proposal. As shown in Table 8, there 
are 353 large, 2,941 small and 2,892 very small federally-inspected 
establishments.

    Table 8.--Size Categories for Federally Inspected Establishments    
------------------------------------------------------------------------
                                                              No. of    
      Establishment  category            Definition       establishments
------------------------------------------------------------------------
                              Proposed Rule                             
------------------------------------------------------------------------
High volume.......................  >$50 million........          849   
Medium volume.....................  $2.5-$50 million....        3,103   
Low volume........................  <$2.5 million.......        2,234   
  Total...........................  ....................        6,186   
------------------------------------------------------------------------
                    Final Rule (Sequencing of HACCP)                    
------------------------------------------------------------------------
Large.............................  <gr-thn-eq>500                353   
                                     Employees.                         
Small <SUP>a...........................  10-499 Employees....        2,941   
Very small <SUP>b......................  <10 Employees or            2,892   
                                     <$2.5 Million.                     
  Total...........................  ....................       6,186    
------------------------------------------------------------------------
<SUP>a New definition of small includes 2,445 establishments that were medium
  volume establishments plus 496 that were high volume for the          
  preliminary analysis.                                                 
<SUP>b New definition of very small includes the 2,234 establishments that   
  were low volume establishments plus 658 that were medium volume       
  establishments for the preliminary analysis.                          

D. Final Cost Estimates

1. Sanitation Standard Operating Procedures
    a. Summary of Requirements. The final rule requires that all 
inspected establishments develop and implement Sanitation SOP's within 
6 months after publication of the final rule. The proposed rule would 
have required the implementation of SOP's within 90

[[Page 38972]]

days. To facilitate the development of SOP's and to provide maximum 
flexibility, the Agency will not prescribe any specific format or 
content but will provide guidelines to assist inspected establishments 
in developing written SOP's. There will not be any FSIS approval of the 
written documents. With the exception of the implementation schedule, 
the requirements for SOP's in the final rule are the same as those in 
the proposed rule.
    b. Review of Preliminary Cost Estimates. The preliminary cost 
analysis identified separate costs for SOP plan development and SOP 
recordkeeping where recordkeeping was defined as observing or verifying 
procedures, recording findings, reviewing records and maintaining 
files. FSIS assumed that the Sanitation SOP's would be developed by a 
quality control manager at a cost of $25.60 per hour. FSIS estimated 
that it would cost an average of $128, $256 and $640 for low, medium 
and high volume establishments to develop Sanitation SOP's.
    The preliminary cost analysis assumed that Sanitation SOP's 
observation and recording for low, medium and high volume 
establishments would take 15, 25 and 45 minutes per day by an employee 
earning $12.87 per hour and that supervisory review of records would 
take 5, 10, and 20 minutes by an employee earning $18.13 per hour. In 
developing these time estimates for recording and reviewing records, 
FSIS recognized that the time required would be influenced by a number 
of factors including the physical size of the establishment, the volume 
of production, the type of production practices and the number of 
production lines. The estimates are based on program judgement of the 
time required to conduct two sets of sanitation observations per day, 
one for preoperational sanitation procedures and one for operational 
sanitation.
    Using the above inputs, the annual costs for recording and 
reviewing Sanitation SOP's records for low, medium and high volume 
establishments would be approx