# Bovine Spongiform Encephalopathy; Minimal-Risk Regions; Importation of Live Bovines and Products Derived From Bovines

> Briefs, arguments, decisions, and more.

URL: https://www.frixlaw.com/law-library/documents/fr%3A07-4595

## Record

- **Collection:** Federal Register
- **Document type:** Rule
- **Published:** September 18, 2007
- **Citation:** 72 FR 53314

## Text

DEPARTMENT OF AGRICULTURE
Animal and Plant Health Inspection Service
9 CFR Parts 93, 94, 95, and 96
[Docket No. APHIS-2006-0041]
RIN 0579-AC01
Bovine Spongiform Encephalopathy; Minimal-Risk Regions; Importation of Live Bovines and Products Derived From Bovines

AGENCY:

Animal and Plant Health Inspection Service, USDA.

ACTION:

Final rule.

SUMMARY:

We are amending the regulations regarding the importation of animals and animal products to establish conditions for the importation of the following commodities from regions that present a minimal risk of introducing bovine spongiform encephalopathy into the United States: Live bovines for any use born on or after a date determined by the Animal and Plant Health Inspection Service to be the date of effective enforcement of a ruminant-to-ruminant feed ban in the region of export; blood and blood products derived from bovines; and casings and part of the small intestine derived from bovines. We are making these amendments after conducting a risk assessment and comprehensive evaluation of the issues and concluding that such bovines and bovine products can be safely imported under the conditions described in this rule. This document also removes the delay in applicability of certain provisions of a final rule published in January 2005.

DATES:

Effective Date:
November 19, 2007.

FOR FURTHER INFORMATION CONTACT:

For information regarding ruminant products, contact Dr. Karen James-Preston, Director, Technical Trade Services, Animal Products, National Center for Import and Export, VS, APHIS, 4700 River Road Unit 38, Riverdale, MD 20737-1231; (301) 734-4356.

For information concerning live ruminants, contact Dr. Lee Ann Thomas, Director, Technical Trade Services, Animals, Organisms and Vectors, and Select Agents, National Center for Import and Export, VS, APHIS, 4700 River Road Unit 38, Riverdale, MD 20737-1231; (301) 734-4935.

For other information concerning this proposed rule, contact Dr. Lisa Ferguson, Senior Staff Veterinarian, National Center for Animal Health Programs, VS, APHIS, 4700 River Road Unit 43, Riverdale, MD 20737-1231; (301) 734-6954.

SUPPLEMENTARY INFORMATION:

Purpose

This document makes final a proposed rule that the Animal and Plant Health Inspection Service (APHIS) of the U.S. Department of Agriculture (USDA or the Department) published in the
Federal Register
on January 9, 2007 (72 FR 1101-1129, Docket No. APHIS-2006-0041). Additionally, it removes the delay of applicability of certain provisions of a final rule APHIS published in January 2005. The removal of delay is discussed below under the heading “Removal of Partial Delay of Applicability of Provisions of January 2005 Final Rule.”

In our January 2007 proposed rule, we proposed to amend the regulations in 9 CFR parts 93, 94, 95, and 96 to establish conditions for the importation of the following commodities from regions that present a minimal risk of introducing bovine spongiform encephalopathy (BSE) into the United States: Live bovines for any use born on or after a date determined by APHIS to be the date of effective enforcement of a ruminant-to-ruminant feed ban in the region of export; blood and blood products derived from bovines; and casings and part of the small intestine derived from bovines.

In this document, we respond to public comments received on the proposed rule and its underlying risk assessment and other supporting analyses. Additionally, we discuss below the history of APHIS rulemaking related to BSE minimal-risk regions.

Background

APHIS regulates the importation of animals and animal products into the United States to guard against the introduction of animal diseases. The regulations in 9 CFR parts 93, 94, 95, and 96 (referred to below as the regulations) govern the importation of certain animals, birds, poultry, meat, other animal products and byproducts, hay, and straw into the United States in order to prevent the introduction of various animal diseases, including BSE, a chronic degenerative disease affecting the central nervous system of cattle.

With some exceptions, APHIS’ regulations prohibit or restrict the importation of live ruminants and certain ruminant products and byproducts from the following three categories of regions with regard to BSE: (1) Those regions in which BSE is known to exist (listed in § 94.18(a)(1) of the regulations); (2) those regions that present an undue risk of introducing BSE into the United States because their import requirements are less restrictive than those that would be acceptable for import into the United States and/or because the regions have inadequate surveillance (listed in § 94.18(a)(2) of the regulations); and (3) those regions that present a minimal risk of introducing BSE into the United States via live ruminants and ruminant products and byproducts (listed in § 94.18(a)(3) of the regulations).

Chronology of Federal Register Publications Regarding BSE Minimal-Risk Regions

We added the § 94.18(a)(3) category (BSE minimal-risk regions) to the regulations in a final rule published in the
Federal Register
on January 4, 2005 (70 FR 459-553, Docket No. 03-080-3). In the final rule, we specified which commodities may be imported from BSE minimal-risk regions and under what conditions, and recognized Canada as a BSE minimal-risk region. (At this time, Canada is the only recognized BSE minimal-risk region.)

The January 2005 final rule was based on a proposed rule we published in the
Federal Register
on November 4, 2003 (68 FR 62386-62405, Docket No. 03-080-1). On December 25, 2003, less than 2 weeks before the close of the comment period for our proposed rule, a case of BSE in a dairy cow of Canadian origin in Washington State was verified by an international reference laboratory.

In response to comments from the public requesting an extension of the comment period and in order to give the public an additional opportunity to comment on the proposed rule in light of this development, on March 8, 2004, we published a document in the
Federal Register
(69 FR 10633-10636, Docket No. 03-080-2) reopening the comment period.

On January 4, 2005, along with the final rule, we published in the
Federal Register
a notice (70 FR 554, Docket No. 03-080-4) announcing the availability of, and requesting comments on, a final environmental assessment (EA) regarding the potential impact on the quality of the human environment due to the importation of ruminants and ruminant products and byproducts from Canada under the conditions specified in the final rule. On January 21, 2005, we published in the
Federal Register
a notice (70 FR 3183-3184, Docket No. 03-080-5) announcing the availability of a corrected version of the EA for public review and comment. On April 8, 2005, we published in the
Federal Register
a finding (70 FR 18252-18262, Docket No. 03-080-7) that the provisions of the final rule would not

have a significant impact on the quality of the human environment.

On March 11, 2005, we published a document in the
Federal Register
that gave notice that the Secretary of Agriculture was delaying until further notice the implementation of certain provisions of the final rule with regard to certain commodities (70 FR 12112-12113, Docket No. 03-080-6).

On November 28, 2005, we published in the
Federal Register
an interim rule (70 FR 71213-71218, Docket No. 03-080-8) that amended certain provisions established by the January 2005 final rule. The interim rule broadened the list of who is authorized to break seals on conveyances and allows transloading under supervision of products transiting the United States.

On March 14, 2006, we published in the
Federal Register
a technical amendment (71 FR 12994-12998, Docket No. 03-080-9) that clarified our intent with regard to certain provisions in the January 2005 final rule and corrected several inconsistencies within the rule.

On August 9, 2006, we published in the
Federal Register
a proposed rule (71 FR 45439-45444, Docket No. APHIS-2006-0026) that proposed to amend the provisions established by the January 2005 final rule by removing several restrictions regarding the identification of animals and the processing of ruminant materials from BSE minimal-risk regions, and by relieving BSE-based restrictions on hide-derived gelatin from BSE minimal-risk regions. We solicited comments concerning our proposal for 60 days ending October 10, 2006. On November 9, 2006, we published a document in the
Federal Register
(71 FR 65758-65759, Docket No. APHIS-2006-0026) reopening and extended the comment period until November 24, 2006. We received a total of 10 comments by that date. We are considering the issues raised by the commenters and will address them in a separate rulemaking document.

Scope of the January 2005 Final Rule

The regulations established by the January 2005 final rule and subsequent amendments have allowed the importation from BSE minimal-risk regions of live bovines that are under 30 months of age when imported and when slaughtered and that have been subject to a ruminant feed ban equivalent to that in place in the United States.

We did not attempt, for that rulemaking, to assess the BSE risk associated with the importation of live bovines 30 months of age or older from a BSE minimal-risk region. Our March 8, 2004, document that reopened the comment period on the November 2003 proposed rule stated that APHIS was evaluating the appropriate approach with regard to the importation of live animals 30 months of age or older from BSE minimal-risk regions, and would address that issue in a supplemental rulemaking proposal in the
Federal Register
. The provisions in our January 9, 2007, proposed rule regarding live bovines were the result of that evaluation.

The regulations established by the January 2005 final rule also provided for the importation of the following commodities derived from bovines of any age: (1) Meat, meat food products, and meat byproducts; (2) whole or half carcasses; (3) offal; (4) tallow composed of less than 0.15 percent insoluble impurities that are not otherwise eligible for importation under § 95.4(a)(1)(i) of the regulations; and (5) gelatin derived from bones of bovines that is not otherwise eligible for importation under § 94.18(c) of the regulations.

The January 2005 final rule and subsequent amendments did not change the regulations concerning the importation of blood and blood products from regions listed in § 94.18(a); the requirements for the importation of blood and blood products from BSE minimal-risk regions remain the same as the requirements for importation of blood and blood products from other regions listed in § 94.18(a)—only serum and serum albumin have been eligible for importation. The January 2005 final rule also did not change the regulations concerning the importation of bovine casings (defined as intestines, stomachs, esophagi, and urinary bladders) from regions listed in § 94.18(a); the requirements for the importation of bovine casings from BSE minimal-risk regions remain the same as the requirements for importation of bovine casings from other regions listed in § 94.18(a)—only bovine stomachs are eligible for importation.

The January 2005 final rule and subsequent amendments allowed trade to resume in many, but not all, of the commodities that had been prohibited importation from Canada following detection of a BSE-infected cow in Canada in May 2003. Following our January 2005 final rule, we continued to consider the BSE risk associated with older bovines and other bovine products from BSE minimal-risk regions—and Canada in particular—including bovine blood and blood products, bovine small intestine other than the distal ileum, and bovine casings, and included provisions in our January 2007 proposed rule for the importation of those commodities.
1

1
The regulations regarding BSE minimal-risk regions apply to bison as well as cattle. In §§ 93.400, 94.0, and 95.1 of the regulations,
bovine
is defined as
Bos taurus, Bos indicus,
and
Bison bison.
Although the research and other data cited in this rulemaking refer to bovines other than bison (i.e., to “cattle”), there is no evidence to indicate that the BSE susceptibility of bison differs from that of cattle. We therefore assume that our conclusions based on cattle-specific evidence discussed in this rulemaking are also applicable to bison. Given that no cases of BSE have been detected in bison, this is likely a conservative assumption. The provisions of this rule apply to bovines as defined in the regulations, which include bison.

Peer Review of APHIS’ Risk Assessment

As part of this rulemaking, APHIS conducted an assessment that evaluated the animal health risk to the United States of BSE—i.e., the likelihood of establishment and the potential impacts of cases that may occur even without establishment—as a result of importing the bovine commodities considered in this rule (APHIS 2006b). Our assessment concluded that, over the 20 years of the analysis, the BSE risk to the United States is negligible. We made the risk assessment available for public review and comment at the time the proposed rule was published.

In addition to making the risk assessment available for review and comment by the general public, we requested an external, formal, independent peer review of the assessment by recognized experts in the field, consistent with guidelines of the U.S. Office of Management and Budget (OMB 2004). The objective of the peer review was to determine whether the risk assessment was scientifically sound, transparent, and consistent with international standards (e.g., those by the OIE); the application of external assessments or models was appropriate; and the assumptions were justified, supported and reasonable. Comments submitted by the public on the proposed rule were submitted to the peer reviewers for their consideration. The peer review process was coordinated by an independent private contractor.

The full peer review report may be viewed at
http://www.aphis.usda.gov/peer_review/peer_review_agenda.shtml.
Additionally, we have included below, under the heading “Final Report from Peer Review of APHIS’ Risk Assessment and Responses to Peer Reviewer Questions and Recommendations,” APHIS’ responses to reviewer comments that we consider representative of the content-related questions and recommendations of the report, and our response to those questions and recommendations. In summary, the

reviewers found that the methods used in the risk assessment were scientifically rigorous in terms of using existing literature and models appropriately and making sound assumptions and that the risk assessment itself adhered to international risk assessment standards. The reviewers also agreed with the conclusion that the likelihood of establishment of BSE in the U.S. cattle population is negligible.

In addition to being supportive of the methods, evidence, and conclusions presented by APHIS in the risk assessment, the reviewers made several useful suggestions for its improvement. We made several clarifications and updates in consideration of these comments. While we expect that the changes improve the transparency and accuracy of the document, they do not alter our conclusion that the risk to the United States of BSE—i.e., the likelihood of establishment and the potential impacts of cases that may occur even without establishment—resulting from the changes outlined in the proposed rule is negligible.

Removal of Partial Delay of Applicability of Provisions of January 2005 Final Rule

Our January 2005 final rule made eligible for importation from Canada meat that is derived from bovines slaughtered in BSE minimal-risk regions, as well as certain other specified commodities derived from such bovines, provided certain specified risk-mitigating conditions have been met. The risk analysis we conducted for that rulemaking indicated a low BSE risk from such commodities derived from bovines of any age if certain conditions are met (APHIS 2004). These conditions include the removal of those tissues considered at particular risk of containing the BSE agent in infected animals (specified risk materials, or SRMs). In that rulemaking, we discussed regulatory requirements implemented by FSIS in 2004 that banned SRMs from the human food supply in the United States, and we stated that the Canadian Government had established similar safeguards in Canada.

Consequently, we provided that meat, meat byproducts, meat food products, and offal derived from bovines are eligible for importation from BSE minimal-risk regions if the following conditions, as well as all other applicable requirements of the regulations, are met:

• The commodity is derived from bovines that have been subject to a ruminant feed ban equivalent to the requirements established by the U.S. Food and Drug Administration at 21 CFR 589.2000;

• The commodity is derived from bovines for which an air-injected stunning process was not used at slaughter; and

• The SRMs and small intestine of the bovines from which the commodity was derived were removed at slaughter.

Additionally we provided that tallow composed of less than 0.15 percent insoluble impurities that is not otherwise eligible for importation under 9 CFR 95.4(a)(1)(i), and gelatin derived from bones of bovines that is not otherwise eligible for importation under 9 CFR 94.18(c) are eligible for importation from BSE minimal-risk regions, provided certain specified conditions are met.

In the economic analysis we conducted for the January 2005 final rule, we evaluated the potential economic effects of implementing that rulemaking, including implementation of the provisions allowing the importation of meat and other commodities derived from bovines slaughtered in BSE minimal-risk regions (APHIS 2004a).

In March 2005, APHIS published a document in the
Federal Register
that, pursuant to an announcement by the Secretary of Agriculture on February 9, 2005, delayed the applicability of the provisions in our January 2005 final rule as they apply to the importation from Canada of the following commodities when derived from bovines 30 months of age or older when slaughtered: (1) Meat, meat food products, and meat byproducts other than liver; (2) whole or half carcasses; (3) offal; (4) tallow composed of less than 0.15 percent insoluble impurities that is not otherwise eligible for importation under 9 CFR 95.4(a)(1)(i); and (5) gelatin derived from bones of bovines that is not otherwise eligible for importation under 9 CFR 94.18(c).

In his February 9, 2005, announcement, the Secretary stated that because ongoing investigations into recent finds of BSE in Canada in animals over 30 months of age were not complete, he felt it prudent to delay the effective date for allowing imports of meat from bovines 30 months of age and over. He also indicated that the delay of applicability would address concerns that the January 2005 final rule allowed the importation of beef from bovines 30 months of age or older, while continuing to prohibit the importation of live cattle 30 months of age or older for processing in the United States. The Secretary stated that the Department would consider and develop a plan—based on the latest scientific information and with the protection of public and animal health as the highest priority—to allow imports of live bovines 30 months of age or older as well as beef from animals 30 months of age and older.

Since the date of the partial delay of applicability of our January 2005 final rule, we have obtained additional information regarding all aspects of the issues that prompted the delay of applicability and have conducted additional analyses in line with the plan as described. The risk assessment for this final rule demonstrates the negligible BSE risk from the importation of additional classes of live cattle, including those 30 months of age or older. This includes acknowledging the potential risk pathway that could be available if the SRMs from infected imported cattle entered the ruminant feed supply in contravention of current feed regulations. The negligible risk from the importation of live older cattle therefore gives further support to the conclusion of the risk analysis conducted for our January 2005 final rule regarding meat and meat products derived from bovines of any age in BSE minimal-risk regions. Specifically, the risk is even lower for the importation of meat and meat products, as the SRMs will be removed in accordance with the regulations, than for live bovines.

Therefore, this document will remove the partial delay of applicability of the January 2005 final rule. The removal of the partial delay of applicability will become effective on the date that the other provisions of this document become applicable. Including the removal of the partial delay of applicability in this final rule and making it effective along with the other provisions of this rule will enable APHIS to more efficiently communicate the necessary implementation instructions to U.S. Customs and Border Protection and to APHIS field personnel. Additionally, it will provide commercial entities more flexibility in carrying out import planning based on the relative economic merits of importing live bovines or meat and other products derived from bovines.

Because, for reasons of efficiency for APHIS and the regulated community, the Secretary has decided to remove the delay in applicability as part of this document, we looked at the economic effects of doing so in combination with allowing the importation of bovines born on or after March 1, 1999. Although we previously analyzed the economic effects of allowing the importation of meat and other products derived from bovines 30 months of age

or older, the economic analysis for this rule provides an updated analysis.

Public Comments on the January 2007 Proposed Rule

We solicited comments concerning our January 2007 proposal for 60 days ending March 12, 2007. We received close to 400 comments by that date. The commenters included cattle industry and farm bureau associations, consumer groups, representatives of the Canadian Government and other foreign countries, State Departments of Agriculture, food processing companies, individual cattle producers, and other members of the public.

Subjects of Comments Received

A number of commenters supported the rule and recommended no changes to the proposed provisions. Other commenters supported the rule in general but recommended certain changes or actions. Other comments consisted only of recommended changes, objections to the rule in general or to specific provisions, or requests for clarification. We discuss below by topic the issues raised by commenters and our response to those comments.

General Opposition to Imports

Issue:
A number of commenters expressed general opposition to the importation of any bovines or bovine products from BSE minimal-risk regions.

Response:
It appears to us that these commenters are not addressing just our January 2007 proposed rule, but, rather, also the January 2005 final rule that recognized the category of BSE minimal-risk regions and established conditions for the importation of certain ruminants and ruminant products from such regions.

As we discussed in the January 2005 final rule, the comprehensive analysis and evaluation we conducted for that rulemaking led to the conclusion that the conditions specified in that rule for the importation of ruminants and ruminant products from BSE minimal-risk regions would be effective and would therefore protect against the introduction of BSE into the United States. Our January 2007 proposed rule considered expansion of the types of commodities allowed importation from BSE minimal-risk regions, based on an evaluation of the risk (i.e., the likelihood of establishment and the potential impacts of cases that may occur even without establishment) of importing from Canada live animals, blood and blood products, and the small intestine excluding distal ileum.) Given the determination of negligible BSE risk associated with the provisions of this final rule, and the findings associated with our 2005 final rule, there is no scientific basis for increasing restrictions from those already in effect or being established in this rule.

Issue:
A number of commenters expressed opposition, without further explanation, to the importation from BSE minimal-risk regions of live bovines 30 months of age or older and to the importation of products derived from such bovines.

Response:
We discussed in our January 2007 proposed rule the rationale for our proposal to allow the importation, under certain conditions, of live bovines 30 months or older from BSE minimal-risk regions. We discussed further the assessment of the disease risk of allowing such imports that we conducted before issuing our proposal. It is not clear to us which factors in our risk assessment or discussion of rationale were being addressed by those commenters who expressed general opposition to the importation of live bovines 30 months of age or older. We continue to consider the BSE risk from importing live bovines under the conditions specified in this rule to be negligible.

Issue:
Several commenters who expressed opposition to the proposed rule expressed concern that the agent that causes BSE has yet to be fully characterized. The commenters stated that what we know about BSE is mostly supposition, which should be a compelling reason not to allow the importation of cattle from a region of known BSE outbreaks. One commenter stated that research recently conducted at Yale University suggests that one of the agents that activates BSE may be viral, which, according to the commenter, implies that a feed ban is effective only when the virus is not present or active.

Response:
As one of the commenters noted, some researchers (Manuelidis
et al.
, 2007) suggest that diseases characterized as transmissible spongiform encephalopathies (TSEs), such as BSE, may be caused by viruses, although, at this point, no infection-specific nucleic acids have been identified.

Experimental data and epidemiological studies strongly suggest that contaminated feed containing ruminant proteins derived from infected animals was the source of the epidemic, and that the epidemic was perpetuated through the use of these materials in ruminant feed. APHIS considers that regardless of the characteristics of the BSE causal agent, it is clear that the epidemic was sustained and amplified by the recycling of BSE infected cattle into cattle feed. Despite the difficulty in definitively determining the causal agent of BSE, risk factors for transmission of the agent have been identified. The identification and characterization of these risk factors through epidemiological and experimental study have allowed the development of effective mitigations to prevent BSE spread. The development and demonstrated effectiveness of those mitigations does not require identification of the agent itself. We consider mitigation measures that address the risk factors for BSE to be effective regardless of the precise nature of the BSE agent.

Prevalence of BSE in Canada

Although the provisions of this rule apply to any region recognized by APHIS as a BSE minimal-risk region, at present APHIS recognizes only one country, Canada, as such a region. Therefore, in evaluating the BSE risk of implementing this rule, we conducted an assessment of the risk of importing bovines and bovine products from Canada under the provisions of our proposed rule (APHIS 2006b). In our risk assessment, we laid out the likely risk pathway (i.e., a series of occurrences or steps necessary for disease to enter and become established).

In conducting our risk assessment, one of the factors we took into account was the prevalence of BSE in Canada, since prevalence is one factor that affects the likelihood of a BSE-infected bovine being imported into the United States. We received a number of comments from the public that addressed our estimate of the prevalence of BSE in Canada. Although some of the comments supported our estimate of BSE prevalence in Canada, in general the commenters maintained that such prevalence is either higher than we estimated, may be increasing, or is uncertain, or that our methods of estimating it were flawed. The methodology we used to arrive at such estimates is discussed in detail in our risk assessment. However, to provide some context for the issues raised by commenters and discussed below, we summarize here the models that we used in conducting our assessment.

The number of BSE cases detected through surveillance understates the disease prevalence because exposed animals may be incubating disease and carrying infectious material in their tissues without presenting clinical symptoms. Like many transmissible spongiform encephalopathies (TSEs),

BSE has an incubation period of several years. Therefore, the disease is not detectable in its early stages with current technology. Moreover, surveillance will miss a proportion of detectable cases. Therefore, we applied statistical methods to the available epidemiologic and surveillance data to estimate, with attendant uncertainty, the prevalence of BSE in Canada.

We used two related, but distinct, methods to estimate BSE prevalence in Canada: the BSurvE model and the Bayesian Birth Cohort (BBC) model. Given its international prominence, we used the European Union (EU) BSurvE model (Wilesmith
et al.
, 2004, 2005), recently developed for the purpose of estimating BSE prevalence in national herds. The BSurvE model is noteworthy for its sound epidemiologic structure, including stratifying cattle by age and cause of death (i.e., healthy slaughter, fallen stock, casualty slaughter, or clinical suspect) and accounting for the relative likelihood of detecting BSE in various strata (EFSA 2004). The BSurvE model structure calculates BSE surveillance point values (random sample size equivalents) represented by targeted Canadian sampling of certain groups of cattle in which BSE cases are more likely to be detected. This approach allows for the inclusion of infected, but undetected, cases (such as young animals in the early stages of incubation) in the estimate, which would be ignored by conventional methods.

The other prevalence estimation model that we used is the BBC model. This model uses the BSurvE model structure and incorporates additional information. Unlike BSurvE, the BBC model adopts a Bayesian statistical framework to incorporate prior information about the decreased incidence of BSE observed in animals born after a feed ban equivalent to the initial ruminant-to-ruminant feed ban introduced in the United Kingdom in 1988.

Issue:
One commenter stated that BSE has become “firmly established” in Canada.

Response:
We disagree with the comment, which we consider to erroneously equate disease presence, which may be transient, with disease establishment. In epidemiology, an infectious disease has become established in a population when the disease is perpetuated in the population without the need for reintroduction from an external source. For example, OIE's sister agency, the international Commission on Phytosanitary Measures (CPM) defines plant pest establishment as “the perpetuation, for the foreseeable future, of a nonindigenous biological agent within an area after entry” (CPM 2001). With the implementation and continuation of a feed ban in Canada, all evidence points toward eventual eradication, rather than perpetuation of BSE in that country.

Issue:
One commenter stated that, since the time APHIS published its January 2005 final rule classifying Canada as a BSE minimal-risk region, the Agency has presented no new evidence that would support allowing the importation from Canada of the additional commodities discussed in the proposed rule. In fact, stated the commenter, evidence points to Canada having a higher prevalence of BSE than APHIS had previously determined.

Response:
As discussed in our January 2007 proposed rule, we revisited our earlier conclusions and policies by conducting a rigorous risk assessment based on current available scientific knowledge of the disease. We used peer reviewed risk assessment models in our analysis to estimate the prevalence of the disease in Canada and to analyze the likelihood of BSE establishment in the United States and the potential impacts of cases that may occur even without establishment as a result of the importation into the United States of the bovine commodities considered in this rule. The risk assessment itself was peer reviewed by experts in the field. As noted above, the reviewers agreed with the conclusion that the risk of establishment of BSE in the U.S. cattle population is negligible and noted that several assumptions in the risk assessment actually over-estimate the risk, so the overall finding that the BSE risk is negligible is reasonable. Based on the results of the risk assessment, we concluded that we could safely import Canadian cattle born on or after March 1, 1999, blood and blood products, and small intestines, excluding the distal ileum.

Issue:
Several commenters raised questions about the ability to statistically determine BSE prevalence “trends” in Canada, but reached different conclusions. Some commenters stated that the trajectory of BSE prevalence in Canada cannot be determined by available surveillance data and that, therefore, BSE prevalence in Canada may be increasing. On the other hand, another commenter requested that APHIS make clear that, despite the Agency's use of the BSurvE Prevalence B estimate, prevalence should not be assumed constant over time. The commenter requested that APHIS emphasize that lack of statistical evidence that prevalence varies from cohort to cohort is likely the result of inadequate statistical power,
2

and that, nevertheless, BSE prevalence in Canada is most likely decreasing.

2
The power of a statistical test is the probability of rejecting the null hypothesis when it is false. The power depends on the test level of significance, the magnitude of effect under the alternative hypothesis, sample size, and variability in the population. Rice (1988, pp.361-364) describes the calculation of statistical power for comparing two independent samples.

Response:
In our risk assessment for this rule, we acknowledge that, given the rarity of BSE cases in Canada, the surveillance data are unlikely to provide adequate statistical power to detect any trend. However, as discussed in the risk assessment, we consider it likely that the prevalence of BSE in Canada will decrease over time. With so few total BSE cases observed in Canada, the statistical power to detect differences in prevalence between cohorts is low. The peer reviewers of our risk assessment concur with our conclusion. (RTI 2007, pp. 6-26, 6-27).

Issue:
One commenter estimated the Canadian BSE prevalence to be 6.4 cases per million cattle. Further, the commenter stated that this prevalence estimate is smaller than the risk estimate provided by one of APHIS’ own risk assessments for a more pessimistic value of the misfeeding rate. The commenter suggested that this discrepancy reflects optimistic modeling assumptions in APHIS’ risk assessment.

Response:
We disagree with the commenter's analysis. Although the commenter's alternative prevalence estimate, based on a simple extrapolation method, falls within the 90 percent confidence interval
3

of APHIS’ BSurvE Prevalence B estimate (2.4 to 6.8 cases per million adult cattle) with an expected value of 3.9 per million case per million adult cattle (APHIS 2006c, table 5), it is based on different assumptions. Based on an analysis of BSE testing in the EU in 2001 and 2002, the commenter's prevalence estimate assumes that targeted “risk cattle” are only 10 times more likely to test positive for BSE than non-targeted routinely slaughtered cattle. Considering the BSE testing conducted in the EU during 2001-2004 (EC 2005a, table 3, p. 23), cattle in the

European BSE risk animals category (emergency slaughter, clinical suspects, and fallen stock) are 22 times more likely to test BSE positive than cattle in the healthy slaughter category. Using the commenter's simple extrapolation method and these more up-to-date data on BSE test positive ratio, the resulting BSE prevalence estimate would be 2.9 per million cattle. Although actually lower than the expected value for the BSurvE estimate, this value also falls within the 90 percent confidence interval of the Agency's BSurvE Prevalence B estimate, described above. APHIS calculated both the BSurvE Prevalence B estimate and the Bayesian Birth Cohort (BBC) prevalence estimate, but judged the latter to better characterize the BSE prevalence in Canada over the next 20 years, due to the expected downward pressure exerted on the disease by a feed ban.

3
A confidence interval is a statistical range with a specified probability that a given parameter lies within the range. For example, the 90 percent confidence interval of a distribution indicates the range of values that we are 90 percent certain include the parameter value of interest. It extends from the 5th percentile, or 5 percent confidence level, at the low end of the distribution of the 95th percentile, or 95 percent confidence level at the high end of the distribution. Similarly, a 95 percent confidence interval would extend from the2.5 percent confidence level to the 97.5 percent confidence level.

With regard to the commenter's suggestion of a discrepancy, the commenter provides no specific reference to “the risk estimate provided by one of APHIS’ own risk assessments,” but appears to refer to the main body of the 2005 report of Cohen and Gray (available at
http://www.fsis.usda.gov/PDF/BSE_Risk_Assess_Report_2005.pdf
), which was prepared for the USDA's Food Safety and Inspection Service (FSIS). Cohen and Gray (2005) do not estimate Canadian BSE prevalence, but rather the effect of introducing 500 BSE-infected cattle into the United States, and the pessimistic misfeeding assumption estimates that introduction would result in an expected 2,600 new cases over 20 years. There is no discrepancy because this aspect of the Cohen and Gray 2005 report is not relevant to our estimate of Canadian BSE prevalence.

Issue:
Based on APHIS” statements that animals are infected within their first year, and that feed produced prior to the feed ban would not be available for longer than a year, one commenter stated that additional undetected infected animals must have existed and been rendered in order to provide infectivity to detected cases. Therefore, stated the commenter, adding in these “undetected” animals raises the number of Canada's known and measurable BSE cases rises from 10 to 14, and APHIS’ estimate of BSE prevalence in Canada based on 10 animals is low.

Response:
We disagree with the commenter's analysis and conclusion, which assumes that we did not take into account the possibility of undetected cases of BSE in arriving at our prevalence estimate. APHIS’ estimate of the prevalence of BSE in Canada was adjusted to account for cases that would not be tested and for false negative test results. Also, although the bulk of feed will be consumed within a year after it is produced, residual infectivity may remain in the feed supply chain for an extended period. For example, examination of BSE cases in animals born in the United Kingdom after the 1996 “reinforced feed ban” suggests that these animals may have been infected from the persistence of the BSE agent in residual feed in storage bins (SEAC 2005).

Issue:
One commenter suggested that it is likely that Canada has numerous cattle over 30 months of age that are presently incubating the BSE disease, rather than just a few (4.1) as suggested by APHIS.

Response:
The estimate of 4.1 BSE-infected animals in the standing Canadian adult cattle population was based on the expected BSE prevalence in Canada under the BBC model. Using the estimated prevalence under BSurvE Prevalence B resulted in an estimate of 23.2 BSE-infected animals in the standing Canadian adult cattle population. Although, quantitatively, our risk assessment did not assume a decline in BSE prevalence over the next 20 years, we qualitatively consider such a decline to be likely because of continued compliance with the feed ban. Therefore, in assessing the BSE risk associated with imports from Canada over the next 20 years, we consider the result of the BBC model to be the more applicable prevalence estimate for use in our quantitative exposure model.

Issue:
One commenter indicated that although it is unclear whether the APHIS estimates of Canadian BSE prevalence included the BSE case confirmed on August 23, 2006, the APHIS estimates certainly do not take into account the case confirmed on February 7, 2007.

Response:
We estimated Canadian BSE prevalence based on a 7-year surveillance period through August 15, 2006. This surveillance period included the detection of nine BSE cases of Canadian origin reported through August 2006. Through surveillance conducted from August 16, 2006, through April 2007, Canada detected one BSE case born in 2000 and another born in 2001 (CFIA 2007). The BSE prevalence estimation methods used by APHIS (2006a) require detailed data to stratify tested cattle by age and cause of death (healthy slaughter, fallen stock, casualty slaughter, or clinical suspect) that are unavailable for the more recent surveillance period. However, we can assess the sensitivity of our previous Canadian BSE prevalence estimates by adding the two additional cases without changing the BSE surveillance points accumulated by Canada during the 7-year surveillance period through August 15, 2006 (APHIS 2006a, table 4).
4

This approach results in a revised table of BSurvE points and BSE cases by birth year cohort that reflects a total of 11 BSE cases of Canadian origin reported through April 2007 (APHIS 2007, table i).

4
In the BsurveE model, specific “point values” are assigned to each test sample, based on the surveillance stream or subpopulation of animals from which it was collected, as well as the likelihood of detecting infected cattle in that subpopulation. A sample from the specific surveillance subpopulation where BSE is most likely to be detected—
i.e.
, a middle adult clinical suspect—provides the most surveillance points. Conversely, a sample from the subpopulation where BSE is least likely to be detected—generally routine slaughter—provides the least points.

Using the same methods described in USDA's estimate of BSE prevalence in Canada (APHIS 2006c), we obtain updated Canadian BSE prevalence estimates:

• BSurvE Prevalence B: 90 percent confidence interval = 3.0-8.0 cases per million adult cattle

• Bayesian Birth Cohort (BBC, Winbugs): 90 percent confidence interval = 0.47-1.2 cases per million adult cattle

Because the updated confidence intervals contain the previous expected value estimates of 0.68 per million (BBC) and 3.9 per million (BSurvE Prevalence B) (APHIS 2006c), we conclude that the prevalence estimate is not sensitive to the addition of the two additional BSE cases discovered in Canada in August 2006 and February 2007.

Issue:
One commenter stated that APHIS’ expectation that the prevalence of BSE in Canada will continue to decline from its present minimal level does not acknowledge that the prevalence of BSE in Canada right now is very uncertain. The commenter's independent estimate of the current Canadian BSE prevalence is “on the order of 4-6 per million.”

Response:
APHIS’ risk assessment addresses the uncertainty in the prevalence of BSE in Canada by considering estimates that differ by more than a factor of five (APHIS 2006b). The BBC prevalence estimate has an expected value of 0.68 cases per million adult cattle.
5

The BSurvE Prevalence B estimate has an expected value of 3.9 per million. The

commenter's own method of estimation—“on the order of 4-6 per million——provides an estimate on the same order of magnitude as the BSurvE Prevalence B estimate of current prevalence. In either case, prevalence is extremely low.

5
The BBC model provides a more precise estimate of BSE prevalence in Canada by combining the epidemiologic theory and application of surveillance data underlying the BSurvE model with additional information about the effect of the feed ban on prevalence.

Issue:
One commenter stated that, although APHIS estimates that BSE prevalence in Canada is about 6.8 or more times greater than in the United States (0.68 vs. 0.1 per million), this does not adjust for the important fact that the first BSE case in the United States was imported from Canada.

Response:
The APHIS October 2006 estimate of BSE prevalence in Canada is based on the nine BSE cases of Canadian origin that had been confirmed in North America as of August 23, 2006. This total includes a case of BSE that was confirmed in Washington State on December 25, 2003 (APHIS 2006c, p. 1). The estimate of BSE prevalence in the United States excludes this case.

Issue:
One commenter stated that the calculation of BSE prevalence in Canada used in APHIS’ risk assessment excluded the European-born case detected in 1993.

Response:
The 1993 Canadian BSE case of European origin was likely part of the original exogenous source of BSE infectivity introduced into Canada that caused the subsequent generation of indigenous cases. Imported cases of BSE reflect an exposure to the disease that occurred elsewhere, and, therefore, are not generally included in estimates of prevalence that reflect native exposure. Similarly, when APHIS estimated the prevalence of BSE in the United States, the BSE-infected cow of Canadian origin that was detected in Washington State in December 2003 was excluded from the analysis, because it was an imported animal. In addition, as noted in APHIS’ estimation of BSE prevalence in Canada (APHIS 2006c, p. 5), in accordance with OIE guidelines (which indicate that surveillance points totals taken into account in assessing a country's BSE risk be accumulated over a maximum of 7 consecutive years), the estimated prevalence of BSE in Canada is based on surveillance data accumulated over a 7-year period beginning August 16, 1999. The 1993 case predates the OIE 7-year period.

Issue:
One commenter indicated that APHIS should not take action on the proposal until real surveillance data (not model-based predictions) show that the BSE problem has abated. The commenter stated further that denying Canada's BSE problem, or assuming it away with unvalidated and incorrect risk modeling assumptions, does not responsibly manage BSE risks to the United States.

Response:
We disagree with the commenter. In low BSE prevalence populations such as Canada, surveillance at levels that meet or even greatly exceed OIE guidelines provide insufficient statistical power to reliably detect changes in BSE prevalence over time. In other words, starting with a very low number of infected animals makes it very difficult to statistically demonstrate decreases in that number, even when testing a relatively large number of animals.

The OIE Guidelines for BSE Surveillance (Type A) call for countries to accumulate 300,000 BSE surveillance points over 7 consecutive years in order to detect with 95 percent confidence a prevalence level of at least one case of BSE per 100,000 animals (OIE 2006, Appendix 3.8.4).

To illustrate the comparative difficulty in demonstrating trends in low versus high prevalence populations, consider two hypothetical countries that have accumulated 1 million BSE surveillance points for each of two cohorts: Animals born before and animals born after the introduction of a ruminant-to-ruminant feed ban. Under this scenario, sampling levels in both countries far exceed the OIE guidelines. Assume, however, that the two countries differ with respect to their initial prevalence—i.e., the initial prevalence in “Country A” is 1 infected animal per 10,000 animals, while that in “Country B” is 1 infected animal per 100,000 animals.

For a given surveillance level, the statistical power of a hypothesis test can be evaluated as a function of the supposed change in BSE prevalence between cohort 1 (pre-feed ban) and cohort 2 (post-feed ban). The conventional minimum statistical power criterion is 80 percent. In other words, the probability that a statistical analysis will detect a true difference across groups should be at least 80 percent. The conventional significance level is 5 percent, meaning that we would conclude that a result was nonrandom if it were 5 percent or less likely to occur by chance alone. In our hypothetical scenario, the power of the surveillance in the country with higher prevalence, Country A, to detect a 50 percent decline in BSE prevalence is 98 percent. In comparison, the power of the surveillance in the lower prevalence Country B to detect a 50 percent decline in BSE prevalence is only 25 percent. In other words, if the Country B feed ban actually led to a 50 percent decline in BSE prevalence and the equivalent of 2 million random samples were collected (6.7 times the level under the OIE guidelines), there would still be a 75 percent chance of concluding that the prevalence was unchanged from its initial level of 1 infected animal per 100,000 animals.

An important implication of the low statistical power of sampling in low prevalence populations is that BSE surveillance data are unlikely to provide a purely statistical basis for making a determination about the date when a specific intervention (e.g., a ruminant-to-ruminant feed ban) becomes effective, even when large amounts of surveillance data are available. For example, according to the OIE (2007a), the annual incidence of reported BSE cases in the Netherlands dropped from 13.2 to 0.8 per million adult cattle from 2001-2005.
6

Despite the EU BSE surveillance requirements for testing all risk animals over 24 months of age and all healthy slaughter cattle over 30 months of age, Figure 1 shows that application of the BSurvE (Prevalence A) model to Netherlands BSE surveillance data does not yield sufficient statistical power to draw clear distinctions among birth year cohorts as prevalence declines (Figure 1).

6
The OIE Terrestrial Animal Code (Chapter 1.1.1., Article 1.1.1.1) defines incidence as “the number of new cases or outbreaks of a disease that occur in a population at risk in a particular geographical area within a defined time interval (OIE 2006b).”

ER18SE07.023

Note that, in figure 1, there is a decrease in estimated prevalence between 1998 birth-year cohorts and 1999 birth-year cohorts, while, at the same time, there is an increase in the upper confidence limit. This apparent paradox is indicative of another shortcoming of relying on surveillance data alone to determine whether BSE prevalence has been reduced. Because fewer animals from the most recent birth year cohorts are tested when sent to slaughter, uncertainty about the prevalence in the most recent cohorts is much greater than in older cohorts. Furthermore, the lower likelihood of detecting BSE in young infected animals means that the young animals that are tested contribute relatively little to reducing uncertainty in the true (as opposed to apparent) BSE prevalence. These two sources of uncertainty in young birth cohorts (low numbers of animals tested, and little value in the surveillance data that are gathered from them) cause an asymmetrical increase in the upper limit of the confidence interval compared to the lower confidence limit. This effect on the upper confidence limit on BSE prevalence is most pronounced for the most recent birth year cohorts which are less likely to be tested and will not have lived long enough to manifest BSE, even if they have been infected. Wilesmith
et al.
(2004, figure 3) further illustrates this same concept.

Consequently, if the effectiveness of a country's safeguards against BSE amplification were determined strictly by setting a tolerance for the upper confidence limit on BSE prevalence associated with the “real surveillance data,” one might reach the incorrect conclusion that prevalence is increasing, when in actuality, the result is simply due to testing fewer and younger animals in the most recent birth year cohorts. Finally, relying solely on surveillance data fails to account for under reporting of disease due to the lack of diagnostic sensitivity to detect BSE at an early stage of disease. By accounting for the possibility of false negative test results, epidemiologic models such as BSurvE are recognized as providing a more accurate estimate of true BSE prevalence than the apparent prevalence measured by surveillance data alone.

Issue:
One commenter stated that the output from the BSurvE model used by Canada in 2005 grossly underestimated Canada's 2006 and 2007 BSE prevalence and, therefore, the BSurvE model is unreliable for estimating Canada's BSE prevalence. The commenter stated further that, at the minimum, APHIS should determine the erroneous inputs that resulted in the failed prediction in 2005 and correct them.

Response:
In the risk assessment conducted for this rulemaking, APHIS used its own prevalence estimate, not that of the Canadian Food Inspection Agency's (CFIA's) 2006 prevalence estimate, which was not based on BSurvE, but on a modified version that appears similar to the APHIS BBC model. The commenter cites CFIA's Assessment of the North American BSE Cases Diagnosed from 2003-2005 (Part II), which states that “when the BSurvE model was recently applied to Canada's statistics and adjusted to account for the effectiveness of the 1997 feed ban (based on experiences with the 1988 feed ban in the United Kingdom), the resulting prediction was that it could be expected that three infected animals remain within the national herd” (CFIA 2006, p. 13).

APHIS’ estimation of BSE prevalence in Canada (APHIS 2006c) is that the expected prevalence values under the BBC and BSurvE Prevalence B models correspond to an expected number of BSE-infected animals in the standing Canadian adult cattle population of 4.1 and 23.2, respectively. APHIS further explains that it is important to note that this range of prevalence estimates represents uncertainty and not variability. BSE-infected animals are recruited into and exit from the adult cattle population over time, but at a given point in time, the number of infected animals in the population is a fixed but uncertain value.

Assuming the overall probability of infection remains constant over time, the actual number of infected cattle in the population at any given point in time would still vary randomly about the mean. This variability is incorporated in the model supporting the exposure assessment for live bovines by means of the Poisson variability distribution. Assuming a fixed mean prevalence of 4.1 and 23.2 BSE infected animals in the standing adult cattle population in Canada, the 95th percentile of the Poisson distribution are 7 and 31 BSE-infected animals in any given year, respectively. We note that these numbers are greater than the

five BSE cases detected in Canada in 2006, which means that the greatest number of Canadian BSE cases identified in a single surveillance year is lower than even the 95th percentile of distribution.

Issue:
One commenter stated that, if the United States were finding BSE cases at the same rate as in Canada, this would translate into roughly 40 BSE cases detected in the United States since January 2006, which would be regarded as a large number. The commenter stated further that, at this time, the BSE situation in Canada does not appear to be improving.

Response:
We do not agree with the commenter. The commenter's conclusion appears to be based on a cursory estimate and does not provide an accurate comparison of BSE cases detected in Canada with a comparable number that would have been detected in the United States, given the larger U.S. cattle population. The commenter's comparison fails to take into account other years of surveillance, as well as the age and surveillance stream of tested animals. These data are extremely important for estimating BSE prevalence. A comparison based solely on the number of detected cases ignores infected animals with unapparent or undetected infections.

Table 1 provides a direct comparison of the estimated BSE prevalence in the current standing adult cattle population of the United States and Canada, respectively, using identical estimation methods (APHIS 2006a; 2006c).

Table 1.—Comparison of Estimated BSE Prevalence in the Current Standing Adult Cattle Population of U.S. and Canada

Country
BSE Prevalence Estimation Method

BSurvE
prevalence B

BBC

Expected value

US

0.18 × 10
−6

0.10 × 10
−6

Canada

3.9 × 10
−6

0.68 × 10
−6

Despite the higher estimated BSE prevalence in the current standing adult cattle population in Canada compared to the prevalence of BSE in the standing adult cattle population in the United States, APHIS finds that, because of the extremely low BSE prevalence in Canada and the high levels of BSE controls in both Canada and the United States, the risk to the United States (i.e., the likelihood of establishment of BSE in the United States and the potential impacts of cases that may occur even without establishment) as a result of importing from Canada the bovine commodities considered in this rule is negligible (APHIS 2006b). Furthermore, as stated in our risk assessment, we expect that the prevalence of BSE in Canada will decrease continuously over the next several years. Peer reviewers of our risk assessment agreed (RTI 2007).

Issue:
One commenter stated that Canada's ratio of positive cases per 10,000 cattle tested exceeds the ratio of 22 of the 25 EU-member countries; that only the ratios for the United Kingdom, Portugal, and Spain exceed Canada's 2006 ratio. The commenter noted further that even the countries of Ireland, Germany, and France, each of which are considered to have had widespread BSE exposure, have a lower ratio for positive cases detected per 10,000 head tested than does Canada. Another commenter stated that Canada's BSE prevalence is higher than that for Denmark, Belgium, and Austria, and is comparable to the rate in Germany. This commenter, who estimated the Canadian BSE prevalence to be 6.4 cases per million cattle, stated further that no one considers countries with a reported BSE rate of 1 to 2 cases per million animals (e.g., Denmark, Belgium and Austria) to have a minimal BSE risk, and that Canada is not a BSE minimal-risk region in any ordinary sense.

Response:
The commenters' statements ignore important differences in BSE surveillance and cattle populations among countries, and a comparison based simply on the proportion of positive cases per number of cattle tested is inconsistent with the prevalence estimate approach taken by one of the commenters, as well as the prevalence estimate used by APHIS. Although calculating the proportion of infected animals detected per number of tested animals can serve as a useful tool, depending on the purpose for the calculation, it is not an estimate of prevalence. Rather, prevalence is defined as the number of infected animals in the total population at a given point in time. On the other hand, the calculation conducted by the commenter who referred to the ratio of positive cases per 10,000 cattle tested is similar to that conducted by the U.S. Department of Health and Human Services, Centers for Disease Control and Prevention (CDC). In May 2007, using data similar to that analyzed by APHIS for this rulemaking, CDC calculated the proportion of Canadian-born BSE cases identified by Canadian authorities in relation to the total number of animals tested in that country. CDC then made a like calculation regarding BSE cases in U.S.-born cattle and compared the Canadian and U.S. results (CDC 2007). Unlike the estimate used by APHIS in the risk assessment for this rule, the CDC calculation is not an estimate of the prevalence of BSE in Canada, nor of the prevalence in the United States. Although the type of calculations conducted by CDC can be useful in comparing relative proportions of BSE detections per number of cattle tested, they do not, as noted above, constitute an estimate of prevalence.

The number of disease detections per total number of animals tested can be influenced by the criteria used for choosing animals for testing. For instance, Canada, like the United States, conducts targeted BSE surveillance, sampling those animals where disease is most likely to be detected if present. In contrast, EU countries routinely test large numbers of healthy animals at slaughter. Approximately 80 percent of cattle tested for BSE in the EU during 2001-2004 were healthy slaughtered animals, but “risk animals” were 22 times more likely to test positive (EC 2005a). One study (Giovannini
et al.
, 2005) estimates the true prevalence of BSE infection in several EU countries. Based on BSE testing in 2001, although Denmark, Finland, and the Netherlands had a lower proportion of positives per test than Canada, the estimated prevalences from this study for those three countries were higher than the expected values of our Canadian BSE prevalence estimates using the BBC estimation method (0.68 cases per million adult cattle) or BSurveE Prevalence B (3.9 cases per million adult cattle). Giovannini
et al.
(2005) estimated the following 90 percent confidence intervals for the prevalence of BSE infection: Denmark, 9 to 38 cases per million animals; Finland, 29 to 110 cases per million animals; and Netherlands, 8 to 34 cases per million animals. The methods used by APHIS to estimate Canada's BSE prevalence, including the BSurvE model developed by the EU Transmissible Spongiform Encephalopathies Community Reference Laboratory, account for the cattle population demographics, the age and surveillance category of animals tested, and the insensitivity of BSE diagnostics with regard to detection of the disease at an early stage of development.

The comments are based on an inappropriate comparison of a statistical estimate of the true BSE prevalence in Canada to the crude rate. Table 2 below compares the crude reported BSE rates in all five countries in 2005. Comparing the reported BSE rate of Canada to those of the countries listed by the commenters shows that Canada's

reported rate is at least an order of magnitude below that of the others.

Table 2.—Reported BSE Rates in 5 Countries

Country

Reported BSE cases per
million adult
cattle—2005

Canada
0.145

Denmark
1.289

Belgium
1.448

Austria
2.114

Germany
4.965

Source: OIE (2007a).

The problem with comparing the crude reported rate of BSE detection to the estimated true BSE prevalence is illustrated by the situation in Belgium. The reported rate of BSE in Belgium peaked in the 2001 surveillance year at 28.22 cases detected per million adult cattle (OIE 2007a). In comparison, Saegerman
et al.
(2004) applied the BSurvE model to the Belgian BSE surveillance data and estimated that the actual BSE prevalence in Belgium peaked at approximately 400 cases per million adult cattle in the 1995 birth year cohort. (The lag between the 1995 birth year and the 2001 surveillance year is consistent with the long BSE incubation period.)

With regard to the comment that countries with 1 to 2 cases per million animals are not considered to present minimal risk, APHIS notes that, prior to the 2005 revisions in the OIE guidelines on BSE, countries with a reported BSE rate of 1 to 2 cases per million animals could satisfy the prevalence criterion for the pre-2005 OIE BSE minimal-risk classification. Under the 2004 OIE Terrestrial Animal Health Code (Article 2.3.13.5), the criteria for a BSE minimal-risk country included a reported rate of less than two cases per million during each of the last four consecutive 12-month periods within the cattle population over 24 months of age. The OIE Code was modified in 2005 to include a revised country categorization system which more accurately reflected current scientific understanding of BSE. These modifications streamlined the number of country categories to three (negligible, controlled, or undetermined BSE risk) and also eliminated the numeric prevalence criteria for classifying the BSE risk status. The previous OIE minimal-risk category is now incorporated into the controlled risk category. We note that in 2007, the OIE recognized Switzerland as a BSE controlled risk region. Switzerland had a reported rate of 5.4 BSE cases per million adult animals in 2006 (OIE 2007a), greater than the 1 to 2 cases per million animals cited by the commenters.

APHIS disagrees with the commenter's statement that Canada does not qualify as a BSE minimal-risk region. APHIS regulations at § 94.0 define the standards for a region to be designated as a minimal-risk region. These include the standard that the region maintain “risk mitigation measures adequate to prevent widespread exposure and/or establishment of the disease.” Canada continues to meet this standard. The commenters provided no specific evidence to document how or why Canada does not meet the APHIS standards.

Issue:
One commenter stated that the prior information [information using data from the United Kingdom feed ban] incorporated into the Bayesian models used to estimate prevalence of BSE-infected cattle in Canada may have resulted in estimates that are biased downward (to a limited degree) from the true burden. However, stated the commenter, the Bayesian models used to estimate prevalence in Canada (as of August 2006) are basically sound and a better approach than relying on the BSurvE Prevalence B estimate. Further, said the commenter, given the proviso that the models could overestimate the effectiveness of the feed ban, it is most likely that the actual prevalence of infected animals is between 0.68 and 3.9 animals per million adult cattle. The commenter stated that because it is likely that the Canadian feed ban was at least as effective as the initial United Kingdom feed ban, and based on available data, the true BSE prevalence in Canada is probably substantially closer to 0.68 cases per million animals than to 3.9 cases per million animals.

Conversely, several commenters suggested that APHIS rejected the higher prevalence estimate of the BSurvE model for the lower prevalence estimate of the BBC model, and that the BBC model prevalence estimate is not realistic in light of recent data.

Response:
Although APHIS considered the results of both the BSurvE and the BBC prevalence estimation models, we consider the result of the BBC model as the more likely prevalence estimate to apply to the assessment of BSE risks associated with imports from Canada over the next 20 years in our quantitative exposure model, for the following reasons. APHIS estimated Canadian BSE prevalence based on surveillance conducted through August 15, 2006. (
Note:
This time period includes all cases of Canadian origin reported through August 2006 (APHIS 2006c).) From August 16, 2006, through April 2007, Canada accumulated approximately 44,980 additional BSE samples and detected two BSE cases (one confirmed on February 7, 2007, and another confirmed on May 2, 2007). Based on the negative binomial likelihood ratio, which considers the number of negative tests prior to one or more positives, the BSurvE Prevalence B estimate (with expected value of 3.9 cases per million animals) is indeed far more likely to be true than is the BBC prevalence estimate (with an expected value of 0.68 cases per million animals) for the current standing Canadian cattle population. However, the primary purpose of characterizing BSE prevalence in Canada's current standing herd (APHIS 2006c) was not to discuss or assume its implications for the present, but rather, to estimate prevalence for use as an input for the Harvard exposure model used in the Exposure Assessment of the analysis. Because BSE has a long amplification cycle (it takes an average of 7 years from the time that one animal is exposed, to the time that another might be exposed from infectivity produced by the first animal), the Harvard model is typically run with 20-year simulations to include roughly 3 amplification cycles. The prevalence estimates contained in APHIS' estimation of BSE prevalence in Canada (APHIS 2006c) are applied, unchanged, to the cattle imports projected over the next 20 years (2007-2026). Since we expect that the true prevalence will drop from its current level (whatever that may be), we anticipate that the lower, BBC estimate is a more realistic prediction (or even an overestimate) of average prevalence levels over this time frame. Consequently, APHIS considers the result of the BBC model, which incorporates the effect of a feed ban, to be better for application to the quantitative assessment of BSE risks associated with imports from Canada over this time period. In order to determine the impact of this assumption on the results, we applied the BSurvE estimate to the exposure model. We note that the likelihood of BSE establishment remained negligible (R
0
of 0.079, which is far less than 1), as did the potential impact of cases even without establishment (less than 4 clinical cases) over the 20 years of the analysis.

Issue:
One commenter suggested that the APHIS risk model is not trustworthy because it has not been shown to have predictive validity and does not explain or predict a sustained flow of BSE cases from one geographic area (the Alberta region in Canada).

Response:
It is not clear to us from the comment which model the commenter is referring to. Consequently, in this response, we discuss the Harvard model and the prevalence models used by APHIS. In either case, we disagree with the commenter's conclusion that the APHIS risk model is not trustworthy. The plausibility of the Harvard model was established by comparing its predictions for Switzerland against the observed progression of BSE within that country's cattle herd (Cohen
et al.
, 2003). It is not clear from the comment how the predictive validity of an infectious disease model is to be demonstrated over a 20-year time horizon, or how the model has failed to explain or predict the observed data. Regarding a sustained flow of BSE cases from one geographic area, assuming a constant proportion of BSE infected cattle in the herd, more BSE cases are found where large cattle populations exist.

As we discuss above in response to another issue raised by commenters, APHIS' estimation of BSE prevalence in Canada (APHIS 2006c) concludes that the expected prevalence values under the BBC and BSurvE Prevalence B models correspond to an expected number of BSE-infected animals in the standing Canadian adult cattle population of 4.1 and 23.2, respectively. Further, the prevalence estimates represent uncertainty and not variability. At any given point in time, the number of infected animals in the population is a fixed (although uncertain) value, although over time the actual number of infected cattle in the population would vary randomly about the mean of the probability distribution, as BSE-infected animals are recruited into and exit from the adult cattle population (i.e., some are newly infected and some die). Even assuming that the probability of infection remains constant, over time the actual number of infected cattle in the population would vary. This variability is incorporated in the model supporting our exposure assessment for live bovines by means of the Poisson variability distribution. Assuming a fixed mean prevalence of 4.1 and 23.2 BSE-infected animals in the standing adult cattle population in Canada, the 95th percentile of the Poisson distribution is respectively 7 and 31 BSE-infected animals in any given year. As we noted above, these numbers are greater than the five BSE cases detected in Canada in 2006, which means that the greatest number of Canadian BSE cases identified in a single surveillance year is lower than even the 95th percentile of distribution. While this observation does not statistically validate (confirm) the APHIS estimates of Canadian BSE prevalence, neither does it invalidate them, as the commenter seems to suggest. Furthermore, the prevalence estimates are applied not only to the current standing population, but also to the next 20 years.

BSE Data From the United Kingdom

In our January 2007 proposed rule and its supporting risk assessment, we discussed data associated with a ruminant-to-ruminant feed ban in the United Kingdom and indicated that experience in the United Kingdom demonstrates that implementation of a ruminant-to-ruminant feed ban causes BSE prevalence to decrease. We noted that animal feed restrictions were implemented in the United Kingdom in 1988, when the use of ruminant MBM in ruminant animal feed was banned. In September 1990, the use of specified bovine offals was banned for use in any animal feed. This ban prohibited the use in any animal feed of bovine tissues with the highest potential concentration of infectivity. In 1994, the use of mammalian protein—not just ruminant protein—was banned from ruminant feed. In 1996, feeding of any farmed livestock, including fish and horses, with mammalian MBM was completely banned. As a result of reducing the recycling of infectivity, the annual incidence of BSE fell by 99.4 percent, from 36,680 in 1992 to 203 in 2005 (DEFRA 2006b). There is, therefore, every reason to expect downward pressure on the prevalence of BSE in any country that implements a feed ban.

Issue:
One commenter stated that, of 180,986 confirmed cases of BSE in Great Britain, the year of birth of the infected animal is unknown in 43,342 cases, and the large percentage of animals whose birth year is unknown casts doubt on the ability to determine the timeframe of an effective feed ban and, and further, makes it doubtful that all BSE-infected cattle in Canada are going to show clinical signs of the disease only if they were born before March 1, 1999. The commenter also stated that Japan has reported cattle as young as possibly 20 months of age or younger as testing positive for BSE.

Response:
It is not clear to us how the information presented by the commenter supports the conclusions the commenter reached. However, we consider it useful to provide some clarification regarding the information presented. With regard to the proportion of BSE cases in Great Britain for which the date of birth is unknown, our risk assessment included a sensitivity analysis that takes into account that general source of uncertainty. (Sensitivity analysis evaluates the degree to which changes in the assumptions used in a model affect the model's results.) We made no assumptions as to whether Great Britain's feed ban is or has been effective, but applied the same proportional drop in cases observed in the United Kingdom to the Bayesian analysis that was performed to estimate BSE prevalence in Canada's standing cattle herd.

The commenter's statement that it is doubtful that only animals born before March 1, 1999, would show clinical signs of BSE indicates a potential confusion between the likelihood of exposure as expressed in terms of the date of the effectively enforced feed ban (and, thus, the potential for exposure) and the likelihood of an exposed animal developing clinical signs (which is based on age and amount of exposure, and the amount of time that has elapsed since exposure). In neither our risk assessment nor our proposed rule do we conclude that only infected animals born before March 1, 1999, would show clinical signs of the disease. Based on Canada's system of regulations, compliance and enforcement, and the length of time we expect pre-feed ban feed to persist in the system, we conclude that animals born on or after March 1, 1999, have an extremely low likelihood of exposure to BSE. Any animal, however, exposed to an infectious dose of the BSE agent and allowed to live to the end of its incubation period, would likely exhibit clinical signs.

Regarding the age of cattle diagnosed with BSE in Japan, the comment did not contain sufficient information for us to determine and respond to the relevance of the statement to the remainder of the comment.

Issue:
One commenter questioned the effectiveness of APHIS' use of United Kingdom surveillance numerators to estimate Canada's BSE prevalence. Specifically, the commenter stated that “Nowhere * * * is incidence reported. Cases (without reference to a population at risk) are used. This may be important because the manner in which BSE cases were counted changed over time in the [United Kingdom].”

Response:
We acknowledge that changes over time in BSE surveillance and in the size and demographics of the cattle population do contribute to the uncertainty about the efficacy of the initial, ruminant-to-ruminant feed ban introduced in the United Kingdom in 1998. However, the United Kingdom's Department for Environment, Food, and

Rural Affairs (DEFRA) does not report BSE surveillance results by birth year and surveillance class (e.g., active or passive surveillance, animal health status). Ideally, such data could be entered into BSurvE or a similar model to estimate true BSE prevalence for all United Kingdom birth year cohorts since the onset of the epidemic. This process would permit not only an improved estimate of the effect of the initial feed ban but also of the incremental impact of additional measures that were subsequently introduced. DEFRA has reported back-calculation model estimates of true BSE prevalence in cohorts born after 1995 to assess the effects of the “reinforced feed ban” introduced by the United Kingdom in August 1996 (DEFRA 2005, 2006b). However, we are unaware of any published estimates of true BSE prevalence in the United Kingdom for the 1987-1995 birth year cohorts based on up-to-date surveillance results.

Issue:
One commenter stated that APHIS is wrong to assume that the United Kingdom data regarding the effectiveness of the feed ban can be applied directly to the situation in Canada.

Response:
We acknowledge that the applicability to Canada of the data from the initial United Kingdom ruminant-to-ruminant feed ban is uncertain. Nonetheless, the United Kingdom's experience and data are important and useful to our risk assessment and analyses. In addition, the Peer Review Report (RTI 2007, p. ES-2) noted that “[all reviewers] agreed that the evidence from the United Kingdom * * * and Europe that the feed ban is effective is reasonable to consider in the case of Canada.”

Issue:
Several commenters noted the differences in the feed bans in the United Kingdom and Canada in stating that it is not valid to draw conclusions about the likely prevalence of BSE in Canada by extrapolating from the rate of decline in BSE cases in the United Kingdom following implementation of a feed ban there. The commenters noted that (until expanded this July) the feed ban in Canada prohibited the feeding of ruminant material to ruminants. In contrast, said one commenter, significant declines in the number of confirmed BSE cases in the United Kingdom did not occur until the United Kingdom took stronger measures, ultimately banning the feeding of all mammalian protein to food animals in 2001. The commenter suggested that the United Kingdom's experience in particular clearly shows that ruminant-to-ruminant feed bans do not drastically curtail the number of confirmed BSE cases and that much stronger measures are needed to eradicate the disease.

Response:
The comments appear to confuse the absolute level of BSE in the United Kingdom with its rate of decline. The comments also ignore the BSE incubation period and the effects of other concurrent measures, trends, and events in the United Kingdom. The number of BSE cases in United Kingdom birth year cohorts (all cattle born in a given year) has continued to decline since peaking in 1987. With the exception of the 1996 birth year cohort, it is not readily apparent that there has been any significant change in the rate of decline in birth year cohort prevalence after the United Kingdom introduced the initial ruminant-to-ruminant feed ban in 1988 (figure 2). As of March 1, 2007, the United Kingdom had confirmed two BSE cases in animals born after 2001, but due to the long BSE incubation period, it is reasonable to expect that ongoing surveillance may detect additional cases in animals born after 1998.

ER18SE07.024

Shortly after the emergence of vCJD was publicly recognized in March 1996, the United Kingdom introduced several BSE-related measures, including the ban on the use of mammalian MBM in feed for all farm animals (the “reinforced feed ban”), a selective cull, and the over-30-month rule limiting the age of animals that could be slaughtered for food. As shown in figure 3, the size of the United Kingdom cattle population began a marked decline in 1996, punctuated by a drop associated with the foot and mouth disease (FMD) outbreak in 2001.

ER18SE07.025

In addition to the declining cattle population size, other confounding variables, such as changes in cattle population demographics and BSE surveillance practices, make it difficult to ascertain the independent or marginal effect of any single measure on the decline of BSE in United Kingdom birth year cohorts. At this time, it appears that the confluence of events and measures of 1996 may have hastened the waning of BSE in the United Kingdom, but the decline was underway in 1988.

Issue:
One commenter indicated that scientific studies in France and Britain have found that, after a ruminant-to-ruminant feed ban was put into place, the subsequent incidence of BSE was correlated to pig density, and that the new Canadian BSE feed rule, to be implemented in July 2007, is, according to the commenter, similar to, but weaker than, the September 1990 United Kingdom SBO [Specified Bovine Offals] ban. The commenter stated that, by not following the lead of the United Kingdom [and banning the feeding of all mammalian protein to food animals], the proposed CFIA SRM ban may reduce but will not eliminate the risk of BSE in Canada.

Response:
Two studies—Abrial
et al.
(2005) and Stevenson
et al.
(2005)—indicate a correlation between cases of BSE born after a ruminant-to-ruminant feed ban was implemented and areas of higher pig density in France and Britain. These studies indicate the potential for cross-contamination of livestock feeds after ruminant-derived protein was excluded from ruminant feed. Eventually, each country and the EU adopted regulations prohibiting the inclusion of any animal protein in livestock feed. At this time, however, it is not possible to ascertain the extent, if any, to which establishment of a more restrictive feed ban had any impact on the rate of BSE decline in EU Member States beyond the feed controls already in effect.

As discussed previously, the number of BSE cases in United Kingdom birth year cohorts began to decline in 1988, the year the initial ruminant-to-ruminant feed ban was introduced. Although France initially introduced a ban on mammalian MBM in cattle feed in July 1990—not a ruminant-to-ruminant feed ban—the European Commission Scientific Steering Committee concluded that the French feed ban adopted in 1990 “was likely not effectively enforced until 1994/1995.” (ECSSC 2000, p. 30). Based on testing in 2001-2002, Bonnardiere
et al.
(2004) found a significant increase in French BSE prevalence between the July 1993-June 1994 and July 1994-June 1995 cohorts, followed by a significant decrease in BSE prevalence in birth cohorts born in France after June 1995. More recently, active surveillance during 2001-2005 also indicates that the number of BSE cases per cohort peaked in France in the 1995 birth year cohort and declined thereafter (EC 2006, table B20).

In Europe more generally, based on active surveillance during 2001-2005, the number of BSE cases per birth year cohort in the original EU Member States (EU 15), excluding the United Kingdom, was on the decline after the 1995 birth year cohort. In June 1994, the EU banned the feeding of mammalian MBM to ruminants. However, among EU members, only Belgium, Germany, Greece, Italy, Luxembourg, and Spain had no feed ban in place prior to the 1994 EU-wide measure (Court of Auditors 2001). In June 2005, the European Commission issued the “Report on the Monitoring and Testing of Ruminants for the Presence of Transmissible Spongiform Encephalopathy in the EU in 2004” and observed that the impact of the 2001 “total feed ban” (EU Regulation 999/2001) cannot yet be assessed due to the long BSE incubation period. As noted in the discussion of the decline of BSE in the United Kingdom, it is reasonable to expect that ongoing surveillance may detect additional cases in animals born after 1998.

The conclusion of our risk assessment that, over the 20 years of the analysis, the risk to the United States (i.e., the likelihood of establishment and the potential impacts of cases that may occur even without establishment of BSE) as a result of importing from Canada the bovine commodities considered in this rule is negligible, is not predicated on the eradication of BSE in Canada.

ER18SE07.026

Issue:
One commenter indicated that year-of-birth data collected by the EU shows that, based on the number of BSE cases detected in the United Kingdom since 2001, there was a steady increase in the number of BSE-positive cattle born in the United Kingdom after its 1988 feed ban, beginning with cattle born in the year 1990.

Response:
We disagree with the commenter. Since July 2001, when the EU-wide active BSE surveillance program commenced, an increasing proportion of the total BSE cases in the United Kingdom have been detected as a result of targeted (active) surveillance (DEFRA 2006b, figure 4.3). However, as shown by the EC (EC 2006, chart B1), the vast majority of BSE cases in the United Kingdom were detected by surveillance prior to 2001. Based on all available United Kingdom BSE surveillance data (DEFRA 2007), the number of BSE cases in United Kingdom birth year cohorts began to decline in 1988, the year the initial ruminant-to-ruminant feed ban was introduced.

For the reasons discussed above, we continue to consider it appropriate to apply our estimates of BSE prevalence in Canada to our risk assessment. As noted above, we used two related, but distinct, methods to estimate BSE prevalence in Canada, and addressed the uncertainty in the prevalence of BSE in Canada by considering prevalence estimates that differ by more than a factor of five. Although we consider the BSurvE Prevalence B estimate to be far more likely to be true than is the BBC estimate for the current standing Canadian cattle population, we consider the result of the BBC model as the more likely prevalence estimate to apply to the assessment of BSE risks associated with imports from Canada over the next 20 years.

Feed Ban in Canada

As discussed above, in our January 2007 proposed rule, we proposed to allow the importation of live bovines from BSE minimal-risk regions if the animals were born on or after a date determined by APHIS to be the date on and after which a ruminant-to-ruminant feed ban in the region of export has been effectively enforced. We noted that experience around the world in countries with BSE has demonstrated that feed bans are effective control measures, and that the incidence of BSE worldwide continues to decline because of these measures (OIE 2007a).

We indicated that, because of the demonstrated efficacy of an effectively enforced feed ban in reducing the possibility of exposure of cattle to the BSE agent, the OIE provides guidelines for trade in live cattle from regions that have reported BSE if such regions have an effective feed ban in place, provided the cattle were born after the date when the feed ban was effectively enforced (OIE
Terrestrial Animal Health Code
, Chapter 2.3.13). We proposed to consider March 1, 1999, as the date on and after which a feed ban has been effectively enforced in Canada. A number of commenters addressed Canadian enforcement of its feed ban, and also addressed the date we proposed to consider as the date of effective enforcement of a feed ban in Canada. Although some commenters specifically supported March 1, 1999, as the date of effective enforcement of a ruminant-to-ruminant feed ban in Canada, a number of other commenters disagreed that Canada was effectively enforcing a feed ban as of that date. Some commenters suggested alternative dates or time frames.

Issue:
Several commenters stated that APHIS' determinations of the level of compliance with the Canadian feed ban and the time at which compliance was achieved are arbitrary and scientifically indeterminable.

Response:
We disagree with the commenters. In January 2005, USDA sent a team to Canada to assess Canada's feed ban and its feed inspection program to determine whether the control measures put in place by the Canadian Government were achieving compliance with that country's regulations. APHIS conducted an extensive review of the feed ban in Canada. As part of its review, APHIS

analyzed CFIA's description of past cases of BSE in Canada, as well as historical inspection and compliance data related to the feed ban for the previous 3 years, educational materials, published notices, and the report of the International Review Team that was submitted to the U.S. Secretary of Agriculture in February 2004. Additionally, the U.S. team accompanied the CFIA inspection staff on inspections of randomly selected commercial feed mills and rendering facilities. At the facilities, the U.S. team observed the application of the inspection standards, observed manufacturing techniques, and discussed processes with facility personnel involved in various steps of feed manufacturing. In its report, the team concluded that Canada has a robust inspection program, that overall compliance with the feed ban in Canada is good, and that the feed ban is reducing the risk of transmission of BSE in the Canadian cattle population (USDA 2005). The team's findings support our conclusions regarding the level of compliance with the feed ban in Canada.

Issue:
In our January 2007 proposed rule, in discussing our rationale for considering March 1, 1999, to be the date of effective enforcement of a feed ban in Canada, we stated that a 12-month period would generally be sufficient to allow purchased feed products that may contain MBM to be completely used. One commenter expressed uncertainty about that estimation and suggested that it might be advisable for APHIS to conduct a quantitative assessment of compliance with the feed ban to determine the date of its effective enforcement.

Response:
We recognize uncertainty regarding the precise date on which Canada achieved effective enforcement of its feed ban, but we note that, given the extremely low prevalence of BSE in Canada along with the safeguards in the United States, the impact on the overall risk of a slightly earlier or later date would be minimal. Although reducing uncertainty can, at times, be achieved by performing more rigorous quantitative analyses, before attempting to reduce the uncertainty regarding any given factor or parameter—such as the precise date on which Canada achieved effective enforcement of its feed ban—it is important to examine the significance of the parameter to the overall risk result.

Issue:
Several commenters stated that APHIS' calculation of the amount of time necessary for ruminant feed to cycle through the Canadian feeding system is irrelevant in the absence of effective enforcement of feed-ban regulations in Canada. The commenters stated that it was not until between 2000 and 2002 that Canada implemented inspections of feed and rendering facilities.

Response:
The commenters' statement is not accurate. Inspections of rendering facilities and feed mills in Canada began immediately with the implementation of the feed ban in that country in August 1997. Rendering facilities were required to obtain an annual permit to operate, and issuance of a permit required an inspection of the facility. In addition, CFIA immediately began a program for inspection of commercial feed mills. All commercial feed mills were inspected in the first year after the implementation of the feed ban, with none of the feed mills found to be including prohibited material in ruminant feed. Thereafter, feed mills were on a 3-year inspection interval until 2002, when annual inspection of commercial feed mills was initiated.

Issue:
A number of commenters stated that the diagnosis of BSE in cattle born after the establishment of a feed ban in Canada demonstrates that Canada's feed ban is either ineffective or not effectively enforced.

Response:
We disagree with the commenters' conclusion. The commenters suggest that, in order for the Canadian feed ban to be considered effective, BSE surveillance data would have to demonstrate that the likelihood of BSE transmission in that country has been eliminated. However, as noted in our risk assessment, Canadian BSE surveillance data do not provide a statistical basis for distinguishing BSE prevalence among birth year cohorts (APHIS 2006b, p. 12); the overall prevalence is so low that distinguishing any difference is nearly impossible. In other words, the data cannot distinguish any significant difference in prevalence among animals born in different years, which would have been one way to demonstrate the effect of a feed ban (e.g., if the feed ban were implemented at the beginning of 1997, surveillance data showing a higher BSE prevalence in animals born in 1996 than in animals born in 1997 would support the effectiveness of the feed ban). However, in the absence of a feed ban that reduced exposure to BSE, we would expect the prevalence of the disease to increase over time. We have no evidence that such an increase has occurred, but we do have data that the feed ban is being enforced.

Furthermore, as we discussed in our risk assessment, detection of BSE in an animal born after the date a feed ban was implemented does not indicate an overall failure of the measures in place to stem transmission of the disease in that country. Most other countries that have experienced cases of BSE, have reported similar cases. Of 25 countries that have reported indigenous BSE cases, only 4 reported no cases in 2005-06 (OIE 2007). Human error is expected, which is why the feed ban is comprised of a number of interrelated measures that have a cumulative effect. Our risk assessment does not assume 100 percent compliance with all measures all of the time. We discussed factors related to the feed ban in Canada since before its implementation in 1997. We considered activities related to inspection and compliance with the feed ban, the rendering industry, the risk of cross-contamination, education activities and industry awareness, and on-farm practices that might contribute to the efficacy of the feed ban. In addition, we highlighted the fact that since the implementation of the feed ban on August 4, 1997, CFIA has continued to revise and strengthen its processes and procedures to further enhance the effectiveness of the feed ban. Canada's July 2007 modification of its feed ban to remove SRMs from all animal feeds, pet food, and fertilizer is a good example of such enhancements. We concluded that compliance with the feed ban measures in Canada continues to increase as the program evolves and that all of these factors have resulted in a cumulative reduction in the risk that Canadian cattle will be exposed to the BSE agent.

Issue:
Several commenters stated that Canada cannot demonstrate that it has effectively prevented the feeding of ruminant material to cattle over the past 8 years. Commenters stated that eight or nine Canadian feedlots were discovered to still be feeding banned bone meal products, and that, because of their violations of the feed ban, 30,000 Canadian cattle were under quarantine. Additionally, one commenter stated that in March 2007, nine farms in Saskatchewan and as many as 8,000 cattle, deer, and other ruminants were quarantined after ruminant MBM was accidentally shipped to those farms from a Saskatoon feed mill. Another commenter stated that, in December 2006, Canada's Minister of Agriculture and Agri-Food acknowledged that up to 10,000 head of Canadian cattle on 113 different farms in the Provinces of Ottawa and Quebec had recently been fed feed contaminated with ruminant material.

Response:
APHIS is aware of the incidents reported in late 2006 and in March 2007 and considered such incidences very carefully in its evaluation of the effectiveness of the

feed ban. However, it is not clear to us what the commenters are referring to regarding 30,000 Canadian cattle under quarantine.

It should be noted that the use of the term “contaminated” above refers to the potential inclusion in ruminant feed of MBM derived from ruminants, but not to the feeding of known BSE-contaminated material to ruminants. Feed control systems, including those in the United States, are inherently subject to human error such as occurred in these incidents. These compliance errors require follow up and correction by CFIA, just as in the United States such incidents would necessitate follow-up by the U.S. Human Health and Services, Food and Drug Administration (FDA). Following detection of these occurrences, CFIA conducted a detailed investigation and traced all potentially contaminated feed. CFIA accounted for and disposed of all feed that did not enter the distribution channels, and feed already distributed to farms was removed, disposed of, and replaced. CFIA conducted risk assessments to help evaluate the possibility that new cases of BSE would occur due to the contamination of feed with prohibited material, and concluded that the overall risk was negligible. Even though this finding indicated that it was highly unlikely that animals exposed to the involved feed would develop BSE in the coming years, in those instances where exposure to the feed could not be ruled out, the CFIA has excluded these animals and their meat and byproducts from export eligibility. This measure was established to meet the technical requirements of various trading partners and does not affect the movement or marketing of these animals within Canada. These findings, together with Canada's rapid and comprehensive response to the incidents, reinforces our confidence in the effective enforcement of Canada's ruminant feed ban.
7

7
In the rulemaking for our 2005 final rule establishing criteria for recognition of a region as a BSE minimal-risk region, we discussed in detail our evaluation of Canada's veterinary infrastructure; disease history; practices for preventing widespread introduction, exposure, and/or establishment of BSE; and measures taken following detection of the disease (APHIS 2005).

Issue:
Some commenters questioned the effectiveness of Canada's feed ban, given evidence of contamination of ruminant feed with MBM derived from ruminants. One commenter stated that, in the five cases of cattle born after March 1, 1999, where investigations of BSE in Canadian cattle have been completed, the reported cause of BSE infectivity centered on ruminant MBM used in non-ruminant feeds cross-contaminating ruminant feeds, either during processing at the feed mill or during transport. Given that four animals were born after March 1, 1999, the commenters indicated that great care must be given to the analysis of these animals in the risk assessment and did not feel that APHIS thoroughly examined the cases.

Response:
We agree with the commenters that the investigations of BSE in animals born in Canada in 2000 and 2002 suggest that these animals were most likely exposed during their first year of life to feed contaminated during processing (CFIA 2006a). Reports of the investigations identified incidents of concern in which ruminant feed was processed or transported immediately following the handling of nonruminant feed containing prohibited material. Such incidents were in contravention of Canadian regulations, which require flushing and/or clean-out between batches if ruminant feed is processed on the same lines as feed containing prohibited material.

We considered the issue of cross-contamination and concluded that Canada has implemented measures to prevent cross-contamination of ruminant feed with prohibited materials in the rendering and feed manufacturing industries are essential for implementation of an effective feed ban. We also considered other factors—including the regulatory actions taken to implement the feed ban, education and industry awareness efforts, inspection and compliance activities, and on-farm feeding practices—in our overall evaluation to determine the date the feed ban was effectively enforced in Canada and, based on those factors, identified March 1, 1999 as the date of effective enforcement of the feed ban.

APHIS did not specifically address each individual case of BSE in Canada in the risk assessment, as the available details of each epidemiological investigation did not contribute to the overall risk estimation. The risk estimation was based on consideration of all factors relevant in the risk pathway. These included consideration of the current Canadian feed ban, with explicit recognition that cases born after the feed ban was implemented in August 1997, or after the March 1, 1999 date have occurred and could continue to occur. The prevalence estimate acknowledges that BSE is present in Canada, albeit at a very low level. The risk reduction factors in the United States, including feed ban regulatory activities similar to those in Canada, were considered in the exposure assessment. The combination of all of these factors, including recognition that human error can occur in any step of the pathway, supported the conclusion that the risk to the United States of BSE—i.e, the likelihood of establishment and the potential impact of cases that may occur even without establishment—as a result of importing from Canada the bovine commodities considered in this rule is negligible.

Issue:
One commenter stated that Canada has experienced an increase in the number of BSE cases since it instituted a feed ban in 1997.

Response:
It appears that the commenter is equating the number of detected cases of BSE with the number of infected animals in a national herd. However, an increased number of detections of BSE does not necessarily mean an increase in prevalence. A BSE detection rate is dependent not only on prevalence, but also on intensity of surveillance. An increased number of BSE cases have been detected in Canada as that country has increased surveillance for the disease. As noted above, an APHIS analysis of the Canadian BSE surveillance data did not find a statistical basis for distinguishing BSE prevalence among birth year cohorts.

Issue:
A number of commenters referred to the number of BSE cases in cattle born in Canada after March 1, 1999, as evidence that the date should not be accepted as the date of an effectively enforced feed ban. Commenters requested that APHIS reassess the proposed rule in light of recent diagnoses of such cattle.

Response:
In the assessment of potential BSE risk we conducted for this rulemaking, we concluded that there is an extremely low likelihood that cattle born in Canada on or after March 1, 1999, will have been exposed to BSE. This conclusion does not mean that effective enforcement necessarily equals no instances of contravention of the feed ban, either accidentally or intentionally, just as isolated transgressions of U.S. laws do not necessarily constitute ineffective enforcement of those laws.

While specific incidents of cross-contamination can, and most likely will, happen, since no regulatory effort can ensure 100 percent compliance, the detection of BSE in several bovines in Canada born after March 1, 1999 does not negate the overall effect of the feed ban in decreasing the opportunities for transmission of disease. Empirical evidence from the United Kingdom has demonstrated, and simulation studies have reinforced, that implementation of a ruminant-to-ruminant feed ban leads to continued decrease in prevalence over time (Cohen,
et al.
, 2001; 2003; DEFRA 2006, EC 2003; 2005). Similar

effects of a feed ban have been seen in other countries in the EU, where there have been continued detections of BSE in cattle born after a feed ban is initially implemented. At the same time, however, the apparent number of cases of BSE identified in the EU-15 Member States has decreased every year since 2001. The available evidence leads firmly to the conclusion that animals born after the date of implementation of a ruminant-to-ruminant feed ban are far less likely to be exposed to the BSE agent (Heim and Kihm, 2003).

Issue:
A number of commenters recommended various alternative dates or timeframes for consideration as the date of effective enforcement of a feed ban in Canada. Most of the commenters who recommended an alternative date expressed concern regarding the detection of BSE in bovines born in Canada after March 1, 1999.

The recommended alternative dates or timeframes included the following: July 1, 2007; the date of birth of the youngest bovine in Canada that has been determined to be BSE-positive; May 1, 2002; 5 to 7 years after the most recently diagnosed case of BSE in Canada; whenever Canada can verify 100 percent compliance with its ruminant-to-ruminant feed ban; a staggered system of dates that would increase the allowable age of bovines intended for importation from Canada as time progressed with no additional diagnoses of BSE in Canada.

Some of the commenters who suggested July 2007 as the date of effective enforcement based their recommendation on the fact that on July 12, 2007, Canada expanded its feed ban to prohibit the inclusion of SRMs in any animal feeds, pet foods, or fertilizers. One commenter asked how APHIS can be satisfied that the United States would be importing a safe product if Canada itself was not satisfied with the safeguards in place at the time the proposed rule was published, and subsequently took additional measures to strengthen its feed ban. A number of commenters recommended that the provisions of the proposed rule not be implemented until Canada bans all feeding of animal material to food animals. One commenter stated that July 2007 would be an appropriate point to begin the importation of breeding animals that have had exposure to processed animal feed, and that March 1, 1999 would be an acceptable date for bovines that have not been exposed to processed animal feeds—such as bison maintained by Parks Canada.

Several commenters, who expressed no animal health concerns with identifying March 1, 1999 as the date of effective enforcement of a feed ban in Canada, recommended that APHIS consider harmonizing the date chosen with the date Canada has identified as the effective date of a ruminant-to-ruminant feed ban in the United States, January 1, 1999.

Response:
In prior rulemaking (APHIS 2005), we evaluated evidence (regulations in place based on statutory authority, adequate infrastructure to implement the regulations, and evidence of implementation and monitoring) in making the determination that compliance with the feed ban in Canada is good and concluded that the feed ban was effectively enforced. In our process of identifying the date of effective enforcement of a ruminant-to-ruminant feed ban in Canada, we considered Canada's implementation guidance and policies. For example, we considered the allowance of grace periods for certain aspects of the industry, in determining the practical implementation period for the feed regulations. Then we considered a sufficient time period subsequent to this implementation period to allow most feed products to cycle through the system, given the management practices in the country. We concluded, based on the above evaluations, that cattle born in Canada on or after March 1, 1999, can be imported into the United States with an extremely low likelihood that they have been exposed to the BSE agent.

As noted, a number of commenters recommended that APHIS consider July 2007, when Canada expanded its feed ban, as the date of effective enforcement of the Canadian feed ban. We consider the July 2007 expansion of the Canadian feed ban to be an enhancement of an already effective ban. CFIA, in explaining its rationale for the enhanced ban, emphasizes that, although surveillance results and investigations of BSE cases indicate that the feed ban in Canada has effectively reduced the spread of BSE since being implemented in 1997, even compliance with the ban's requirements left limited opportunities for contamination during manufacture, transportation, and storage that CFIA considered worth eliminating. In addition, the accidental misuse of feed on farms with multiple species could not be discounted. With the enhanced ban, CFIA projects that the eradication of BSE in Canada will be accelerated. Following such a regulatory path does not indicate that the feed ban in Canada prior to July 2007 was not effective or effectively enforced.

With regard to the recommendation that the date of effective enforcement of the Canadian feed ban be identified as the date of birth of the youngest bovine in Canada that has been determined to be BSE-positive, we do not consider such a change to be necessary or justified. The risk assessment we conducted for this rulemaking acknowledged that BSE exists in Canada and that there would likely be additional cases detected. March 1, 1999 was never intended to be an absolute cut-off point after which no new cases of BSE would be acceptable. The risk assessment concluded that, despite the likelihood of additional diagnoses of BSE in Canadian cattle, the proposed amendments would pose negligible risk to animal health and food safety in the United States. If an infected cow were to be imported into the United States, a series of strong safeguards would have to fail—in sequence—for that animal to pose any risk.

With regard to the recommendation that APHIS harmonize its identification of the effective enforcement date of a Canadian feed ban with the date identified by Canada as the date of effective enforcement in the United States, we do not agree that such a change would be appropriate or necessary. APHIS arrived at the March 1, 1999 date for effective enforcement of the feed ban in Canada by considering not only the date the feed ban was established in that country but also information provided by Canada regarding its implementation timetable, as well as feeding practices in that country. It does not necessarily follow that implementation events in the United States followed precisely the same track as those in Canada.

Issue:
In our January 2007 proposed rule, we discussed the diagnosis of BSE in cattle in Canada born after March 1, 1999, and stated that “such isolated incidents are not epidemiologically significant and do not contribute to further spread of BSE, especially when considered in light of the entire risk pathway and its attendant risk mitigations.”

Several commenters took issue with APHIS’ description of the cases as “isolated.” Some commenters stated that “isolated” implies a solitary or separated condition, which cannot be said of the BSE cases recently confirmed in Canada. Further, other commenters stated the cases are linked by a trend in geographic location, with the last three cases occurring in the Province of Alberta. One commenter stated that of the nine cases of BSE detected in Canada, four occurred in cattle born after March 1, 1999, and that four of nine cases—or 44 percent—do not represent isolated cases and strongly disagreed that this date corresponds to

when Canada's feed ban became effectively enforced.

Response:
We disagree with the comments, although we acknowledge that the term “isolated” could be interpreted in several ways. The use of the term in our proposed rule was not intended to imply that the cases were “solitary or separated.” Our use of the term “isolated” was intended to characterize the cases as being small in number and not indicative of a systemic failure of the feed ban in Canada, but rather the result of individual instances of error in contravention of the feed ban (
e.g.
, inadequate cleaning between handling of feed for non-ruminants and feed for ruminants).

For the reasons discussed above, we consider our determination that March 1, 1999 be deemed the date of effective enforcement of the feed ban in Canada to be reasonable, grounded firmly in the regulatory basis and operations of the ban in Canada, and entirely consistent with the science and with OIE guidelines. Accordingly, we are making no changes based on the comments.

Likelihood of Exposure of Cattle in the United States to BSE

The assessment is designed to estimate the likelihood of each of the multiple steps. Although we analyzed the likelihood of each individual step in the process occurring, we interpreted its significance in the context of the entire process.

As part of the risk assessment we conducted for our January 2007 proposed rule, we evaluated both the likelihood of “release” of the BSE agent into the United States and the likelihood of susceptible animals being exposed, given such release. We evaluated the pathways by which infected Canadian cattle, if imported, might expose U.S. cattle to BSE, and the likelihood that these pathways might lead to the establishment of the disease in the U.S. cattle population.

Several steps must take place for BSE to be transmitted to cattle in the United States from a bovine imported live from another country. A BSE-infected bovine must be imported into the United States; the infected bovine must die or be slaughtered; tissues from that animal that contain the infectious agent must be sent to a rendering facility; the infectivity present in these tissues must survive inactivation in the rendering process; the resulting meat-and-bone meal (MBM) containing the abnormal prion protein must be incorporated into feed; and this feed must be fed to cattle at a level adequate to infect the cattle. (The amount of infectious material required in feed for cattle to become infected is dependent on the age of the cattle; younger cattle are more susceptible to BSE and require less BSE-contaminated feed to become infected (Arnold and Wilesmith, 2004). We indicated in our risk assessment that the nature and likelihood of these pathways depend in large part on mitigations acting in series and in parallel that reduce the likelihood that BSE will be established in the United States.

A number of commenters addressed the issues of the likelihood of release of the BSE agent into the United States and the likelihood of exposure of U.S. cattle to BSE due to the importation of bovines from Canada. In general, the commenters said that we had underestimated the likelihood of release and/or exposure, or questioned one or more elements of our assessment.

Issue:
One commenter, whose statements were referenced and supported by a second commenter, discussed the geographic distribution of BSE cases in Canada and expressed concern that Canada's experience demonstrates that certain locations in the United States might be more susceptible to BSE establishment than others. The commenter stated that events in Canada indicate that an average risk estimate is meaningless for BSE and demonstrates how “hot spots” (
i.e.
, locations that are more susceptible to spread of disease and, therefore, that have a localized higher BSE prevalence) allow BSE to propagate and spread. The commenter stated that the model-based predictions in APHIS’ risk assessment are useless because the models do not account for geographic and other sources of heterogeneity and pointed to Alberta as a BSE hot spot. Further, the commenter indicated that the APHIS risk assessment has not provided any real data or relevant analyses related to BSE hot spot development and that APHIS has not quantified the risks that imports will create localized BSE hot spots in the United States. The commenter calculated that, if 5 percent of U.S. locations are potential hot spots, and 1 million animals are imported each year with six of them BSE-positive, the expected probability of at least one hot spot being activated in the United States is at least 77.7 percent.

Response:
We disagree with the commenters. The available evidence provides no basis for distinguishing BSE prevalence among Canadian provinces. The commenter who singled out Alberta provides no analysis to support the hypothesis that the BSE prevalence in Alberta is higher than in other provinces. Through May 2007, reported BSE cases have originated in three western Provinces: Alberta (8 cases), British Columbia (2 cases), and Manitoba (1 case). No cases have been reported through May 2007 in the eastern Provinces. Intuition might suggest that the BSE prevalence is higher in Alberta. However, Alberta contains approximately 40 percent of the Canadian cattle herd. Other factors being equal, BSE is more likely to be detected in regions with large cattle populations.

Apart from the detected cases, geographically disaggregated data on BSE surveillance and Canadian cattle population demographics are not available. However, assuming that the total BSurvE points accumulated through August 15, 2006 (APHIS 2006c, table 4) were collected proportionally to the cattle population size in each province, table 3 presents the allocation of the random sample size equivalents (BSurvE points).

Table 3.—Allocation of BSurvE Points among Provinces Proportional to Herd Size

Province

Cattle
(000)*

Percent

BSurvE
points

BSE
cases**

Alberta
6,300.0
38.8
594,858.4
7

Manitoba
1,720.0
10.6
162,405.8
1

British Columbia
830.0
5.1
78,370.2
1

Saskatchewan
3,450.0
21.2
325,755.8
0+

Ontario
2,203.9
13.6
208,096.6
0

Quebec
1,455.0
9.0
137,384.0
0

Nova Scotia
107.0
0.7
10,103.2
0

New Brunswick
90.5
0.6
8,545.2
0

Prince Edward Island
84.5
0.5
7,978.7
0

Newfoundland
9.1
0.1

Labrador

859.2

Total
16,250.0

1,534,357
9

*Source: Statistics Canada (2007).
**BSE cases reported through August 2006 were included in APHIS (2006c).
+The BSE case confirmed in May 2003 was born in Saskatchewan but reported in Alberta.

Based on this allocation of evidence, a binomial likelihood ratio test (Fleiss
et al.
, 2003) fails to reject the hypothesis that the provinces have the same BSE prevalence. That is, the result provides no basis for concluding that BSE prevalence varies among provinces. Depending on the method used to estimate provincial BSE prevalence, the test indicates that 11 to 20 BSE cases would have to have been observed in Alberta (or 4 to 7 cases in British Columbia) before rejection of the hypothesis.

The commenters provide no data or analysis related to BSE hot-spot development. APHIS’ risk assessment discusses the apparent geographic clustering of Canadian BSE cases reported through August 2006 in three western provinces: Alberta, British Columbia, and Manitoba (APHIS 2006b, pp. 12-13). (In addition, the May 2003 case reported in Alberta was born in Saskatchewan.) However, APHIS also noted that the Manitoba BSE case was phenotypically different than the previously detected BSE cases of Canadian origin (APHIS 2006b). In addition, in its risk assessment, APHIS considered the CFIA report

[Text truncated at 120,000 characters. The full text is on the page linked above.]

---

Source: Frix Law Library, https://www.frixlaw.com/law-library/documents/fr%3A07-4595. Public record. Not legal advice.