August 10, 2016 Importation of Bone-in Ovine Meat from Uruguay Docket No. APHIS-2015-0050 We are commenting on this proposed rule on behalf of the American Sheep Industry Association (ASI). ASI is the national trade association for the U.S. sheep industry representing nearly 90,000 sheep producers and allied organizations through its affiliated state organizations. ASI is a proponent of free and fair trade; our borders have generally been open to lamb imports from all qualifying countries, essentially duty and quota free for decades. Currently about half of lamb consumption in the United States is imported, primarily from Australia and New Zealand. ASI and our members are interested in and concerned with meat safety, quality and disease risk from all sources of lamb products. The proposed rule to allow the importation of bone-in lamb products from Uruguay includes some specific intriguing FMD risk-control and mitigation measures that we believe could be used rather effectively. Further, Uruguay has a history of a highly functional inspection system. However, we find that some elements of the risk assessments are deficient and that too much time has passed since rigorous FMD risk assessments have been done for them to be used as valid support for the current proposed changes in the import regulations. While comments on the accompanying risk assessment documentation are not invited, we provide the following recommendations gleaned from the material provided: It appears that the 2002 risk assessment, while not available to the public in detail, was rather thorough and included quantitative measurements and estimates. However, this was done 14 years ago and should not be used directly for this proposed action. The information provided is not very clear but the text of the documents imply that the 2007 assessment was a site visit and it is now nine years old; the 2014 site visit appears to be an update of the 2007 site visit. ASI also notes that no assessment of economic risk is mentioned in any of the documents. It is both known and intuitive that when a product has increased value (potentially in this case bone-in lamb meat sales to the U.S. from Uruguay) and there are like products in other zones, regions or areas of lower value because they can t export, there is an opportunity for transshipment or smuggling. ASI believes that such economic risk should be measured and included in a quantitative risk assessment. Therefore we believe that before action is taken on this matter, an updated and 1
comprehensive quantitative risk assessment should be conducted and the results made available to the public for comment before this proposal is acted upon. Specific comments primarily on the Import Risk Analysis supporting document: 3. Background Condition no. 2: The ability to comply with this requirement hinges in part on surveillance and the testing intervals. (See below for comments on surveillance under 6.1.3.1.) Condition no. 3: A) This condition and condition no. 2 appear not to be in alignment. Condition no. 3 can be interpreted to mean the herd of origin only needs to be free of FMD for 6 months after birth of a cohort of lambs, if lambs are weaned at 3 months, sent to the facility and slaughtered 3 months later. The lifetime in this case is 6 months. B) Condition 2 states there is no FMD diagnosis within 12 months, whereas condition 3 refers instead to the presence of FMD, which is entirely different. One cannot say if FMD (the disease) is present or not. Rather, one can only use an imperfect (in this case considerably imperfect) sampling scheme and imperfect diagnostic assays to suggest whether the disease is present above 10 percent or above 5 percent. This is not simply a semantics issue. One should not equate presence of disease (or of disease agent either) only with diagnosed cases. Condition no. 5: Particular attention is given to examination for vesicles of the head and feet. This requirement is inconsistent with the published science on natural infections, particularly in sheep. The literature (most recently regarding the UK 2001 epidemic) tells us that under natural conditions (non experimental), vesicles are almost never observed in sheep with FMD. Instead, erosions/ulceration and secondary infections of burst vesicles are observed, which can easily be misdiagnosed as trauma, footrot, foreign body, Orf, and other look-alike diseases. Any lesions consistent with FMD should be highlighted, particularly those seen under field conditions. Condition no. 10: It is unclear how this certification would function. How would official verification of ph readings be conducted to assure that they were accurate, that diagnostic tests were accurate, and all other points from no. 1 through no. 9 have been addressed? 4.1 Risk Analysis Methods: A fundamental need met by well-conducted risk analyses is an understanding of the likely magnitude of a risk (i.e. an estimate of the probability of FMD entering the US during some time period) and an understanding of confidence in that estimate. Thus, risk assessments necessarily require that methodologies should yield 2
quantified (calculated) estimates of risk (probability) and uncertainty in the estimate. The APHIS risk analysis methods provide no quantification; rather, it states scaled qualifiers of risk, without offering scientific data. Scientific risk assessments should also be verifiable through validation. We believe that the model proposed for the exportation of bone-in lamb meat to the U.S. would need to be validated through some other process to assess likelihood that it accurately predicts no movement of FMD from Uruguay to the U.S. The purpose of exposure assessment is to provide convincing data and risk estimates regarding the risk of exposure of lambs to FMD. The OIE standards typically represent minimal guidelines; that is, procedures, proportion tested, sample sizes, etc. should at least meet the minimal OIE standards. As is stated in the document: The OIE standards for import risk analysis require that risk be expressed in terms of probability, rather than mere possibility (OIE 2010). There is no justification presented for not following minimal standards. Estimates of probability are necessary to permit all interpretations to be based on a consistent, quantifiable measurement scale, and not on opinion. 6.1.1 Official veterinary infrastructure: While the veterinary infrastructure in Uruguay is well-respected for their aggressive integrated programs to control and eradicate FMD, more information should be published specifically regarding the procedures used by DILAVE. No information was provided regarding the quality control, assessments, use of proper positive and negative controls, and other procedures that should be in place to routinely assess quality and accuracy of diagnostic testing. Typically, laboratory error, which always exists at some level, should be included in the sequence of events that drive the probabilities of a false negative test result, in addition to published values for a particular assay, surveillance sampling schemes, etc. Thus, in order to obtain estimates of risk of FMD in Uruguay, and of FMD in exported lambs, one must know the probability that a specific assay will fail to detect antibodies and the additional probability that the specific laboratory conducting the assay will impose more inaccuracy and false negative results by improperly conducting or interpreting the assay. Even though the test kits are validated by an OIE approved lab (see 6.1.5), labs using the test kits should undergo annual or more frequent blind testing by an independent agency. 6.1.3.1 Serosurveillance: Surveillance, by definition, is a strategic, active, and purposeful process to intentionally seek out and find infected animals. Typically, it does not involve random selection of test herds/flocks or subjects; rather, high risk populations and animals are targeted. It is unclear why random selection would be used, unless there is no difference in risk of acquiring FMD among the Uruguay herds/flocks, which would be unlikely. The term systematic sampling is not defined, and should be. Depending on the type of systematic sampling used, significant bias could be introduced that would lessen the likelihood of selecting an infected animal. 3
The assumption of 0.5% prevalence among herds means that the sampling scheme could miss testing an infected herd/flock for every 200 herds sampled, and that a very large number of herds/flocks will have to be sampled to ensure that the population does not include a few infected herds. If there were only 900 herds, a 0.5 percent prevalence would mean at least 4 are infected. If an infected herd is selected for testing, the assumption of 10 percent prevalence among cattle within a herd (e.g. 10 out of 100 animals infected) means that a very large number of animals (large sample size) will have to be tested to confidently claim there is no infection (for example, on average, 100 animals would have to be tested to find the one animal infected if prevalence is 1 percent). APHIS claims that because FMD is such a highly contagious disease, most animals in a herd would be infected. This may not likely be the case for sheep raised in a country with a reasonably aggressive vaccination program in cattle. It is conceivable that any circulating field viruses will have adapted to the vaccine strain, where such changes could affect virulence; we simply do not know. For sheep, an even larger sample of animals would have to be detected to find one or a few infected animals in a flock with a low prevalence of 5 percent. This type of sampling is not appropriate for surveillance in that negative results do not tell us there is no infection. It only tells us that infection prevalence is less than the stipulated statistical limits, namely less than 0.5 percent of herds are infected and in those infected herds, less than 10 percent (cattle) and 5 percent (sheep) are infected. We should not interpret negative results from testing 3600 animals in 900 herds to mean there is no FMD or FMD virus in Uruguay. Given the statistical threshold values assumed, it would be reasonable to conclude that up to 4 (actually 4.5=3600x0.005) herds could be infected but not detected. Testing of only 3600 animals from 900 herds over the course of one year seems inadequate for the entire country, given the importance the country (and the US) would be placing on this plan. If the country has 2.5 million susceptible animals in 10,000 herds, testing less than 10 percent of the herds and less than 2 percent of the animals per year is inadequate to ensure FMD is absent (see 6.1.3.2). Ideally, very short test intervals are preferred in surveillance in order to allow rapid action should the agent be found. The plan APHIS presented does not provide information on surveillance testing intervals that would help us understand how well they might be able to respond to a positive test result. 6.2.4 Testing of select lambs: Claims of sensitivity of the FMDv antibody test for sheep are not supported by the studies cited. The study by Sharma et al 2014 did not examine sheep, and, therefore, there is no scientific reason based on that study to believe the assay would have a 99 percent sensitivity in sheep. The study by Brocchi et al 2006 did examine sheep, but reported in the abstract a 99 percent sensitivity only for cattle. The authors state: Conclusive data concerning sensitivity cannot be derived for ovine and swine 4
species given the small number of experimental samples---. There were 9 sheep in the experiment, which all became seropositive 25-28 days after infection. Thus, the assay cannot be said to detect on average more than 9 out of 9 after such a long incubation period of 4 weeks. That is, one out of 10 could still be missed, or a sensitivity of only 90 percent. More importantly, they did not report results after 6-8 days post infection (even though they did so with vaccinated sheep), which would be a critical and necessary period to evaluate for the serologic assay. They did not report titer, whether it was high, low, or marginal, which also is important. This paragraph also presents an argument for why lambs would not likely be infected with FMD. The argument is reasonable as far as it goes, but no estimates or risk (probability) are derived that would offer a quantitative measure of the likelihood and confidence level of the estimate. 6.2.5 Management of lambs: In the appendix, APHIS notes appropriately that infected sheep typically have very subtle or no lesions, or clinical signs of FMD. The clinical FMD examination of lambs in the Uruguay facility could be further compromised or confused by a high prevalence of look-alike conditions, namely footrot and internal parasites. Foot lesions caused by FMD, as seen under field conditions, seldom if ever present as vesicles. Rather, one observes eroded or ulcerated skin with secondary infection, easily misclassified as footrot. Similarly, sheep with FMD infection can have diarrhea, typically mild, which could be misclassified as an internal parasite infection in this particular facility. Consequently, less confidence should be given to clinical assessments, and more to serologic testing. Further, the European Commission (EC) published a report on the findings of their audit in 2012 concerning FMD and certification procedures in Uruguay. Following is an excerpt from the executive summary. We believe the EC findings are consistent with several of our stated concerns: some outstanding issues still weaken the effectiveness of some of the pillars of the system the insufficient targeting of official on-the-spot controls on FMD vaccination that can not ensure adequate vaccination coverage in all areas with an increased risk of FMD appearance, and the deficient official reporting of those controls; a very limited contribution of passive surveillance to the detection and notification of suspect cases of vesicular diseases, and the non-validated sensitivity of the combination of diagnostic tests used to carry out the sero- epidemiological checks carried out since 2007 aimed at proving the absence of virus circulation in the cattle and ovine populations. 5
In the Regulatory Impact Analysis under a description of the U.S. Sheep Industry it is stated: Historically, lamb and mutton were viewed as byproducts of wool production, even though wool receipts accounted for a smaller share of revenue. As wool revenues have declined, producers have turned their attention to lamb and mutton production and the possibility of other byproducts such as sheep leather. While it is a quote from the USDA/ERS website, it is very old history at best and erroneous. We recommend the following publication for a more accurate history and description of the U.S. Sheep Industry as USDA conducts any economic impact analyses regarding this industry: National Academy of Science, National Research Council 2008, Changes in the Sheep Industry in the United States. We believe that as USDA works with trading partners and potential trading partners and examine their animal disease control and management systems, our own preparedness and readiness systems should be the best in the world as well as improving and evolving continuously using current scientific methods. A readily available and up to date FMD vaccine bank of a size that is appropriate for the U.S. livestock population needs to be a priority for USDA. In conclusion, we appreciate the willingness of Uruguay s sheep industry participants and of their regulatory agencies to propose to adopt several interlocking measures to help assure that exported lamb meat would not contain FMD virus. However, given the concerns listed above, we believe APHIS should withdraw this proposed rule and conduct the following: A) Conduct a current quantitative and qualitative risk assessment and present the findings in a transparent manner to the public. B) Reexamine the proposed surveillance-sampling plan including serological sampling tests and protocols with evidence for effectiveness. Once these tasks are accomplished APHIS could publish a new proposed rule if warranted by the findings. Respectfully submitted, Burton Pfliger, ASI President 6