Accepted Manuscript Title: Estimating the BSE infection and detectable prevalence in cattle born after 2000 in Japan Author: Katsuaki Sugiura Takeshi Haga Takashi Onodera PII: DOI: Reference:
S0167-5877(14)00163-9 http://dx.doi.org/doi:10.1016/j.prevetmed.2014.04.008 PREVET 3570
To appear in:
PREVET
Received date: Revised date: Accepted date:
25-12-2013 10-4-2014 12-4-2014
Please cite this article as: Sugiura, K., Haga, T., Onodera, T.,Estimating the BSE infection and detectable prevalence in cattle born after 2000 in Japan, Preventive Veterinary Medicine (2014), http://dx.doi.org/10.1016/j.prevetmed.2014.04.008 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
i cr us an
Katsuaki Sugiura
Tokyo 113-8657, Japan Tel: +81-3-5841-5383
Fax: +81-5841-5191
M
Department of Global Agricultural Sciences, Graduate school of Agricultural and Life Sciences, The University of Tokyo, Yayoi 1-1-1, Bunkyo-ku,
ed
E-mail address: [email protected]
pt
Takeshi Haga
ce
Department of Veterinary Medical Sciences,
Ac
Graduate school of Agricultural and Life Sciences, The University of Tokyo, Yayoi 1-1-1, Bunkyo-ku, Tokyo 113-8657, Japan
Takashi Onodera
Research Center for Food Safety, Graduate school of Agricultural and Life Sciences, The University of Tokyo, Yayoi 1-1-1, Bunkyo-ku, Tokyo 113-8657, Japan
Abstract
Page 1 of 26
i cr us
surveillance data of these birth cohorts.
an
We estimated the infection prevalence of BSE in Japanese cattle born in the period 2000-2012, using maximum likelihood methods and BSE From this, we predicted the number of infected cattle and test positives in years 2004-2020. Assuming
M
that the infection prevalence decayed exponentially over time from 2000, the infection prevalence of the 2000 birth cohort was estimated to be 0.00058 which declined exponentially by 0.115 times per year in the following years. The number of infected cattle was calculated to have peaked The number of test positives was calculated to have peaked in 2005 and would be zero by 2012. The number
ed
in 2005 and would be zero by 2020.
of BSE cases actually detected was within the 95% confidence interval of the predicted numbers. The detectable prevalence (predicted number of
The detectable prevalence would decline exponentially to zero in the subsequent years.
Ac
Key words
In this year it was predicted that one animal out of 160,000 tested
ce
would test positive.
pt
test positives/number of cattle tested) was predicted to be highest in 2005.
BSE, maximum likelihood methods, prediction model, prevalence, Japan
Page 2 of 26
i cr us an
1.
Introduction
on the use of processed animal protein
M
The first case of BSE in Japan was confirmed in September 2001. As a result the Japanese Government immediately introduced a compulsory ban in feed for ruminant animals and in fertilizers. The purpose was to stop the exposure of these food
ed
animals to transmissible spongiform encephalopathy (TSE) agents (including the BSE agent) via the oral route since processing methods to convert animal-derived raw materials to edible protein could not be guaranteed to destroy TSE agents.
In 2001 surveillance for BSE was enhanced in two
pt
main ways. First, reporting of fallen-stock >=24months of age and all clinical BSE suspects from all herds irrespective of type became mandatory
ce
so that brain testing (for PrPBSE) could be used to confirm or exclude BSE. Secondly, BSE testing was introduced for cattle of all ages slaughtered for human consumption. Positive cases were destroyed and did not enter food or feed chains.
Japan obtained the negligible BSE risk status at the World Assembly
Ac
cases had been detected, with the last case being detected in January 2009.
As a result, by the end of 2012, 35 additional BSE
of the World Organisation for Animal Health (OIE) in May 2013 (OIE, 2013a). slaughter-cattle has been raised so that all slaughter-cattle >=48 months are tested.
Since 1 July 2013, the target age for BSE testing of
Although Japan is a net importer of cattle and beef and exports
no cattle and only a small amount of beef, it is important for Japan to ensure an effective BSE surveillance that demonstrates a continuing negligible risk from BSE to maintain the confidence of consumers and trading partners. Sugiura and Murray (2007) and Sugiura et al. (2009a) previously estimated the prevalence of BSE-infected animals within birth cohorts of dairy cattle born between 1992-2001 using Bayesian inference and the surveillance data.
In the present study, we estimated the infection and detectable
Page 3 of 26
i cr us
an
prevalence in the birth cohorts of beef and dairy cattle from 2000-2012, using an exponential decay model similar to the C-TSEMM developed by Adkin et al. (2012) and the surveillance data of these birth cohorts, and predicted the number of infected and test positive animals that would occur
2. Materials and methods 2.1. Assumptions
ed
M
until 2020. We also predicted the detectable prevalence until 2020 to provide data to enable the design of a future surveillance programme.
pt
Our previous studies indicated that there were two peaks of BSE infection in Japan, the first in the 1996 birth cohort and the second in the 2000 The source of infection was likely to
ce
birth cohort (Sugiura and Murray, 2007; Sugiura et al., 2009a; Sugiura et al., 2009b; Sugiura et al., 2011).
have been BSE-infected concentrate feed consumed in these years. Based on the results of these studies and with the knowledge that a strictly We
Ac
enforced ban on the use of ruminant protein in feed has been in place since 2001, we focused on the birth cohorts born in 2000-2012.
assumed that the infection prevalence decayed exponentially over time from the 2000 birth cohort through 2012 birth cohort. This seems reasonable because, with no source of infection in the feed produced from 2001, the only significant source of BSE infection would be from infected feed produced before 2001 that remained in the delivery chain such as transport, feed stores, and the like. We assumed that when cattle leave the population they are either slaughtered for human consumption, or reported as fallen stock. In Japan, no distinction is made between clinical suspects that are killed and fallen-stock that have died on farm. We also assumed that the number of spontaneous cases is so low that it would not affect the model outputs.
Page 4 of 26
i cr us an
2.2. Exit model
M
The exit model uses the power of simulation to estimate both the probability that a BSE-infected animal leaves (exits) the cattle population as a
2.2.1. Input parameters
ed
result of death or being culled for slaughter at a certain age in relation to the stage of its incubation period.
The input parameters used for the exit model are: the age that an animal is infected; the age at which an infected animal dies or is slaughtered; the
pt
incubation period; and the duration of clinical signs.
The age that an animal becomes infected was modeled as a uniform distribution over its first The age
ce
year of life, because cattle are mostly likely exposed and infected during this time (Wilesmith et al., 1988; Arnold and Wilesmith, 2004).
at slaughter or death was estimated using statistics derived from data on the number of cattle slaughtered and the number of fallen stock in Japan,
Ac
provided by the Cattle Identification Scheme Database in 2011 (Anonymous, 2012a). A distribution of the incubation period was modeled as a gamma distribution with mean 5 and variance 1.6, as estimated by Anderson et al. (1996). The duration of clinical signs was modeled as a uniform distribution between 2 and 6 months. 2.2.2. Output variables The output variables, which are used as inputs in the BSE epidemic model described in Section 2.3 are as follows: P_FSa is the probability that an infected animal dies at age a during the clinical onset or within 6 months before the clinical onset. P_SCa is the probability that an infected animal is culled for slaughter within 6 months before the clinical onset or during clinical stage at age
Page 5 of 26
i cr us an
a.
I_FSa is the probability that an infected animal dies at age a irrespective of the reason.
M
I_SCa is the probability that an infected animal is slaughtered for human consumption at age a. The input parameters are combined using a series of IF statements to determine whether or not an animal leaves (exits) the cattle population at a
ed
certain age as a result of death or being culled for slaughter relative to the stage of its incubation period. The probability estimates are derived as point values, which represent the proportion of successful iterations. A probability distribution can then be defined by combining the estimates for The simulation was undertaken with the software @Risk 6.0 (Palisade) added into the spreadsheet software Excel
pt
each age and exit pathway.
ce
14.0 (Microsoft Corporation). To account for rare events in the model, for example the likelihood of an animal not being slaughtered or dying from BSE before reaching 12 years of age, 100,000 iterations were undertaken.
This ensured that a sufficient number of successful iterations were
Ac
obtained for such rare events from which reliable estimates could be obtained. The obtained probability distributions for P_FSa and P_SCa and I_FSa and I_SCa are shown in Fig. 1.
2.3. BSE epidemic model Maximum likelihood methods based on those used by Adkin et al. (2012), but adjusted to incorporate the input parameters estimated in Section 2.2, were used to estimate the infection prevalence (proportion of infected animals) in birth cohorts born in 2000 through to 2012.
Page 6 of 26
i cr us
The infection prevalence of birth cohort c ( )
an
2.3.1.
(1) where
M
We assumed that the infection prevalence decayed over time and gave the equation of the curve by:
is the estimated proportion of animals infected in birth cohort c (c=0 for 2000 birth cohort, c=1 for 2001 birth cohort, …, c=12 for 2012
2.3.2. Log-likelihood model
ed
birth cohort), and A and K are model parameters which were estimated using the log-likelihood model.
pt
Since we are incorporating both fallen-stock and slaughter-cattle data into the back-calculation model, the log-likelihood functions consist of two
ce
parts: log-likelihood functions of the data for the number of clinical test positives (fallen-stock) and preclinical test positives (slaughter-cattle). 2.3.2.1. Log-likelihood model for fallen-stock
Ac
The testing data for fallen-stock cases in cohort c, in year y (y=0 for 2000, y=1 for 2001, …, y=12 for 2012) arises from a binomial distribution. The joint log-likelihood function for the number of test positive fallen-stock is given by:
where
is the size of the birth cohort c,
(2) is the number of test positives in birth cohort c in year y, and
is the probability of an
animal in cohort c exiting the population and being detected as fallen-stock in year y. was modeled by
, where
is the proportion of fallen-stock tested in year y,
Page 7 of 26
is infection
i M
2.3.2.2. Log-likelihood model for slaughter-cattle
us
cr or within 6 months before the clinical onset.
is the probability that an infected animal dies at age a (=y-c) during the clinical onset
an
prevalence of birth cohort c (see Section 2.3.1.), and
The joint log-likelihood function for the number of test positive slaughter-cattle is given by:
ed
(3)
where Hc is the size of the birth cohort c, D_SCc,y is the number of test positives in birth cohort c in year y, and E_SCc,y is the probability of an
pt
animal in cohort c exiting the population and being detected as slaughter-cattle in year y. , where T_SCy is the proportion of slaughter-cattle tested in year y,
ce
E_SCc,y was modeled by
prevalence of birth cohort c (see Section 2.3.1.), and P_SCa is the probability that an infected animal is culled for slaughter at age a
is infection (=y-c) during
Ac
the clinical onset or within 6 months before the clinical onset. 2.3.2.3. Overall joint log-likelihood function From joint log-likelihood functions 2 and 3, the overall joint log-likelihood function for the number of test positive fallen-stock and slaughter-cattle animals of birth cohorts born in 2000 through 2012 is given: (4) 2.3.3. Input variables
Page 8 of 26
i cr us an
2.3.3.1. The size of the birth cohort c (Hc)
The numbers of cattle born in each year from 2004 through 2012 were 1,419,977, 1,432,766, 1,413,234, 1,421,145, 1,411,683, 1,389,139, As
there is no data available for 2000-2003 birth cohorts, we assumed that the size of 2000-2003 cohorts is the same as the 2004 birth cohort.
The
M
1,377,578, 1,331,785 and 1,282,505 respectively, obtained from the National Cattle Identification Scheme Database (Anonymous, 2013).
observed in the following years.
ed
deviation of this assumed size of 2000-2003 birth cohorts from the actual size would be small considering the stable milk and beef production
pt
2.3.3.2. The proportion of slaughter-cattle tested in year y (T_FSy and T_SCy)
ce
The proportion of fallen-stock and slaughter-cattle tested in 2000-2012 is shown in Table 1. The probability that fallen-stock >= 24 months of age are tested for BSE was determined by dividing the number of fallen-stock >= 24 months tested each
Ac
year from 2000 to 2005 by the number tested in 2005. Since all fallen-stock >=24 months have been tested for BSE since 1 April 2004, the number tested in 2005 is assumed to be the same as the number of fallen stock reported in that year under the National Cattle Traceability Scheme (Anonymous, 2012a). Likewise, the probability that a slaughter animal was tested was determined by dividing the number of slaughter-cattle tested each year from 2000 to 2002 by the number of slaughter-cattle tested in 2002, since all slaughter cattle have been tested since 18 October 2001. The number of slaughter animals tested in 2002 was obtained from the Ministry of Health, Labour and Welfare’s database (Anonymous, 2012).
2.3.3.3. The number of fallen-stock and slaughter-cattle test positives in birth cohort c in year y (D_FSc,y and D_SCc,y) The number of BSE cases detected in 2003-2009 by year of detection and exit-streams is shown in Table 2. No case had been detected in animals
Page 9 of 26
i cr us
an
born in 2002 and subsequent years, except for one case born in 2001 and another born in 2002, both detected in 2003. We did not consider these cases in estimating the infection prevalence, because the confirmatory diagnosis for these cases was made by Western blot only, which detected an
M
accumulation of unusually small amounts of core fragments of PrPSc (PrPcore) and the transmission study conducted later by inoculating bovine PrP-overexpressing transgenic mice (TgBoPrP) with brain material from these cases resulted in none of these mice developing evidence of BSE
ed
infection (Yamakawa et al., 2003; Yokoyama et al., 2007, Ono et al., 2011). 2.3.4. Estimating the infection prevalence of birth cohort c ( )
pt
As discussed in Section 2.2 (input parameters), values for all the parameters in overall joint log-likelihood function in Section 2.3.2.3, apart from
ce
model parameters A and K in Section 2.3.1, can be obtained from existing sources of information. In this situation, A and K were estimated by maximizing the overall joint log-likelihood in Section 2.3.2.3. using RiskOptimizer (Palisade).
RiskOptimizer is software added into the
Ac
spreadsheet software Excel 14.0 (Microsoft Corporation) that enables estimation of two or more unknown variables by genetic algorithm. We obtained the values of A and K that maximize the joint log-likelihood by undertaking 10,000 simulations of 1000 iterations each. The infection prevalence of birth cohort c ( ) was estimated by inserting the values of A and K into the model in Section 2.3.1.
2.4. Predicting the number of BSE infected animals and test positives The number of BSE infected animals originating from birth cohort c that left the population in year y in either of the fallen-stock or slaughter-cattle streams was estimated respectively by:
Page 10 of 26
i cr us an
(5) (6)
that
is the estimated infection prevalence of birth cohort c. In calculating the number of infected cattle in these born after 2013, we assumed
M
where
declines after 2013 with the same trend as up to 2013.
ed
The number of test positives of fallen-stock and slaughter-cattle originating from birth cohort c in year y was estimated respectively by:
pt
(7)
(8)
ce
The number of BSE infected fallen-stock and slaughter-cattle and the number of test positives in fallen-stock and slaughter-cattle in a particular year was calculated by summing over the birth cohorts for the respective exit streams. The total number of BSE infected cattle culled and the total
Ac
number of test positives in a particular year was computed by summing over the two exit streams and birth cohorts. To incorporate the uncertainty of the model results arising from the binomial function, we put these estimates into a Poisson function as a mean number of events per year and used Monte Carlo simulation and ran 100,000 iterations.
The simulations were performed using software @Risk
6.0 (Palisade) added into the spreadsheet software Excel 14.0 (Microsoft Corporation).
2.5. Predicting the detectable prevalence The detectable prevalence in a particular year was calculated by dividing the total number of test positives by the total number of animals tested in
Page 11 of 26
i cr us
an
that year. In calculating the number of cattle tested, we assumed that active surveillance composed of testing of all fallen-stock >=24 months and
M
all slaughter-cattle >=0 month will continue until 2020.
3. Results and discussion
ed
Fig. 2 shows the estimated infection prevalence for each of the 13 birth cohorts from 2000 to 2012.
The infection prevalence of cattle born in
2000 was estimated to be 0.00058 and declined from 2001 birth cohort exponentially by e-2.2 (=0.115) times per year. This infection prevalence in
pt
the 2000 birth cohort is higher than the infection prevalence of slaughter-cattle in the 2000 birth cohort of dairy cattle that we estimated previously, This is because the infection prevalence that we estimated in the current study
ce
namely 0.00016 (95%CI: 0.00008-0.00028) (Sugiura et al., 2009a).
is the average of fallen-stock and slaughter-cattle and takes account of dairy and beef cattle, and that the infection prevalence ratio of fallen-stock
Ac
over slaughter-cattle is higher in the current study. This reflects the surveillance result that in cattle born after 2000, most of BSE cases are in fallen-stock (Table 2).
Fig. 3 shows the predicted total number of infected fallen-stock and infected cattle slaughtered for human consumption during the years 2004-2020. It is predicted that there will be no infected animals culled after 2020. The results show that the presence of infected animals would last longer in the current study than in our previous studies (Sugiura and Murray, 2007; Sugiura et al., 2009a).
This is because in the previous study we
assumed that cattle born after 2002 are not infected, while in the current study we assumed that the infection prevalence decays exponentially over time from 2002.
Page 12 of 26
i cr us
an
Fig. 4 shows the predicted total number of test positives (fallen-stock and slaughter-cattle) during the years 2004-2020, assuming that active surveillance composed of BSE testing of all fallen-stock animals >=24 months and all slaughter-cattle >=0 months is in place, enforced and
M
effective. The number of BSE cases actually detected is within the 95% confidence interval of the predicted numbers. Also, the predicted mean number of test positives (BSE cases) will be zero in the years after 2010, which is consistent with the fact that no BSE case has been detected since
ed
the last case detected in January 2009. Fig. 4(a) indicates that the probability that a BSE case is detected will be practicably zero after 2012. Our model assumes that the infection prevalence delays exponentially over time through birth cohorts, as does the C-TSEMM model (Adkin et al. This assumption is reasonable because the only significant source of BSE infection after a rigidly enforced feed ban would be from
pt
2012).
In fact, the number of BSE infected
ce
infected feed produced prior to 2001 and which had contaminated feed vehicles, feed stores and the like.
animals detected declined exponentially over time between different birth cohorts in the UK, France and other EU members after the feed ban in
Ac
1996 (European Commission, 2013), indicating the exponential decline in infection prevalence over time between birth cohorts. In Japan, although the number of BSE cases detected is too low to fit an exponential decay curve, the results of official on-site inspections conducted after 2002 indicated nearly full compliance with the feed ban by feed importers, manufacturers, distributors and end-users (farmers) (Kusama et al. 2009). One advantage of our model, compared to the C-TSEMM model, is that our model does not have to use the differential slaughter parameter in the model, as the parameter is essentially incorporated into the probability estimates that an infected animal exits in the fallen-stock or slaughter streams (P_FSa and P_SCa), which were obtained using the exit model. Our model can be used to simulate the effectiveness of different surveillance programs by changing the target age of BSE testing and predicting the
Page 13 of 26
i cr us
an
number of test positives. For example, the predicted number of test positives will not change, even if we raise the target age of BSE testing from zero to 48 months for slaughter-cattle from 2013 (results not shown).
M
Fig. 5 shows the predicted detectable prevalence during 2004-2020. The detectable prevalence was predicted to be highest in 2005, with one BSE animal detected in 16,000 fallen-stock or in 0.9 million slaughter-cattle tested. On average one in 0.2 million cattle tested can be detected. From
ed
2006, the detectable prevalence starts to decline exponentially as the infection prevalence declines and most of the cattle in 2000 and 2001 birth cohorts with relatively high infection prevalence had been culled.
The detectable prevalence in all cattle, fallen-stock and slaughter-cattle falls
pt
below one in one million animals after 2008, 2010 and 2006 respectively.
ce
Countries recognized as negligible or controlled BSE risk by the OIE are required to conduct surveillance that is capable of detecting at least one BSE animal if there is one detectable animal in 50,000 or 100,000 respectively if they want to maintain their respective status (OIE, 2013b). Adkin
Ac
et al. (2012) evaluated the effectiveness of the current BSE monitoring schemes in the EU members using the C-TSEMM model, and concluded that no healthy slaughtered cattle need to be tested in order to meet a design prevalence of one detectable case in 100,000 adult cattle with a 95% confidence level. Our model can also be used to estimate the number of fallen-stock and slaughter-cattle required to be tested. For example, under the current BSE surveillance scheme in Japan, the number of fallen-stock and slaughter-cattle to be tested for BSE in 2014 is 90,000 and 220,000 respectively. However, using the detectable prevalence in fallen-stock and slaughter-cattle estimated in our study and the binomial formulae used by Cannon and Roe (1982), the number of animals required to be tested to detect one animal in 100,000 cattle is calculated to be 56,000 fallen-stock and no slaughter-cattle need to be tested. Likewise, the number of cattle to be tested from 2015 onward to meet such a design
Page 14 of 26
i cr us
an
prevalence can be calculated using the detectable prevalence in the respective years estimated in this study.
M
Acknowledgement
We thank Dr. Ray Bradley for carefully reading the manuscript and providing us with useful information and suggestions. We also thank Dr. Noel
Ac
ce
pt
ed
Murray, Canadian Food Inspection Agency for correcting mathematical errors in our model.
Page 15 of 26
i cr us an
References
Encephalopathies
in
cattle
in
M
Adkin, A., Simons, R., Arnold, M., 2012. Model for the evaluation of different options or the monitoring of Transmissible Spongiform the
European
Union
(C-TSEMM),
Supporting
Publications
2012:
EN-349.
55pp
,
ed
www.efsa.europa.eu/publications
Anderson, R.M., Donnelly, C.A., Ferguson, N.M., Woolhouse, M.E.J., Watt, C.J., Udy, H.J., MaWhinney, S., Dunstan, S.P., Southwood, T.R.E.,
pt
Wilesmith. J.W., Ryan, J.B.M., Hoinville, L.J., Hillerton, J.E., Austin, A.R., Wells, G.A.H., 1996. Transmission dynamic and epidemiology of
Anonymous,
ce
BSE in British cattle. Nature 382, 779-788. 2012.
Result
of
BSE
Testing
of
Cattle
in
Abattoirs,
Ministry
of
Health,
Labour
and
Welfare,
Tokyo,
Ac
http://www.mhlw.go.jp/houdou/0110/h1018-6.html Anonymous,
2013.
National
Cattle
Identification
Scheme
Database,
National
Livestock
Breeding
Center
(NLBC),
https://www.id.nlbc.go.jp/data/toukei.html
Arnold, M.E., Wilesmith, J.W., 2004. Estimation of the age-dependent risk of infection to BSE of dairy cattle in Great Britain. Prev. Vet. Med. 66(1-4), 35-47. Cannon, R.M., Roe, R.T., 1982. Livestock disease surveys -a field manual for veterinarians. Australian Government Publishing Service, Canberra, 35pp
Page 16 of 26
i cr us
an
European Commission, 2013. Report on the monitoring of ruminants for the presence of transmissible spongiform encephalopathies (TSEs) in the EU in 2012. Publication Office of the European Union, Luxembourg, 65pp
encephalopathy. Vet. Ital. 45, 287-295. (World
Organisation
for
Animal
ed
OIE
M
Kusama, T., Hibino, H., Onodera, T., Sugiura, K., 2009. Animal feed controls implemented in Japan or the eradication of bovine spongiform
Health),
2013a.
List
of
members’
BSE
risk
status,
http://www.oie.int/en/animal-health-in-the-world/official-disease-status/bse/list-of-bse-risk-status/
pt
OIE (World Organisation for Animal Health), 2013b. Chapter 11.5 bovine spongiform encephalopathy, in: Terrestrial Animal Health Code 22nd Ed.
ce
Volume II. OIE, Paris, pp. 531-546.
Ono, F., Tase, N., Kurosawa, A., Hiyaoka, A., Ohyama, A., Tezuka, Y., Wada, N., Sato, Y., Tobiume, M., Hagiwara, K. 2011. Aypical L-type
Ac
bovine spongiform encephalopathy (L-BSE) transmission to Cynomolgus macaques, a non-human primate. Jpn. J. Infect. Dis, 64, 81-84.
Sugiura, K., Kikuchi, E., Onodera, T., 2009a. Updated prediction for the BSE epidemic in dairy cattle in Japan, Prev. Vet. Med. 89, 272-276. Sugiura, K., Murray, N., 2007. Estimating the prevalence of BSE in dairy birth cohorts and predicting the incidence of BSE cases in Japan, Prev. Vet. Med. 82, 213-235. Sugiura, K., Murray, N., Tsutsui, T., Kikuchi, E., Onodera, T., 2011. Simulating the bovine spongiform encephalopathy situation in Japan Rev. sci. tech. Off. int. Epiz. 30, 627-634
Page 17 of 26
i cr us
an
Sugiura, K., Onodera, T., Bradley, R., 2009b. Epidemiological features of the bovine spongiform encephalopathy epidemic in Japan Rev. sci. tech. Off. int. Epiz. 28, 945-956
M
Yamakawa, Y., Hagiwara, K., Nohtomi, K., Nakamura, Y., Nishijima, M., Higuchi, Y., Sata, Y., Sata, T. and the Expert Committee for BSE diagnosis, Ministry of Health, Labour and Welfare of Japan, 2003. Atypical proteinase K-resisitant prion protein (PrPre) observed in an
ed
apparently health 23-month-old Holstein steer. Jpn. J. infect. Dis., 56, 221-222. Yokoyama, T., Masujin, K., Yamakawa, Y., Sata, t., Murayama, Y., Shu, Y., Okada, H., Mohri, S., Shinagawa, M., 2007. Experimental
ce
Dis, 60, 31-320.
pt
transmission of two young and one suspended bovine spongiform encephalopathy (BSE) cases into bovinized transgenic mic. Jpn. J. infect.
Wilesmith, J.W., Wells, G.A.H., Cranwell, M.P., Ryan, J.B.M., 1988. Bovine spongiform encephalopathy: epidemiological studies. Vet. Rec. 123,
Ac
638-644.
Page 18 of 26
Fig. 1
17
Probability that a bovine spongiform encephalopathy infected animal in Japan dies at
18
different ages during the clinical onset or within 6 months before the clinical onset (a, black
19
bars), and probability that an infected animal in Japan is culled for slaughter within 6 months
20
before the clinical onset or during clinical stage (a, white bars). Probability that an infected
21
animal in Japan dies at different ages (b, black bars), and probability that an infected animal
22
in Japan is slaughtered for human consumption (b, white bars).
23
obtained using the demographic data of Japanese dairy and beef cattle and the exit model
24
described in Section 2.2.
ip t
16
cr
These probabilities were
us
25 Fig. 2
27
Estimated prevalence of bovine spongiform encephalopathy infection in different birth
28
cohorts of cattle in Japan.
an
26
29 Fig. 3
31
Predicted number of bovine spongiform encephalopathy infected animals in Japan in total
32
(a), fallen-stock (b) and slaughter cattle (c) that would be culled in years from 2004 to 2020.
33
Solid lines indicate the mean estimates, and dashed lines the 95% confidence intervals.
M
30
d
34 Fig. 4
36
Predicted number of bovine spongiform encephalopathy test positives in total (a),
37
fallen-stock (b) and slaughter cattle (c) in years from 2004 to 2020. Solid lines indicate the
38
mean estimates, and dashed lines the 95% confidence intervals, assuming a surveillance
39
program composed of BSE testing of all fallen-stock >=24 months and all slaughter-cattle
40
>=0 month was in effect until 2020 and is rigidly enforced. Dotted lines indicate the number
41
of BSE cases actually detected.
Ac ce p
42
te
35
43
Fig. 5
44
Predicted bovine spongiform encephalopathy detectable prevalence of all cattle (solid line
45
with △), fallen-stock (dashed line with ◇) and slaughter-cattle (dotted line with □) during
46
2004-2020, assuming a surveillance program composed of BSE testing of all fallen-stock
47
>=24 months and all slaughter-cattle >=0 month was in effect until 2020.
48 49 50
21
Page 19 of 26
Highlights We made editorial corrections as suggested by the reviewer. ▶We inserted one sentence explain the effect of the assumption for the 2000-2003 birth cohort size.▶We added one reference to the OIE terrestrial animal health code.
an
us
cr
ip t
50 51 52 53
55
Ac ce p
te
d
M
54
56
Fig. 1
57
Probability that a bovine spongiform encephalopathy infected animal in Japan dies at
58
different ages during the clinical onset or within 6 months before the clinical onset (a, black
59
bars), and probability that an infected animal in Japan is culled for slaughter within 6 months
60
before the clinical onset or during clinical stage (a, white bars). Probability that an infected
61
animal in Japan dies at different ages (b, black bars), and probability that an infected animal
62
in Japan is slaughtered for human consumption (b, white bars).
These probabilities were
22
Page 20 of 26
63
obtained using the demographic data of Japanese dairy and beef cattle and the exit model
64
described in Section 2.2.
65 66
te
d
M
an
us
cr
ip t
67
Ac ce p
68 69 70
Fig. 2
71
Estimated prevalence of bovine spongiform encephalopathy infection in different birth
72
cohorts of cattle in Japan.
73
23
Page 21 of 26
us
cr
ip t
73
75
Ac ce p
te
d
M
an
74
76 77
Fig. 3
78
Predicted number of bovine spongiform encephalopathy infected animals in Japan in total 24
Page 22 of 26
(a), fallen-stock (b) and slaughter cattle (c) that would be culled in years from 2004 to 2020.
80
Solid lines indicate the mean estimates, and dashed lines the 95% confidence intervals.
us
cr
ip t
79
te
d
M
an
81
Ac ce p
82
83 84
Fig. 4
85
Predicted number of bovine spongiform encephalopathy test positives in total (a),
86
fallen-stock (b) and slaughter cattle (c) in years from 2004 to 2020. Solid lines indicate the
87
mean estimates, and dashed lines the 95% confidence intervals, assuming a surveillance 25
Page 23 of 26
88
program composed of BSE testing of all fallen-stock >=24 months and all slaughter-cattle
89
>=0 month was in effect until 2020 and is rigidly enforced. Dotted lines indicate the number
90
of BSE cases actually detected.
M
an
us
cr
ip t
91
Fig. 5
94
Predicted bovine spongiform encephalopathy detectable prevalence of all cattle (solid line
95
with △), fallen-stock (dashed line with ◇) and slaughter-cattle (dotted line with □) during
96
2004-2020, assuming a surveillance program composed of BSE testing of all fallen-stock
97
>=24 months and all slaughter-cattle >=0 month was in effect until 2020.
99 100
te
Ac ce p
98
d
92 93
26
Page 24 of 26
cr
ip t
Table
us
Table 1
The proportion of fallen-stock >=24 months and slaughter-cattle tested for bovine spongiform encephalopathy in Japan in 2000-2012. 2000
2001
2002
2003
2004
2005
Fallen-stock
0.003
0.010
0.036
0.379
0.873
1.000
Slaughter-cattle
0.000
0.178
1.000
1.000
1.000
1.000
2006
2007
2008
2009
2010
2011
2012
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
an
Years
The proportion of fallen-stock >= 24 months of age tested for bovine spongiform encephalopathy was determined by dividing the number of
M
fallen-stock >= 24 months tested each year from 2000 to 2005 by the number tested in 2005. The proportion of slaughter-cattle tested was
Ac c
ep te
d
determined by dividing the number of slaughter-cattle tested each year from 2000 to 2002 by the number of slaughter-cattle tested in 2002.
Page 25 of 26
ip t cr us
Table 2
Number of bovine spongiform encephalopathy cases recorded in Japan for birth cohorts 2000-2001 by year of detection and by exit streams. Year of detection
Birth cohort
2005
2006
2007
2008
Fallen-stock
0
1
2
4
1
1
1
10
Slaughter-cattle
0
0
1
2
0
0
0
3
0
1
0
0
0
1
0
0
1
0
0
1
an
2004
2001 birth cohort
2002 birth cohort Fallen-stock Slaughter-cattle
0 a
(1)
d
Slaughter-cattle
0
0
ep te
Fallen-stock
M
2000 birth cohort
Total
Total
2003
0
2009
0
0
0
0
0
0
0
a
0
0
0
0
0
0
0
a
1
3
7
2
1
1
15
(1) (2)
No bovine spongiform encephalopathy cases had been detected after 2009 or in birth cohorts born in 2003 and later. One case born in 2001 and another born in 2002 were detected in 2003. We did not consider these cases in estimating the infection
Ac c
a
prevalence, because the confirmatory diagnosis was made by Western blot only, which detected an accumulation of unusually small amounts of core fragments of PrPSc (PrPcore) and the transmission study conducted later by inoculating bovine PrP-overexpressing transgenic mice (TgBoPrP) with brain materials from these cases resulted in none developing evidence of bovine spongiform encephalopathy infection (Yamakawa et al., 2003; Yokoyama et al., 2007).
Page 26 of 26
S0167-5877(14)00163-9 http://dx.doi.org/doi:10.1016/j.prevetmed.2014.04.008 PREVET 3570
To appear in:
PREVET
Received date: Revised date: Accepted date:
25-12-2013 10-4-2014 12-4-2014
Please cite this article as: Sugiura, K., Haga, T., Onodera, T.,Estimating the BSE infection and detectable prevalence in cattle born after 2000 in Japan, Preventive Veterinary Medicine (2014), http://dx.doi.org/10.1016/j.prevetmed.2014.04.008 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
i cr us an
Katsuaki Sugiura
Tokyo 113-8657, Japan Tel: +81-3-5841-5383
Fax: +81-5841-5191
M
Department of Global Agricultural Sciences, Graduate school of Agricultural and Life Sciences, The University of Tokyo, Yayoi 1-1-1, Bunkyo-ku,
ed
E-mail address: [email protected]
pt
Takeshi Haga
ce
Department of Veterinary Medical Sciences,
Ac
Graduate school of Agricultural and Life Sciences, The University of Tokyo, Yayoi 1-1-1, Bunkyo-ku, Tokyo 113-8657, Japan
Takashi Onodera
Research Center for Food Safety, Graduate school of Agricultural and Life Sciences, The University of Tokyo, Yayoi 1-1-1, Bunkyo-ku, Tokyo 113-8657, Japan
Abstract
Page 1 of 26
i cr us
surveillance data of these birth cohorts.
an
We estimated the infection prevalence of BSE in Japanese cattle born in the period 2000-2012, using maximum likelihood methods and BSE From this, we predicted the number of infected cattle and test positives in years 2004-2020. Assuming
M
that the infection prevalence decayed exponentially over time from 2000, the infection prevalence of the 2000 birth cohort was estimated to be 0.00058 which declined exponentially by 0.115 times per year in the following years. The number of infected cattle was calculated to have peaked The number of test positives was calculated to have peaked in 2005 and would be zero by 2012. The number
ed
in 2005 and would be zero by 2020.
of BSE cases actually detected was within the 95% confidence interval of the predicted numbers. The detectable prevalence (predicted number of
The detectable prevalence would decline exponentially to zero in the subsequent years.
Ac
Key words
In this year it was predicted that one animal out of 160,000 tested
ce
would test positive.
pt
test positives/number of cattle tested) was predicted to be highest in 2005.
BSE, maximum likelihood methods, prediction model, prevalence, Japan
Page 2 of 26
i cr us an
1.
Introduction
on the use of processed animal protein
M
The first case of BSE in Japan was confirmed in September 2001. As a result the Japanese Government immediately introduced a compulsory ban in feed for ruminant animals and in fertilizers. The purpose was to stop the exposure of these food
ed
animals to transmissible spongiform encephalopathy (TSE) agents (including the BSE agent) via the oral route since processing methods to convert animal-derived raw materials to edible protein could not be guaranteed to destroy TSE agents.
In 2001 surveillance for BSE was enhanced in two
pt
main ways. First, reporting of fallen-stock >=24months of age and all clinical BSE suspects from all herds irrespective of type became mandatory
ce
so that brain testing (for PrPBSE) could be used to confirm or exclude BSE. Secondly, BSE testing was introduced for cattle of all ages slaughtered for human consumption. Positive cases were destroyed and did not enter food or feed chains.
Japan obtained the negligible BSE risk status at the World Assembly
Ac
cases had been detected, with the last case being detected in January 2009.
As a result, by the end of 2012, 35 additional BSE
of the World Organisation for Animal Health (OIE) in May 2013 (OIE, 2013a). slaughter-cattle has been raised so that all slaughter-cattle >=48 months are tested.
Since 1 July 2013, the target age for BSE testing of
Although Japan is a net importer of cattle and beef and exports
no cattle and only a small amount of beef, it is important for Japan to ensure an effective BSE surveillance that demonstrates a continuing negligible risk from BSE to maintain the confidence of consumers and trading partners. Sugiura and Murray (2007) and Sugiura et al. (2009a) previously estimated the prevalence of BSE-infected animals within birth cohorts of dairy cattle born between 1992-2001 using Bayesian inference and the surveillance data.
In the present study, we estimated the infection and detectable
Page 3 of 26
i cr us
an
prevalence in the birth cohorts of beef and dairy cattle from 2000-2012, using an exponential decay model similar to the C-TSEMM developed by Adkin et al. (2012) and the surveillance data of these birth cohorts, and predicted the number of infected and test positive animals that would occur
2. Materials and methods 2.1. Assumptions
ed
M
until 2020. We also predicted the detectable prevalence until 2020 to provide data to enable the design of a future surveillance programme.
pt
Our previous studies indicated that there were two peaks of BSE infection in Japan, the first in the 1996 birth cohort and the second in the 2000 The source of infection was likely to
ce
birth cohort (Sugiura and Murray, 2007; Sugiura et al., 2009a; Sugiura et al., 2009b; Sugiura et al., 2011).
have been BSE-infected concentrate feed consumed in these years. Based on the results of these studies and with the knowledge that a strictly We
Ac
enforced ban on the use of ruminant protein in feed has been in place since 2001, we focused on the birth cohorts born in 2000-2012.
assumed that the infection prevalence decayed exponentially over time from the 2000 birth cohort through 2012 birth cohort. This seems reasonable because, with no source of infection in the feed produced from 2001, the only significant source of BSE infection would be from infected feed produced before 2001 that remained in the delivery chain such as transport, feed stores, and the like. We assumed that when cattle leave the population they are either slaughtered for human consumption, or reported as fallen stock. In Japan, no distinction is made between clinical suspects that are killed and fallen-stock that have died on farm. We also assumed that the number of spontaneous cases is so low that it would not affect the model outputs.
Page 4 of 26
i cr us an
2.2. Exit model
M
The exit model uses the power of simulation to estimate both the probability that a BSE-infected animal leaves (exits) the cattle population as a
2.2.1. Input parameters
ed
result of death or being culled for slaughter at a certain age in relation to the stage of its incubation period.
The input parameters used for the exit model are: the age that an animal is infected; the age at which an infected animal dies or is slaughtered; the
pt
incubation period; and the duration of clinical signs.
The age that an animal becomes infected was modeled as a uniform distribution over its first The age
ce
year of life, because cattle are mostly likely exposed and infected during this time (Wilesmith et al., 1988; Arnold and Wilesmith, 2004).
at slaughter or death was estimated using statistics derived from data on the number of cattle slaughtered and the number of fallen stock in Japan,
Ac
provided by the Cattle Identification Scheme Database in 2011 (Anonymous, 2012a). A distribution of the incubation period was modeled as a gamma distribution with mean 5 and variance 1.6, as estimated by Anderson et al. (1996). The duration of clinical signs was modeled as a uniform distribution between 2 and 6 months. 2.2.2. Output variables The output variables, which are used as inputs in the BSE epidemic model described in Section 2.3 are as follows: P_FSa is the probability that an infected animal dies at age a during the clinical onset or within 6 months before the clinical onset. P_SCa is the probability that an infected animal is culled for slaughter within 6 months before the clinical onset or during clinical stage at age
Page 5 of 26
i cr us an
a.
I_FSa is the probability that an infected animal dies at age a irrespective of the reason.
M
I_SCa is the probability that an infected animal is slaughtered for human consumption at age a. The input parameters are combined using a series of IF statements to determine whether or not an animal leaves (exits) the cattle population at a
ed
certain age as a result of death or being culled for slaughter relative to the stage of its incubation period. The probability estimates are derived as point values, which represent the proportion of successful iterations. A probability distribution can then be defined by combining the estimates for The simulation was undertaken with the software @Risk 6.0 (Palisade) added into the spreadsheet software Excel
pt
each age and exit pathway.
ce
14.0 (Microsoft Corporation). To account for rare events in the model, for example the likelihood of an animal not being slaughtered or dying from BSE before reaching 12 years of age, 100,000 iterations were undertaken.
This ensured that a sufficient number of successful iterations were
Ac
obtained for such rare events from which reliable estimates could be obtained. The obtained probability distributions for P_FSa and P_SCa and I_FSa and I_SCa are shown in Fig. 1.
2.3. BSE epidemic model Maximum likelihood methods based on those used by Adkin et al. (2012), but adjusted to incorporate the input parameters estimated in Section 2.2, were used to estimate the infection prevalence (proportion of infected animals) in birth cohorts born in 2000 through to 2012.
Page 6 of 26
i cr us
The infection prevalence of birth cohort c ( )
an
2.3.1.
(1) where
M
We assumed that the infection prevalence decayed over time and gave the equation of the curve by:
is the estimated proportion of animals infected in birth cohort c (c=0 for 2000 birth cohort, c=1 for 2001 birth cohort, …, c=12 for 2012
2.3.2. Log-likelihood model
ed
birth cohort), and A and K are model parameters which were estimated using the log-likelihood model.
pt
Since we are incorporating both fallen-stock and slaughter-cattle data into the back-calculation model, the log-likelihood functions consist of two
ce
parts: log-likelihood functions of the data for the number of clinical test positives (fallen-stock) and preclinical test positives (slaughter-cattle). 2.3.2.1. Log-likelihood model for fallen-stock
Ac
The testing data for fallen-stock cases in cohort c, in year y (y=0 for 2000, y=1 for 2001, …, y=12 for 2012) arises from a binomial distribution. The joint log-likelihood function for the number of test positive fallen-stock is given by:
where
is the size of the birth cohort c,
(2) is the number of test positives in birth cohort c in year y, and
is the probability of an
animal in cohort c exiting the population and being detected as fallen-stock in year y. was modeled by
, where
is the proportion of fallen-stock tested in year y,
Page 7 of 26
is infection
i M
2.3.2.2. Log-likelihood model for slaughter-cattle
us
cr or within 6 months before the clinical onset.
is the probability that an infected animal dies at age a (=y-c) during the clinical onset
an
prevalence of birth cohort c (see Section 2.3.1.), and
The joint log-likelihood function for the number of test positive slaughter-cattle is given by:
ed
(3)
where Hc is the size of the birth cohort c, D_SCc,y is the number of test positives in birth cohort c in year y, and E_SCc,y is the probability of an
pt
animal in cohort c exiting the population and being detected as slaughter-cattle in year y. , where T_SCy is the proportion of slaughter-cattle tested in year y,
ce
E_SCc,y was modeled by
prevalence of birth cohort c (see Section 2.3.1.), and P_SCa is the probability that an infected animal is culled for slaughter at age a
is infection (=y-c) during
Ac
the clinical onset or within 6 months before the clinical onset. 2.3.2.3. Overall joint log-likelihood function From joint log-likelihood functions 2 and 3, the overall joint log-likelihood function for the number of test positive fallen-stock and slaughter-cattle animals of birth cohorts born in 2000 through 2012 is given: (4) 2.3.3. Input variables
Page 8 of 26
i cr us an
2.3.3.1. The size of the birth cohort c (Hc)
The numbers of cattle born in each year from 2004 through 2012 were 1,419,977, 1,432,766, 1,413,234, 1,421,145, 1,411,683, 1,389,139, As
there is no data available for 2000-2003 birth cohorts, we assumed that the size of 2000-2003 cohorts is the same as the 2004 birth cohort.
The
M
1,377,578, 1,331,785 and 1,282,505 respectively, obtained from the National Cattle Identification Scheme Database (Anonymous, 2013).
observed in the following years.
ed
deviation of this assumed size of 2000-2003 birth cohorts from the actual size would be small considering the stable milk and beef production
pt
2.3.3.2. The proportion of slaughter-cattle tested in year y (T_FSy and T_SCy)
ce
The proportion of fallen-stock and slaughter-cattle tested in 2000-2012 is shown in Table 1. The probability that fallen-stock >= 24 months of age are tested for BSE was determined by dividing the number of fallen-stock >= 24 months tested each
Ac
year from 2000 to 2005 by the number tested in 2005. Since all fallen-stock >=24 months have been tested for BSE since 1 April 2004, the number tested in 2005 is assumed to be the same as the number of fallen stock reported in that year under the National Cattle Traceability Scheme (Anonymous, 2012a). Likewise, the probability that a slaughter animal was tested was determined by dividing the number of slaughter-cattle tested each year from 2000 to 2002 by the number of slaughter-cattle tested in 2002, since all slaughter cattle have been tested since 18 October 2001. The number of slaughter animals tested in 2002 was obtained from the Ministry of Health, Labour and Welfare’s database (Anonymous, 2012).
2.3.3.3. The number of fallen-stock and slaughter-cattle test positives in birth cohort c in year y (D_FSc,y and D_SCc,y) The number of BSE cases detected in 2003-2009 by year of detection and exit-streams is shown in Table 2. No case had been detected in animals
Page 9 of 26
i cr us
an
born in 2002 and subsequent years, except for one case born in 2001 and another born in 2002, both detected in 2003. We did not consider these cases in estimating the infection prevalence, because the confirmatory diagnosis for these cases was made by Western blot only, which detected an
M
accumulation of unusually small amounts of core fragments of PrPSc (PrPcore) and the transmission study conducted later by inoculating bovine PrP-overexpressing transgenic mice (TgBoPrP) with brain material from these cases resulted in none of these mice developing evidence of BSE
ed
infection (Yamakawa et al., 2003; Yokoyama et al., 2007, Ono et al., 2011). 2.3.4. Estimating the infection prevalence of birth cohort c ( )
pt
As discussed in Section 2.2 (input parameters), values for all the parameters in overall joint log-likelihood function in Section 2.3.2.3, apart from
ce
model parameters A and K in Section 2.3.1, can be obtained from existing sources of information. In this situation, A and K were estimated by maximizing the overall joint log-likelihood in Section 2.3.2.3. using RiskOptimizer (Palisade).
RiskOptimizer is software added into the
Ac
spreadsheet software Excel 14.0 (Microsoft Corporation) that enables estimation of two or more unknown variables by genetic algorithm. We obtained the values of A and K that maximize the joint log-likelihood by undertaking 10,000 simulations of 1000 iterations each. The infection prevalence of birth cohort c ( ) was estimated by inserting the values of A and K into the model in Section 2.3.1.
2.4. Predicting the number of BSE infected animals and test positives The number of BSE infected animals originating from birth cohort c that left the population in year y in either of the fallen-stock or slaughter-cattle streams was estimated respectively by:
Page 10 of 26
i cr us an
(5) (6)
that
is the estimated infection prevalence of birth cohort c. In calculating the number of infected cattle in these born after 2013, we assumed
M
where
declines after 2013 with the same trend as up to 2013.
ed
The number of test positives of fallen-stock and slaughter-cattle originating from birth cohort c in year y was estimated respectively by:
pt
(7)
(8)
ce
The number of BSE infected fallen-stock and slaughter-cattle and the number of test positives in fallen-stock and slaughter-cattle in a particular year was calculated by summing over the birth cohorts for the respective exit streams. The total number of BSE infected cattle culled and the total
Ac
number of test positives in a particular year was computed by summing over the two exit streams and birth cohorts. To incorporate the uncertainty of the model results arising from the binomial function, we put these estimates into a Poisson function as a mean number of events per year and used Monte Carlo simulation and ran 100,000 iterations.
The simulations were performed using software @Risk
6.0 (Palisade) added into the spreadsheet software Excel 14.0 (Microsoft Corporation).
2.5. Predicting the detectable prevalence The detectable prevalence in a particular year was calculated by dividing the total number of test positives by the total number of animals tested in
Page 11 of 26
i cr us
an
that year. In calculating the number of cattle tested, we assumed that active surveillance composed of testing of all fallen-stock >=24 months and
M
all slaughter-cattle >=0 month will continue until 2020.
3. Results and discussion
ed
Fig. 2 shows the estimated infection prevalence for each of the 13 birth cohorts from 2000 to 2012.
The infection prevalence of cattle born in
2000 was estimated to be 0.00058 and declined from 2001 birth cohort exponentially by e-2.2 (=0.115) times per year. This infection prevalence in
pt
the 2000 birth cohort is higher than the infection prevalence of slaughter-cattle in the 2000 birth cohort of dairy cattle that we estimated previously, This is because the infection prevalence that we estimated in the current study
ce
namely 0.00016 (95%CI: 0.00008-0.00028) (Sugiura et al., 2009a).
is the average of fallen-stock and slaughter-cattle and takes account of dairy and beef cattle, and that the infection prevalence ratio of fallen-stock
Ac
over slaughter-cattle is higher in the current study. This reflects the surveillance result that in cattle born after 2000, most of BSE cases are in fallen-stock (Table 2).
Fig. 3 shows the predicted total number of infected fallen-stock and infected cattle slaughtered for human consumption during the years 2004-2020. It is predicted that there will be no infected animals culled after 2020. The results show that the presence of infected animals would last longer in the current study than in our previous studies (Sugiura and Murray, 2007; Sugiura et al., 2009a).
This is because in the previous study we
assumed that cattle born after 2002 are not infected, while in the current study we assumed that the infection prevalence decays exponentially over time from 2002.
Page 12 of 26
i cr us
an
Fig. 4 shows the predicted total number of test positives (fallen-stock and slaughter-cattle) during the years 2004-2020, assuming that active surveillance composed of BSE testing of all fallen-stock animals >=24 months and all slaughter-cattle >=0 months is in place, enforced and
M
effective. The number of BSE cases actually detected is within the 95% confidence interval of the predicted numbers. Also, the predicted mean number of test positives (BSE cases) will be zero in the years after 2010, which is consistent with the fact that no BSE case has been detected since
ed
the last case detected in January 2009. Fig. 4(a) indicates that the probability that a BSE case is detected will be practicably zero after 2012. Our model assumes that the infection prevalence delays exponentially over time through birth cohorts, as does the C-TSEMM model (Adkin et al. This assumption is reasonable because the only significant source of BSE infection after a rigidly enforced feed ban would be from
pt
2012).
In fact, the number of BSE infected
ce
infected feed produced prior to 2001 and which had contaminated feed vehicles, feed stores and the like.
animals detected declined exponentially over time between different birth cohorts in the UK, France and other EU members after the feed ban in
Ac
1996 (European Commission, 2013), indicating the exponential decline in infection prevalence over time between birth cohorts. In Japan, although the number of BSE cases detected is too low to fit an exponential decay curve, the results of official on-site inspections conducted after 2002 indicated nearly full compliance with the feed ban by feed importers, manufacturers, distributors and end-users (farmers) (Kusama et al. 2009). One advantage of our model, compared to the C-TSEMM model, is that our model does not have to use the differential slaughter parameter in the model, as the parameter is essentially incorporated into the probability estimates that an infected animal exits in the fallen-stock or slaughter streams (P_FSa and P_SCa), which were obtained using the exit model. Our model can be used to simulate the effectiveness of different surveillance programs by changing the target age of BSE testing and predicting the
Page 13 of 26
i cr us
an
number of test positives. For example, the predicted number of test positives will not change, even if we raise the target age of BSE testing from zero to 48 months for slaughter-cattle from 2013 (results not shown).
M
Fig. 5 shows the predicted detectable prevalence during 2004-2020. The detectable prevalence was predicted to be highest in 2005, with one BSE animal detected in 16,000 fallen-stock or in 0.9 million slaughter-cattle tested. On average one in 0.2 million cattle tested can be detected. From
ed
2006, the detectable prevalence starts to decline exponentially as the infection prevalence declines and most of the cattle in 2000 and 2001 birth cohorts with relatively high infection prevalence had been culled.
The detectable prevalence in all cattle, fallen-stock and slaughter-cattle falls
pt
below one in one million animals after 2008, 2010 and 2006 respectively.
ce
Countries recognized as negligible or controlled BSE risk by the OIE are required to conduct surveillance that is capable of detecting at least one BSE animal if there is one detectable animal in 50,000 or 100,000 respectively if they want to maintain their respective status (OIE, 2013b). Adkin
Ac
et al. (2012) evaluated the effectiveness of the current BSE monitoring schemes in the EU members using the C-TSEMM model, and concluded that no healthy slaughtered cattle need to be tested in order to meet a design prevalence of one detectable case in 100,000 adult cattle with a 95% confidence level. Our model can also be used to estimate the number of fallen-stock and slaughter-cattle required to be tested. For example, under the current BSE surveillance scheme in Japan, the number of fallen-stock and slaughter-cattle to be tested for BSE in 2014 is 90,000 and 220,000 respectively. However, using the detectable prevalence in fallen-stock and slaughter-cattle estimated in our study and the binomial formulae used by Cannon and Roe (1982), the number of animals required to be tested to detect one animal in 100,000 cattle is calculated to be 56,000 fallen-stock and no slaughter-cattle need to be tested. Likewise, the number of cattle to be tested from 2015 onward to meet such a design
Page 14 of 26
i cr us
an
prevalence can be calculated using the detectable prevalence in the respective years estimated in this study.
M
Acknowledgement
We thank Dr. Ray Bradley for carefully reading the manuscript and providing us with useful information and suggestions. We also thank Dr. Noel
Ac
ce
pt
ed
Murray, Canadian Food Inspection Agency for correcting mathematical errors in our model.
Page 15 of 26
i cr us an
References
Encephalopathies
in
cattle
in
M
Adkin, A., Simons, R., Arnold, M., 2012. Model for the evaluation of different options or the monitoring of Transmissible Spongiform the
European
Union
(C-TSEMM),
Supporting
Publications
2012:
EN-349.
55pp
,
ed
www.efsa.europa.eu/publications
Anderson, R.M., Donnelly, C.A., Ferguson, N.M., Woolhouse, M.E.J., Watt, C.J., Udy, H.J., MaWhinney, S., Dunstan, S.P., Southwood, T.R.E.,
pt
Wilesmith. J.W., Ryan, J.B.M., Hoinville, L.J., Hillerton, J.E., Austin, A.R., Wells, G.A.H., 1996. Transmission dynamic and epidemiology of
Anonymous,
ce
BSE in British cattle. Nature 382, 779-788. 2012.
Result
of
BSE
Testing
of
Cattle
in
Abattoirs,
Ministry
of
Health,
Labour
and
Welfare,
Tokyo,
Ac
http://www.mhlw.go.jp/houdou/0110/h1018-6.html Anonymous,
2013.
National
Cattle
Identification
Scheme
Database,
National
Livestock
Breeding
Center
(NLBC),
https://www.id.nlbc.go.jp/data/toukei.html
Arnold, M.E., Wilesmith, J.W., 2004. Estimation of the age-dependent risk of infection to BSE of dairy cattle in Great Britain. Prev. Vet. Med. 66(1-4), 35-47. Cannon, R.M., Roe, R.T., 1982. Livestock disease surveys -a field manual for veterinarians. Australian Government Publishing Service, Canberra, 35pp
Page 16 of 26
i cr us
an
European Commission, 2013. Report on the monitoring of ruminants for the presence of transmissible spongiform encephalopathies (TSEs) in the EU in 2012. Publication Office of the European Union, Luxembourg, 65pp
encephalopathy. Vet. Ital. 45, 287-295. (World
Organisation
for
Animal
ed
OIE
M
Kusama, T., Hibino, H., Onodera, T., Sugiura, K., 2009. Animal feed controls implemented in Japan or the eradication of bovine spongiform
Health),
2013a.
List
of
members’
BSE
risk
status,
http://www.oie.int/en/animal-health-in-the-world/official-disease-status/bse/list-of-bse-risk-status/
pt
OIE (World Organisation for Animal Health), 2013b. Chapter 11.5 bovine spongiform encephalopathy, in: Terrestrial Animal Health Code 22nd Ed.
ce
Volume II. OIE, Paris, pp. 531-546.
Ono, F., Tase, N., Kurosawa, A., Hiyaoka, A., Ohyama, A., Tezuka, Y., Wada, N., Sato, Y., Tobiume, M., Hagiwara, K. 2011. Aypical L-type
Ac
bovine spongiform encephalopathy (L-BSE) transmission to Cynomolgus macaques, a non-human primate. Jpn. J. Infect. Dis, 64, 81-84.
Sugiura, K., Kikuchi, E., Onodera, T., 2009a. Updated prediction for the BSE epidemic in dairy cattle in Japan, Prev. Vet. Med. 89, 272-276. Sugiura, K., Murray, N., 2007. Estimating the prevalence of BSE in dairy birth cohorts and predicting the incidence of BSE cases in Japan, Prev. Vet. Med. 82, 213-235. Sugiura, K., Murray, N., Tsutsui, T., Kikuchi, E., Onodera, T., 2011. Simulating the bovine spongiform encephalopathy situation in Japan Rev. sci. tech. Off. int. Epiz. 30, 627-634
Page 17 of 26
i cr us
an
Sugiura, K., Onodera, T., Bradley, R., 2009b. Epidemiological features of the bovine spongiform encephalopathy epidemic in Japan Rev. sci. tech. Off. int. Epiz. 28, 945-956
M
Yamakawa, Y., Hagiwara, K., Nohtomi, K., Nakamura, Y., Nishijima, M., Higuchi, Y., Sata, Y., Sata, T. and the Expert Committee for BSE diagnosis, Ministry of Health, Labour and Welfare of Japan, 2003. Atypical proteinase K-resisitant prion protein (PrPre) observed in an
ed
apparently health 23-month-old Holstein steer. Jpn. J. infect. Dis., 56, 221-222. Yokoyama, T., Masujin, K., Yamakawa, Y., Sata, t., Murayama, Y., Shu, Y., Okada, H., Mohri, S., Shinagawa, M., 2007. Experimental
ce
Dis, 60, 31-320.
pt
transmission of two young and one suspended bovine spongiform encephalopathy (BSE) cases into bovinized transgenic mic. Jpn. J. infect.
Wilesmith, J.W., Wells, G.A.H., Cranwell, M.P., Ryan, J.B.M., 1988. Bovine spongiform encephalopathy: epidemiological studies. Vet. Rec. 123,
Ac
638-644.
Page 18 of 26
Fig. 1
17
Probability that a bovine spongiform encephalopathy infected animal in Japan dies at
18
different ages during the clinical onset or within 6 months before the clinical onset (a, black
19
bars), and probability that an infected animal in Japan is culled for slaughter within 6 months
20
before the clinical onset or during clinical stage (a, white bars). Probability that an infected
21
animal in Japan dies at different ages (b, black bars), and probability that an infected animal
22
in Japan is slaughtered for human consumption (b, white bars).
23
obtained using the demographic data of Japanese dairy and beef cattle and the exit model
24
described in Section 2.2.
ip t
16
cr
These probabilities were
us
25 Fig. 2
27
Estimated prevalence of bovine spongiform encephalopathy infection in different birth
28
cohorts of cattle in Japan.
an
26
29 Fig. 3
31
Predicted number of bovine spongiform encephalopathy infected animals in Japan in total
32
(a), fallen-stock (b) and slaughter cattle (c) that would be culled in years from 2004 to 2020.
33
Solid lines indicate the mean estimates, and dashed lines the 95% confidence intervals.
M
30
d
34 Fig. 4
36
Predicted number of bovine spongiform encephalopathy test positives in total (a),
37
fallen-stock (b) and slaughter cattle (c) in years from 2004 to 2020. Solid lines indicate the
38
mean estimates, and dashed lines the 95% confidence intervals, assuming a surveillance
39
program composed of BSE testing of all fallen-stock >=24 months and all slaughter-cattle
40
>=0 month was in effect until 2020 and is rigidly enforced. Dotted lines indicate the number
41
of BSE cases actually detected.
Ac ce p
42
te
35
43
Fig. 5
44
Predicted bovine spongiform encephalopathy detectable prevalence of all cattle (solid line
45
with △), fallen-stock (dashed line with ◇) and slaughter-cattle (dotted line with □) during
46
2004-2020, assuming a surveillance program composed of BSE testing of all fallen-stock
47
>=24 months and all slaughter-cattle >=0 month was in effect until 2020.
48 49 50
21
Page 19 of 26
Highlights We made editorial corrections as suggested by the reviewer. ▶We inserted one sentence explain the effect of the assumption for the 2000-2003 birth cohort size.▶We added one reference to the OIE terrestrial animal health code.
an
us
cr
ip t
50 51 52 53
55
Ac ce p
te
d
M
54
56
Fig. 1
57
Probability that a bovine spongiform encephalopathy infected animal in Japan dies at
58
different ages during the clinical onset or within 6 months before the clinical onset (a, black
59
bars), and probability that an infected animal in Japan is culled for slaughter within 6 months
60
before the clinical onset or during clinical stage (a, white bars). Probability that an infected
61
animal in Japan dies at different ages (b, black bars), and probability that an infected animal
62
in Japan is slaughtered for human consumption (b, white bars).
These probabilities were
22
Page 20 of 26
63
obtained using the demographic data of Japanese dairy and beef cattle and the exit model
64
described in Section 2.2.
65 66
te
d
M
an
us
cr
ip t
67
Ac ce p
68 69 70
Fig. 2
71
Estimated prevalence of bovine spongiform encephalopathy infection in different birth
72
cohorts of cattle in Japan.
73
23
Page 21 of 26
us
cr
ip t
73
75
Ac ce p
te
d
M
an
74
76 77
Fig. 3
78
Predicted number of bovine spongiform encephalopathy infected animals in Japan in total 24
Page 22 of 26
(a), fallen-stock (b) and slaughter cattle (c) that would be culled in years from 2004 to 2020.
80
Solid lines indicate the mean estimates, and dashed lines the 95% confidence intervals.
us
cr
ip t
79
te
d
M
an
81
Ac ce p
82
83 84
Fig. 4
85
Predicted number of bovine spongiform encephalopathy test positives in total (a),
86
fallen-stock (b) and slaughter cattle (c) in years from 2004 to 2020. Solid lines indicate the
87
mean estimates, and dashed lines the 95% confidence intervals, assuming a surveillance 25
Page 23 of 26
88
program composed of BSE testing of all fallen-stock >=24 months and all slaughter-cattle
89
>=0 month was in effect until 2020 and is rigidly enforced. Dotted lines indicate the number
90
of BSE cases actually detected.
M
an
us
cr
ip t
91
Fig. 5
94
Predicted bovine spongiform encephalopathy detectable prevalence of all cattle (solid line
95
with △), fallen-stock (dashed line with ◇) and slaughter-cattle (dotted line with □) during
96
2004-2020, assuming a surveillance program composed of BSE testing of all fallen-stock
97
>=24 months and all slaughter-cattle >=0 month was in effect until 2020.
99 100
te
Ac ce p
98
d
92 93
26
Page 24 of 26
cr
ip t
Table
us
Table 1
The proportion of fallen-stock >=24 months and slaughter-cattle tested for bovine spongiform encephalopathy in Japan in 2000-2012. 2000
2001
2002
2003
2004
2005
Fallen-stock
0.003
0.010
0.036
0.379
0.873
1.000
Slaughter-cattle
0.000
0.178
1.000
1.000
1.000
1.000
2006
2007
2008
2009
2010
2011
2012
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
an
Years
The proportion of fallen-stock >= 24 months of age tested for bovine spongiform encephalopathy was determined by dividing the number of
M
fallen-stock >= 24 months tested each year from 2000 to 2005 by the number tested in 2005. The proportion of slaughter-cattle tested was
Ac c
ep te
d
determined by dividing the number of slaughter-cattle tested each year from 2000 to 2002 by the number of slaughter-cattle tested in 2002.
Page 25 of 26
ip t cr us
Table 2
Number of bovine spongiform encephalopathy cases recorded in Japan for birth cohorts 2000-2001 by year of detection and by exit streams. Year of detection
Birth cohort
2005
2006
2007
2008
Fallen-stock
0
1
2
4
1
1
1
10
Slaughter-cattle
0
0
1
2
0
0
0
3
0
1
0
0
0
1
0
0
1
0
0
1
an
2004
2001 birth cohort
2002 birth cohort Fallen-stock Slaughter-cattle
0 a
(1)
d
Slaughter-cattle
0
0
ep te
Fallen-stock
M
2000 birth cohort
Total
Total
2003
0
2009
0
0
0
0
0
0
0
a
0
0
0
0
0
0
0
a
1
3
7
2
1
1
15
(1) (2)
No bovine spongiform encephalopathy cases had been detected after 2009 or in birth cohorts born in 2003 and later. One case born in 2001 and another born in 2002 were detected in 2003. We did not consider these cases in estimating the infection
Ac c
a
prevalence, because the confirmatory diagnosis was made by Western blot only, which detected an accumulation of unusually small amounts of core fragments of PrPSc (PrPcore) and the transmission study conducted later by inoculating bovine PrP-overexpressing transgenic mice (TgBoPrP) with brain materials from these cases resulted in none developing evidence of bovine spongiform encephalopathy infection (Yamakawa et al., 2003; Yokoyama et al., 2007).
Page 26 of 26
