Accelerat ing t he world's research.

Factors associated with bovine

tuberculosis confirmation rates in
suspect lesions found in cattle at
routine sla...
Aminu Shittu

Related papers Download a PDF Pack of t he best relat ed papers 

Chapt er 14. 1971 onwards: M. bovis fight s back

Pet er At kins

Est imat ion of t he Relat ive Sensit ivit y of t he Comparat ive Tuberculin Skin Test in Tuberculous Cat t le …
Tony Goodchild

Est imat ion of t he Relat ive Sensit ivit y of t he Comparat ive Tuberculin Skin Test in Tuberculous Cat t le …
Tony Goodchild, Christ l Donnelly
 Preventive Veterinary Medicine 110 (2013) 395–404

Contents lists available at SciVerse ScienceDirect

Preventive Veterinary Medicine

journal homepage: www.elsevier.com/locate/prevetmed

Factors associated with bovine tuberculosis confirmation

rates in suspect lesions found in cattle at routine
slaughter in Great Britain, 2003–2008
A. Shittu a,∗ , R.S. Clifton-Hadley b , E.R. Ely c , P.U. Upton d , S.H. Downs c
a
The Royal Veterinary College, University of London, 4 Royal College Street, Camden, London NW1 0TU, United Kingdom
b
Department for Bovine Tuberculosis, Animal Health and Veterinary Laboratories Agency, Woodham Lane, New Haw, Addlestone,
Surrey KT15 3NB, United Kingdom
c
Epidemiology and Risk Group, Animal Health and Veterinary Laboratories Agency, Woodham Lane, New Haw, Addlestone, Surrey KT15
3NB, United Kingdom
d
Data Systems Group, Animal Health and Veterinary Laboratories Agency, Woodham Lane, New Haw, Addlestone, Surrey KT15 3NB,
United Kingdom

a r t i c l e i n f o a b s t r a c t

Article history: Bovine tuberculosis (bTB) is one of the most complex and intractable animal health prob-
Received 11 July 2012 lems facing the British cattle industry today. The inspection of carcasses from cattle sent to
Received in revised form 26 February 2013
slaughter is part of routine surveillance for bTB in Great Britain (GB). Tissue with suspect
Accepted 7 March 2013
lesions from cattle from herds previously considered uninfected with bTB is sent to the Ani-
mal Health and Veterinary Laboratories Agency (AHVLA) for culture and histopathological
Keywords:
examination for Mycobacterium bovis infection. In this study, risk factors for confirmation
Abattoir
Mycobacterium bovis of infection in suspect bTB lesions found at routine slaughter of cattle from officially bTB-
Bovine tuberculosis free (OTF) herds in GB were investigated. The study sample included the first record of a
Cattle suspect lesion in a bovine from any OTF herd identified during post-mortem inspection
Surveillance between 2003 and 2008. There were 3663 submissions from 151 slaughterhouses of which
2470 (67.4%) were confirmed as culture positive for M. bovis. Logistic regression analysis
with a random intercept for slaughterhouse was used to investigate relationships between
bTB confirmation and animal and herd-level risk factors. Slaughterhouse of post mortem
and the following factors related to bTB prevalence were significant predictors of confir-
mation probability: region of farm of origin of the animal, the testing interval for routine
field surveillance for bTB on the farm, number of reactors in the last bTB incident on the
farm within the last 4 years, if applicable, the animal’s date of birth and the year of animal’s
slaughter. The modelled predicted population averaged probabilities for confirmation var-
ied from 0.14 to 0.90 between slaughterhouses. Differences in the detection of cattle with
bTB between British slaughterhouses warrant further study.
© 2013 Elsevier B.V. All rights reserved.

1. Introduction

M. bovis, a slow-growing non-photochromogenic

∗ Corresponding author. Tel.: +44 7428 673702.
E-mail addresses: ashittu@rvc.ac.uk, ameen vet@yahoo.com,
aminu.shittu@liverpool.ac.uk (A. Shittu),
r.s.clifton-hadley@vla.defra.gsi.gov.uk (R.S. Clifton-Hadley),
elizabeth.ely@ahvla.gsi.gov.uk (E.R. Ely), paul.upton@ahvla.gsi.gov.uk
(P.U. Upton), sara.downs@ahvla.gsi.gov.uk (S.H. Downs).
Gram-positive acid-fast bacillus, is the principal cause of
bTB (Theon and Barletta, 2004). The organism has a wide
host range including humans and multiple wildlife species
(Morris et al., 1994). Bovine TB in cattle is regarded as one
of the GB’s most serious animal health problems (House

0167-5877/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.prevetmed.2013.03.001
396 A. Shittu et al. / Preventive Veterinary Medicine 110 (2013) 395–404
of Commons, 2008). The national control programme
includes tuberculin skin testing, removal of reactors,
removal of other cattle from herds undergoing a bTB
incident that are deemed to be at high risk of infection
and movement restrictions on cattle from infected herds.
The incidence of bTB in British herds has been steadily
increasing from very low levels in the early 1980s. Between
2003 and 2008, the number of new bTB incidents in herds
which were confirmed by pathology or culture of the
bacteria increased by 59% and the number of cattle culled
as part of bTB control increased by 67% (Defra, 2012). Since
2008 the number of incidents and cattle culled annually
has decreased but there were 4703 new incidents in GB
herds and a total of 33.000 cattle culled in 2010. The
test and slaughter policy to date has not achieved bTB
control and the disease may spread geographically into
areas previously considered as free from the disease. It is
recognised that a vigorous and multi-faceted surveillance
and control strategy is essential if bTB control is to be
eventually achieved (Defra, 2011).
Slaughterhouse surveillance is an important means
of detecting infection in herds (Meah and Lewis, 1999;
Kaneene et al., 2006) and has contributed to the eradi-
cation of bTB from several developed countries (Caffrey,
1994; Clifton-Hadley and Wilesmith, 1995; Cousins et al.,
1998; Cousins, 2001). O’Keeffe and White (1999) and
Frankena et al. (2007) reported that between 27% and
46% new herd breakdowns were detected in 1993 and
2001, respectively, through slaughterhouse surveillance
in Ireland. During 2008, slaughterhouse surveillance was
responsible for detecting almost 16% of new bTB incidents
(confirmed by culture or pathology) in herds in GB as a
whole and 35% of incidents in non-bTB endemic areas
where the routine field testing for bTB is every four years
(Veterinary Laboratories Agency, 2008). Failure to detect
infection in the slaughterhouse will contribute to the
spread of disease in a herd, particularly in low incidence
areas where the frequency of routine surveillance of cattle
herds using the tuberculin skin test, the main diagnostic
test used in routine field surveillance, is less than annual.
The proportion of infected animals detected will depend
on the thoroughness of the post-mortem examination and
other factors related to the operation of the slaughter-
house (Corner et al., 1990; Kaneene et al., 2006). Recent
studies of slaughterhouse surveillance in Ireland have
found significant differences in rates of detection of
infected cattle between slaughterhouses that cannot be
explained by difference in the characteristics of the animals
slaughtered (Frankena et al., 2007; Olea-Popelka et al.,
2012).
The post-mortem inspection procedure of cattle is
described in the Fresh Meat (Hygiene and Inspection
Regulations) (as amended) SI 1995 No. 539. The procedure
consists of visual and manual examination of the carcass
and incisions of a defined range of tissues with further
visual examination. There is additional inspection where
a meat inspector or Official Veterinary Surgeon (OVS)
has any reason to suspect any part of the carcass or offal
is infected. Suspect lesions from all slaughtered cattle
from herds that were Officially Tuberculosis free (OTF) at
the time of slaughter are sent to the Animal Health and
Veterinary Laboratories Agency (AHVLA, formerly the
Veterinary Laboratories Agency) for histopathological
investigation and culture to determine whether M. bovis
can be isolated. Movement restrictions are placed on
the herd and if infection with M. bovis is confirmed the
restrictions on a herd are extended and the cattle in the
herd will undergo rounds of tuberculin skin testing until
infection is cleared which is defined by two negative tests
on all animals in the herd, that are required to be tested,
conducted 60 days apart.
The aim of this study was to investigate the herd and
cattle factors associated with the proportion of lesions con-
firmed as positive for infection with M. bovis amongst those
that were identified as suspect during routine post-mortem
inspection of cattle from OTF herds in GB between 2003 and
2008. The standard for confirmatory evidence was bacterial
isolation of M. bovis as required by the European Directive
(EU Council Directive 64/623/EEC, as amended).
2. Materials and methods
2.1. The dataset
The study used data for the period of 2003–2008
extracted from several databases including the Cattle Trac-
ing System (CTS), bTB surveillance data from the AHVLA
(Formerly Animal Health) recorded on VetNet, and the
AHVLA TB Culture System database. Each database pro-
vided a unique and different set of data for the study. For
example, VetNet data provided bTB test results, CTS pro-
vided the slaughterhouse locations where each bovine was
slaughtered and inspected for infection, and the TB culture
system provided the M. bovis culture results. Data from the
different databases were linked by animal eartags which
are unique identifiers of each bovine.
Potential risk factors for confirmation of M. bovis infec-
tion investigated in the analysis were slaughterhouse, year
of slaughter, season of slaughter and lesion submission,
region (West, North, and East England, Wales or Scotland)
in which the farm of origin of the animal was located, herd
type (dairy, beef or mixed), herd size, parish testing inter-
val (PTI) (the testing interval for routine field surveillance
using the single intradermal comparative cervical tuber-
culin (SICCT)) test, in the herd of origin of the slaughtered
animal, which ranges between 1 and 4 years depending
on local bTB incidence), number of reactors on the ani-
mal’s farm of origin in the most recent TB incident in the
last 4 years (if any), number of previous breakdowns (con-
firmed and unconfirmed) on the animal’s farm of origin in
the last 4 years (if any), time since last bTB test conducted
on the herd, reason for last bTB test of herd (e.g. routine
test for surveillance, pre-movement test, etc.) and age of
animal at slaughter. These factors were selected primar-
ily as being possibly associated with the prevalence of bTB
in the farm of origin of the slaughtered animal which could
affect the probability of detection of lesions at post mortem
and the probability that the suspect lesions were confirmed
as infected. The aim of the analysis was to identify fac-
tors associated with confirmation of M. bovis infection in
lesions detected during post-mortem inspection in slaugh-
terhouses.
 A. Shittu et al. / Preventive Veterinary Medicine 110 (2013) 395–404
2.2. Animal and herd identifications
The population sample included in the study was sub-
missions of tissue with macroscopic lesions typical of M.
bovis infection, from 3663 cattle from 151 GB slaughter-
houses between 2003 and 2008. The submissions of tissue
with lesions where M. bovis infection was suspected had
been excised during post-mortem inspection of slaugh-
tered cattle. During the study period, GB cattle herds were
tested using the single intradermal comparative cervical
tuberculin (SICCT) test, for evidence of infection with M.
bovis every 1, 2, 3 or 4 years. SICCT testing frequency
is determined by EU Council Directive 64/632/EEC (as
amended) and is based upon a calculation of historical inci-
dence of herd breakdowns with post-mortem evidence of
bTB infection within a defined area. All study submissions
were from cattle from herds that had tested negative to
the SICCT test and were OTF. Where there were submis-
sions from more than one animal with lesions from the
same herd at the same time from a slaughterhouse, the
submission included in the population sample was from
the first animal from a herd in the slaughterhouse recorded
on VetNet as having a bTB-like lesion.
2.3. Detection of lesions and laboratory confirmations
All confirmations were based on bacteriological isola-
tion of M. bovis. The bacterial culture technique followed an
AHVLA standard operating procedure. After a preparatory
phase where the tissue samples were homogenised using a
stomacher the resulting inoculum was sown onto 12 diag-
nostic solid-media slopes that were then incubated at 37 ◦ C.
The slopes were examined after six weeks for evidence of
growth of mycobacteria. M. bovis was discriminated from
M. avium and other mycobacterium by characteristic pro-
files. Histopathology is also conducted and positive results
are reported to the farmer. However, if the histopathology
result is negative, the farmer still has to wait for a negative
culture result before cattle movement restrictions on the
herd can be lifted.
2.4. Data management and statistical analyses
The dataset was stored as a Microsoft® Office Excel
for Windows 2007 spreadsheet which was used to iden-
tify errors and missing values through observation of the
descriptive data. Data were transferred for further analyses
to StataIC 10® (StataCorp, Lakeway Drive, College Station,
TX, USA). In the first step, descriptive analyses including
cross-tabulations of proportions of lesions confirmed by
possible explanatory variables and Chi-square tests were
conducted to identify variables that might be of value for
further investigation.
In the logistic regression modelling, confirmation of
bTB infection (confirmed/unconfirmed) was the dependent
variable and the predictors identified during the descrip-
tive analyses were tested in the model one at a time.
Initially a fixed effect model was built and various candi-
date models evaluated using the Hosmer and Lemeshow
goodness-of-fit (GOF) test (Hosmer et al., 1991), and the
Akaike Information Criterion (AIC). The Likelihood ratio test
397
(LRT) was also used to evaluate the difference between
nested models. A correlation matrix of all independent
variables was constructed after modelling to check that
all correlation coefficients were ≤0.8. Next slaughterhouse
was incorporated as a random effect to address poten-
tial data clustering at the slaughterhouse level and to
address omitted slaughterhouse specific covariates that
could affect detection rates. Variables that were significant
predictors in the fixed effect model but omitted because
their inclusion did not improve model fit were re-examined
in the random effect model. The effect of the incorporation
of farm of origin of cattle as a random effect term was also
investigated. Model fit in the random effects modelling was
assessed using the AIC and LRTs. All modelling was con-
ducted using logistic procedures and xtlogit in Stata except
for the final extraction of population averaged probabilities
implied by the random intercept for slaughterhouse. The
latter were obtained by averaging subject specific proba-
bilities over the random intercept distribution using the
marginal post estimation command after rerunning the
model using the gllamm procedure. Results from the fixed
effect models are not shown, as it was decided that the
hierarchical model was the most appropriate due to clus-
tering of detection and confirmation in cattle sent to the
same slaughterhouse.
Choropleth maps showing confirmation by county of
animal’s origin were constructed using ArcGIS 9.2 by ESRI,
Redlands, CA, USA. Both crude and “smoothed” confirma-
tion rates were mapped; the latter to adjust for extreme
variation caused by small numbers of observations within
an area rather than genuine geographical heterogeneity.
Smoothed rates were calculated using shrinkage estima-
tors based on the variance of the crude rate and the overall
regional rate for the region in which the county was situ-
ated (Marshall, 1991).
3. Results
Between 2003 and 2008, the overall throughput of cattle
through slaughterhouses in GB was approximately 14 mil-
lion with approximately 2.3 million animals slaughtered
per year. Throughput varied by region with the North of
England having the highest (36.60%), followed by Scotland
(24.14%) and the West (23.46%) then the East (8.45%) and
Wales (7.36%).
The study sample included results from submissions of
suspect lesions from cattle of ages ranging from 6 months
to 19 years, submitted throughout the years between 2003
and 2008. A minimum of 1 and a maximum of 289 tis-
sues with suspect lesions were submitted by each of the
151 slaughterhouses between 2003 and 2008 (mean = 24,
median = 4, IQR = 16). A total of 3663 tissues were submit-
ted over the 6-year period from unique animals originating
from 2799 OTF herds. The cattle were from beef (2170,
60.90%), dairy (881, 24.73%), or mixed (612, 14.37%) herds
subject to varying routine bTB field testing frequencies.
Table 1 shows the number and proportion of samples
with lesions submitted by slaughterhouses from OTF herds
aggregated by year of animal’s slaughter. The number of
suspect tissues submitted for testing increased by 200%
between 2003 and 2008.
398 A. Shittu et al. / Preventive Veterinary Medicine 110 (2013) 395–404

Table 1 testing (see Table 3) and the proportion confirmed infected

Number of officially tuberculosis free herds where at least one suspect
almost double that in herds subject to quadrennial testing.
lesion was detected in a bovine at post mortem in a slaughterhouse
and submitted for confirmation by culture of M. bovis, Great Britain
Herds of origin had had between 0 and 6 and 0 and 3 con-
(2003–2008). firmed and unconfirmed breakdowns respectively in the
previous 4 years and the number of reactors detected in
Year Total submission Proportion of all submissions (%)
the most recent breakdown (if any) on the farm, ranged
2003 323 8.82 between 0 and 210. Confirmation increased with the num-
2004 401 10.95
ber of previous bTB incident in the herds and number of
2005 576 15.72
2006 669 18.26 reactors in most recent incident within the last 4 years.
2007 740 20.20 Age at slaughter showed a positively skewed distri-
2008 954 26.04 bution with a long tail in the positive direction and age
Total 3663 100.00 categories were selected to provide a reasonable represen-
tation of the distribution of the data. The age at slaughter
ranged from less than 6 months to 19.4 years with a
Of the 3663 tissues with lesions submitted during the median age at slaughter of 2.5 years. Sixty-seven per-
study period, 2470 (67.43%) were confirmed as culture pos- cent of the bovine animals were slaughtered before age
itive for infection with M. bovis. Of the lesions confirmed 3 years, with forty-five percent slaughtered between ages
as negative for M. bovis (n = 1193), 14 were positive for M. 2 and 3 years. In the descriptive analyses, age was nega-
avium, 1 was positive for M. kansasii, 6 were unclassified tively associated with the probability that the lesion from
mycobacterium, 7 were without a final result, 1161 were the animal would be confirmed as infected with lesions
negative to all and in 4 cases the sample was contaminated. from older animals were least likely to be confirmed (see
The proportion of lesions from which M. bovis that could Table 3). However, age was negatively correlated with
be isolated varied by slaughterhouse: Of the 151 slaugh- year of birth (Spearman’s rho = −0.72, p < 0.001) with older
terhouses that submitted suspect samples 55 (36.42%), 29 animals more likely to have been born in earlier years.
(19.21%), 26 (17.22%), and 41 (27.15%) had confirmation Animal born in earlier years were less likely to have
rates of between 76% and 100%, 51% and 75%, 26% and 50%, lesions confirmed as infected. The proportion of lesions
and 0% and 25%, respectively. Table 2 shows results from confirmed in animals born between 2006 and 2007 was
21 slaughterhouses where suspect lesions from at least 50 over double the proportion confirmed in animals born 1986
cattle were identified and submitted for culture over the to 1989.
study time period. Both the reason for the last bTB test on the herd of ori-
The probability that a lesion would be confirmed as gin of the bovine and the time since the last herd test were
infected with M. bovis increased over the time period and associated with lesion confirmation in the slaughterhouse.
varied geographically with the highest proportions con- The longer the length of time since the last test on the herd
firmed in the west of GB (see Table 3 and Fig. 1). The number the lower the probability that a lesion from a bovine from
of lesions submitted was greatest in herds subject to annual the herd would be confirmed. However, the time interval

Table 2
Confirmation rates for M. bovis infection for slaughterhouses where 50 or more cattlea from attested herds with lesions typical of bTB were identified
(2003–2008).

Slaughterhouse Throughput Total submissionsb M. bovis positive Confirmation by culture(%) 95%CI

1 233,278 65 25 38.46 26.31; 50.61

2 60,361 54 27 50.00 36.08; 63.92
3 89,939 92 51 55.43 45.09; 65.78
4 155,521 198 115 58.08 51.15; 65.01
5 188,328 66 40 60.61 48.50; 72.71
6 211,518 287 176 61.32 55.66; 66.99
7 114,426 63 39 61.90 49.58; 74.23
8 36,165 55 36 65.45 52.48; 78.43
9 303,665 180 121 67.22 60.29; 74.15
10 549,432 180 126 70.00 63.24; 76.76
11 172,194 177 127 71.75 64.01; 78.44
12 97,708 77 57 74.03 64.01; 84.04
13 372,176 58 43 74.14 62.52; 85.75
14 517,962 289 219 75.78 70.81; 80.67
15 168,288 107 82 76.64 68.49; 84.78
16 446,790 265 208 78.49 73.51; 83.47
17 103,825 68 54 79.41 69.55; 89.27
18 52,874 57 46 80.70 70.14; 91.27
19 188,399 135 116 85.93 79.98; 91.87
20 154,191 71 62 87.32 79.39; 95.25
21 444,220 171 157 91.81 87.66; 95.96

Total 4,661,260 2715 1927 70.78 69.68; 73.11

a
Only one submission per OTF herd per herd incident was included in the database.
b
Tissues with macroscopic lesions typical of M. bovis infection from OTF herds.
 A. Shittu et al. / Preventive Veterinary Medicine 110 (2013) 395–404 399

Table 3
Categorical variables tested in univariable analyses of risk factors for confirmation of infection by culture of M. bovis in suspect lesionsa in cattle from
officially tuberculosis free (OTF) herds submitted by slaughterhouses in Great Britain (2003–2008).

Type of explanatory variable Level Total submissions M. bovis positive Confirmation by culture (%) P-value for difference

Year of slaughter of bovine 2003 323 218 67.49 0.004

2004 401 275 68.58
2005 576 366 63.54
2006 669 431 64.42
2007 740 491 66.35
2008 954 689 72.22
Total 3663 2.470 67.43
Region East 280 144 51.43 <0.001
North 487 257 52.77
Scotland 153 45 29.41
Wales 584 347 59.42
West 2159 1677 77.67
Total 3663 2470 67.43
Herd type Mixed 612 418 68.3 0.288
Beef 2170 1477 68.06
Dairy 881 575 65.27
Total 3663 2470 67.43
Herd size (animals) 0–56 722 508 70.36 0.295
57–123 709 491 69.25
124–210 712 476 66.85
211–254 712 466 65.45
>354 713 486 68.16
Total 3568 (95 missing) 2427 68.02
Testing interval 1 year 2242 1744 77.79 <0.001
2 years 561 352 62.75
3 years 28 17 60.71
4 years 831 356 42.84
Total 3662 (1 missing) 2469 67.42
Number of reactors in most 0 2233 1407 63.01 <0.001
recent previous bTB incident 1 344 244 70.93
2 154 113 73.38
3 100 77 77.00
4 82 67 81.71
>5 536 440 82.09
Total 3449 (214 missing) 2348 68.08
Age (years) 0–3 2373 1706 71.89 <0.001
4–7 564 360 63.83
8–12 422 256 60.66
13–15 171 75 43.86
16–19 19 9 47.37
Total 3549 (114 missing) 2406 67.79
Previous confirmedb bTB 0 2248 1397 62.14 <0.001
incident (2003–2008) 1 848 657 77.48
≥2 353 294 83.29
Total 3449 (214 missing) 2348 68.08
Previous unconfirmedb bTB 0 3005 2028 67.49 0.053
incident (2003–2008) 1 444 320 72.07
Total 3449 (214 missing) 2348 68.08
Season Autumn 1162 800 68.85 0.538
Spring 775 509 65.68
Summer 793 533 67.21
Winter 933 628 67.31
Total 3663 2470 67.43
Last herd testc VE-WHT 790 557 70.51 <0.001
VE-RHT 815 392 48.10
VE-SI 501 409 81.64
VE-6M 432 323 74.77
VE-12M 270 196 72.59
VE-CON 382 280 73.30
VE-CT 225 159 70.67
VE-IFN 17 15 88.24
Other 17 17 100.00
Total 3449 (214 missing) 2348 68.08
Days since last herd test for bTB 0– 691 487 70.48 <0.001
(quintiles) 79– 690 493 71.45
181– 689 504 73.15
289 691 463 67.00
561 688 401 58.28
3449 (214 missing) 2348 68.08
400 A. Shittu et al. / Preventive Veterinary Medicine 110 (2013) 395–404

Table 3 (Continued)

Type of explanatory variable Level Total submissions M. bovis positive Confirmation by culture (%) P-value for difference

Year of birth of bovine 1986–1989 21 7 33.33 <0.001

1990–1994 303 139 45.87
1995–1999 663 434 65.46
2000–2005 2060 1456 70.68
2006–2007 502 370 73.71
Total 3549 (114 missing) 2406 67.79
a
Tissues with macroscopic lesions typical of M. bovis infection from OFT herds.
b
A confirmed incident indicates infection detected in the herd due to a bovine reactor to a field test for bTB confirmed by observation of macroscopic
lesions typical of bTB in the slaughterhouse and/or culture of the M. bovis bacterium from tissues form a bovine animal in the herd suspected of being
infected.
c
Single Intradermal Comparative Cevical Tuberculin tests (SICCT) except VE-IFN. VE-WHT: Whole herd test (generally conducted in a bTB high incidence
endemic areas, VE-RHT: Routine herd test (generally conducted in lower incidence non bTB endemic areas, VE-SI: Short interval test conducted during
a bTB incident in a herd, VE-6M and VE-12M: 6 month check and 12 month check tests conducted after a herd has undergone a bTB incident and has
regained Officially Tuberculosis free status, VE-CON: Contiguous test conducted on a herd on land contiguous to land with another herd that is undergoing
a bTB incident, VE-CT: Check test conducted on a herd that is thought to have higher epidemiological risk for bTB, VE-IFN: Interferon gamma blood test
conducted as an ancillary test after a SICCT.

between last bTB SICCT test and slaughter was associated
with PTI (p < 0.001). Almost 55% of animals from herds with
a PTI of 4 years (characteristic of areas where bTB is not
endemic), were in the quintile for the longest time inter-
val between date of last bTB test in the herd and date of
slaughter.
Predictors for confirmation in the multivariable random
effects logistic model with the best fit to the data (GOF
X2 = 10.09, d.f. = 8 and p = 0.259) included region of the farm
of origin of the animal, bTB PTI, number of reactors in most
recent previous bTB incident on the farm of origin, year of
animal’s slaughter, date of birth of the animal and slaugh-
terhouse incorporated as a random intercept (see Table 4).
The effects from year of birth and age were indistinguish-
able and the former was retained in the final model because
the distribution of the variable was more normal than age.
Fig. 2 shows how the probability of confirmation of infec-
tion increased with year of birth of the bovine. The time
interval between the last herd test for bTB and the date the
animal was slaughtered and type of test conducted were
significant predictors in the multivariable logistic model
but were not retained in the final model because their

Fig. 1. Choropleth map showing the distribution of the confirmation rate of lesions submitted by slaughterhouses aggregated by county of animal’s origin
in Great Britain. Counties are shown using 1974 county boundaries. Smoothed rates could not be calculated where there were fewer than two submissions.
 A. Shittu et al. / Preventive Veterinary Medicine 110 (2013) 395–404 401

Table 4
Final multivariable modela of potential risk factors for confirmations of lesions typical of BTB submitted by slaughterhouses in Great Britain (2003–2008).

Variable Level OR (adjusted) P-valueb 95%CI

Region of farm of origin of bovine Scotland Reference <0.001

East 2.35 1.24; 4.47
North 1.88 1.02; 3.47
Wales 2.14 1.13; 4.04
West 4.33 2.32; 8.11

Parish testing interval for farm of origin 4 years Reference <0.001

1 year 3.14 2.40; 4.10
2 years 1.97 1.48; 2.62
3 years 1.40 0.60; 3.27

Year of slaughter of bovine 2003 Reference <0.001

2004 1.23 0.85; 1.79
2005 0.86 0.61; 1.22
2006 0.68 0.49; 0.97
2007 0.60 0.43; 0.86
2008 0.73 0.51; 1.03

Number of reactors in most recent previous None Reference 0.009

breakdown on farm in last 4 years 1 0.95 0.71; 1.26
2 0.92 0.61; 1.37
3 1.22 0.72; 2.05
4 1.48 0.79; 2.75
5 or more 1.63 1.24; 2.14

Year of birth of bovine 1.00 0.001 1.00; 1.00

a
Random effects logistic regression model with 3337 observations and random intercept term for 146 slaughterhouses.
b
For difference in effect between levels of parameter.

inclusion adversely affected model fit and had little influ-
ence on other coefficients. Intra-slaughterhouse correla-
tion of effects was 0.085 and there was significant variation
in confirmation by slaughterhouse (LRT p < 0.001). The vari-
ance of the random intercept over the whole population
sample of slaughterhouses was 0.32 and the predicted pop-
ulation averaged probabilities for confirmation by slaugh-
terhouse ranged from 0.14 to 0.90. Fig. 3 shows the distri-
bution of the predicted probabilities for confirmation for
slaughterhouses that had submitted at least 50 lesions. In
this sample, the slaughterhouse with the lowest predicted
confirmation rates (median 21%) was based in Scotland
and all suspect lesions were from cattle from herds in
Scotland except two. The slaughterhouse with the highest
predicted confirmation rates (median 81%) was based in
the west of England and all suspect lesions were from
cattle from herds in the west except 13.

Fig. 3. Box plots showing distribution of predicted probability of con-

Fig. 2. Probability of confirmation of infection with M. bovis by date of
birth and Parish bovine tuberculosis testing interval of farm of origin
of bovine with suspect lesion. Lowess smoothers derived from random
effects logistic model controlling for region of farm of origin of bovine
with suspect lesion, parish bovine tuberculosis testing interval for farm
of origin, number of reactors in the most recent bTB incident on the farm
in the last 4 years, year of birth of bovine and with a random intercept for
slaughterhouse.
firmation of infection with M. bovis in suspect lesions submitted by
slaughterhouses who had submitted at least 50 suspect lesions in GB,
2003–2008. From random effects logistic model controlling for region of
farm of origin of bovine with suspect lesion, parish bovine tuberculosis
testing interval for farm of origin, number of reactors in the most recent
bTB incident on the farm in the last 4 years, year of birth of bovine and with
a random intercept for slaughterhouse. Numbers to the left of the figure
are the same numerical identifiers for slaughterhouses listed in Table 2.
402 A. Shittu et al. / Preventive Veterinary Medicine 110 (2013) 395–404
4. Discussion
Overall, the study showed that M. bovis was isolated
from 67% of the suspect lesions identified during slaugh-
terhouse inspection of cattle from OTF herds between 2005
and 2008 in GB. Studies in Ireland have reported similar
confirmation rates (Dolan et al., 1993; Frankena et al., 2007)
with 67% estimated most recently for the 2005–2007 time
period (Olea-Popelka et al., 2012). Confirmation was high-
est in lesions from cattle from regions of GB where bTB
is known to be endemic and from herds with a history of
bTB incidents. After adjustment for region of farm of origin
of the animal slaughtered and other confounding factors,
confirmation rates differed significantly between slaugh-
terhouses indicating that that there may be factors related
to inspection procedures or operation of slaughterhouses
that affect the probability of detecting M. bovis infection in
cattle. Detection procedures may be less specific in some
slaughterhouses resulting in a higher probability of false
positive lesions submitted.
The number of submissions of suspect lesions for lab-
oratory culture varied between slaughterhouses, and was
likely to be related to the probability of infection in the
herds of origin of the cattle sent to slaughter and the size
and throughput of the slaughterhouse among other factors.
Clustering of submissions in time related to herd of origin
was addressed by only including one submission from the
slaughterhouse from animals from same herd with suspect
lesions detected at the same time point. The submission
included in the analysis was the first submission recorded
on the surveillance database at the same time point from
the same herd in the same slaughterhouse.
There was a threefold increase in submissions of suspect
lesions from cattle in OTF herds as a result of slaugh-
terhouse surveillance over the time period. At the same
time the incidence of bTB in GB cattle herds as a result
of routine field tests as well slaughterhouse surveillance
increased. Cognizance of the bTB epidemic in cattle may
have influenced detection rates. However, the probability
of confirmation of infection with M. bovis in the samples
also increased implying that the rise in slaughterhouse sub-
missions reflects either the increasing spread of bTB and/or
better detection but cannot be attributed to increased num-
bers of false positive submissions.
The geographical distribution of bTB in GB is heteroge-
neous with highest incidence and prevalence to the west
of GB and Wales. The spatial analysis showed that the pro-
portions of suspect lesions confirmed culture positive for
M. bovis were higher in cattle from areas where bTB is
endemic (southwest of Britain) compared to other regions.
This could be anticipated since the predictive probability
of lesions from cattle in areas with a higher prevalence of
bTB will be higher than for lesions from cattle areas where
bTB is less prevalent. Crude confirmation rates from suspect
lesions were however unexpectedly high in the counties of
Cleveland and Kent in the north and east of GB, respec-
tively where bTB is not endemic. However there was only
one submission of a suspect lesion originating from a herd
in Cleveland and four submissions from Kent between 2003
and 2008 and the high rates are most likely to be artefacts
of the small numbers of submissions. The map of smoothed
rates adjusted for some of the heterogeneity by shrinking
disease rates towards the overall rate for the region based
on the assumption that risk factors and hence disease rates
are likely to be geographically correlated (Marshall, 1991).
If herds were infected at random, the probability that
an animal with suspect lesions is detected should increase
with time since the last herd test for bTB (Goodchild and
Clifton-Hadley, 2001). We attempted to address the effect
of time interval between tests by controlling for routine
testing frequency (PTI). However, in our analysis, the prob-
ability of bTB confirmation decreased with time. A possible
reason is the fact that testing frequency is much lower in
the non-endemic areas of GB (every four years as opposed
to annually) and the lower force of infection may counter-
act the effect of time interval between tests. Inclusion of
time since last bTB test as a parameter in the model did not
improve model fit.
Preliminary analyses suggested that confirmation rates
for M. bovis infection decreased with increasing age of
the slaughtered bovine animal. However further investiga-
tion suggested that this association could be confounded
by date of birth. Since the prevalence of bTB increased
between the 1986 (the earliest year of birth in the study
sample) and 2007 (latest year of birth) animals born later
would have had a higher probability of being exposed
to bTB than animals of equivalent ages born earlier. This
means that the probability of infection being confirmed
in suspect lesions (i.e. the predictive value of slaughter-
house detection), would be higher in animals born later.
However, the age distribution of animal born later will be
younger than in animals born earlier, because of censoring
of the sample at 2008. Therefore, any increase in the risk
of bTB with age in the cohort born later will not be repre-
sented in the analysis. A higher probability of submission of
post-mortem tissues from older animals has been reported
elsewhere (Frankena et al., 2007) but the relation between
age and the probability of confirmation was not reported.
The study found significant differences in confirmation
rates between slaughterhouses that could not be explained
by differences in the prevalence of disease in the areas
where the slaughtered cattle originated. Similar findings
have been reported in studies comparing confirmation
rates across slaughterhouses in Ireland (Frankena et al.,
2007; Olea-Popelka et al., 2012). Efficiency in the detection
of macroscopic lesions typical of infection at post mortem
may be influenced by factory characteristics such as speed
of the slaughter line, lighting, etc. but will also be influenced
by the thoroughness of the examination, training and expe-
rience of the meat inspectors (McIlroy et al., 1986; Corner
et al., 1990; Collins, 1997; Lees, 2004; Frankena et al., 2007).
Pathological signs can be difficult to detect particularly
when an animal is in the early stage of infection and a
number of studies have shown that exhaustive inspection
procedures reduce the possibility of failing to detect incip-
ient lesions (De Kantor et al., 1987; Corner, 1994; Whipple
et al., 1996). Systems can be developed to maintain and
foster improvement of meat inspection (Toma et al., 1999;
Doherr and Audige, 2001; Enoe et al., 2003) and ensure that
the sensitivity of inspection is at par or above the minimum
acceptable standards, within a bTB control or eradication
plan (Anonymous, 1992; Koller-Jones, 2002; Anonymous,
 A. Shittu et al. / Preventive Veterinary Medicine 110 (2013) 395–404
2002). Work in the USA has demonstrated that assignment
of target submission rates and performance awards can
also improve bTB detection (Kaneene et al., 2006).
While one hundred percent confirmation of infections
would have been desirable, it may be indicative of a highly
specific bTB test which is generally accompanied by a loss in
sensitivity and hence missed infection. A balance needs to
be struck between the proportion of cattle correctly iden-
tified as infected and the proportion incorrectly identified
(false positives) as a result of inspection procedures. This
balance is likely to vary with the degree of endemicity
in the area of origin of the cattle. More sensitive detec-
tion procedures may be particularly important in low bTB
prevalence areas where infected animals represent a new
incursion into an area or the tail end of a successful erad-
ication campaign (Corner, 1994). Within herd spread will
occur if bTB is not detected and removed and may be a
particular problem in areas where routine skin testing is
relatively infrequent. An outbreak of bTB due to insuffi-
cient routine meat inspection was reported by Fischer et al.
(2005) in the Netherlands as a result of an economic opti-
misation analysis of six surveillance strategies. Increases
in the incidence of bTB amongst unrestricted herds have
been associated with reduction in cattle testing elsewhere
(Kaneene and Pfeiffer, 2006).
A number of researchers have reported that keeping
dairy cattle (Asseged et al., 2000; Miliano-Suazo et al.,
2000; Radostits et al., 2000; Olea-Popelka et al., 2008) and
the size of the herd (Green and Cornell, 2005; Reilly and
Courtenay, 2007; Pollock and Keeling, 2009) increases the
probability of infection and disease propagation. Morris
et al. (1994), Hancox (2000) and Bernard et al. (2005) also
reported a higher prevalence of bTB in dairy husbandry
than in other management systems. This study did not
identify differences in confirmation associated with herd
management type or the number of cattle in the herd. How-
ever, this could be due to the fact that his study is based on
an analysis of one suspect lesion per herd which may not
be indicative of the level of infection within the herd.
This study represents a snapshot of post-mortem detec-
tion of M. bovis infection confirmation rates with culture in
cattle from OTF herds for the period 2003–2008. Slaughter-
house inspection plays an important role in the detection of
bTB in GB both in high incidence and low incidence areas of
the disease. The differences in confirmation rates observed
between slaughterhouses could not be explained by dif-
ferences in background prevalence of bTB alone and the
study suggests that investigation of factors related to the
operation of slaughterhouses and inspection procedures
may highlight areas that could improve the efficiency of
detection of bTB. Given that there generally is a trade-off
between sensitivity and specificity of any diagnostic test,
ideally the sensitivity of detection in the slaughterhouse
should be tailored to enhance the ongoing disease control
and surveillance programs.
Acknowledgements
We would like to acknowledge Josh Mawdesley for
assistance with compiling the datasets with all the
403
variables we requested and Robin Sayers for comments on
the modelling of random effects.
References
Anonymous, 1992. Assessment of Risk Factors for Mycobacterium bovis
in the United States. USAD: APHIS: VS, Centers for Epidemiology and
Animal Health, Fort Collins.
Anonymous, 2002. National Granuloma Submission Program Annual
Report 2001/02. http://www.aahc.com.au
Asseged, B., Lubke-Becker, A., Lemma, E., Taddele, K., Britton, S., 2000.
Bovine TB: a cross-sectional and epidemiological study in and around
Addis Ababa. Bull. Anim. Health Prod. Africa 48, 71–80.
Bernard, F., Vincent, C., Mathhieu, L., David, R., James, D., 2005. Tuberculo-
sis and brucellosis prevalence survey on dairy cattle in Mbarara milk
basin (Uganda). Prev. Vet. Med. 67, 267–281.
Caffrey, J.P., 1994. Status of bovine tuberculosis eradication programmes
in Europe. Vet. Microbiol. 40, 1–4.
Clifton-Hadley, R.S., Wilesmith, J.W., 1995. An epidemiological outlook
on bovine tuberculosis in the developed world. In: Griffin, F., de
Lisle, G. (Eds.), Proceedings of the Second International Conference
on Mycobacterium bovis. University of Otago Press, Otago, Dunedin,
New Zealand.
Collins, J.D., 1997. Meat Plant Surveillance and Its Role in the Eradica-
tion of Tuberculosis in Cattle. In selected papers 1996. Tuberculosis
Investigation Unit, University College Dublin, Dublin55–59.
Corner, L.A., Melville, L., McCubbin, K., Small, K.J., McCormick, B.S., Wood,
P.R., Rothel, J.S., 1990. Efficiency of inspection procedures for the
detection of tuberculous lesions in cattle. Aust. Vet. J. 67, 389–392.
Corner, L.A., 1994. Post mortem diagnosis of Mycobacterium bovis infection
in cattle. Vet. Microbiol. 40, 53–63.
Cousins, D.V., Corner, L.A., Tolson, J.W., Jones, S.L., Wood, P.R., 1998. Erad-
ication of Bovine Tuberculosis from Australia: Key Management and
Technical Aspects. CSL Limited, Parkville, Australia45.
Cousins, D.V., 2001. Mycobacterium bovis infection and control in domestic
livestock. Rev. Sci. Technol. 20, 71–85.
Defra (2012) GB National statistics: 1998–2010. http://www.defra.gov.uk/
statistics/foodfarm/landuselivestock/cattletb/national/ (accessed
12.04.12).
Defra (2011) Bovine TB Eradication Programme for England. July 2011.
http://www.defra.gov.uk/publications/files/pb13601-bovinetb-
eradication-programme-110719.pdf (accessed 20.06.12).
De Kantor, I.N., Nader, A., Bernardelli, A., Giron, D.O., Man, E., 1987. Tuber-
culosis infection in cattle not detected by slaughterhouse inspection.
J. Vet. Med. 34 (3), 202–205.
Doherr, M.G., Audige, L., 2001. Monitoring and surveillance for rare health-
related events: a review from the veterinary perspective. Philos. Trans.
R. Soc. B 356, 1097–1106.
Dolan, L.A., Costello, E., Quigley, F., 1993. Confirmed Tuberculous Lesions
in Non-tuberculin-reactor Cattle Slaughtered at Export Licensed Meat
Plants in the Republic of Ireland (1988 to 1993). Centre for Veterinary
Epidemiology and Risk Analysis (CVERA) Selected Reports.
Enoe, C., Christensen, G., Andersen, S., Willeberg, P., 2003. The need for
built-in validation of surveillance data so that changes in diagnostic
performance of post-mortem meat inspection can be detected. Prev.
Vet. Med. 57, 117–125.
Fischer, E.A., van Roermund, H.J., Hemerik, L., van Asseldonk, M.A., de Jong,
M.C., 2005. Evaluation of surveillance strategies for bovine tuberculo-
sis (Mycobacterium bovis) using an individual based epidemiological
model. Prev. Vet. Med. 67, 283–301.
Frankena, K., White, P.W., O’Keeffe, J., Costello, E., Martin, S.W., van
Grevenhof, I., More, S.J., 2007. Quantification of the relative efficiency
of factory surveillance in the disclosure of tuberculosis lesions in
attested Irish cattle. Vet. Rec. 161, 679–684.
Goodchild, A.V., Clifton-Hadley, R.S., 2001. Cattle-to-cattle transmission
of Mycobacterium bovis. Tuberculosis 81 (1/2), 23–41.
Green, L., Cornell, S., 2005. Investigations of cattle herd breakdowns with
bovine tuberculosis in four counties of England and Wales using VET-
NET data. Prev. Vet. Med. 70 (3–4), 293–311.
Hancox, M., 2000. Cattle TB schemes: control or eradication?: a critical
appraisal. Appl. Microbiol. 31, 87–93.
Hosmer, D.W., Taber, S., Lemeshow, S., 1991. The importance of assessing
the fit of logistic regression models: a case study. Am. J. Public Health
81, 1630–1635.
House of Commons (2008). Badgers and cattle TB: the final report of the
Independent Scientific Group on Cattle TB. Environment, Food and
Rural Affairs Committee, Fourth Report of Session 2007–2008. HMSO
Norwich, UK.
404 A. Shittu et al. / Preventive Veterinary Medicine 110 (2013) 395–404
Kaneene, J.B., Miller, R., Meyer, R.M., 2006. Abattoir surveillance: the US
experience. Vet. Microbiol. 112, 273–282.
Kaneene, J.B., Pfeiffer, D., 2006. Epidemiology of Mycobacterium bovis. In:
Theon, C.O., Steele, J.H. (Eds.), Mycobacterium bovis Infection in Ani-
mals and Humans. Iowa State Press, Ames, p. 34, 35, 37 & 38.
Koller-Jones, M. (2002). Bovine Tuberculosis Eradication Program in
Canada, November, 2003. Available at: http://www.inspection.gc.ca
Lees, V.W., 2004. Learning from outbreaks of bovine tuberculosis near Rid-
ing Mountain National Park: applications to a foreign animal disease
outbreak. Can. Vet. J. 45, 28–34.
Marshall, R.J., 1991. Mapping disease and mortality rates using empirical
bayes estimators. Appl. Stat., 283–294.
Miliano-Suazo, F., Salmar, M.D., Ramirez, C., Payeur, J.B., Rhyan, J.C., San-
tillan, M., 2000. Identification of TB in cattle slaughtered in Mexico.
Am. J. Vet. Res. 61, 86–89.
McIlroy, S.G., Neill, S.D., McCracken, R.M., 1986. Pulmonary lesions and
Mycobacterium bovis excretion from the respiratory tract of tuberculin
reacting cattle. Vet. Rec. 118, 718–721.
Meah, M.N., Lewis, G.A., 1999. A Review of Veterinary Surveillance in
England and Wales: The Report to the Permanent Secretary. Ministry
of Agriculture, Fisheries and Food, p. 37.
Morris, R.S., Pfeiffer, D.U., Jackson, R., 1994. The epidemiology of Mycobac-
terium bovis infections. Vet. Microbiol. 40, 153–177.
O’Keeffe, J.J., White, P.W., 1999. An Overview of Herd Restriction Episodes
Initiated by the Disclosure of a Tuberculous Lesion in Cattle from Clear
Herds During 1989 to 1997. In Selected Papers 1998. Tuberculosis
Investigation Unit, University College Dublin, Dublin44–50.
Olea-Popelka, F.J., Costello, E., White, P., McGrath, G., Collins, J.D., O’Keeffe,
J., Kelton, D.F., Berke, O., More, S., Martin, S.W., 2008. Risk factors for
disclosure of additional tuberculous cattle in attested-clear herds that
had one animal with a confirmed lesion of tuberculosis at slaughter
during 2003 in Ireland. Prev. Vet. Med. 85, 81–91.
Olea-Popelka, F., Freeman, Z., White, P., Costello, E., O’Keeffe, J., Frankena,
K., Martin, W., More, S., 2012. Relative effectiveness of Irish factories in
the surveillance of slaughtered cattle for visible lesions of tuberculosis,
2005–2007. Irish Vet. J. 65, 2.
Pollock, E.B., Keeling, M., 2009. Herd size and bovine tuberculo-
sis persistence in cattle farms in Great Britain. Prev. Vet. Med.,
http://dx.doi.org/10.1016/j.prevetmed.2009.08.022.
Radostits, O.M., Gay, C.C., Blood, D.C., Hinchcliff, K.W., 2000. In: Deborah,
R. (Ed.), A Text Book of the Diseases of Cattle, Sheep, Pigs, Goats and
Horses. , ninth ed. Harcourt, London, pp. 909–919.
Reilly, L., Courtenay, O., 2007. Husbandry practices, badger sett density
and habitat composition as risk factors for transient and persistent
bovine tuberculosis on UK cattle farms. Prev. Vet. Med. 80 (2–3),
129–142.
Theon, C.O., Barletta, R.G., 2004. Mycobacterium. In: Gyles, C.L., Prescot,
J.F., Songer, J.G., Theon, R.G. (Eds.), Pathogenesis of Bacterial Infections
in Animals. Blackwell Publishing, Ames, I.A, pp. 69–76.
Toma, B., Dufour, B., Sanaa, M., Benet, J.J., Moutou, F., Louza, A., Ellis, P.,
Shaw, A., 1999. Applied Veterinary Epidemiology and the Control of
Disease in Populations. AEEMA, Maisons-Alfort, France, 536 pp.
Whipple, D.L., Bolin, C.A., Miller, J.M., 1996. Distribution of lesions in cattle
infected with M. bovis. J. Vet. Diagn. Invest. 8 (3), 351–354.