Preventive Veterinary Medicine 80 (2007) 129–142

www.elsevier.com/locate/prevetmed

Husbandry practices, badger sett density and

habitat composition as risk factors for transient
and persistent bovine tuberculosis
on UK cattle farms
L.A. Reilly, O. Courtenay *
Ecology & Epidemiology Group, Department of Biological Sciences, University of Warwick,
Coventry CV4 7AL, UK
Received 28 April 2006; received in revised form 5 February 2007; accepted 6 February 2007

Abstract

Bovine tuberculosis (bTB) is a persistent problem in cattle herds in Great Britain and Ireland.
Farm management and cattle husbandry practices can influence the risk of transmission of bTB and
hence the likelihood of bTB breakdown (1 reactor to the tuberculin skin test). Biological differences
are expected in the transmission dynamics, and hence risk factors for bTB breakdown, on farms
where infection persists in the herd compared to farms where infection is more sporadic or short-
lived.
Comparative case–control studies were performed to test farm management practices as potential
risk factors for transient (under breakdown restrictions for 6 months) and persistent (under
breakdown restrictions for >6 months) bTB breakdown over 5 years (1995–1999) on 179 and
171 UK cattle farms, respectively. Farms were characterised for badger sett density and farm habitat
composition by ground survey, farmers were questioned retrospectively on management practices,
and cases and controls were identified from national tuberculin test records.
Controlling for routine tuberculin testing interval, log-transformed herd size, regional location,
badger sett density and farm habitat complexity, multivariable logistic regression identified increased
odds of both transient and persistent breakdown on farms that bought-in cows (odds ratio (OR)  4.9;
95% confidence interval (CI)  1.1;22.8). In addition, the purchase of >50 head of cattle (OR = 4.0,
95% CI = 1.0;16.0) and the storage of manure for 6 months (OR = 4.4; 95% CI = 1.3;15.4) were
risk factors for transient breakdown, whereas the use of silage clamps (OR = 9.1; 95% CI = 2.0;40.8)

* Corresponding author. Tel.: +44 24 7652 4550; fax: +44 24 7652 4619.
E-mail address: orin.courtenay@warwick.ac.uk (O. Courtenay).

0167-5877/$ – see front matter # 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.prevetmed.2007.02.002
130 L.A. Reilly, O. Courtenay / Preventive Veterinary Medicined 80 (2007) 129–142

increased the risk of persistent breakdown. Decreased odds of both transient and persistent break-
down were associated with higher stocking densities (>3 cattle/ha) (OR  0.2; 95% CI  0.1;0.9),
and running mixed herd enterprises compared to only beef or dairy (OR = 0.1; 95% CI = 0.0;0.7) was
an additional protective factor against persistent breakdown. In these analyses, the covariates log herd
size and tuberculin testing interval were significant predictors of both transient and persistent
breakdown, whereas active badger sett density and regional location only affected the risk of
persistent breakdown.
The collective results suggest that the probability of transient breakdown is most strongly
influenced by the purchase of cattle over other management variables and covariates, whereas
the probability of persistent breakdown appears to be mostly affected by management factors relating
to type of herd enterprise and silage storage in addition to the relative density of badgers. Implications
for bTB management are discussed.
# 2007 Elsevier B.V. All rights reserved.

Keywords: Bovine tuberculosis; Mycobacterium bovis; Case–control study; Risk factor; Epidemiology; Cattle;
Farm management; Badger

1. Introduction

Bovine tuberculosis (bTB) is a persistent economic and veterinary problem in cattle

herds in the UK despite the policy of cattle herd testing and slaughter (Defra, 2005a).
Badgers (Meles meles) are implicated as an important wildlife reservoir and an obstacle to
reducing bTB incidence (Krebs et al., 1997; Donnelly et al., 2005; Griffin et al., 2005;
Woodroffe et al., 2006). There is general consensus that cattle husbandry and farm
management practices also contribute to heightened risk of herd breakdown (1 animal
reacting to the tuberculin skin test) (Krebs et al., 1997; Phillips et al., 2000, 2003; Johnston
et al., 2005). The perceived risks relate to (i) environmental biosecurity including farm
waste management and foodstuff storage (Scanlon and Quinn, 2000; Garnett et al., 2002),
(ii) contact between cattle and badgers or their excreta at pasture (White et al., 1993;
Hutchings and Harris, 1997; Courtenay et al., 2006), (iii) herd characteristics such as herd
size, farm enterprise and animal movements (Griffin et al., 1996; Marangon et al., 1998;
Green and Cornell, 2005; Johnston et al., 2005; Gilbert et al., 2005; Carrique-Mas et al.,
2005; Gopal et al., 2006), and (iv) animal management including stocking densities,
feeding and grazing regimes (Benham and Broom, 1989; Scantelbury et al., 2004).
Preventive husbandry measures aimed to reduce some of these associated risks are advised
(Phillips et al., 2000).
In the UK, routine ‘‘whole-herd’’ tuberculin skin tests are performed at intervals of 1, 2,
3 or 4 years depending on bTB incidence within the parish. Upon the discovery of a skin
test reactor, the animal is culled and infection classified as confirmed on detection of visible
lesions at post-mortem and/or Mycobacterium bovis growth in cultures of lymph node
material (Corner, 1994). The remaining herd is placed under movement restriction until
one (if the breakdown is unconfirmed) or two consecutive (if the breakdown is confirmed)
‘‘whole-herd’’ 60-day tuberculin skin tests prove negative. Additional reactors disclosed at
follow-up 60-day check tests result in delays of months to years in lifting movement
restrictions, with clear consequences to farm production.
 L.A. Reilly, O. Courtenay / Preventive Veterinary Medicined 80 (2007) 129–142 131

Cattle movement is thought to be one of the principle routes of transmission leading to

herd breakdown, however in regions where a wildlife reservoir exists, herd breakdown is
also attributed to this additional pathogen source, and thus management practices that
facilitate this transmission pathway may differ on farms that experience persistent
compared to those that experience transient breakdowns. Spatial cluster analysis (Kulldorf,
1997) of all bTB-tested farms in the UK between 1995 and 1999 prior to this study (Reilly
and Courtenay, unpublished data), demonstrated that farms under prolonged breakdown
restriction (>6 months during this period) were clustered, and significantly more so than all
breakdown farms. This indicates a common source of exposure to M. bovis and/or local
environmental factors influencing the likelihood of persistent versus transient breakdown.
This supports the logic behind the introduction of the highly sensitive gamma-interferon
test alongside the tuberculin test in parishes on 3- and 4-yearly testing intervals, aimed to
prevent bTB becoming endemic in the local wildlife (Anon., 2006).
Few studies have been conducted in the British Isles to investigate the simultaneous
associations of husbandry practices, badger activity and habitat composition on the
longevity of bTB movement restrictions or severity of breakdown incidence (Griffin et al.,
1993; Denny and Wilesmith, 1999). We have shown from related studies of cattle farms in
bTB endemic regions of Britain that farms with greater herd incidence had higher densities
of active badger setts, and that both bTB herd incidence and sett density were significantly
influenced by farm habitat composition (Courtenay et al., 2006, in preparation).
Here we report the results of two case–control studies performed on those farms to
identify cattle husbandry practices that, in addition to the effects of badger density and
habitat characteristics, significantly alter the likelihood of bTB breakdown. Second, we
identify any differences in the husbandry practices that are associated with herds that suffer
persistent breakdowns compared to those that experience transient breakdowns.

2. Materials and methods

2.1. Farm selection and case definition

The ratio of bTB breakdown herds to tested herds from 1995 to 1999 was obtained from
the national herd testing records (VetNet supplied by Defra) and interpolated in a GIS to
define adjacent regions of high (hotspot) and low bTB incidence, onto which sampling
transects measuring 10 km  3 km were purposively positioned to maximise the incidence
gradient of bTB along the transect length as described in a related study (Courtenay et al.,
in preparation). From each sampling transect, cattle farms were identified that fulfilled the
criteria that (i) they had 10 head of cattle between 1995 and 1999, (ii) a routine ‘‘whole-
herd’’ test was performed on at least two occasions during this period, (iii) correct farm
contact details were available in the database, and that (iv) the farm’s geo-reference in the
database represented the working farm location.
Two sets of case farms were recruited from this selection. Farms that had been under
breakdown restrictions for up to 6 months between 1995 and 1999 were classed as transient
breakdown cases, and farms that had been under breakdown restrictions for more than 6
months between 1995 and 1999 were classed as persistent breakdown case farms. The
132 L.A. Reilly, O. Courtenay / Preventive Veterinary Medicined 80 (2007) 129–142

Table 1
Numbers of transient bTB case, persistent bTB case, and control farms in each of ten UK study regions, and the
percentages on an annual routine testing interval
Region Transient case Persistent case Control farms
farmsa farmsb
N % tested N % tested N % tested
annually annually annually
Cornwall 13 92 8 88 29 52
Derbyshire 0 – 7 100 14 64
N. Gloucestershire 4 100 5 100 9 78
S. Gloucestershire 6 100 3 100 13 92
Herefordshire 3 100 8 100 10 100
Monmouthshire 11 91 7 100 18 56
Pembrokeshire 8 100 1 100 8 88
Somerset 4 75 2 50 6 33
E. Sussex 4 25 3 67 4 75
Wiltshire 5 100 6 100 10 90
Total 58 90 50 94 121 69
a
6 months under movement restriction during the study period 1995–1999.
b
>6 months under movement restriction during the study period 1995–1999.

farms that had not experienced a breakdown during this period (nor subsequently up to
2004) were identified as the control group for both sets of case farms. Farms on a routine
parish testing interval of >2 years were excluded from the analyses, as were farms that had
undergone major changes in farm management during the study period. In total, 229 farms
met all the above conditions (Table 1).

2.2. Data collection

Between March 2000 and February 2003 (with an interruption of 10 months in 2001 due
to the foot and mouth disease epidemic), trained personnel obtained information on cattle
husbandry practices during the 5-year period 1995–1999 inclusive by interviewing
farmers. Interviews included open and closed questions concerning herd demographics,
indoor and outdoor management of cattle, and storage of feed and manure. Variables were
selected on the basis of a prior literature review of published bTB risk factors (see
references above). The farm boundaries indicated by farmers were marked onto large scale
paper Ordnance Survey maps and subsequently digitised in a GIS (MapInfo v.6.0, Mapinfo
Cooperation, 1992–2000). Farms comprised of one to seven (median 1) separate land
fragments; a fragment was defined as an area used by the farm (e.g. for cattle grazing or
growing crops) that was separated from the main farm site by at least one field not utilised
by that farm. Data for all farm fragments were aggregated.

2.3. Badger density estimation

With informed consent, active badger setts within the farm were located by ground
survey of all field boundaries and patches of woodland and scrub by trained personnel
between 2000 and 2003, and setts classified as active or inactive as described in a related
 L.A. Reilly, O. Courtenay / Preventive Veterinary Medicined 80 (2007) 129–142 133

study (Courtenay et al., in preparation). Active setts were mapped and densities per km2 per
farm calculated. Whilst it is acknowledged that badger activity at the time of ground survey
may have differed to the period over which husbandry practises were assessed (1995–
1999), we found no evidence by analysis of the badger removal operations records (VetNet)
to indicate potential bias in badger abundance on these farms prior to, or during, the two
time periods.

2.4. Farm habitat characterisation

Farm habitat maps were created by ground survey to classify and delineate the
vegetation into five categories: (i) grassland, (ii) woodland, (iii) arable land, (iv) scrub
(including bracken and gorse) and (v) marsh or moorland, as described in a related study
(Courtenay et al., in preparation). Briefly, the area and proportion of each habitat type on a
farm was estimated to the nearest hectare by overlaying a 100 m  100 m grid onto
digitised maps of the farm using a GIS (Mapinfo v.6.0). Each grid square was assigned a
vegetation type code that represented the largest area within the hectare, and habitat areas
totalled across all farm fragments.

2.5. Explanatory variables

The explanatory variables investigated in this study are listed in Table 2. Continuous
explanatory variables (number of cattle purchased, stocking density, field size, and number
of land fragments) were each converted to three categorical groups of approximately equal
sample sizes. Ages of purchased animals were categorised as calves (stores or replacement
breeding animals <12 m), yearlings (stores or replacement breeding animal-
s 12 m < 24 m) or cows (female breeding animals 24 m). Where purchased animals
per farm comprised >1 age-class, the maximum age-class was entered into the analysis,
based on the observation that bTB prevalence tends to increase with age (Benham, 1985
cited in Phillips et al., 2000). Categories of herd type considered here included dairy only,
beef (beef suckler or store/finisher cattle), and mixed (dairy plus beef). All other variables
were treated as binary data. ‘Fed in field’ refers to cattle that were fed any foodstuff in the
field, regardless of whether or not they were also fed in the shed or yard when housed.
Variables relating to food type apply to all feeding locations. Average stocking densities
per farm over the 5 years were calculated relative to the area of grassland. Average field
size was calculated by dividing the total farm area by the total number of fields. Few farms
practiced a sole regime of strip-grazing, hence in analysis, strip-grazing could include set-
stocking for a portion of the herd or for a proportion of the year.

2.6. Covariates

A base model was constructed including covariates considered likely to influence the
probability of bTB breakdown, including log-transformed annual mean herd size, parish
testing interval and regional location. Additional covariates entered into the base model
were active badger sett density, farm habitat composition and habitat variety, which each
showed significant associations with herd incidence in a related study on the same farms
134 L.A. Reilly, O. Courtenay / Preventive Veterinary Medicined 80 (2007) 129–142

Table 2
Results of logistic regression analyses showing the effects of farm management variables on the risk of transient
vs. persistent bTB herd breakdown for UK cattle herds
Explanatory factors Transient TBa Persistent TBb
Odds ratio N cases: Odds ratio N cases:
(95% CI) controls (95% CI) controls

Age-class purchased (max)

None 1 16:47 1 13:47
Calves 1.1 (0.3;4.0) 7:23 1.4 (0.3;6.0) 9:23
Yearlings 2.2 (0.6;7.5) 12:17 1.9 (0.5;7.2) 10:17
Cows 10.8 (2.3;50.6)** 11:11 5.1 (1.2;22.4)* 12:11
Bull hire
No 1 44:83 1 39:83
Yes 1.2 (0.5;3.2) 14:38 0.4 (0.1;1.2) 11:38
Concentrates fed
No 1 9:17 1 7:17
Yes 0.4 (0.1 1.2)* 41:97 0.7 (0.1;3.2) 39:97
Farm fragments
1 1 31:67 1 30:67
2 1.3 (0.5;3.4) 16:29 0.5 (0.1;1.6) 8:29
>2 1.0 (0.3;3.2) 11:25 1.0 (0.3;3.7) 12:25
Fed in field
No 1 11:21 1 10:21
Yes 1.1 (0.4;3.1) 47:100 0.7 (0.2;2.4) 40:100
Field size (ha)
2.4 1 19:48 1 10:48
>2.4, 3.7 0.7 (0.2;2.1) 15:39 1.2 (0.3;5.0) 17:39
>3.7 1.5 (0.4;5.1) 24:34 1.0 (0.2;5.1) 23:34
Grass silage fed
No 1 2:17 1 0:17
Yes 6.6 (1.0;42.6)* 55:104 – 50:104
Hay fed
No 1 20:27 1 19:27
Yes 0.5 (0.2;1.3) 36:94 0.4 (0.1;1.1)* 31:94
Herd type
Beef 1 26:66 1 24:66
Dairy 0.7 (0.2;2.4) 11:19 1.5 (0.4;6.1) 14:19
Mixed 0.5 (0.2;1.5) 21:36 0.3 (0.1;1.1)* 12:36
Manure storage
<6 m 1 20:58 1 17:58
6 m 2.5 (1.1;5.9)* 36:59 1.4 (0.5;3.7) 31:59
Maize silage fed
No 1 38:106 1 28:106
Yes 1.6 (0.5;4.7) 20:15 3.9 (1.1;13.1)* 22:15
Minerals fed
No 1 14:33 1 11:33
 L.A. Reilly, O. Courtenay / Preventive Veterinary Medicined 80 (2007) 129–142 135

Table 2 (Continued )
Explanatory factors Transient TBa Persistent TBb
Odds ratio N cases: Odds ratio N cases:
(95% CI) controls (95% CI) controls
Yes 0.9 (0.4;2.3) 44:87 1.3 (0.4;3.8) 39:87
Molasses fed
No 1 38:97 1 34:97
Yes 2.4 (0.9;6.2)* 20:24 1.4 (0.5;3.9) 16:24
Number of cattle purchased
0 1 17:47 1 13:47
1–50 1.8 (0.7;5.2) 16:43 1.6 (0.5;5.1) 16:43
>50 3.0 (0.9;9.5)* 19:25 2.4 (0.7;8.4) 17:25
Silage clamp
No 1 21:75 1 13:75
Yes 1.6 (0.6;4.4) 37:46 4.7 (1.5;14.1)** 37:46
Stock density (cattle/ha)
0.8 1 16:38 1 18:38
>0.8, 3.0 0.5 (0.2;1.5) 23:45 0.5 (0.2;1.7) 19:45
>3.0 0.2 (0.1;1.8)* 19:38 0.2 (0.1;0.8)* 13:38
Strip-grazing regime
No 1 31:75 1 27:75
Yes 1.1 (0.5;2.5) 27:46 1.9 (0.6;5.6) 23:46
All analyses control for covariates: log mean annual herd size, routine testing interval, regional location, active
badger sett density, farm habitat composition and habitat variety. Significance levels: *P  0.10; **P  0.01;
***P  0.001.
a
6 months under movement restriction during the study period 1995–1999.
b
>6 months under movement restriction during the study period 1995–1999.

(Courtenay et al., in preparation). To maintain consistency with that study, active sett
density was entered as a categorical variable: zero, intermediate (>0–3 setts km 2) and
high (>3 setts km 2), and farms were grouped according to their habitat complexity, the
baseline group including farms that consisted of grassland only and subsequent groups
formed by adding one or more habitat type (defined above) in all possible combinations. A
semi-quantitative score of habitat variety was calculated as the sum number (1–5) of the
habitat types recorded on the farm.

2.7. Statistical procedures

Analysis of potential farm management risk factors was performed in a two-step

process. First, explanatory variables were added in turn to the base model and analysed
using logistic regression to identify those that altered the odds of bTB breakdown with
90% probability (a  0.10) whilst controlling for the covariates. The significant
variables were then entered into a full multivariable logistic regression model, where the
minimal adequate model (MAM) was achieved by backward step-wise maximum-
likelihood estimation until only the base model plus variables that were significant at
a  0.05 remained. A Spearman’s rank correlation matrix of all ordinal or binary
136 L.A. Reilly, O. Courtenay / Preventive Veterinary Medicined 80 (2007) 129–142

independent variables that were to be entered into the multivariable models was constructed
prior to multivariable analysis and all coefficients were 0.54. The goodness of fit of the
models was assessed using the Hosmer–Lemeshow statistic, which compares observed versus
predicted counts of outcome events (breakdowns). Likelihood ratio tests were used to assess
the improvement in model fit attained by including the significant explanatory variables
compared to just the base model. The predictive power of the models was evaluated using
post-estimation commands, which following logistic regression calculates the probability of
each farm being a case farm. All analyses were performed in Stata (StataCorp, 2003).

3. Results

3.1. Farm sample

The study of transient herd breakdowns included 58 case and 121 control farms, and the
study of persistent herd breakdowns included 50 case farms and the same 121 controls
(Table 1). The percentage of farms that were on an annual routine testing regime were 90%
(transient breakdown), 94% (persistent breakdown), and 69% (control). The transient
breakdown farms were under movement restrictions for a median of 4 months (range 2–6 m),
which was significantly less than the median restriction time of 10 months (range 7–28 m) for
the persistent breakdown farms (Wilcoxon rank-sum test: z = 8.97, P < 0.001). Transient
breakdown farms underwent a median of five independent whole-herd tests (range 2–9),
persistent breakdown farms a median of four (range 2–9), and control farms a median of three
(range 2–6) independent tests, during the 5-year study period. Thirty seven (63%) of the 59
transient breakdown case farms and 47 (92%) of the 51 persistent breakdown case farms had
at least one of their breakdowns confirmed at post-mortem or by cultivation of M. bovis.

3.2. Screening of risk factors

In analyses adding each explanatory variable in turn to the base model, purchasing cows
(compared to no purchases) over the 5-year study period was independently associated
with increased odds of transient and persistent breakdown, whilst having a high (>3 cattle/
ha) stocking density was protective against both severities of breakdown (Table 2).
Additional increased risk of transient breakdown was associated with purchasing >50
cattle (compared to no purchases), the provision of grass silage to cattle and the storage of
manure for at least 6 months (compared to <6 m). The feeding of concentrates to cattle was
protective against transient breakdown. The feeding of maize silage and the use of a silage
clamp on the farm increased the risk of persistent breakdown but not transient breakdown.
Farming a mixed herd (compared to beef only) and the provision of hay to cattle were each
protective against persistent breakdown (Table 2).

3.3. Minimum adequate models

Controlling for all potential covariates in multivariable analyses (Table 3), the
significant risk factors for transient breakdown were buying in cows (compared to purchase
 L.A. Reilly, O. Courtenay / Preventive Veterinary Medicined 80 (2007) 129–142 137

Table 3
Multivariable logistic regression models showing the effects of farm management variables on the risk of transient
vs. persistent bTB breakdown in UK cattle herds
Explanatory variables Transient TBa Persistent TBb
Odds ratio (95% CI) Odds ratio (95% CI)
Age-class purchased (max)
All other 1 1
Cows 22.5 (4.0;124.9)*** 4.9 (1.1;22.8)*
Herd type
Dairy or beef 1
Mixed 0.1 (0.0;0.7)*
Manure storage
<6 m 1
6 m 4.4 (1.3;15.4)*
Number of cattle purchased
50 1
>50 4.0 (1.0;16.0)*
Silage clamp
No 1
Yes 9.1 (2.0;40.8)**
Stock density
3 1 1
>3 0.2 (0.1;0.8)* 0.2 (0.0;0.9)*

Log herd sizec 5.0 (1.6;15.7)** 5.2 (1.4;18.6)*

Testing intervalc 0.1 (0.0;0.6)* 0.1 (0.0;0.4)**
Regional locationc
Region 1 1 1
All others NS 0.01–0.2d
Active badger sett densityc
0 1 1
>0–3 km 2 NS NS
>3 km 2 NS 9.6 (1.5;61.2)*
Habitat composition c NS NS
Habitat varietyc NS NS
Statistics are from the minimum adequate model including only significant variables and all forced covariates.
Significance levels: *P  0.05; **P  0.01; ***P  0.001.
a
6 months under movement restriction during the study period 1995–1999.
b
>6 months under movement restriction during the study period 1995–1999.
c
Covariates forced into all analyses.
d
Seven of nine study regions had significantly different odds of persistent breakdown compared to region 1.

of other age-groups or none), the purchase of >50 (compared to 50) cattle during the
study period and storing manure for at least 6 months. Farms with relatively high stocking
densities (3 head of cattle/ha versus <3 head of cattle/ha) were protected against
transient breakdown. Buying in cows and the use of a silage clamp on the farm were
each shown to be risk factors for persistent breakdown in the final multivariable model. In
138 L.A. Reilly, O. Courtenay / Preventive Veterinary Medicined 80 (2007) 129–142

contrast, mixed herds (compared to beef or dairy only), and farms with relatively high
stocking densities (3 head of cattle/ha versus <3 head of cattle/ha), each protected
against persistent breakdown.
Some qualitative differences in the covariates between the two multivariable models
were observed. Farms with a relatively high active badger sett density were at increased
risk of persistent but not transient breakdown, and regional location was a significant factor
in the risk of persistent breakdown but not transient breakdown.

3.4. Model validation

The Hosmer–Lemeshow statistic showed that the number of cases and controls in each
subgroup predicted by the models for both transient and persistent breakdowns were not
significantly different from those observed (x2 = 5.16, df = 8, P = 0.74 and x2 = 5.11,
df = 8, P = 0.75, respectively). The fit of the base models were significantly improved by
adding the farm management explanatory variables to predict the likelihood of both
transient and persistent breakdowns (LR x2 = 27.3, df = 4, P < 0.001 and LR x2 = 26.8,
df = 4, P < 0.001, respectively). The predictive performance of the model for transient
breakdown was maximised at a cut-off probability of 0.40, at which the sensitivity and
specificity to correctly classify cases and controls were 78% and 77%, respectively,
whereas the equivalent cut-off, sensitivity and specificity of the model for persistent
breakdown were 0.29, 80% and 80%, respectively.

4. Discussion

This study identified husbandry related factors that significantly alter the risk of
breakdown when analysed in conjunction with non-husbandry related environmental
factors (badger sett density and habitat composition) that are known to be associated with
bTB outcome on these farms (Courtenay et al., in preparation). Some similarities and some
differences in the risk factors associated with persistent versus transient breakdown were
also revealed. The adjusted odds ratios from multivariable analyses indicate that variables
associated with cattle purchase were the most important risk factors for transient
breakdown, whereas the risk of persistent breakdown was additionally associated with herd
type, method of silage storage, higher badger sett densities and regional location. Some of
the significant factors identified (e.g. manure storage >6 m, mixed herd enterprise, and
higher stocking density) are counter-intuitively related to increased breakdown probability,
nevertheless interesting parallels with studies elsewhere add biological integrity to the
general results of this study.
Two published studies have addressed the additive risks associated with farm
environment and farm management. Griffin et al. (1993) performed a matched case–
control study in the Republic of Ireland (RoI) to evaluate a broad range of farm
environment and management variables to explain persistent breakdown, defined there as
farms put under movement restrictions at least twice, or for a continuous period of more
than 12 months, during a 4.5-year study period. They concluded that neither the purchase
nor the frequency of cattle purchase, or likely contact with contiguous herds, were
 L.A. Reilly, O. Courtenay / Preventive Veterinary Medicined 80 (2007) 129–142 139

significant risk factors, whereas the presence of at least one badger sett on the farm was
significantly associated with persistent breakdown. Unfortunately, that study did not
compare the effects of these variables on persistent versus transient breakdown events. In
the second, Denny and Wilesmith (1999) evaluated the odds of confirmed breakdown
versus no breakdown in dairy herds in Northern Ireland over 3 years. Controlling for
confounders, they attributed 40% of breakdowns to both the presence of contiguous herds
with a contemporary confirmed breakdown, and to badger presence (sett/carcass). That
study purposefully excluded records of purchased animals (to eliminate ‘‘away’’
infections), and included few herd management measures to compare to the results of this
study.
More recently, cattle movement onto farms from markets or farm sales has been shown
to increase the risk of breakdown amongst herds within the Randomised Badger Control
Trial areas in the UK (Johnston et al., 2005), and there is evidence that bought-in cattle are
the source of the majority of breakdowns in regions previously free of the disease (Gopal
et al., 2006). Considering the origin of purchased animals, Carrique-Mas et al. (2005)
demonstrated that animals purchased from high (versus low) bTB incidence counties in the
UK increased the hazard of herd breakdown, consolidating the conclusions of modelling
studies (Gilbert et al., 2005), that cattle movements from bTB endemic areas were
significant predictors of bTB occurrence. In the present study, we did not have access to the
cattle movement records to test for such relationships, nor consequently could we
differentiate between breakdowns associated with previous exposure on source farms
versus exposure on the study farms. If the majority of transient breakdowns actually
belonged to the former category, we might expect cattle purchase variables to be more
likely risk factors of transient breakdown than persistent breakdown. Cruder analyses using
numbers and ages of replacement animals indicated that buying in cows increased the risk
of both transient and persistent breakdown when compared to farming a closed herd;
however the risk appeared to be much greater for transient breakdown (Table 3). Moreover,
the purchase of larger numbers of cattle (>50 compared to <50 head) was identified as a
risk factor for transient, but not persistent breakdown after adjusting for effects of the other
significant factors (Table 3). Older animals may have an increased likelihood of failing the
tuberculin test compared to young stock (Griffin et al., 1996) possibly due to greater
cumulative exposure with age or to age-related immunogenic factors. The purchase of bulls
was not tested as an explanatory factor in the current study as too few of the study farms
bought-in bulls; however we did not detect any significant effects relating to the hire of
bulls. In contrast, recurrent bTB was associated with bull purchase in RoI (Griffin et al.,
1993).
Badger presence on a farm, or proximity to badger setts, have previously been identified
as risk factors for bTB breakdown or incidence (Griffin et al., 1993; Martin et al., 1997;
Denny and Wilesmith, 1999; Courtenay et al., in preparation). In the present study, a
relatively high active sett density (>3 km 2) was associated with persistent breakdown in
the presence of significant management related variables, which adds to the growing
evidence that badgers play a part in maintaining the disease in hotspot areas. Routes of
transmission from badgers to cattle are likely to involve cattle foodstuffs and
environmental contamination with M. bovis in urine, faeces or sputum of excreting
badgers that are attracted to the food (Garnett et al., 2002), and we have previously shown
140 L.A. Reilly, O. Courtenay / Preventive Veterinary Medicined 80 (2007) 129–142

that soil from 41% of badger setts on sample farms was positive for M. bovis detected by
PCR (Courtenay et al., 2006). If this proves to be an important route of transmission, the
observed univariable increase in both transient and persistent breakdown on farms that feed
maize silage, grass silage or molasses to cattle may be explained by their attractiveness to
badgers. Badgers are known to cause significant damage to maize crops (Moore et al.,
1999), and to enter farmyards to forage cattle feed and silage in cattle troughs, feed sheds,
barns and cowsheds (Garnett et al., 2002). Silage clamps are likely to be more accessible to
foraging badgers than baled silage, which may be a reason for the increased risk of
persistent breakdown on farms with silage clamps. One hundred percent of the farms that
experienced persistent breakdown fed grass silage to cattle, and consequently a statistic
could not be assigned to this variable, whereas feeding hay was protective against
persistent breakdown at the screening stage. Griffin et al. (1993) reported that the quantity
of silage and concentrates fed per cow was higher, and quantity of hay fed was less, on
farms with chronic bTB, and hypothesised that providing greater quantities of higher
energy feed (i.e. silage as opposed to hay) may be an indirect measure of more intensive
management. We found no independent evidence of this in the current study; on the
contrary, feeding concentrates were protective against transient breakdown at the screening
stage, and higher stocking densities were shown to be protective against both types of
breakdown in multivariable analysis.
Greater herd size was associated with an increased risk of breakdown (transient and
persistent), as identified in studies elsewhere in the UK (Green and Cornell, 2005) and in
the RoI (Griffin et al., 1996). The interpretation of this relationship is complicated since a
greater number of animal tuberculin tests is likely to result in an increased probability of a
positive test due to <100% sensitivity of the test (Monaghan et al., 1994), and herd size
may be a proxy for other management related risk factors such as herd turnover rates,
stocking density, farm enterprise, and foodstuffs. Indeed we found positive correlations
between log herd size and stocking density (rs = 0.32), field size (rs = 0.40); presence of
dairy herd only (rs = 0.45) or mixed herd (rs = 0.38), feeding of maize silage (rs = 0.43) and
having a silage clamp (rs = 0.54).
The final multivariable models showed reasonably good predictive performance, with
the unexplained variation likely due to the existence of risk factors that were either not
measured or that were the result of complex interactions. Based on the farm sample sizes,
the study had a 90% power to detect an increase in univariable odds of 1.8 and a decrease of
0.7 with 95% confidence, given an exposure among controls of 39%, which was the mean
exposure for all tested variables. There are no standard definitions of transient and
persistent (or chronic) breakdown (Griffin et al., 1993), however for the purposes of this
study, classification of farms into breakdown type was based on statistical differences in
median time under breakdown restriction. Re-defining the bTB outcome variable from
unconfirmed to confirmed breakdowns (i.e. reactors with confirmation of clinical lesions or
positive culture growth), gave qualitatively similar results on re-analysis of the data (results
not shown). The 5-year study period was chosen as the minimum time period over which
persistent breakdowns could be identified in herds under 2-year routine testing interval, and
as the maximum period over which we envisaged farmers could recall (to the detail
required) any major changes in farm management which might confound the analysis.
There were few cases (n = 12) of the latter, and these farms were omitted from the study.
 L.A. Reilly, O. Courtenay / Preventive Veterinary Medicined 80 (2007) 129–142 141

The general findings of this study highlight the importance of risk factors already
considered under current management recommendations aimed to reduce bTB as
documented (Phillips et al., 2000; Defra, 2005a,b): the recently introduced pre-movement
testing is designed to prevent the spread of bTB infection in purchased cattle to non-
endemic areas. To reduce the risk of bTB from purchased cattle, farmers are advised to
manage closed herds, or to minimise the numbers purchased. Limiting contact between
cattle and infected badger sources by fencing, building maintenance and design, and/or
grazing and cropping management is also advised; this includes restricting badger access to
farm buildings, feed troughs and other cattle feed storage areas such as silage clamps.
There is some evidence to suggest that different management tools may be necessary to
reduce the risk of persistent bTB from those required to reduce the risk of any herd
breakdown. For farms that do not have a recent history of bTB breakdown, limiting
movements of cattle onto the farm or purchasing from non-endemic areas may be sufficient
to reduce the risk of infection, whereas for herds that experience chronic breakdown,
reducing the probability of re-infection from a wildlife or environmental reservoir may also
be required. The findings of this study support the anecdotal evidence that management of
persistent herd breakdown is likely to require intelligent control of badger infection (e.g. by
protective vaccine) to reduce the likelihood of (re)-introduction of infection into clean
herds.

Acknowledgements

We thank the farmers and landowners for their participation in the study: John Field,
Lindsay Mann and Stephen Wilson for coordination of fieldwork, and all the volunteer field
staff. Financial support was provided by the Department for Environment, Food and Rural
Affairs.

References

Anon., 2006. TB testing: gamma-interferon use to be increased. Vet. Rec. 158, 222.
Benham, P.F.J., Broom, D.M., 1989. Interactions between cattle and badgers at pasture with reference to bovine
tuberculosis transmission. Br. Vet. J. 145, 226–241.
Carrique-Mas, J.J., Medley, G.F., Green, L.E., 2005. Risk of bovine tuberculosis breakdowns in post-foot and
mouth disease restocked cattle herds in Great Britain. In: Proceedings of the SVEPM Conference, Nairn, pp.
27–41.
Corner, L.A., 1994. Post-mortem diagnosis of Mycobacterium bovis infection in cattle. Vet. Microbiol. 40, 53–56.
Courtenay, O., Reilly, L.A., Sweeney, F.P., Hibberd, V., Bryan, S., Ul-Hassan, A., Newman, C., Macdonald, D.W.,
Delahay, R.J., Wilson, G.J., Wellington, E.M.H., 2006. Is Mycobacterium bovis in the environment important
for the persistence of bovine tuberculosis? Biol. Lett. 2, 460–462.
Courtenay, O., Reilly, L.A., Woodroffe, R., in preparation. Inter- and intra-regional associations between badger
density, habitat complexity and bovine tuberculosis on UK farms.
Defra, 2005a. Government Strategic Framework for the Sustainable Control of Bovine Tuberculosis (bTB) in
Great Britain. Defra, London, UK.
Defra, 2005b. Veterinary Advice on Methods to Reduce the Likelihood of the Transmission of Bovine
Tuberculosis from Badgers to Cattle. Defra, London, UK.
142 L.A. Reilly, O. Courtenay / Preventive Veterinary Medicined 80 (2007) 129–142

Denny, G.O., Wilesmith, J.W., 1999. Bovine tuberculosis in Northern Ireland: a case–control study of herd risk
factors. Vet. Rec. 144, 305–310.
Donnelly, C.A., Woodroffe, R., Cox, D.R., Bourne, J.F., Cheeseman, C.L., Clifton-Hadley, R.S., Wei, G., Gettinby,
G., Gilks, P., Jenkins, H., Johnston, W.T., Le Fevre, A.M., McInerney, J.P., Morrison, W.I., 2005. Positive and
negative effects of widespread badger culling on tuberculosis in cattle. Nature 437, 1299–1320.
Garnett, B.T., Delahay, R.J., Roper, T.J., 2002. Use of cattle farm resources by badgers (Meles meles) and risk of
bovine tuberculosis (Mycobacterium bovis) transmission to cattle. Proc. R. Soc. Lond. B 269, 1487–1491.
Gilbert, M., Mitchell, A., Bourn, D., Mawdsley, J., Clifton-Hadley, R., Wint, W., 2005. Cattle movements and
bovine tuberculosis in Great Britain. Nature 435, 491–496.
Gopal, R., Goodchild, A., Hewinson, G., Domenech, de la Rua, Clifton-Hadley, R., 2006. Introduction of bovine
tuberculosis to north-east England by bought-in cattle. Vet. Rec. 159, 265–271.
Green, L.E., Cornell, S.J., 2005. Investigations of cattle herd breakdowns with bovine tuberculosis in four counties
of England and Wales using VETNET data. Prev. Vet. Med. 70, 293–311.
Griffin, J.M., Hahesy, T., Lynch, K., Salman, M.D., McCarthy, J., Hurley, T., 1993. The association of cattle
husbandry practices, environmental factors and farmer characteristics with the occurrence of chronic bovine
tuberculosis in dairy herds in the Republic of Ireland. Prev. Vet. Med. 17, 145–160.
Griffin, J.M., Martin, S.W., Thorburn, M.A., Eves, J.A., Hammond, R.F., 1996. A case–control study of selected
risk factors with the occurrence of bovine tuberculosis in the Republic of Ireland. Prev. Vet. Med. 27, 75–87.
Griffin, J.M., Williams, D.H., Kelly, G.E., Clegg, T.A., O’Boyle, I., Collins, J.D., More, S.J., 2005. The impact of
badger removal on the control of tuberculosis in cattle herds in Ireland. Prev. Vet. Med. 67, 237–266.
Hutchings, M.R., Harris, S., 1997. Effects of farm management practices on cattle grazing behaviour and the
potential for transmission of bovine tuberculosis from badgers to cattle. Vet. J. 153, 149–162.
Johnston, W.T., Gettinby, G., Cox, D.R., Donnelly, C.A., Bourne, J., Clifton-Hadley, R., Le Fevre, A.M.,
McInerney, J.P., Mitchell, A., Morrison, W.I., Woodroffe, R., 2005. Herd-level risk factors associated with
tuberculosis breakdowns among cattle herds in England before the 2001 foot-and-mouth disease epidemic.
Biol. Lett. 1, 53–56.
Krebs, J.R., Anderson, R., Clutton-Brock, T., Morrison, I., Young, D., Donnelly, C., 1997. Bovine Tuberculosis in
Cattle and Badgers. MAFF, London.
Kulldorf, M., 1997. A spatial scan statistic. Commun. Stat. Theory Meth. 26, 1481–1496.
Marangon, S., Martini, M., Pozza, M.D., Neto, J.F., 1998. A case–control study on bovine tuberculosis in the
Veneto Region (Italy). Prev. Vet. Med. 34, 87–95.
Martin, S.W., Eves, J.A., Dolan, L.A., Hammond, R.F., Griffin, J.M., Collins, J.D., Shoukri, M.M., 1997. The
association between the bovine tuberculosis status of herds in the East Offaly Project Area, and the distance to
badger setts, 1988–1993. Prev. Vet. Med. 31, 113–125.
Monaghan, M., Doherty, M., Collins, J., Kazda, J., Quinn, P., 1994. The tuberculin test. Vet. Microbiol. 40, 111–
124.
Moore, N., Whiterow, A., Kelly, P., Garthwaite, D., Bishop, J., Langton, S., Cheeseman, C., 1999. Survey of
badger Meles meles damage to agriculture in England and Wales. J. Appl. Ecol. 36, 974–988.
Phillips, C.J.C., Foster, C., Morris, P., Teverson, R., 2000. The Role of Cattle Husbandry in the Development of a
Suitable Policy to Control M. bovis Infection in Cattle. MAFF, London.
Phillips, C.J.C., Foster, C.R.W., Morris, P.A., Teverson, R., 2003. The transmission of Mycobacterum bovis
infection to cattle. Res. Vet. Sci. 74, 1–15.
Scanlon, M.P., Quinn, P.J., 2000. The survival of Mycobacterium bovis in sterilized cattle slurry and its relevance
to the persistence of this pathogen in the environment. Ir. Vet. J. 53, 412–415.
Scantelbury, M., Hutchings, M.R., Allcroft, D.J., Harris, S., 2004. Risk of disease from wildlife reservoirs:
badgers, cattle and bovine tuberculosis. J. Dairy Sci. 87, 330–339.
White, P.C.L., Brown, J.A., Harris, S., 1993. Badgers (Meles meles), cattle and bovine tuberculosis (Mycobacter-
ium bovis): a hypothesis to explain the influence of habitat on the risk of disease transmission in southwest
England. Proc. R. Soc. Lond. B 253, 277–284.
Woodroffe, R., Donnelly, C.A., Jenkins, H.E., Johnston, W.T., Cox, D.R., Bourne, F.J., Cheeseman, C.L., Delahay,
R.J., Clifton-Hadley, R.S., Gettinby, G., Gilks, P., Hewinson, R.G., McInerney, J.P., Morrison, W.I., 2006.
Culling and cattle controls influence tuberculosis risk for badgers. PNAS 103, 14713–14717.