The impact of animal introductions during herd restrictions on future herd-level bovine tuberculosis risk

Preventive Veterinary Medicine 109 (2013) 246–257

Contents lists available at SciVerse ScienceDirect

Preventive Veterinary Medicine journal homepage: www.elsevier.com/locate/prevetmed

The impact of animal introductions during herd restrictions on future herd-level bovine tuberculosis risk T.A. Clegg a,∗ , M. Blake b , R. Healy b , M. Good b , I.M. Higgins a , S.J. More a a Centre for Veterinary Epidemiology and Risk Analysis, UCD School of Agriculture, Food Science and Veterinary Medicine, University College Dublin, Belfield, Dublin 4, Ireland b Department of Agriculture, Food and the Marine, Agriculture House, Kildare St., Dublin 2, Ireland

a r t i c l e

i n f o

Article history: Received 23 February 2012 Received in revised form 10 October 2012 Accepted 14 October 2012 Keywords: Mycobacterium bovis Tuberculosis Animal movement Ireland

a b s t r a c t In Ireland new cases of bovine tuberculosis (bTB) are detected using both field (with the single intradermal comparative tuberculin test (SICTT)) and abattoir surveillance. Once a new case has been detected, herd restrictions, including restrictions on animal movements into and out of the herd, are implemented until the herd has passed two consecutive clear tests. While a herd is restricted, there may be several reasons why it may be desirable to introduce new stock, such as enabling routine management practices to continue ‘as near to normal’. In Ireland, introduction of animals during a bTB episode is permitted under specific conditions, with permission from the local veterinary office. The objectives of this study were (1) to provide an overview of movement events associated with each bTB episode, (2) to determine whether introduction of animals during a bTB episode is associated with increased future bTB risk and (3) to identify the practices relating to the introduction of animals that are the most risky. All herds that were not restricted at the start of 2006, but experienced a bTB episode during 2006 with 2 or more SICTT standard reactors (the eligible bTB episode) were included in the study. We calculated the number of extended eligible bTB episodes and subsequent bTB episodes that could be directly attributed to introduced animals. The main outcome of interest was the time from de-restriction of the eligible bTB episode to the start of a subsequent bTB episode or the date of the last test prior to the end of the study (31 December 2010). Cox proportional-hazard models were developed, each using a different introduction variable: introduced animals during an episode (yes/no), introduced animals prior to the first retest/first clear test, time from start of episode until first animals introduced and number of animals introduced during the episode. Only a small proportion of subsequent bTB episodes (1.8%) or extended eligible bTB episodes (2.7%) could be directly attributed to introduced animals. The results highlight an increased risk of a subsequent bTB episode among only a subset of herds that introduced animals during the eligible bTB episode. Specifically, herds that introduced animals early during the eligible bTB episode were at significantly greater future bTB risk than herds where animals were only introduced later. To illustrate, herds that introduced animals after the first retest did not have a significantly different risk compared to herds that did not introduce animals at all. In contrast, herds that did introduce animals prior to the first retest had 1.5 times higher risk of a subsequent bTB episode. Future practices concerning the introduction of animals during an episode now need to be reviewed. © 2012 Elsevier B.V. All rights reserved.

1. Introduction ∗ Corresponding author. Tel.: +353 1 716 6142; fax: +353 1 716 6147. E-mail address: [email protected] (T.A. Clegg). 0167-5877/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.prevetmed.2012.10.005

In Ireland, the national programme to control bovine tuberculosis (bTB) began in 1954. After initial success, the

T.A. Clegg et al. / Preventive Veterinary Medicine 109 (2013) 246–257

number of reactors in Ireland remained at a constant level of between 30,000 and 50,000 annually, before falling more recently to below 30,000 since 2002 and below 20,000 reactors in 2011. Every herd i.e. a distinct epidemiological unit, is tested at least annually using the single intradermal comparative tuberculin test (SICTT) and herds with one or more reactor animals are classified as bTB positive. Herds may also be classified as bTB positive following the discovery of a confirmed tuberculous lesion in an attested animal at slaughter (More and Good, 2006). Once a new case has been detected, herd restrictions, including restriction on animal movements into and out of the herd, are implemented until the herd has passed two consecutive clear SICTT tests (retests where no reactors were detected) at least 42-days apart, the second being at least 4-months after the last reactor left the herd. The protocol for the management of Mycobacterium bovis infected herds is described in the ‘Veterinary handbook for herd management in the bovine TB eradication programme’ (Good et al., 2010) and meets the terms of EU Directive 64/432/EEC and the Bovine Tuberculosis Order 1989–2007. In Ireland, introduction of animals during a bTB episode (the time period when movement restrictions are imposed on the herd) is permitted under specific conditions. There may be several reasons why the introduction of new stock may be desirable, such as replacing animals that were lost as a result of bTB, moving point-of-calving heifers from a rearing herd to a herd where milking facilities are available, and enabling routine management practices to continue ‘as near to normal’ despite the bTB restrictions imposed. Generally animals may not be moved into a restricted herd unless the herd owner has permission from the local veterinary office. The bTB eradication programme in Ireland is informed and underpinned by science, with scientific findings and analysis contributing to a broad range of areas, including management of infected herds. In view of the implications for farm income of restricting the movement of animals into a restricted herd and the absence of research on the risks posed by the introduction of animals during a bTB episode, it is appropriate that the nature and extent of the risk be examined. Several questions are relevant: Is this a risky practice? Should this practice continue? Are there times during a bTB episode when introducing animals is a more/or less risky practice? Would it be prudent to impose certain risk management measures, prior to the movement of animals into a restricted herd? Therefore, the objectives of this study were: Objective 1: • To provide an overview of events associated with each bTB episode, during which new animals were introduced into the herd, and • To clarify the infection status of animals that were introduced during bTB episodes. Objective 2: • To determine whether introducing animals during a bTB episode was associated with increased future herd-level bTB risk.

247

Objective 3: (on the proviso that there is evidence [from objective 2] of an increased bTB risk): • To identify the practices relating to the introduction of animals that increase future bTB risk and to determine whether the increased risk is associated with the source or herd into which the animals had been introduced. 2. Materials and methods 2.1. Background information 2.1.1. The study population All Irish herds that were not restricted at the start of 2006 but experienced a bTB episode during 2006 with 2 or more SICTT standard reactors were considered for inclusion in the study. An animal with an increase in skin thickness at the bovine site more than 4 mm greater than the increase at the avian site is classed as a standard reactor. Total reactors include standard reactors and any other animal deemed a reactor based on ancillary testing, epidemiological grounds or a more severe interpretation of the SICTT. A bTB episode commences with a bTB ‘breakdown’ event, associated with bTB detection (detected during either field or abattoir surveillance), and encompasses the subsequent period of trading restriction. The herd becomes ‘de-restricted’ (trading restrictions are lifted) following two consecutive clear tests at least 42-days apart, the second being at least 4months after the last reactor has left the herd. If a herd had more than one bTB episode in 2006, only the first was considered eligible in this study. We excluded all known and suspected dealer/feedlot herds, including: • herds with a herd size outside of the 10–90% herd size distribution for all eligible bTB episodes in the study. Herd size was taken as the mean herd size during the episode, • herds that sent >200 animals to slaughter in 2006, and • herds identified on the Animal Health Computer System (AHCS) as agents, dealers or feedlots. In summary, we observed a single ‘eligible bTB episode’ for each enrolled study herd. 2.1.2. Period and unit of interest The herd was the epidemiological unit of interest. Each study herd was observed throughout the study period, from the start of the eligible bTB episode in 2006 until the end of 2010. 2.1.3. Defining introduction status For all eligible bTB episodes, ‘introduction status’ was defined as: 1 = if cattle were introduced into the herd during the eligible bTB episode (that is, there was an introduction of one or more animals – excluding calves aged <42 days when introduced – from another herd during the eligible bTB episode), or 0 = if no cattle were introduced into the herd (except calves aged <42 days) during the eligible bTB episode.

248

T.A. Clegg et al. / Preventive Veterinary Medicine 109 (2013) 246–257

2.2. Overview of events associated with the introduction of animals during a bTB episode and the infection status of introduced animals 2.2.1. Extended or subsequent bTB episodes attributable to introduced animals We defined ‘extended eligible bTB episodes attributable to introduced animals’ as eligible bTB episodes where introduced animals were the only bTB reactors identified at one of the bTB tests during the episode. In addition, for each study herd during the study period, we calculated the number of ‘subsequent bTB episodes attributable to introduced animals’ (episodes with 1 or more reactor animals or an animal identified with a confirmed tuberculous lesion at slaughter), these being subsequent bTB episodes (following resolution of the eligible bTB episode) where all of the bTB positive animals at the first test of the subsequent bTB episode were animals introduced during the earlier eligible bTB episode. Fig. 1 illustrates two herd examples, highlighting an extended eligible bTB episode attributable to introduced animals and a subsequent bTB episode attributable to introduced animals. In both herds, animals were introduced prior to the first retest during the eligible bTB episode. Herd A had 4 retests; on the second retest, a single reactor (one of the introduced animals) was detected, leading to an extension of the eligible bTB episode. Herd A was de-restricted after the 4th retest and did not have any further bTB episodes before the end of the study period. Herd B had 2 retests during the eligible bTB episode; each was clear leading to herd de-restriction. A bTB episode commenced at some point after the herd had been de-restricted following resolution of the eligible bTB episode, with the only reactor at this initial test of the episode being an animal that had been previously introduced during the eligible bTB episode.

2.2.2. Time of infection for introduced animals that become reactors Among the animals introduced during the eligible bTB episode, we identified and examined two animal-level time periods of interest: • Interval between (a) the last bTB test prior to introduction and (b) introduction into the study herd. This time period was calculated for all introduced animals, and compared by infection status at the end of the study using a Wilcoxon rank-sum test, adjusted for clustering by herd (Rosner et al., 2006). • Interval between (a) introduction into the study herd and (b) subsequent bTB detection. This time period was calculated for the subset of introduced animals that became reactors, and presented separately for those animals that became reactors during or after the eligible bTB episode.

2.2.3. Trends over time The proportion of herds that introduced an animal during a bTB episode that started between 2004 and 2009 was compared annually to identify trends.

2.3. Analysis to determine whether introducing animals during an eligible bTB episode was associated with an increased future herd-level risk of bTB and to identify the most risky practices 2.3.1. Univariable analysis The proportion of study herds with at least one subsequent bTB episode (with 1 or more reactor animal(s) or an animal identified with a confirmed tuberculous lesion at slaughter; during the period from resolution of the eligible bTB episode through to the end of 2010) was compared for each of the categorical risk factors listed below, using a chi-square test. 2.3.2. Multivariable analysis 2.3.2.1. Outcome of interest. The outcome of interest was the time from de-restriction of the eligible bTB episode to the start of a subsequent bTB episode (with 1 or more reactor animal(s) or an animal identified with a confirmed tuberculous lesion at slaughter) or the date of the last test prior to the end of the study (31 December 2010), whichever occurred first. 2.3.2.2. Independent variables. A range of independent variables were considered, including variables related to the introduction of animals during the eligible bTB episode and other risk factors. Variables related to the introduction of animals during the eligible bTB episode: • Whether animals were introduced during the eligible bTB episode (Intro; 0, 1). • Timing of the first introduction event during the eligible bTB episode: ◦ Introduced prior to first retest (rr) (Intro priorrr: 0, no animals introduced; 1, introduced after first retest; 2, introduced prior to first retest). ◦ Status of the first retest (rr status: 0, negative; 1, positive). ◦ Introduced prior to first clear test (Intro priorclear: 0, no animals introduced; 1, introduced after first clear test; 2, introduced prior to first clear test). ◦ Time (days) from start date of the eligible bTB episode to the date the first animal was introduced during the episode (Intro time: 0, no animals introduced; 1, ≤57 days; 2, 58–109 days; 3, 110–175 days; 4, >175 days; note the categories were based on the quartiles of the distribution of the time for herds that introduced animals). • Number of animals introduced: as a continuous variable (Animals introduced), and 2 categorical variables (Intro3: 0, 1, >1; Intro4: 0, 1, 2–5, 6–25, 26+); the latter grouping was based on the quartiles of the number of animals introduced. Other risk factors: • Average herd size (Herdhs; average size of all full herd tests during the eligible bTB episode).

T.A. Clegg et al. / Preventive Veterinary Medicine 109 (2013) 246–257

249

Fig. 1. An example of an extended eligible bTB episode attributable to introduced animals and a subsequent bTB episode attributable to introduced animals. Herd A had an extended eligible bTB episode attributable to introduced animals, with one of the introduced animals being identified as a reactor (Rx) at the second herd retest. Herd B had a subsequent bTB episode attributable to introduced animals. In this herd, a bTB episode has commenced at some point after the herd had been de-restricted following resolution of the eligible bTB episode, with the only reactor at the initial test of the episode being an animal that had been previously introduced during the eligible bTB episode.

• Proportion of cows in the herd (Propcows; proportion of the herd that were cows at the start of the eligible bTB episode). • Infection status during the eligible bTB episode:

◦ Number of reactors during the eligible bTB episode (Tot rct). ◦ Number of standard reactors during the eligible bTB episode (Tot stdrct).

250

T.A. Clegg et al. / Preventive Veterinary Medicine 109 (2013) 246–257

• Time (days) between the end of the previous bTB episode and the start of the eligible bTB episode (Time last; for herds that did not have a previous episode the time was set at 7300 days (20 years)). • Enterprise type of the herd (Herd type: dairy, suckler, beef or other). • Infection status of contiguous herds (within 150 m). In Ireland, farms are often very fragmented. A contiguous herd was defined as any farm with at least one fragment of land adjacent to the study herd. ◦ Number of reactors in contiguous herds in 2006 (Contig rx). ◦ Number of standard reactors in contiguous herds in 2006 (Contig stdrx). • Badger activity, defined according to whether a badger sett had been surveyed as part of an epidemiological investigation, within 1 km of the study herd in 2006 (Bad06). A survey will seek to identify any sett located within 1 km of any land parcel attributed to the study herd. • Neighbourhood risk in 2006 (Surface of disease – kernel density based on the number of standard reactors using a 10 km search radius). The kernel densities were calculated using the centroid of the largest fragment of land for each farm. Smoothing at a national level with a 10 km search radius effectively eliminates the noise created by fragmentation. Further details are given in McGrath et al. (2009). ◦ Rank (1–10) of standard reactor density in the neighbourhood during 2006 (Rct dense), due to small numbers in the categories 7–10, these categories were combined to create a rank with 7 categories (Rct dense7). ◦ Rank (1–10) of cattle population density in the neighbourhood during 2006 (Pop dense). 2.3.2.3. Multivariable modelling. All study herds: Survival time to the end of the study was initially compared for each of the ‘introduction’ variables using a log-rank test and a Wilcoxon test. The Wilcoxon test is weighted according to sample size and is therefore more sensitive to differences early in the time period when the sample size is larger (Dohoo et al., 2009). Seven Cox proportional-hazard models, each using the above-mentioned outcome of interest, were developed using STCOX in STATA version 11 (StataCorp LP, College Station, TX, USA), each with a different introduction variable (i.e. whether herds introduced animals during the eligible bTB episode (Intro), timing of first introduction event during the eligible bTB episode (Intro priorrr, Intro priorclear, Intro time), number of animals introduced (Animals introduced, Intro3, Intro4)). Note that the status of the first retest variable (rr status) was also included in the model that included the variable ‘introduction prior to first retest’ (Intro priorrr) along with an interaction term between these two variables. For each model, a univariable screening approach was initially used, where all variables with p < 0.2 at the univariable stage became candidates for the multivariable model. A backward selection procedure was then used to eliminate terms from the model based on a likelihood ratio test

(p > 0.05). The correlation between covariates was evaluated using a chi-square test between nominal variables and kendall’s tau-b assessment of correlation between binary, ordinal and continuous variables. Continuous variables were also categorised into 4 groups based on the corresponding quartiles. Whether to treat variables as continuous, categorical or to transform the variable was tested by comparing univariable models using the AIC (Akaike information criteria). To examine the appropriate functional form of a variable, a plot of the lowess smooth of martingale residuals against transformations of the covariate was used. Additionally, the choice of whether to use total reactors or standard reactors was determined by comparing the AIC from the univariable models. Confounding effects on the ‘introduction’ variables were assessed by evaluating the changes in the categorical coefficients for changes of greater than 20% upon deletion of potential confounders from the final model. The proportional-hazard assumption was checked visually using a plot of −log(−log) survival lines to examine whether the different covariate groups were parallel. In addition, the chi-squared Schoenfeld residuals were tested (p < 0.05) to examine whether the hazard ratio varied over time; if significant, the risk factor was included as a time-varying covariate. The model was checked by examining the martingale, influence and Schoenfeld residuals. Study herds that introduced animals during an eligible bTB episode: The multivariable model was repeated for just the herds that introduced animals during an eligible bTB episode using the risk factors described above with the exception of the variable: Intro and with the addition of a variable describing ‘Interval between the last bTB test prior to introduction and introduction into the study herd (last test)’. The minimum, median and maximum time was considered for the last time (last test) of testing for all animals introduced. Separate models were developed using one of the variables describing the timing of the first introduction event during the eligible bTB episode (Intro priorrr, Intro priorclear) and with the addition of the variable describing the number of animals introduced (Animals introduced, Intro3, Intro4). All other risk factors were considered within the multivariable model if the p-value in the univariable analysis was <0.2. The neighbourhood risk variables (Rct dense and Pop dense) were combined into 5 groups (Rct dense5 and Pop dense5) due to the small number of herds in the higher groups; Rct dense ≥ 5 and Pop dense ≥ 5 were combined into one group for each variable. 3. Results 3.1. Study population 3.1.1. Study herds The following highlights the number of herds at each stage of the selection process. Herds restricted in 2006: Excluding herds with a herd size <10th and >90th percentile: (The herd size of the 10th percentile = 16 and 90th percentile = 218.) Excluding herds already restricted at the start of 2006:

6394 5139

4879

T.A. Clegg et al. / Preventive Veterinary Medicine 109 (2013) 246–257 Excluding herds with <2 standard reactors: Of these: Excluding the 2nd or further episode from the same herd: Excluding herds that slaughtered more than 200 animals in 2006: Excluding dealer/feedlot herds:

251

3.2.2. Extended or subsequent bTB episodes attributable to introduced animals Of the 442 herds that introduced an animal during the eligible bTB episode:

1902 1864 1838 1829

Therefore 1829 study herds were enrolled.

3.1.2. Introduced animals In total, 9134 animals were introduced during an eligible bTB episode into 442 study herds. (Note of the 9134 animals introduced, 30 were introduced by more than one study herd. These animals have been allocated to the latter study herd that they were in.) Of the 442 herds that introduced animals, 152 (34%) introduced only 1 animal. The number of animals introduced per herd ranged from 1 to 303 animals (median = 5 animals).

3.2. Overview of events associated with the introduction of animals during an eligible bTB episode and the infection status of introduced animals 3.2.1. Future bTB status of introduced animals Of the 9134 animals introduced, 143 animals (from 59 herds) were subsequently identified as reactors, either during or after the eligible bTB episode. Of these, 60 animals (0.6%, from 22 herds) became reactors during the course of the eligible bTB episode. Six of the 143 animals had moved to another herd before they became reactors. Of the animals introduced, 7970 were slaughtered by the end of 2010. Of these, 968 were slaughtered during the eligible bTB episode. An additional 13 (from 9 additional herds) of the introduced animals had lesions due to M. bovis at slaughter. Of these, 1 (from one additional herd) had lesions during the eligible bTB episode. Two of the 13 animals had moved to another herd before they were detected M. bovis positive at post mortem. In total, 156 (1.7%) of the 9134 introduced animals were positive for bTB by the end of the study period. Of the 442 herds that introduced an animal, 197 (44.6%) had a subsequent bTB episode before the end of the study. Only 42 of these herds had a subsequent bTB episode which involved one or more of the introduced animals; a further 8 herds (not the study herds) had a bTB episode involving one of the introduced animals which had moved after the eligible bTB episode.

• 20 herds had either an extended eligible bTB episode or a subsequent bTB episode that involved only an introduced animal on at least one of the tests (48 reactors in 28 tests). • Of these, 12 herds (2.7% of herds that introduced animals during the eligible bTB episode) had an extended eligible bTB episode involving only introduced animals on a retest. • The remaining 8 herds (1.8% of herds that introduced animals during the eligible bTB episode) involved just the introduced animal(s) at the initial test of the subsequent bTB episode. 3.2.3. Time of infection for introduced animals that became reactors Interval between (a) the last bTB test prior to introduction and (b) introduction into the study herd. On average, there were 97 days (median) since an animal was last tested before being introduced into the study herd (Table 1). The time since last tested prior to introduction was longer for animals that were subsequently deemed positive for bTB compared with those that were not deemed positive (positive animals: median = 131 days; negative animals: median = 96 days), however, this difference was not significant after accounting for clustering within herds (p = 0.085). Interval between (a) introduction into the study herd and (b) subsequent bTB detection. On average (median), there were 242 days from the date of introduction until an animal was declared as positive either at a future SICTT test or at post mortem (Table 2). For animals that became positive during the eligible bTB episode, the median time between being declared as positive and being introduced was 75 days. For animals that survived the eligible bTB episode but were later positive, there was on average (median) 385 days between being declared as positive and the end of the eligible bTB episode (Table 2). 3.2.4. Trends over time The proportion of herds that introduced an animal during an eligible bTB episode, either before the end of the episode or by the end of the preceding year (whichever

Table 1 bTB status of 9134 cattle in Ireland at the end of 2010 and the interval (days) between the last bTB test prior to introduction and introduction into 1829 study herds. Final bTB status

No. of animals

Interval between the last bTB test prior to introduction and introduction into the study herd (days)

Positive Negative

156 8978

43 26

131 96

212 182

All

9134

26

97

182

Q25

median

p-Valuea

Q75 0.085

Q25: 25% of animals had a time lower than this value. Median: 50% of animals had a time lower than this value. Q75: 75% of animals had a time lower than this value. a Wilcoxon rank-sum test, adjusted to account for clustering within herds.

252

T.A. Clegg et al. / Preventive Veterinary Medicine 109 (2013) 246–257

Table 2 Interval (days) between introduction into the study herd and subsequent bTB detection, for 156 bTB positive animals in Ireland as at the end of 2010 and 63 animals deemed positive during the eligible bTB episode, and between the end of the eligible bTB episode and subsequent bTB detection for 93 animals deemed positive following the eligible bTB episode. No. of animals

Time (days) Q25

Time (days) between study herd introduction and bTB detection All positive animals Animals deemed bTB positive during the eligible bTB episode Time (days) between end of eligible bTB episode and bTB detection Animals deemed bTB positive after the eligible bTB episode

Median

Q75

156 63

79 38

242 75

485 103

93

196

385

566

Table 3 Number of herds in Ireland that had a new bTB episode (with ≥2 standard reactors) each year (excluding dealer/feedlot herds) and the percentage of herds that introduced animals (with animals > 42 days old) during the eligible bTB episode and by the end of the year following the start of the episode. Year eligible bTB episode began

No. of eligible bTB episodes

No. of herds that introduced animals

% of herds that introduced animals

2004 2005 2006 2007 2008 2009

1949 2062 1829 2098 2051 1671

549 497 440 528 459 344

28.2 24.1 24.1 25.2 22.4 20.6

0.75 0.50 0.25 0.00

500

1000 analysis time (days)

no introduction introduction before 1st clear test

1500

2000

introduction after 1st clear test

Fig. 2. Kaplan–Meier probability of a subsequent bTB episode, by time of introduction in relation to the first clear test during the eligible bTB episode.

1.00

(Intro) p = 0.003, introduced prior to first clear test (Intro priorclear) p = 0.001, introduced prior to first retest (Intro priorrr) p = 0.001, time to first animals introduced (Intro time) p = 0.008, and number of animals introduced (Intro3) p = 0.010 and (Intro4) p = 0.043). The results of the Wilcoxon test were similar with the exception of number of animals introduced as a categorical variable with 4 groups (Intro4) (p = 0.064), which was borderline significant. Figs. 2 and 3 show the Kaplan–Meier survival curves

0.75

3.3.2. Multivariable analysis 3.3.2.1. All study herds. Of the 1829 study herds, 1 had an eligible bTB episode that was ongoing at the end of the study period and an additional 13 herds were not tested again after the eligible bTB episode. A further 5 herds were excluded from the analysis due to missing data regarding neighbouring herds. In total, 1810 study herds were included in the multivariable analysis. There was a significant difference in the time to a subsequent bTB episode for all the introduction variables (log-rank test: animals introduced into herd

0

0.50

3.3.1. Univariable analysis The proportion of herds that had at least one subsequent bTB episode (following resolution of the eligible bTB episode) by the end of the study period is shown for each of the categorical risk factors in Table 4. A significantly (p = 0.005) higher proportion of herds that introduced animals had a subsequent bTB episode by the end of the study compared with herds that did not. Of the introduced animals, 1942 (21.3%) were introduced prior to the first retest into 137 herds (31.0%). Prior to the first clear test, 2895 (31.7%) animals were introduced by 219 (49.5%) herds.

0.25

3.3. Analysis to determine whether introducing animals during an eligible bTB episode was associated with an increased future herd-level risk of bTB and to identify the most risky practices

0.00

was the shorter period), for episodes starting 2004–2009 is shown in Table 3.

1.00

Q25: 25% of animals had a time lower than this value. Median: 50% of animals had a time lower than this value. Q75: 75% of animals had a time lower than this value.

0

500

1000 analysis time (days)

no introduction introduction before retest

1500

2000

introduction after retest

Fig. 3. Kaplan–Meier probability of a subsequent bTB episode, by time of introduction in relation to the first retest.

T.A. Clegg et al. / Preventive Veterinary Medicine 109 (2013) 246–257

253

Table 4 The number and percentage of study herds in Ireland with a subsequent bTB episode (up to end of 2010) following de-restriction of the eligible bTB episode, by categorised risk factors. Variable

Category

No. of herds

No.

%

Whether animals were introduced during the eligible bTB episode (Intro)

No Yes

1387 442

514 197

37.1 44.6

0.005

1387 305 137

514 124 73

37.1 40.7 53.3

0.001

Timing of first introduction event during the eligible bTB episode: No animals introduced Introduced prior to first retest Introduction after 1st retest (Intro priorrr) Introduction before 1st retest

Herds with a subsequent bTB episode

p-Value

Introduced prior to first clear test (Intro priorclear)

No animals introduced Introduction after 1st clear test Introduction before 1st clear test

1387 223 219

514 90 107

37.1 40.4 48.9

0.003

Time (days) from start date of the eligible bTB episode to the date the first animal was introduced during the episode (Intro time)

No animals introduced ≤57 days 58–109 days 110–175 days >175 days

1387 111 115 106 110

514 56 50 42 49

37.1 50.5 43.5 39.6 44.5

0.030

Status of the first retest (Rr status)

−ve at 1st retest +ve at 1st retest

1117 712

432 279

38.7 39.2

0.673

Number of animals introduced (Intro3)

0 1 >1

1387 152 290

514 65 132

37.1 42.8 45.5

0.016

Number of animals introduced (Intro4)

0 1 2–5 6–25 >25

1387 152 71 112 107

514 65 34 47 51

37.1 42.8 47.9 42.0 47.7

0.055

Average herd size (herdhs)

<46 46–78 79–122 >122

461 470 444 454

150 180 173 208

32.5 38.3 39.0 45.8

0.001

Proportion of cows in the herd (Propcows)

<0.27 0.27–0.35 0.36–0.44 >0.44

458 456 458 457

181 177 178 175

39.5 38.8 38.9 38.3

0.986

Number of standard reactors during the eligible bTB episode (Tot stdrct)

2 3–4 5–7 >7

595 492 314 428

200 191 134 186

33.6 38.8 42.7 43.5

0.006

Time (days) since the previous bTB episode prior to the eligible bTB episode (Time last)

<973 973–2363 2364–6215 >6215

457 457 458 457

220 160 174 157

48.1 35.0 38.0 34.4

< 0.001

Enterprise type of the herd (Herd type)

Beef Dairy Other Suckler

226 670 12 921

93 290 4 324

41.2 43.3 33.3 35.2

0.009

Number of standard reactors in contiguous herds in 2006 (Contig stdrx)

0 1–4 5–12 >12 Missing

491 509 392 432 5

165 195 160 190 1

33.6 38.3 40.8 44.0 20.0

0.011

Badger activity within 1 km of the study herd in 2006 (Bad06)

No Yes Missing

469 1355 5

168 542 1

35.8 40.0 20.0

0.110

Standard reactor density in neighbourhood in 2006 (Rct dense)

1: <0.26 2: 0.26–0.53 3: 0.54–0.79 4: 0.80–1.06

463 639 369 165

145 251 159 65

31.3 39.3 43.1 39.4

0.001

254

T.A. Clegg et al. / Preventive Veterinary Medicine 109 (2013) 246–257

Table 4 (Continued) Variable

Category

No. of herds

Herds with a subsequent bTB episode No.

5: 1.07–1.32 6: 1.33–1.59 7: 1.60–1.85 8: 1.86–2.12 9: 2.13–2.38 10: >2.38 Missing Cattle population density in the neighbourhood in 2006 (Pop dense)

1: <22.1 2: 22.1–44.2 3: 44.3–66.3 4: 66.4–88.4 5: 88.5–110.5 6: 110.6–132.6 7: 132.7–154.7 8: 154.8–176.8 9: 176.9–198.9 10: >198.9 Missing

for the introduction of animals prior to first clear test ‘Intro priorclear’ and introduction of animals prior to first retest ‘Intro priorrr’ variables. Based on the results of the univariable analysis, the variables tested in the survival models were the log of herd size and the following categorical variables: time since previous bTB episode (Time last), neighbourhood risk (Rct dense7, Pop dense), infection status during the eligible bTB episode (tot stdrct), contiguous standard reactors (contig stdrx), badger activity (Bad06), herd type and one of the introduction variables. In the final survival model (Table 5), introduction prior to the first retest (Intro priorrr) was significant (p = 0.018). Herds that introduced animals after the first retest had a similar risk of a subsequent bTB episode as the herds that did not introduce animals (p = 0.659). Herds that introduced animals prior to a first retest had 1.5 times the risk of a subsequent bTB episode compared with herds that did not introduce animals (p = 0.003) (Table 5). During model construction, the status of the first retest (positive or negative) was not significant (p = 0.673), nor was an interaction term between the neighbourhood risk (rct dense5) and the introduction of animals prior to a first retest variable (intro priorrr) (p = 0.103). In separate models based on other introduction variables, we found either borderline (introduction prior to the first clear test (Intro priorclear), p = 0.060) or no significance (introduction status (Intro) p = 0.087, time to introduction (Intro time) p = 0.091, number of animals introduced (Intro3) p = 0.193, (Intro4) p = 0.471).

3.3.2.2. Study herds that introduced animals during an eligible bTB episode. The introduction variables that were borderline significant were introduction prior to the first retest (intro priorrr, p = 0.051) and introduction prior to the first clear test (intro priorclear, p = 0.069). The final model incorporating the introduction prior to the first retest (intro priorrr) variable is shown in Table 6. None of the

p-Value

%

84 49 29 8 7 11 5

37 26 13 2 6 6 1

44.0 53.1 44.8 25.0 85.7 54.5 20.0

24 71 107 194 372 343 363 221 113 16 5

12 24 33 81 120 141 144 93 56 6 1

50.0 33.8 30.8 41.8 32.3 41.1 39.7 42.1 49.6 37.5 20.0

0.020

other variables related to introduction were significant in the final models (intro3 p = 0.628, intro4 p = 0.871, intro time p = 0.286, time last p = 0.335).

4. Discussion The introduction of animals into herds during a bTB episode is not an uncommon practice in Ireland, and is also allowed under licence in the UK (Karolemeas et al., 2011). As yet, however, there has been no published work on the impact of this practice on future bTB risk. This paper seeks to address this gap in knowledge, with a number of key findings. Firstly, of the herds that introduced animals during a bTB episode, only a small proportion had an extended eligible bTB episode (2.7%) or subsequent bTB episode (1.8%) involving only introduced animals on a test. Secondly, the results highlight an increased risk of a subsequent bTB episode among only a subset of herds that introduced animals during the eligible bTB episode. Specifically, herds that introduced animals early during the eligible bTB episode were at significantly greater future bTB risk than herds where animals were only introduced later. To illustrate, herds that introduced animals after the first retest did not have a significantly different risk compared to herds that did not introduce animals at all. In contrast, herds that did introduce animals prior to the first retest had 1.5 times higher risk of a subsequent bTB episode. Early introduction of animals appears to be important regardless of the disease status at the first retest. Finally, among the herds that did introduce animals during the eligible bTB episode, time since last test was not a significant predictor of risk of a subsequent bTB episode. We consider three possible explanations for the results obtained: residual infection within the study herds, introduction of infection with the introduced animals and farm-related differences. As indicated below, the first and third explanations may be consistent with the study findings, whereas the second is not.

T.A. Clegg et al. / Preventive Veterinary Medicine 109 (2013) 246–257

255

Table 5 Final Cox proportional hazard model of the time to a subsequent bTB episode following 1810 eligible bTB episodes in Ireland. In this model, the timing of the first introduction event during the eligible bTB episode was described using Intro priorrr (animals introduced either prior to or following the first retest).

Introduced animals prior to first retest (Intro priorrr)

HR

Lower

Upper

No animals introduced Introduction after 1st retest Introduction before 1st retest

Referent 1.05 1.46

0.86 1.14

1.28 1.88

0.659 0.003

1.15

1.00

1.32

0.043

Log average herd size (herdhs)

95% confidence interval

p-Valuea

Category

0.018

Time since previous episode prior to the eligible bTB episode (Time last)

<973 973–2363 2364–6215 >6215

Referent 0.66 0.78 0.73

0.54 0.64 0.59

0.81 0.95 0.90

<0.001 0.015 0.003

Standard reactors in the eligible bTB episode (Tot stdrct)

2 3–4 5–7 >7

Referent 1.22 1.43 1.42

1.00 1.14 1.15

1.50 1.78 1.75

0.048 0.002 0.001

Reactor density in neighbourhood in 2006 (Rct dense7)

1: <0.26 2: 0.26–0.53 3: 0.54–0.79 4: 0.80–1.06 5: 1.07–1.32 6: 1.33–1.59 7: 1.60–1.85

Referent 1.26 1.37 1.14 1.48 2.00 1.60

1.02 1.09 0.85 1.03 1.31 1.05

1.55 1.72 1.54 2.14 3.06 2.43

0.031 0.007 0.388 0.035 0.001 0.028

Enterprise type of the herd (Herd type)

Beef Dairy Other Suckler

Referent 0.93 0.80 0.75

0.72 0.29 0.59

1.19 2.19 0.95

0.552 0.667 0.016

a b

p-Valueb

0.042 <0.001

0.002

0.012

0.031

Wald test. Likelihood ratio test.

Diagnostic tests for bTB have imperfect sensitivity (de la Rua-Domenech et al., 2006; Clegg et al., 2011), which leads to difficulties when seeking to identify all infected animals in infected herds. In national bTB control programmes, residual herd infection, the presence of infected but testnegative animals in infected herds, is generally addressed through repeated whole-herd testing and removal of reactor animals. Residual herd infection was a well-recognised feature of the Australian control programme (Radunz, 2006), and has been identified in several recent studies from Ireland (Clegg et al., 2008; More, 2009; Wolfe et al., 2009; Kelly and More, 2011; Berrian et al., 2012) and the

UK (Carrique-Mas et al., 2008; Green et al., 2012). In the current study, residual infection, and associated infection risk for introduced animals, is likely to have been higher earlier (compared with later) during eligible bTB episodes, prior to repeated whole-herd testing and removal of reactor animals. Further, it is conceivable that introduced cattle are most-susceptible to infection shortly after introduction, due to stress. For these reasons, residual infection could provide a plausible explanation for the observed herd-level effect, namely an increased future bTB risk among herds where animals were introduced early versus late or not at all.

Table 6 Final Cox proportional hazard model of the time to a subsequent bTB episode among the 442 Irish herds that introduced animals during the eligible bTB episodes. In this model, the timing of the first introduction event during the eligible bTB episode was described using Intro priorrr (either prior to or following the first retest). Category

Introduced animals prior to first retest (Intro priorrr)

Introduction after 1st retest Introduction before 1st retest

Log average herd size (herdhs) Reactor density in neighbourhood in 2006 (Rct dense5)

a b

Wald test. Likelihood ratio test.

1: <0.26 2: 0.26–0.53 3: 0.54–0.79 4: 0.80–1.06 5: >1.07

HR

95% confidence interval

p-Valuea

Lower

Upper

Referent 1.34

1.00

1.79

0.048

1.40

1.08

1.81

0.011

Referent 1.42 1.58 1.98 1.97

0.94 1.00 1.18 1.21

2.15 2.48 3.30 3.22

0.096 0.048 0.009 0.006

p-Valueb

0.051

0.009 0.032

256

T.A. Clegg et al. / Preventive Veterinary Medicine 109 (2013) 246–257

It is possible that some of the introduced animals were in fact infected at the time of introduction into herds during the eligible bTB episodes, with these animals then contributing to an increase in future bTB risk in the study herds. In other words, infection persists in some herds despite herd-level controls, with the potential for infected animals being moved to another herd following de-restriction. This was confirmed in a recent Irish study, where Berrian et al. (2012) found that the animallevel odds of bTB were 1.91 times higher among animals sold out from bTB ‘exposed’ herds compared with animals sold out from bTB ‘non-exposed’ herds. Certainly, there is some evidence from previous studies of the role of introduced infection in bTB episodes, with 7% of bTB episodes in Ireland (Clegg et al., 2008) and 16% in Great Britain (Green et al., 2008) attributed to introduced animals. Although introduced infection is of importance in the national programme, specifically with respect to future herd-level bTB risk, it is unlikely to be an explanation for the effects observed in the current study. The observed effects could only be attributed to introduced infection if the infection risk of animals introduced prior to the first retest were systematically different (in terms of past bTB history of the herd of origin; Berrian et al., 2012) in comparison to animals introduced subsequently during the eligible bTB episode. Given our understanding of the nature of animal purchases in Ireland, this would seem very unlikely. The final possible explanation for the observed increased risk associated with time of introduction relates to as-yet-unmeasured factors related to farmer behaviour. To illustrate, it is possible that early introduction of animals during a bTB episode may be more common among farmers who also engage in other high risk practices, with a low general aversion to bTB risk, in comparison to other farmers. A better understanding, using appropriate methodologies from the social science, of farmers’ reasons for early introductions of animals during a bTB episode would seem justified, to best inform future policy changes. Social science methods have been applied to some animal health issues, including bTB (Christley et al., 2011), to better understand farmer behaviour. To our knowledge, there has been only one previous study investigating the impact of introduction on episode recurrence. In a recent UK study, Karolemeas et al. (2011) evaluated a range of risk factors associated with recurrence of a bTB episode within 12 months. Only one variable related to animal introductions: the number of animals introduced during the ‘study’ episode. In both this and the current study, this factor was not significant in the final model. Karolemeas et al. (2011) did not consider the timing of introductions during the episode, and did not compare herds that did and did not introduce animals during an episode. Several studies have evaluated risk factors for bTB recurrence following de-restriction, identifying similar risk factors to those in the current study. Severity of the initial episode and a previous history of bTB in the herd have been consistently identified as important risk factors for bTB recurrence (Olea-Popelka et al., 2004; Abernethy et al., 2010; Wolfe et al., 2010; Karolemeas et al., 2011); herd size was also identified by all but the latter author. The infection history in the location has been identified as a risk factor for

recurrent bTB in Ireland by Olea-Popelka et al. (2004) and Wolfe et al. (2010). In the current study, the hazard ratio did not increase linearly with reactor density in an area, however, the lowest hazard ratio was in the area with the lowest density of positive animals. The increased risk due to introducing animals was the same regardless of the neighbourhood risk, as shown by a non-significant interaction between neighbourhood risk (rct dense) and introduction prior to the first retest (intro priorrr). A range of methodological issues were considered during study design and analysis. We elected to focus on episode recurrence, rather than episode duration, as our preferred outcome of interest. With the latter outcome, it is difficult to disentangle cause (introduction preceding an extended episode) and consequence (introduction occurring subsequent to an extended episode). Of the study herds that introduced animals 12 (2.7%) had an extended episode as a direct result of introduced animals i.e. there were no other bTB positive animals on a test date other than those introduced. Therefore, introduction of animals can affect the duration of an episode for a small proportion of herds. The study was confined to herds that were identified as unlikely to be “dealer” or “feedlot” herds. Such herds have unusual trading patterns and are subjected to different movement controls. For example, dealer herds may voluntarily de-stock the infected premises and/or resume trading from an alternative epidemiologically distinct premises. Similarly, feedlot herds, subject to an approval system, whose business and livelihood depends on buying and finishing animals for slaughter in a short timeframe (<6 months) may continue this practice provided they satisfy the competent authority with responsibility for the bTB eradication programme that they pose minimum risk of transmission of disease either within herd or to the locality. Movement of calves under the age of 42 days was also excluded as an introduction event from the study. The replacement of dead suckler calves is allowed during a bTB episode; these are very low risk animals and it was decided to exclude these movements from the study. The study herds were restricted to episodes with two or more standard reactors in order to exclude episodes due to possible ‘false-positive’ reactors. Under Irish conditions, the specificity of the SICTT has been estimated to range from 99.8 to 99.9% (O’Reilly, 1993) to 99.5% (Clegg et al., 2011). In order to address the incomplete specificity issue, a ‘Singleton Protocol’ was established (Good and Duignan, 2011). This protocol allows the early restoration of trading status when the herd was not confirmed as infected by epidemiological investigation, post-mortem or a further test. In order to exclude possible ‘false-positive’ episodes the eligible bTB episodes were restricted to those with two or more standard reactors. The subsequent bTB episodes include any episode regardless of the number of reactors, all these herds have a prior history of bTB therefore they are unlikely to be ‘false-positive’ reactions although there is still a possibility that this may have occurred. The time since animals were previously tested until the date of introduction was longer for animals that were subsequently deemed positive for bTB, however, when other risk factors were taken into account this variable was not significant in the final multivariable model of herds that

T.A. Clegg et al. / Preventive Veterinary Medicine 109 (2013) 246–257

introduced animals during a bTB episode. All herds within Ireland are tested at least annually and it is illegal to move animals that have not been individually tested within the previous 365 days, so it is unlikely that an animal would not be tested for a prolonged period of time. Therefore, the time since last tested appeared to have little effect on the risk of a subsequent episode. In Ireland, future policy concerning the introduction of animals during a bTB episode should be reviewed in the light of these findings. Although the introduced animals only directly resulted in a small number of extended episodes or subsequent episodes, possible amendments could be developed to restrict introduction of animals until at least one retest has taken place on the herd. Further work is needed to better understand the impact of farmer behaviour on bTB risk, and of the impact of stress on susceptibility to bTB infection in cattle. Acknowledgements The assistance of Guy McGrath from the Centre for Veterinary Epidemiology and Risk Analysis, University College Dublin, regarding spatial data analysis is gratefully acknowledged. References Abernethy, D.A., Graham, D., Skuce, R., Gordon, A., Menzies, F., Robinson, P., Harwood, R., Clarke, N., 2010. The bovine tuberculosis eradication scheme. In: Proceedings of the Society for Veterinary Epidemiology and Preventive Medicine, Nantes, France, pp. 167–173. Berrian, A.M., O’Keeffe, J., White, P.W., Norris, J., Litt, J., More, S.J., OleaPopelka, F.J., 2012. Risk of bovine tuberculosis for cattle sold out from herds during 2005 in Ireland. Vet. Rec. 170, 620. Carrique-Mas, J.J., Medley, G.F., Green, L.E., 2008. Risks for bovine tuberculosis in British cattle farms restocked after the foot and mouth disease epidemic of 2001. Prev. Vet. Med. 84, 85–93. Christley, R.M., Robinson, S.E., Moore, B., Setzkorn, C., Donald, I., 2011. Responses of farmers to introduction in England and Wales of premovement testing for bovine tuberculosis. Prev. Vet. Med. 100, 126–133. Clegg, T.A., More, S.J., Higgins, I.M., Good, M., Blake, M., Williams, D.H., 2008. Potential infection-control benefit for Ireland from premovement testing of cattle for tuberculosis. Prev. Vet. Med. 84, 94–111. Clegg, T.A., Duignan, A., Whelan, C., Gormley, E., Good, M., Clarke, J., Toft, N., More, S.J., 2011. Using latent class analysis to estimate the test characteristics of the gamma-interferon test, the single intradermal comparative tuberculin test and a multiplex immunoassay under Irish conditions. Vet. Microbiol. 151, 68–76.

257

de la Rua-Domenech, R., Goodchild, A.T., Vordermeier, H.M., Hewinson, R.G., Christiansen, K.H., Clifton-Hadley, R.S., 2006. Ante mortem diagnosis of tuberculosis in cattle: a review of the tuberculin tests, gamma-interferon assay and other ancillary diagnostic techniques. Res. Vet. Sci. 81, 190–210. Dohoo, I., Martin, W., Stryhm, H., 2009. Veterinary Epidemiologic Research, 2nd ed. VER Inc., Charlottetown, PEI, Canada. Good, M., Duignan, A., Maher, P., O’Keeffe, J., 2010. Veterinary Handbook for herd management in the bovine TB eradication programme. Department of Agriculture, Fisheries and Food, Dublin. http://agriculture.gov.ie/media/migration/animalhealthwelfare/ diseasecontrols/tuberculosistbandbrucellosis/diseaseeradication policy/TBHandbook2010070710.pdf Good, M., Duignan, A., 2011. An evaluation of the Irish Single Reactor Breakdown protocol for 2005–2008 inclusive and its potential application as a monitor of tuberculin test performance. Vet. Microbiol. 151, 85–90. Green, D.M., Kiss, I.Z., Mitchell, A.P., Kao, R.R., 2008. Estimates for local and movement-based transmission of bovine tuberculosis in British cattle. Proc. R. Soc. B 275, 1001–1005. Green, L.E., Carrique-Mas, J.J., Mason, S.A., Medley, G.F., 2012. Patterns of delayed detection and persistence of bovine tuberculosis in confirmed and unconfirmed herd breakdowns in cattle and cattle herds in Great Britain. Prev. Vet. Med. 106, 266–274. Karolemeas, K., McKinley, T.J., Clifton-Hadley, R.S., Goodchild, A.V., Mitchell, A., Johnston, W.T., Conlan, A.J.K., Donnelly, C.A., Wood, J.L.N., 2011. Recurrence of bovine tuberculosis breakdowns in Great Britain: risk factors and prediction. Prev. Vet. Med. 102, 22–29. Kelly, G.E., More, S.J., 2011. Spatial clustering of TB-infected cattle herds prior to and following proactive badger removal. Epidemiol. Infect. 139, 1220–1229. McGrath, G., Abernethy, D., Stringer, L., More, S.J., 2009. An all-island approach to mapping bovine tuberculosis. Irish Vet. J. 62, 192–197. More, S.J., Good, M., 2006. The tuberculosis eradication programme in Ireland: a review of scientific and policy advances since 1988. Vet. Microbiol. 112, 239–251. More, S.J., 2009. What is needed to eradicate bovine tuberculosis successfully: an Irish perspective. Vet. J. 180, 275–278. Olea-Popelka, F.J., White, P.W., Collins, J.D., O’Keeffe, J., Kelton, D.F., Martin, S.W., 2004. Breakdown severity during a bovine tuberculosis episode as a predictor of future herd breakdowns in Ireland. Prev. Vet. Med. 63, 163–172. O’Reilly, L.M., 1993. Specificity and sensitivity of tuberculin tests: a review. In: A Monograph Presented at Post-Graduate Studies Course, Faculty of Veterinary Medicine, University of Zaragosa, Spain. Radunz, B., 2006. Surveillance and risk management during the latter stages of eradication: experiences from Australia. Vet. Microbiol. 112, 283–290. Rosner, B., Glynn, R.J., Lee, M.T., 2006. Extension of the Rank Sum Test for clustered data: two-group comparison with group membership defined at the subunit level. Biometrics 62, 1251–1259. Wolfe, D.M., Berke, O., More, S.J., Kelton, D.F., White, P.W., O’Keeffe, J.J., Martin, S.W., 2009. The risk of a positive test for bovine tuberculosis in cattle purchased from herds with and without a recent history of bovine tuberculosis in Ireland. Prev. Vet. Med. 92, 99–105. Wolfe, D.M., Berke, O., Kelton, D.F., White, P.W., More, S.J., O’Keeffe, J., Martin, S.W., 2010. From explanation to prediction: a model for recurrent bovine tuberculosis in Irish cattle herds. Prev. Vet. Med. 94, 170–177.